See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/232600895

Structural analysis of social behavior (SASB): Circumplex analyses and structural relations with the Interpersonal Circle and the Five-Factor Model of personality

Article in Journal of Personality and Social Psychology · June 1998 DOI: 10.1037/0022-3514.74.6.1629

CITATIONS READS 95 3,975

3 authors, including:

Aaron L. Pincus Michael B. Gurtman Pennsylvania State University University of Wisconsin - Parkside

210 PUBLICATIONS 12,565 CITATIONS 25 PUBLICATIONS 1,970 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

An EMA study of narcissistic personality states in daily life. View project

All content following this page was uploaded by Aaron L. Pincus on 01 June 2014.

The user has requested enhancement of the downloaded file. Journal of Personality and Social Psychology Copyright 1998 by the American Psychological , Inc. 1998, Vol. 74, No. 6, 1629-1645 0022-3514/98/S3.00

Structural Analysis of Social Behavior (SASB): Circumplex Analyses and Structural Relations With the Interpersonal Circle and the Five-Factor Model of Personality

Aaron L. Pincus Michael B. Gurtman Pennsylvania State University University of Wisconsin—Parkside

Mark A. Ruiz Pennsylvania State University

The circumplexity of Structural Analysis of Social Behavior (SASB) and the interpersonal circle (IPC) were evaluated in a large adult sample. SASB exhibits elliptical structure, whereas multiple measures of the IPC exhibit circumplex structure. SASB results were replicated in a sample of psychiatric inpatients. Structural relations between the IPC and SASB indicate one common dimen- sion (affiliation). Structural relations between SASB and the five-factor model (FFM) of personality indicate SASB has both convergent relations with all FFM dimensions and unique aspects of personal- ity description captured by the autonomy dimension. SASB introject is related to Neuroticism and Conscientiousness. Autonomy-taking is related to Openness. Autonomy-granting was unrelated to the IPC or FFM. Integration of SASB, IPC, and FFM models identifies new dimensions of personality description and articulates the interpersonal influence of "noninterpersonal" traits.

As Wiggins (1991a, 1996a, 1996b) pointed out, circular rep- for a variety of domains including mood (Russell, 1980), per- resentations of human nature can be traced from ancient astrol- sonality disorders (Millon, 1987), and vocational interests (Tra- ogy and philosophy to Leonardo da Vinci to contemporary rep- cey & Rounds, 1997), it is in the domain of interpersonal func- resentations of interpersonal behavior. The centrality of circular tioning that they have been most widely used. (i.e., circumplex) structural models for the study of personality is evident in the recent publication of two major compendiums Individual-Differences and Dyadic Representations of research and theory (Kiesler, 1996; Plutchik & Conte, 1997). Regardless of conceptual focus, all circumplex models must of the Interpersonal Domain meet several geometric assumptions. In order of increasing spec- The most general property required for circumplexity in- ificity, these are (a) differences among variables are reducible volves the representation of the domain of interest in two funda- to differences in two dimensions, (b) all variables have equal mental dimensions. We propose that the circumplex representa- projections (the constant radius property), and (c) discretely tion of interpersonal functioning has been conceptualized in two measured variables are uniformly distributed along the circle's related, but nonredundant, ways. We refer to the first as the circumference (the equal-spacing property; Gurtman, 1994). individual-differences approach and the second as the dyadic The advantages of circumplexity include the application of approach. As detailed below, these two approaches to the study both continuous measurement (Gurtman, 1994) and geometric of interpersonal functioning emerged from slightly different re- classification of persons and constructs (e.g., Wiggins, 1991b; search traditions and emphasize slightly different research ques- Wiggins & Broughton, 1985, 1991) to the area of interest. Al- tions. We suggest that these two approaches be considered not though contemporary circumplex models have been proposed as competitors but as possible complements—with each poten- tially adding to the other. In fact, each approach clearly incorpo- rates elements of the other in its theoretical system. These ap- proaches differ in the fundamental unit (persons or dyads) to Aaron L. Pincus and Mark A. Ruiz, Department of Psychology, Penn- sylvania State University; Michael B. Gurtman, Department of Psychol- be represented by their two-dimensional structural models. This ogy, University of Wisconsin—Parkside. difference in the basic representational unit has led to a differ- We sincerely thank Lorna Smith Benjamin for making her inpatient ence in conceptualization of the two fundamental dimensions data available to us. We also are very grateful to Michael Browne for of interpersonal functioning. providing us with his model-fitting software and guiding us in its use and interpretation. Finally, we thank Jim Hollender, Michelle Isaac, Jen Heslin, and Aaron Jacoby for help in data collection. The Individual-Differences Approach Correspondence concerning this article should be addressed to Aaron L. Pincus, Department of Psychology, Pennsylvania State University, 542 The circumplex model we use to represent the individual- Moore Building, University Park, Pennsylvania 16802-3104. Electronic differences approach is a single two-dimensional circular sur- mail may be sent to [email protected]. face often referred to as the Leary circle (Freedman, Leary,

1629 1630 PINCUS, GURTMAN, AND RUIZ

Ossorio, & Coffey, 1951; Leary, 1957) or the interpersonal tual and statistical opposites (Gurtman, 1994; Wiggins, 1982). circle (IPC; Kiesler, 1983, 1996; Pincus, 1994). The original There are several additional descriptive components of the IPC, IPC was derived by Leary and his associates at the Kaiser Foun- including levels of behavioral analysis and discrimination of dation Psychology Research Group by observing interactions adaptive and maladaptive functioning (Kiesler, 1996; Pincus, among group members from a particular func- 1994). However, in the present discussion, we limit ourselves tional standpoint based on principles of Sullivan's (1953) inter- to issues pertaining to structure and the representational unit of personal theory. Specifically, they asked "What is the subject that structure. of the activity, e.g., the individual whose behavior is being rated, Regarding the latter, Carson (1969, 1991) has observed that doing to the object or objects of the activity?" (Freedman et the IPC has remained predominantly a social interactionist al., 1951, p. 149). This cataloguing of individuals' interpersonal model on observable descriptions of individual differ- behaviors eventually led to an empirically derived two-dimen- ences in interpersonal functioning. This point is further sup- sional structure that has changed little from the IPC presented ported by the recent recognition that the IPC dimensions of in Figure 1. The IPC is based on the two underlying dimensions dominance and nurturance represent variants of the two interper- of dominance-submissiveness on the vertical axis and nurtur- sonal dimensions (Extraversion and Agreeableness) of a more ance-coldness on the horizontal axis. comprehensive descriptive model of individual differences—the The IPC model is a geometric representation of individual five-factor model (FFM) of personality (McCrae & Costa, differences in a variety of interpersonal domains including inter- 1989; Trapnell & Wiggins, 1990). In general, it is assumed that personal traits (Wiggins, 1979, 1991b), interpersonal problems the FFM dimensions of Neuroticism, Extraversion, Openness (Alden, Wiggins, & Pincus, 1990), verbal and nonverbal inter- to Experience, Agreeableness, and Conscientiousness tap the personal acts (Blackburn & Renwick, 1996; Gifford, 1991; reliable variance in personality description, both interpersonal Kiesler, 1985, 1987), and covert interpersonal impacts (Kiesler, and noninterpersonal (McCrae & Costa, 1997), and have lexi- Schmidt, & Wagner, 1997; Wagner, Kiesler, & Schmidt, 1995). cal, evolutionary, personological, and clinical significance Thus, all qualities of individual differences within these domains (Costa & Widiger, 1994; Wiggins, 1996a). can be described as blends of the two underlying dimensions. The recent development of an IPC-based model of covert Blends of the IPC dimensions can be trigonometrically located impacts (Kiesler et al., 1997; Wagner et al., 1995) is a promising along the 360° perimeter of the circle. Interpersonal qualities new expansion of the IPC tradition. In general, however, the close to one another on the perimeter are conceptually and statis- structure of the IPC (including its representation of covert im- tically similar, qualities at 90° angles are conceptually and statis- pacts) was derived from observations and descriptions of indi- tically independent, and qualities on opposite points are concep- viduals. Maintaining an individual-differences stance, theorists

Assured- Dominant/ Domineering

Arrogant- (PA) Gregarious- Calculating / Extraverted / Vindictive (BC) (NO) Intrusive

Cold- Warm- (DE) IURTURANCE I (LM) Hearted Agreeable/ Overly Nurturant

(FG) (JK)

Aloof- Unassuming- Introverted / (HI) Ingenuous/ Socially Unassured- Exploitable Avoidant Submissive / Nonasserttve

Figure 1. The Interpersonal Circle. From ' 'Parental Representations and Dimensions fo Personality: Empir- ical Relations and Assessment Implications," by A. L. Pincus and M. A. Ruiz, 1997, Journal of Personality Assessment, 68, p. 441. Copyright 1997 by Lawrence Erlbaum Associates, Inc. Adapted with permission. STRUCTURE OF SASB 1631 have retained fundamental dimensions of dominance and nurtur- Lichtenstein & Cassidy, 1991) assesses two dimensions of ance and have assumed that the structure of covert interpersonal adults' recalled qualities of attachment to their childhood care- impacts mirrors that of overt interpersonal behavior. Although givers: rejection versus love and enmeshment versus autonomy. the individual described is presumed to be in a transactional Although all these models converge with regard to the two relationship with a real or generalized other (Mead, 1932; Sulli- basic dimensions of dyadic representation, none (other than van, 1953), the IPC structure does not include specific structural Schaefer's 1959, 1961) are circumplex models. or contextual references to the interacting other. Thus, IPC-based However, one widely used dyadic model that is represented as research has most often focused on describing the qualities of a circumplex is Structural Analysis of Social Behavior (SASB; a single person. For example, Horowitz, Rosenberg, and Kaleh- Benjamin, 1974, 1984, 1996a, 1996b). SASB is a complex, zan (1992) reported that psychotherapy patients exhibiting hos- three-plane circumplex model that operationally defines inter- tile-dominant interpersonal problems may have deficits in their personal and intrapsychic interactions (see Figure 2). The di- ability to describe others and relationships clearly, and this defi- mensions underlying SASB include autonomy (i.e., en- cit may have a negative impact on treatment response. These meshment-differentiation on the vertical axis), affiliation (i.e., results focus on the qualities of the individual patient interacting love-hate on the horizontal axis), and interpersonal focus (i.e., with a generalized therapist. transitive actions towards others represented by the top circle, We wish to make clear that although the IPC structure is intransitive reactions to others represented by the middle circle, essentially monadic, the model has been used to generate and and introjected actions directed toward the self represented by test hypotheses regarding interpersonal transactions and rela- the bottom circle). Benjamin (1996b) described the develop- tionships (Kiesler & Goldston, 1988; Kiesler & Watkins, 1989). ment of SASB as an effort ' 'to combine the prevailing clinical The theoretical system in which the IPC is central provides a wisdom about attachment with the descriptive power of the more comprehensive account of transactional processes than the circumplex as Schaefer had envisioned it" (p. 1204). The structure of the IPC itself. Kiesler and his colleagues' work unique multiplane structure of SASB is based on corresponding (Kiesler, 1988, 1991, 1996; Kiesler et al., 1997; Wagner et al., circumplex surfaces of parentlike (transitive) behaviors and 1995) has been an ongoing effort to incorporate both overt childlike (intransitive) behaviors. SASB also incorporates Sulli- behavior and covert experiences as components of interpersonal van's (1953) concept of introjection—that is, the expected im- transaction cycles. This process model is a flexible account of pact of caregiver behavior on the self-concept—by proposing potentially adaptive and maladaptive interpersonal transactions a third corresponding circumplex that reflects individuals' rela- and deserves more significant empirical attention than it has so tionships with themselves. Thus, unlike typical self-reports of far received. traits, the introject surface is rated with regard to an intrapsychic relationship, that is, how one relates to self.

The Dyadic Approach By separating parentlike and childlike behaviors into two planes, SASB incorporates both the vertical dimension of Soon after Leary (1957) and his colleagues began their semi- Schaefer's model (dominate vs. emancipate) and that of the IPC nal investigations, Schaefer (1959, 1961) also proposed a cir- (dominate vs. submit). The transitive surface represents the cumplex model of interpersonal behavior. However, this model former, whereas the intransitive surface opposes submission with was based on observations of mother-child dyads rather than autonomy-taking. Thus, according to circumplex geometry, sub- on extraction of the individual's behavior from the transactional mission and autonomy-taking are opposite interpersonal reac- context. Schaefer's original circumplex was similar in structure tions, whereas dominance and autonomy-granting are opposite to the IPC with a single difference. The fundamental dimensions interpersonal actions (Lorr, 1991). Dominance and submission were hostility-love on the horizontal axis and autonomy-con- are placed at comparable locations on different surfaces to re- trol on the vertical axis. Schaefer (1997) noted that additional flect the fact that they are complementary, not opposing, posi- investigations of both adult parent-child dyads and marital dy- tions (Benjamin, 1996b, p. 1204). Thus, SASB expands inter- ads consistently revealed two fundamental dimensions that were personal space by including the concepts of friendly and hostile labeled autonomy and relatedness (Schaefer & Burnett, 1987; differentiation (e.g., affirming, ignoring). These concepts are Schaefer, Edgerton, & Burnett, 1992). These two dimensions not defined within the IPC structure. In addition, the introject have early developmental origins but are important across the surface corresponds to the parentlike transitive surface, in that life span and across relationships. Early needs for love and one actively relates to the self. Although the vertical dimensions acceptance are complemented by needs for individuation and and complexity of SASB set it apart from the IPC, the same autonomy (Guisinger & Blatt, 1994; Mahler, Pine, & Bergman, geometric and trigonometric assumptions and principles are ap- 1975; Schaefer, 1997). plicable. Interpersonal behaviors located along the perimeters of There are now several widely used instruments available to the SASB circles (identified as clusters in SASB terminology) assess actual and recalled parent-child relationships that reflect represent blends of the basic dimensions of autonomy and affil- these two basic dyadic dimensions. The Parental Bonding Instru- iation, and conceptual and statistical similarities and differences ment (PBI; Parker, Tupling, & Brown, 1979) assesses care versus among clusters are based on the same circular geometry as rejection and overprotection versus encouragement of auton- described for the IPC. omy. The Parent-Child Relations Questionnaire (PCR-II; To complete our description, we note that recently Benjamin Siegelman & Roe, 1979) assesses two dimensions of parenting (1993a, 1995) has explicitly incorporated attachment concepts style, loving versus rejecting and casual versus strict control. into SASB structure. Boxes in Figure 2 denote that transactional The Inventory of Adult Attachment (INVAA; Cassidy, 1995; elements on the right side of the circles (affirm-disclose, recip- 1632 PINCUS, GURTMAN, AND RUIZ

emancipate 1-1 TRANSITIVE: ignore Action Toward Other 1-8

Disrupted Attachment Attachment Group Group

INTRANSITIVE: Reaction to Other

Disrupted Attachment reactive love Attachment Group 2-3 Group

submit 2-5 self-emancipate INTRO.IECT 3-1 self-neglect self-affirm 3-8 3-2

Disrupted Attachment self-attack Affiliation 1 self-love Attachment Group 3-7 3-3 Group

self-blame self-protect 3-6 3-4 self-control 3-5

Figure 2. Structural Analysis of Social Behavior. From "Parental Representations and Dimensions of Personality: Empirical Relations and Assessment Implications," by A. L. Pincus and M. A. Ruiz, 1997, Journal of Personality Assessment, 68, p. 441. Copyright 1997 by Lawrence Erlbaum Associates, Inc. Adapted with permission.

rocal love, protect-trust) represent the attachment group (AG). the patient-therapist dyad as the unit of investigation (e.g., Transactional elements on the left side of the circle (blame- Henry, Schacht, & Strupp, 1990). However, as mentioned earlier, sulk, attack-recoil, ignore-wall off) represent the disrupted the dyadic and individual-differences approaches to interper- attachment group (DAG). Individuals whose relations with their sonal representation are not mutually exclusive competitors. It attachment figures are permeated with AG elements should inter- should be noted that just as the IPC has, at times, been used in nalize positive working models of self and others and exhibit investigations of interpersonal transactions and relationships, healthy attachment. Individuals whose developmental history in- the SASB introject surface has been used in a few studies focus- volves relations permeated with DAG elements should exhibit ing exclusively on description of individuals or diagnostic attachment problems (Dickinson, Pincus, & Ruiz, 1996). groups (Alpher, 1996; Henry, 1992). Using this expanded interpersonal space, SASB describes a dyadic interpersonal unit—that is, a real or internalized relation- Rationale and Goals for the Present Study ship rather than the qualities of a single interactant. For example, Both SASB and the IPC are interpersonal circumplex models psychotherapy research using SASB has consistently looked at that have been used to guide programmatic research in clinical STRUCTURE OF SASB 1633 and personality psychology (e.g., Benjamin & Wonderlich, tional dimensions of individual differences beyond those of the 1994; Henry, Schacht, & Strupp, 1986; Horowitz, Rosenberg, & IPC. This topic was investigated for several reasons. First, al- Bartholomew, 1993; Pincus & Gurtman, 1995). The IPC and though we can predict that autonomy will not be related to the SASB are also, respectively, the central organizing features of dimensions of the IPC, there is no "gold standard" with regard two interpersonal psychotherapy approaches: therapeutic meta- to the assessment of autonomy. Thus it is unclear what the exact communication (Kiesler, 1988) and reconstructive learning ther- relationship may be between this fundamental dimension of the apy (Benjamin, 1987, 1993b). Finally, both models also allow dyadic approach and basic dimensions of individual differences. for the articulation of transactional principles for interpersonal Second, there has been an increased recognition of the need to complementarity (Benjamin, 1994; Carson, 1969; Kiesler, articulate the interpersonal aspects of purportedly noninterper- 1983). sonal traits (e.g., McCrae, 1996; Pincus & Wiggins, 1992). Given that both SASB and the IPC are readily identified as Finally, SASB's introject surface is an interesting hybrid model. interpersonal by the psychological community at large and share It describes the self but is not a typical trait model. It describes many common research focuses, it is surprising that there has a relationship but is not an explicitly interpersonal model. Al- been little research investigating the interrelations between these though the SASB model implies that the introject surface struc- two interpersonal systems. In addition, whereas the IPC has ture mirrors that of the dyadic surfaces, this has never been been subject to empirical examinations of its structural validity examined using external structural referents. It is possible that (Wiggins, 1991b), structural convergence across instruments the introject dimensions are not interpersonal at all. We selected and observers (Alden et al., 1990; Bartholomew & Horowitz, the FFM as our structural referent because it incorporates the 1991; Wagner et al., 1995), and structural relations with broader IPC and the relations among these two models is well-estab- models of personality description (McCrae & Costa, 1989; lished. This provides a detailed system of referents in which to Trapnell & Wiggins, 1990), Wiggins (1994) noted that "SASB investigate and clarify SASB structure. has enjoyed a friendly independence from the mainstream of Benjamin (1994) presented the first data in which SASB, the personality structure research" (p. 333). IPC, and the FFM were concurrently assessed on the same One consequence of this independence is that SASB's cir- sample. The major focus of her study was to demonstrate the cumplex structure has not been thoroughly evaluated. The ma- utility of the three models in differentiating psychopathological jority of evidence for the structural validity of SASB is reported groups. Demonstrating structural relations among the models in works detailing the development of the model (Benjamin, was clearly not the major goal of her investigation, and the data 1974) and the construction of a long and short form of the self- presented were not representative of standard analytical methods report Intrex questionnaires (Benjamin, 1983, 1988). Benjamin for such a purpose (Costa, 1994; Van Denburg, Kiesler, (1988) reported several exploratory factor analytic results for Wagner, & Schmidt, 1994; Wiggins, 1994). In addition, her selected surfaces that supported a two-dimensional factorial study was hampered by small sample size (n = 73) and impre- structure with an appropriate circumplex ordering of clusters cise measurements of the IPC (using a modified version of the around the two axes. However, Benjamin (1994) reported sev- Interpersonal Adjective Scales [IAS]; Wiggins, 1991b) and the eral factor analyses in which a third factor was identified in her FFM (using a five-factor solution of Jackson's [1974] Personal- data. In all cases, sample size was lower than optimal (many ity Research Form). Wiggins (1994) concluded that "although with ns < 75), and factor loadings suggested deviation from convergences between the FFM and IPC have been fairly well the equal-spacing circumplex criterion. Given these diverse re- established, the relations between these models and the three sults, it seemed useful to test the circumplexity of SASB with planes of SASB are still not clear" (p. 334). Thus the second a larger sample of participants. Because SASB is used for re- goal of the present study was to clarify the structural relations search in both clinical and nonclinical populations and can be between the three models by using appropriate analytic methods, assessed by means of both long and short forms of the Intrex, we employing standard IPC and FFM instrumentation, and using a investigated circumplexity in both young adult college students large sample of participants. Structural clarification may guide (using the short form) and adult psychiatric inpatients (using the integration of individual-differences and dyadic approaches the long form) in order to maximize generalizability. The goals to interpersonal research and theory. of these analyses were to determine, using confirmatory factor methods (Browne, 1992), how well an explicit circumplex Method model fit the SASB data and to assess the degree to which SASB clusters conformed to the equal-spacing criterion of Participants circumplexity. Normal sample. Three hundred seventy-six introductory psychology Beyond the circumplexity of SASB, we also investigated the students (125 men, average age = 19.3 years, and 238 women, average relations between the dimensions underlying individual-differ- age = 19.1 years; 13 missing values for gender) participated in the ences and dyadic representations of interpersonal space. Both study for course credit. Participants completed a battery of self-report approaches have a long empirical history of deriving and validat- measures in groups for two semesters in the first author's laboratory. ing two fundamental dimensions, and theoretically they share Inpatient sample. Scores of selected SASB surfaces for 187 psychi- atric inpatients with a variety of Diagnostic and Statistical Manual of at least one dimension in common: nurturance-affiliation-love. Mental Disorders (DSM-W; American Psychiatric Association, 1980) Considering the virtue of parsimony, is it necessary to split diagnoses (106 women and 81 men, average age = 31.9 years) were the IPC dimension of dominance-submission into two separate generously provided for the current analyses by Lorna Smith Benjamin. dimensions as proposed by SASB? Details regarding the sample can be found in Benjamin (1994) and Finally, we examined the relations between SASB and addi- Benjamin and Wonderlich (1994). 1634 PINCUS, GURTMAN, AND RUIZ

Measures 60-item self-report inventory designed to assess the five personality dimensions of Neuroticism, Extraversion, Openness to Experience, Intrex long form. Inpatient participants completed the long form of Agreeableness, and Conscientiousness. Raters endorse how self-descrip- the Intrex questionnaires (Benjamin, 1983) for selected relationships. tive each item is on a 5-point Likert scale (0 = strongly disagree, 4 = In the present study, we analyzed responses to introject at best, introject strongly agree). It is a short version of the Revised NEO Personality at worst, mother acted toward participant when participant was a child Inventory. Previous studies have demonstrated high reliability and valid- (i.e., mother to participant—transitive), participant reacted to mother ity of the NEO-FFI (see Costa & McCrae, 1992, for a review). (i.e., participant to mother—intransitive). This allowed us to assess structure on all three planes and a complementary (action-reaction) relationship for each inpatient. Results The Intrex long form uses 36 items per surface to represent the full Circumplex Analyses of SASB: Student Sample SASB model. A given interpersonal relationship (e.g., the relationship with one's mother) is assessed by rating 144 items. There are 36 items The Intrex questionnaire yielded 14 eight-item surfaces (po- to measure the mother's transitive focus on the rater and 36 items for tential circumplex structures) for possible analysis. TVvo of the the mother's intransitive reactions to the rater. Thirty-six more items are surfaces were of the introject (at best, at worst), and the re- devoted to the rater's transitive focus on the mother, and 36 assess the maining 12 involved 4 surfaces each for the mother-self, fa- rater's intransitive reaction to the mother. For example, the item "She ther-self, and mother-father relationships. We chose 10 of the accuses and blames me and tries to get me to admit I'm wrong" is 14 surfaces for analysis, omitting the 4 surfaces that did not transitive, and it contributes to Cluster 6 on the transitive plane. The complementary item "I bury my rage and resentment and scurry to involve the self (i.e., the 4 mother-father surfaces). In order appease her'' is intransitive and contributes to Cluster 6 on the intransi- to have a realistic yardstick for comparisons, we also analyzed tive surface. Each item was rated on a 0 (not at all characteristic of the IAS and the IIP-C; previous studies suggest that these two the relationship) to 100 (perfectly characteristic of the relationship) measures have a close fit to the circular model (Kiesler, 1996; scale marked at 10-point intervals. Reliabilities for cluster scales range Tracey, Gurtman, & Rounds, 1996; Tracey & Schneider, 1995; from .67 to .90 (Benjamin, 1983). Wiggins, Steiger, & Gaelick, 1981). Intrex short form. Our normal sample completed the short version We subjected each correlation matrix to a confirmatory factor of the Intrex (Benjamin, 1988). In the short form, a single item repre- analysis, fitting the factor model for an equally spaced circum- sents each cluster of each SASB surface. Participants completed ratings plex to each of the obtained correlation patterns. This factor of introject at best state of functioning and worst state of functioning, model is a circulant matrix (Browne, 1992; Guttman, 1954; relationship with mother when rater was 5 to 10 years old, relationship with father when rater was 5 to 10 years old, and relationship between Tracey & Rounds, 1995; Wiggins et al., 1981) as illustrated in mother and father when rater was 5 to 10 years old. Test-retest reliabili- Table 1. The table reveals that, for an eight-element circumplex, ties of the Intrex short form range from an average of .66 across the the correlation structure for an equally spaced circumplex re- eight introject-at-worst clusters to an average of .92 across the eight duces to a four-parameter model with the inequality constraint subject-acted-towards-father clusters over a 4-week interval (average Pl > Pl > P3 > Pi- test-retest reliability for all SASB clusters = .87), and substantial In an important contribution, Browne (1992) showed that it is convergence between long and short forms has been reported (Benjamin, possible to "reparametricize" this circulant matrix as a Fourier 1988). series of the general form Interpersonal Adjective Scales (IAS). The IAS (Wiggins, 1991b; Wiggins, Trapnell, & Phillips, 1988) consists of 64 adjectives, 8 re- Pij = /30 + I ft cos(k*ed) (1) flecting each octant of the IPC. Each adjective is rated on an 8-point

Likert scale (1 = extremely inaccurate, 8 = extremely accurate), in terms with k = 1 tow components, and where piS is the common- of how characteristic it is of the rater. Research indicates that the IAS factor correlation of variables i, j; Od is the angular discrepancy has acceptable internal consistency (alphas range from .75 for octant JK (Qd = 0, — 0;) of the variables' polar angles (for a p-element to .86 for octant LM), and it exhibits meaningful correlations with other circumplex, necessarily a multiple of 360°//?); and m s, p/2. self-report measures of personality (Wiggins & Broughton, 1985, 1991) A constraint is imposed that /?s are nonnegative and sum to 1. and with ratings of nonverbal interpersonal behavior (Gifford & O'Con- nor, 1987). When the IAS scales are subjected to a principal-components To test the circumplex model, we can fit this function to the analysis and the scales are plotted on the two extracted components, a clear circumplex structure emerges (Wiggins et al., 1988). Inventory of Interpersonal Problems Circumplex Scales (IIP-C). Table 1 The IIP-C (Alden et al., 1990) is a 64-item subset of the 127-item Equally Spaced Eight-Element Circulant Correlation Matrix Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer, Ureflo, & Villasefior, 1988), developed from verbatim analyses of the 0° 45° 90° 135° 180° 225° 270° 315° presenting complaints of interviewees for psychotherapy. The IIP-C consists of eight 8-item scales intended to operationalize the octants of a circumplex of interpersonal problems. Participants indicated their de- Pl — gree of distress or difficulty for each problem (e.g., ' 'I am too aggressive pl Pl — toward other people," "It is hard for me to trust other people") on a Pl Pl Pl — 4-point scale (0 = not at all, 4 = extremely). Alpha reliabilities range Pi pl Pl Pl from .72 for octant NO to .85 for octants FG and HI. Previous research Pl pi Pl Pl Pl Pl Pl pi P3 Pl Pl has demonstrated that the IAS and IIP-C octant scores show excellent Pi Pl Pl Pi Pl Pl Pl — convergence in circumplex space (Alden et al., 1990).

NEO Five Factor Inventory (NEO-FFI). The FFM was assessed Note. Elements are represented by their polar angles, p, > p2 > using the NEO-FFI (Costa & McCrae, 1992). The NEO-FFI is a Pi > Pi- STRUCTURE OF SASB 1635 correlation pattern. One advantage of doing so is that the first spaced model does not constrain the polar angles but conse- two beta parameters for the model are essentially factor weights. quently is a less parsimonious model (with p - 1 = 7 more Specifically, /30 indexes the magnitude of the general factor of free parameters). The analyses were conducted using AUTFIT the circumplex, and /3i indexes the combined size of the two (Browne & DuToit, 1992), a program designed to fit nonstan- succeeding factors—for the interpersonal circumplex, theoreti- dard models, in combination with CIRCUM (Browne, 1992), cally, dominance and love, and, for SASB, affiliation and auton- which is a satellite program for fitting circumplex models. The omy (or interdependence). Note that if a meaningful representa- maximum-likelihood criterion was employed. tion can be modeled in two dimensions, then /3l will be relatively Table 2 summarizes the results of these analyses, contrasting, large; however, a large /?i does not guarantee circumplex struc- in each instance, the equally spaced and unequally spaced mod- ture, nor does it indicate the optimal number of dimensions els. The table reports various model fit measures (see Loehlin, needed to fit the data. 1992, for a further discussion) as follows: (a) the x2 likelihood We fit an m = 3 version of Browne's (1992) Fourier model ratio; (b) F, the maximum-likelihood discrepancy function be- to each matrix. All 16 matrices that were analyzed in this article tween the obtained and implied matrix, which is a measure of are available from the authors. We started with the equally absolute fit; (c) the goodness-of-fit index (GFI; Joreskog & spaced, equal-communality model, then relaxed the equal- Sorbom, 1986), which is analogous to a squared multiple corre- spacing constraint. The equally spaced model assumes that vari- lation and can range from 0 to 1; (d) the GFI adjusted for ables are uniformly distributed around the perimeter of the circle degrees of freedom (AGFI; Joreskog & Sorbom, 1986), which (i.e., their polar angles are at 45° intervals); the unequally is a parsimony-weighted measure of model fit (both GFI and

Table 2 Summary of Confirmatory Factor Analyses and Parameter Estimates of Beta Weights Confirmatory factor analyses Beta weights

Surface or circle Spacing n df p x2 F GFI AGFI NFI ft ft ft ft Data: Student sample, Intrex short form

Introject at best Equal 373 24 12 226.76 0.610 .862 .793 .213 .730 .039 .018 Unequal 373 17 19 103.46 0.278 .937 .866 .544 .202 .667 .052 .078 Introject at worst Equal 374 24 12 215.49 0.578 .867 .801 .215 .748 .010 .027 Unequal 374 17 19 70.27 0.188 .956 .907 .674 .189 .620 .061 .130 Mother-self transitive Equal 376 24 12 340.44 0.908 .840 .760 .077 .920 .000 .004 Unequal 376 17 19 55.61 0.148 .965 .925 .837 .066 .745 .006 .183 Mother-self intransitive Equal 371 24 12 411.00 1.111 .795 .692 .196 .764 .021 .019 Unequal 371 17 19 103.66 0.280 .934 .861 .748 .101 .505 .000 .394 Self-mother transitive Equal 374 24 12 336.25 0.901 .842 .762 .147 .750 .049 .054 Unequal 374 17 19 105.38 0.283 .935 .862 .687 .000 .616 .384 .000 Self-mother intransitive Equal 372 24 12 421.55 1.136 .111 .666 .081 .872 .020 .028 Unequal 372 17 19 76.14 0.205 .952 .900 .819 .033 .652 .000 .316 Father-self transitive Equal 376 24 12 476.55 1.271 .755 .632 .102 .819 .028 .051 Unequal 376 17 19 94.85 0.253 .940 .874 .801 .069 .647 .022 .262 Father-self intransitive Equal 372 24 12 545.55 1.470 .732 .598 .124 .775 .063 .038 Unequal 372 17 19 122.82 0.331 .923 .838 .775 .056 .759 .000 .185 Self-father transitive Equal 375 24 12 402.43 1.076 .827 .741 .158 .795 .000 .047 Unequal 375 17 19 82.92 0.222 .945 .884 .794 .124 .612 .095 .168 Self-father intransitive Equal 372 24 12 426.13 1.149 .782 .673 .084 .893 .023 .000 Unequal 372 17 19 121.51 0.328 .932 .855 .715 .050 .695 .014 .242 IAS Equal 357 24 12 138.76 0.390 .924 .886 .095 .768 .014 .124 Unequal 357 17 19 63.59 0.179 .962 .919 .542 .095 .751 .015 .139 IIP-C Equal 364 24 12 127.79 0.352 .925 .888 .574 .383 .028 .015 Unequal 364 17 19 63.59 0.175 .959 .914 .502 .544 .384 .034 .037

Comparison data: Inpatient sample, Intrex long form Introject at best Equal 187 24 12 245.61 1.320 .780 .671 .215 .781 .003 .001 Unequal 187 17 19 28.93 0.156 .965 .926 .882 .196 .703 .000 .101 Introject at worst Equal 187 24 12 316.66 1.702 .744 .616 .145 .841 .015 .000 Unequal 187 17 19 52.88 0.284 .937 .867 .833 .131 .762 .000 .108 Mother-self transitive Equal 184 24 12 371.30 2.029 .713 .569 .154 .801 .023 .021 Unequal 184 17 19 82.87 0.453 .902 .793 .777 Ml .807 .015 .000 Self-mother intransitive Equal 184 24 12 167.62 0.916 .845 .767 .293 .658 .032 .017 Unequal 184 17 19 36.14 0.198 .953 .901 .784 .302 .643 .043 .013

Note. IAS = Interpersonal Adjective Scales; IIP-C = Inventory of Interpersonal Problems Circumplex Scales; SASB = Structural Analysis of Social Behavior; p = parameters; F = maximum-likelihood discrepancy function; GFI = goodness-of-fit index; AGFI = adjusted goodness-of-fit index; NFI = normed fit index. 1636 PINCUS, GURTMAN, AND RUIZ

AGFI were computed from formulas presented in Maiti & ses limit examination to circular structures. It cannot be deter- Mukherjee, 1990); and (e) the normed fit index (NFI; Bentler, mined from the previous analyses whether the SASB data are 1990), which is a measure of incremental fit (NFI = xl ~ better represented by an alternative structure. As a further exami- X?)/ X?) often used to compare nested models (here the un- nation of the structure of SASB and the other instruments, we equally spaced vs. equally spaced models). also subjected the set of correlation matrices to a multidimen-

The estimates of the beta parameters, 0O, Pu 02, and Pi, are sional scaling (MDS) analysis. Davison (1985, 1994) among shown in the last four columns of Table 2. Each SASB surface others (see also Gurtman, 1994) has shown that MDS may be was characterized by a relatively large /3i weight, indicating that an especially valuable tool for capturing and representing both a meaningful two-dimensional representation of the data could circumplex and quasi-circular structure when it exists. In addi- be modeled (although, as noted previously, this does not neces- tion, MDS provides stress values, which allows the number of sarily suggest equal weight for the two factors). These values dimensions needed to best describe the data to be evaluated. are equivalent to those describing the well-established IAS cir- This could not be quantified in the previous analyses. cumplex. The 0O value for the IIP-C is greater than its px For each matrix, we compared one-, two-, and three- value, reflecting the large general factor that has been previously dimensional solutions for representing the dissimilarities im- described in its construction (Alden et al., 1990; Horowitz et plied by the correlation matrix. Kruskal's nonmetric MDS al., 1988). method, as implemented in SYSTAT (Wilkinson, 1990), was The confirmatory analyses revealed that the equally spaced used. Table 3 shows the stress values (badness of fit) for each circumplex model was generally not a good fit for the SASB of these solutions. As can be seen, each measure could be fit in data, with GFIs ranging from .732 to .867 and AGFIs ranging two dimensions (stress values were close to 0). However, the from .598 to .801. This inadequate fit is also apparent when the gain in fit from a one-dimensional to a two-dimensional model correlation patterns are simply compared to the circulant ideal was smaller for the SASB scales than for the IAS and IIP-C; displayed in Table 1. Use of Browne's (1992) CIRCUM pro- indeed, for 4 of the 14 SASB surfaces (students' self-mother gram made it possible to also test a circular model in which intransitive, students' father-self intransitive, patients' introject the equal-spacing requirement is relaxed. Removing the equal- at worst, patients' mother-self transitive) the dissimilarities spacing requirement led to a relatively large gain in model fit could be adequately represented in a one-dimensional space (see NFIs), and all GFIs improved to beyond a .9 standard. (stress < .02 as the criterion). However, most AGFIs, which weight parsimony, were still below As the next step, for each of the 16 measures, we fit the this standard. In contrast to the results for SASB, the IAS and model of an ellipse—with the circle being a special case— IIP-C data conformed more closely to an equally spaced cir- to the two-dimensional configurations obtained from the MDS cumplex model (GFIs were .924 and .925, and AGFIs were .888 analysis. This involved fitting the theoretical X, Y coordinates and .886, respectively). Fit improved somewhat when the equal- for an ellipse, X = a*cos(8), Y = b*sin(Q), a s b, to the spacing requirement was relaxed (GFIs were .962 and .959, and obtained X, Y coordinates as given by the MDS solution. The AGFIs were .919 and .914, respectively). parameters, a and b, were obtained by minimizing, according to the least squares criterion, the discrepancy function

Circumplex Analyses of SASB: Inpatient Sample 2 2 D = ZJ IXi - a*cos(6) Y + \Y{ - b*sin(6) | (2) For comparison, we also subjected another data set to a con-

firmatory analysis. These data were from 187 inpatients who for i = 1 to 8, where Xt and Yt are the dimensional coordinates responded to the long-form version of the Intrex (Benjamin, from the MDS solution for scale, i (i = 1 to 8), and 8t is the 1983). Data from a subset of this sample were previously pre- angular orientation of scale in the two-dimensional space. A sented in Benjamin (1994). Four surfaces were selected for parameter to allow for decentering of the figures was also in- analysis—the introject at best, introject at worst, mother-self cluded. The model's goodness of fit was calculated as R2 = 1 transitive, and self-mother intransitive. As previously, the re- — D /SSTOTAL. sults are also summarized in Table 2. As can be seen, the results Each of the 16 analyses then involved fitting the model to 8 were fairly similar to those obtained for the primary sample points (clusters or scales). We also computed an eccentricity (students completing the short-form measure). Again, px values index, e, indicating the degree to which the modeled elliptical were all relatively large. The equally spaced model was generally structure departed from that of a circle (the special case: a = not a good fit. Model fit improved considerably when equal b), or spacing was relaxed, but AGFIs for two of the four surfaces (introject at worst and mother-self transitive) did not reach e = (a2 - b2yl2/a (3) acceptable standards. A value of e = 0 would indicate a perfect circle (i.e., in the Spatial Analysis and Representation sense of projected points being equidistant from the center); higher eccentricity values (up to a maximum of 1) would indi- The accumulation of results suggests that SASB data do not cate a more elongated or "flattened" figure. meet stringent circumplex criteria. Although results suggest that To index departures from equal spacing, we also calculated for many SASB surfaces an unequally spaced circular structure a "gap statistic" (Upton & Fingleton, 1989) for each measure. is a closer approximation to the data, the combined AUTFIT Given an equally spaced, circularly arranged eight-octant mea- (Browne & DuToit, 1992) and CIRCUM (Browne, 1992) analy- sure, each scale would be separated by a gap of 45° from its STRUCTURE OF SASB 1637

Table 3 Summary of Multidimensional Scaling Analyses

Stress Ellipse model fit statistics

Surface DIM = 1 DIM = 2 a b D2 R2fit e G

SASB student sample Introject at best .134 .012 1.069 .805 .154 .981 .658 24.33 Introject at worst .148 .022 1.033 .897 .094 .988 .496 24.47 Mother-self transitive .108 .007 1.048 .823 .040 .995 .620 38.08 Mother—self intransitive .137 .010 1.098 .774 .126 .984 .710 40.60 Self-mother transitive .123 .002 1.016 .874 .252 .968 .510 37.53 Self-mother intransitive .001 .001 1.095 .701 .134 .983 .768 36.59 Father-self transitive .154 .010 1.013 .954 .029 .996 .336 37.11 Father-self intransitive .010 .002 1.078 .685 .214 .973 .772 35.61 Self-father transitive .151 .007 1.010 .967 .014 .998 .289 42.84 Self-father intransitive .106 .032 1.054 .792 .218 .973 .660 30.57 IAS .326 .001 1.003 .997 .000 1.000 .112 0.07 IIP-C .279 .019 1.019 .968 .068 .991 .313 12.40 SASB patient sample Introject at best .081 .002 1.060 .783 .050 .994 .674 32.24 Introject at worst .000 .001 1.050 .731 .111 .986 .718 35.36 Mother-self transitive .003 .003 1.040 .836 .032 .996 .594 36.53 Self-mother intransitive .120 .004 1.046 .879 .018 .998 .543 27.21

Note. DIM = dimension; a = major axis weight; b = minor axis weight; D2 = sum of squares error; e = eccentricity of ellipse; G = gap statistic; SASB = Structural Analysis of Social Behavior; IAS = Interpersonal Adjective Scales; IIP-C = Interpersonal Problems Circumplex.

nearest neighbor. The gap statistic, G, is based on the average 1 and 5, reflecting autonomy. This general pattern also occurs absolute discrepancy from that ideal, or for the other SASB surfaces when fit in two dimensions.

G = X Abs(di - 45°)/8 (4) Dimensional Relations for (' = 1 to 8, where d, = 0,- + 1 — 0, (except when i = 8, d, Results of the spatial analyses indicate that, although not a = 6t + 9i - 360°), and #, is the angular orientation, in degrees, circumplex, SASB data can be represented in two dimensions for scale, i, with scales ordered so that 6, + 1 a #,. It follows reflecting dyadic conceptions of autonomy and affiliation. In that, for an equally spaced distribution of points, G would equal order to evaluate the relations between individual-differences 0; higher values of G would indicate greater deviations from and dyadic approaches to description of the interpersonal do- the equal-spaced ideal. main and to clarify the nature of SASB's hybrid introject sur- The continuation of Table 3 shows the results of these further face, a series of conjoint principal-components analyses (PCAs) analyses conducted on the 16 matrices. Of main interest are the were conducted on dimensions of SASB, the IPC, and the FFM. eccentricity (e), gap (G), and model fit (D2, R2) statistics. For each of 14 surfaces, respondents' affiliation and autonomy Looking first at the results for the IAS, these statistics together scores were computed using a weighted sum of item responses suggest that the measure conforms well to the model of an (Benjamin, 1988; Paivio & Greenberg, 1995; Pincus, Newes, equally spaced circle; the IIP—C configuration offers a slightly Dickinson, & Ruiz, 1998; Pincus & Ruiz, 1997; Quintana & poorer, but still reasonable, fit to this ideal. In contrast, the Meara, 1990). Unlike computation of IPC dimensions based on SASB measures tended to be best characterized by an elliptical trigonometric weights (Wiggins, Phillips, & Trapnell, 1989), (eccentric) form, rather than a circular form. Their relatively weights for the derivation of SASB dimensional scores from the high gap values also indicate unequal spacing of the scales. Intrex short form are based on the number and prototypicality of As illustration of these results, Figures 3 and 4 show the long-form items that were collapsed into single, short-form MDS configurations for two sample SASB measures: mother- items marking each cluster (Benjamin, 1988). Although un- self transitive surface (student group, short form) and the self- usual, this approach yields normally distributed, orthogonal di- mother intransitive surface (patient group, long form). In each mensions in large samples (Pincus et al., 1998). figure, the curve for the best-fit ellipse is also plotted. As can We conducted all analyses using two sets of dimensional be seen in both examples, the points (clusters) can be closely scores—one including all 10 SASB surfaces used in the previ- modeled by an elongated elliptical curve. In addition, the clus- ous analyses and one using only those surfaces that specifically ters are not evenly distributed but, for six of the eight items, represent self-ratings (respondents' actions and reactions to are joined in one of two major groups at opposite ends of a mother, respondents' actions and reactions to father, introject at single dimension: the 2-3-4 AG elements and the 6-7-8 DAG best and at worst). In order to simplify our presentation, we elements. The second, smaller dimension is marked by Clusters report the analyses conducted using only the smaller set of 1638 PINCUS, GUKTMAN, AND RUIZ

1.50

-1.50

Figure 3. Elliptical structure of the Structural Analysis of Social Behavior rater-reacts-to-mother (intransi- tive) surface for the inpatient sample, based on Intrex long form. Cluster 3 is fixed at 0°.

scores. Please note, however, that results were highly compara- and introject autonomy dimensions, but affiliation is not dis- ble for both sets and those involving all SASB surfaces are criminated by interpersonal focus. Introject affiliation is sepa- available from the authors. rate from the relationship affiliation ratings. SASB. The multiplane model of SASB suggests that auton- SASB and the IPC model. In order to directly investigate the omy and affiliation should be further discriminated by interper- relations among dyadic and individual-differences dimensions of sonal focus (transitive and intransitive). The SASB model im- interpersonal representation, we conducted a conjoint PCA of plies that separable dimensions should be identified for transitive the SASB self-ratings and the dominance and nurturance dimen- and intransitive affiliation and autonomy. This is particularly sions of the IAS. Results are presented in Table 5. Examination important for the vertical axis, in that this distinction creates of the scree plot suggested a five-component solution accounting the structural difference between the IPC and SASB models. for 63.9% of the variance. The first component is an affiliation Thus, PCA should reveal transitive and intransitive components component, marked by all SASB affiliation-with-parents dimen- for affiliation and autonomy, as well as separable components sions and IAS love. The second component is marked by intro- for the introject dimensions. The results of this analysis are ject affiliation and IAS dominance, indicating that introject af- presented in Table 4. filiation is not the same as interpersonal affiliation on the dyadic Examination of the scree plot suggested a six-component SASB surfaces. The final three components maintain the distinc- solution accounting for 78.8% of the variance. The first two tions among transitive, intransitive, and introject autonomy di- components represented separate affiliation components for spe- mensions identified in the previous PCA. As expected, the dy- cific relationships (self with mother, self with father) rather adic planes of SASB and the IPC share a common affiliation- than the expected distinction between transitive and intransitive nurturance dimension, whereas SASB autonomy is not related dimensions. The next two components did support the distinc- to the fundamental IPC dimensions. tion between transitive and intransitive autonomy. The last two SASB and the FFM. As there are some large secondary components reflected the two dimensions of the SASB introject. loadings in the previous factor space and no discernible relations From this analysis, the multiplane SASB model is only partially between transitive, intransitive, and introject autonomy and the supported. The results indicate separate transitive, intransitive, IPC, we conducted a third conjoint PCA to investigate the rela- STRUCTURE OF SASB 1639

1.50

-1.50 1.50

-1.50

Figure 4. Elliptical structure of the Structural Analysis of Social Behavior mother-acts-toward-rater (transi- tive) surface for the student sample, based on Intrex short form. Cluster 3 is fixed at 0°.

Table 4 Principal Components for SASB Self-Ratings Only

SASB dimensions I II III IV V VI

Self s intransitive affiliation towards father .92 .16 -.01 -.04 .06 -.04 Self s transitive affiliation towards father .92 .20 .05 -.11 .07 -.02 Self s intransitive affiliation towards mother .16 .89 .07 -.05 .17 -.09 Self s transitive affiliation towards mother .22 .89 .05 -.04 .16 .07 Self's transitive autonomy-granting to father -.03 .00 .89 -.06 .01 -.02 Self's transitive autonomy-granting to mother .07 .11 .86 .09 .10 .07 Self s intransitive autonomy-taking from mother .10 -.16 .02 .85 -.01 .14 Self s intransitive autonomy-taking from father -.28 .08 .01 .81 .05 .01 Introject affiliation at worst -.02 .09 .03 .10 .84 -.03 Introject affiliation at best .14 .19 .07 -.06 .78 .04 Introject autonomy at worst -.11 .10 .02 .05 -.09 .85 Introject autonomy at best .05 -.25 .02 .09 .11 .76

Note. N = 371; 78.8% of variance, varimax rotation. Loadings s.33 are indicated in boldface. SASB = Structural Analysis of Social Behavior. 1640 PINCUS, GURTMAN, AND RUIZ

Table 5 Conjoint Principal Components for SASB Self-Ratings and IAS Dimensions

SASB and IAS dimensions I II III IV V

Self s transitive affiliation towards father .89 .05 .08 -.01 -.21 Self s intransitive affiliation towards father .87 .05 .10 .06 -.18 Self s transitive affiliation towards mother .59 .37 -.43 .10 .08 Self s intransitive affiliation towards mother .53 .39 -.49 .09 .08 IAS love .41 .08 -.11 .14 .15 Introject affiliation at best .17 .75 -.03 .12 -.06 Introject affiliation at worst -.01 .72 -.05 .07 .11 IAS dominance .15 .69 .27 -.13 .00 Introject autonomy at best .04 -.01 .72 .08 .12 Introject autonomy at worst -.06 .16 .63 .06 .08 Self's transitive autonomy-granting to father -.04 .04 .04 .87 -.09 Self's transitive autonomy-granting to mother .16 .05 .09 .85 .13 Self s intransitive autonomy-taking from father -.15 .03 -.03 .05 .85 Self s intransitive autonomy-taking from mother .03 .02 .33 -.02 .74

Note. N = 357; 63.9% of variance, varimax rotation. Loadings a | .331 are indicated in boldface. SASB = Structural Analysis of Social Behavior; IAS = Interpersonal Adjective Scales.

tions among dyadic autonomy, the SASB introject, and a more yses indicate poor structural validity of SASB surfaces (with comprehensive model of personality description, the FFM. The regard to circumplexity), the SASB model fared better at the results of a conjoint PCA of SASB self-ratings, dominance and dimensional level as five of the six predicted dimensions were nurturance dimensions of the IAS, and the FFM (as measured recovered using PCA. Examination of the underlying dimensions by the NEO-FFI) are presented in Table 6. of SASB partially support its multiplane structure in that (a) Examination of the scree plot suggested a seven-component transitive, intransitive, and introject autonomy are consistently solution accounting for 70.4% of the variance. The first compo- identified as discrete components and (b) introject affiliation is nent was marked by low Neuroticism, dominance, introject af- distinct from transitive and intransitive affiliation. However, the filiation, and Extraversion. We interpret this component as re- distinction between transitive and intransitive affiliation failed flecting social confidence or social agency rather than nurtur- to emerge. Conjoint PCA indicated that, as expected, SASB's ance. The next three components reflect state (i.e., relationship) dyadic planes and the IPC share only a single interpersonal and trait factors reflecting interpersonal affiliation. The fifth dimension (affiliation-nurturance). SASB introject affiliation component contrasts introject autonomy with Conscientious- is unexpectedly related to IPC dominance. The autonomy di- ness, reflecting self-control or constraint. The sixth component mension of each SASB plane is distinct from the IPC, implying reflects transitive autonomy-granting. The last component is unique conceptions of interpersonal and self-representation. Ex- marked by intransitive autonomy-taking and Openness to Expe- amination of SASB's structural relations with the FFM indi- rience, suggesting that going one's own way (i.e., autonomy- cated that (a) the introject surface is most strongly associated taking) in relationships may be more related to what is tradition- with Neuroticism and Conscientiousness, (b) intransitive auton- ally considered a noninterpersonal trait than to interpersonal omy-taking is most strongly related to Openness to Experience, traits. and (c) transitive autonomy-granting is not well-captured by other superordinate dimensions of personality description. Im- plications of these results are discussed below. Discussion

Our results do not support the circumplex structure of SASB Structure of SASB surfaces as they only meet the first of three geometric criteria for circumplexity—that is, differences in variables for each SASB surfaces. The best fitting two-dimensional model for surface are reducible to differences in two dimensions. SASB SASB surfaces was an unequally spaced ellipse. As the eccen- clusters fail to exhibit a constant radius and equal spacing, pro- tricity of an elliptical structure increases, the ability to "ac- ducing elliptical structures when MDS is applied to the data. In count' ' for relationships among its variables in a single dimen- contrast to SASB, our results support the circumplex structure sion clearly increases. From a geometric standpoint, the Intrex of the IPC as multiple measures generally met all three geomet- appears to be a flawed representation of the SASB model as ric assumptions of circumplexity. Although our circumplex anal- compared to the IAS and IIP-C representations of the IPC. At STRUCTURE OF SASB 1641

Table 6 Conjoint Principal Components for SASB Self-Ratings, IAS Dimensions, and NEO-FFI Dimensions

SASB, IAS, and FFM dimensions I II III IV V VI VII

NEO-FFI Neuroticism -.84 -.03 -.04 -.10 -.03 -.08 .01 IAS dominance .73 .18 -.06 -.04 .07 -.14 .10 Introject affiliation at best .73 .09 .14 .13 -.06 .16 -.04 Introject affiliation at worst .66 -.07 -.02 .17 -.04 .09 .08 NEO-FFI Extraversion .62 .12 .50 .08 .03 -.13 -.03 Self s transitive affiliation towards father .10 .89 .13 .21 -.05 .03 -.08 Self s intransitive affiliation towards father .13 .89 .12 .16 -.08 -.02 -.04 IAS love .06 .04 .86 .09 .01 .04 .01 NEO-FFI Agreeableness .05 .17 .85 .15 -.05 .05 .01 Self s intransitive affiliation towards mother .20 .17 .13 .87 -.11 .04 -.03 Self s transitive affiliation towards mother .18 .22 .18 .85 -.09 .05 -.05 Introject autonomy at worst .14 .10 -.09 .07 .79 -.04 .06 Introject autonomy at best .02 .07 .13 .24 .72 .06 .08 NEO-FFI Conscientiousness .41 .19 .13 .08 -.61 -.11 -.13 Self s transitive autonomy- granting to father .04 -.06 -.04 -.01 .04 .86 -.04 Self s transitive autonomy- granting to mother .06 .07 .09 .09 .03 .85 .14 Self s intransitive autonomy- taking from mother .11 -.01 -.08 -.10 .05 -.05 .80 Self s intransitive autonomy- taking from father .04 -.28 .04 .15 .02 .05 .78 NEO-FFI Openness to Experience -.08 .21 .08 -.16 .24 .19 .52

Note. N = 357; 70.4% of variance, varimax rotation. Loadings s | .331 are indicated in boldface. SASB = Structural Analysis of Social Behavior; IAS = Interpersonal Adjective Scales; NEO-FFI = NEO Five- Factor Inventory; FFM = Five Factor Model.

best, the Intrex allows for simple structure assessment of the min, 1993b, 1995; Florsheim, Henry, & Benjamin, 1996; Henry, autonomy and affiliation dimensions of each surface. Applica- 1994). Clusters 2, 3, and 4 representing the AG and Clusters 6, tions of continuous measurement methodology (Gurtman, 1994) 7, and 8 representing the DAG fell close together as sub- and geometric classification (e.g., Wiggins & Broughton, 1985, groupings and marked the horizontal axis of SASB surfaces. 1991; Wiggins et al., 1989) require a more circular structure Empirical investigations have also indicated that AG and DAG than those recovered in the present study. It should be noted clusters' links to adult attachment styles are distinct from those that earlier IPC-based instruments such as LaForge and Suczek's of Clusters 1 and 5 marking the vertical axis (Dickinson et al., (1955) Interpersonal Check List exhibited poor structural corre- 1996). This suggests that although SASB surfaces do not exhibit spondence with the IPC model (Paddock & Nowicki, 1986). circumplex structure, they can be represented in two distinct Significant effort has been invested in fine-grained item selec- dimensions. As the theoretical bases underlying dyadic represen- tion for contemporary IPC-based instruments (Alden et al., tations of the interpersonal domain are valuable to understanding 1990; Wagner et al., 1995; Wiggins, 1979, 1991b). personality development, psychopathology, and adaptive func- Although our results do not support the structural validity of tioning (e.g., Blatt, 1990; Blatt & Blass, 1990; Schaefer, 1997), SASB as a circumplex, a distinction can be made between the we also attempted to clarify the substantive meaning of SASB factor structure of the current Intrex questionnaires and the con- dimensions. ceptual model it strives to operationalize. Using an alternative The SASB model. The full SASB model is not simply two methodology, Benjamin (1988) presented data suggesting that dimensional. The third "implied" dimension, interpersonal fo- when naive judges are asked to rate Intrex items for their under- cus, generates six possible measurable dimensions: transitive lying dimensionality, this procedure yielded much more equally affiliation, transitive autonomy, intransitive affiliation, intransi- spaced angular placements that better conformed to the model tive autonomy, introject affiliation, and introject autonomy. than factor analyses. The deviations in dispersion of SASB clus- PCA analysis of SASB dimensions revealed support for all ters in our data (for both the long and the short forms of the three distinct autonomy dimensions, but affiliation only sepa- Intrex and in both psychiatric and nonpsychiatric samples) mir- rated into interpersonal (transitive and intransitive) and intro- ror recent integrations of attachment theory and SASB (Benja- ject components. The interpersonal affiliation dimensions split 1642 PINCUS, GURTMAN, AND RUIZ into relationship factors, which we interpret as "state" effects, a new dimension, autonomy in various forms (not limited to in that affiliation in one relationship does not imply affiliation SASB), which appears to be important and not wholly ac- in another. This result also has implications for the structural counted for by current models of individual differences. In the validity of the SASB model. Either the Intrex fails to ade- present study, transitive autonomy-granting (i.e., letting others quately distinguish active and reactive affiliation, or this dimen- be free to "do their own thing") was unrelated to individual sion simply permeates interpersonal functioning. Our data sup- differences assessed by the IPC and the FFM. Beyond the level port the existence of five of six SASB dimensions. Of interest, of personality description, autonomy is a major dimension in the lack of a transitive-intransitive distinction occurs for the theories of human and social development (Guisinger & Blatt, dimension that is shared with individual-differences concep- 1994; Mahler et al., 1975). Its continued investigation in person- tions of interpersonal representation. ality research may lead to additional facets of personality de- The SASB introject. The introject surface is proposed to scription and link descriptive models and developmental represent an intrapsychic relationship with the self that reflects theories. the impact on the self-concept of (typically early) interpersonal The integration of dyadic and individual-differences ap- relationships with important others. This hybrid surface is, per- proaches to the interpersonal domain also provides a framework haps, SASB's most unique feature. Although the SASB model to understand the interpersonalness of presumed noninterper- suggests that the structure of the introject surface parallels that sonal traits. Earlier, Pincus and Wiggins (1992) suggested that of the dyadic surfaces, our results indicate that its structure does individuals with different levels of Conscientiousness, Openness not mirror that of the interpersonal planes. When viewed in to Experience, and Neuroticism would have different interper- relation to the FFM of personality traits, the introject represents sonal qualities, even if their IPC profiles were identical. This a model of self-esteem and agency on the affiliation axis and seems illogical if a strictly individual-differences approach to ego-control on the autonomy axis. Self-love reflects low emo- interpersonal representation is taken. If the dyadic approach is tional distress, high dominance, and high Extraversion; that is, taken, we can first ask, ' 'Where will these interpersonal differ- those who accept themselves can more confidently engage and ences be manifested?" Not in the structural representation of adapt in the social world and take charge of things when it is traits, but vis-a-vis the dyadic partners involved in ongoing required. Introject autonomy reflects aspects of Conscientious- interpersonal transactions. In the present study, intransitive au- ness—that is, the ability to modulate self-control, contrasting tonomy-taking (i.e., going one's own way and doing one's own constraint and delay with free expression of the self. These two thing) was related to Openness to Experience. This suggests aspects of the self—control and esteem-agency—are often that openness isn't all in the mind or about the self. This is considered primary targets of psychotherapeutic change (Benja- actually quite obvious in the facet structure of openness min, 1996a; Henry, 1996). This suggests that the SASB introject (Costa & McCrae, 1992) in that a person high in openness to dimensions are mislabeled as they are not internalized equiva- values, ideas, feelings, and actions is bound to react autono- lents of the dyadic dimensions. However, it is of note that the mously rather than submit (i.e., its SASB opposite) consistently convergence of SASB introject dimensions with the noninterper- to the values, ideas, feelings, and actions of another. Costa and sonal FFM dimensions of Neuroticism and Conscientiousness McCrae (1995) suggested that future personality research suggests that the SASB introject does represent aspects of the should focus on intraindividual functioning to account for ob- self-concept that are distinct from interpersonal traits. One ad- servable individual differences. To this suggestion, we add that vantage of SASB may be the generation of testable hypotheses further investigation of the interpersonal influence of all individ- regarding the developmental links between introject dimensions ual differences is a useful step in the integration of personality and important early relationships or the links between therapeu- psychology (see also McCrae, 1996). tic change and therapy relationships. In SASB terminology, these Affiliation, or relatedness, is common to both dyadic and are referred to as predictive principles (Benjamin, 1993b, 1994; individual-differences approaches to interpersonal representa- Schacht, 1994). tion. The current study found no evidence to split this dimension into transitive and intransitive variants. Perhaps the most funda- Integrating Dyadic and Individual-Differences mental of human qualities, it has the evolutionary significance Approaches to the Interpersonal Domain of ensuring the safety of offspring and bonding humans into social groups (Buss, 1996; Hogan, 1996). Given its clear evolu- The ultimate goal of this research is to clarify the relations tionary significance and its consistent identification in both dy- between models of interpersonal representation. The data pre- adic and individual-differences approaches, the distinctions sented here support distinctions between dyadic and individual- made in the SASB model may provide a conceptual symmetry, differences approaches. The data cannot be used to determine but not a true empirical or conceptual distinction. This dimen- which, if either, model is in some way better than the other sion involves the nature and quality of attachment, first as it (i.e., comparisons based on criterion validity; Messick, 1989). relates to early relationships with caregivers (the dyadic aspect) However, we believe personality psychology can benefit from and then to the orientation the person brings to the ongoing examining the potential integration of multiple perspectives participation in, and experience of, all subsequent relationships (e.g., Bellak, 1993; Blatt & Maroudas, 1992). Such efforts may (the individual-differences aspect). result in the expanded understanding of current traits (e.g., Before ending, we wish to alert the reader to three main McCrae, 1996) and the identification of new personality traits caveats. First, one limitation of the current study involves the to complement existing taxonomies (Schaefer, 1997). relationships assessed by SASB. These were retrospective re- The dyadic approach to interpersonal representation provides ports of childhood relations with parents and are susceptible to STRUCTURE OF SASB 1643 bias. Although we do not believe this bias affects the dimension- personality configurations and their implications for psychopathology ality of SASB ratings, it would be important for future research and psychotherapy. In J. L. Singer (Ed.), Repression and dissociation: to obtain SASB ratings on current relationships and compare Implications for personality theory, psychopathology, and health (pp. these with other interpersonal measures. Second, the Intrex short 299-335). Chicago: University of Chicago Press. form uses single-item measures of each cluster, which can Blatt, S. J., & Blass, R. B. (1990). Attachment and separateness: A clearly affect reliability. However, we obtained highly similar dialectical model of the products and processes of psychological de- velopment. The Analytical Study of the Child, 44, 107-127. results in our psychiatric sample, which used the Intrex long Blatt, S. J., & Maroudas, C. (1992). Convergences of psychoanalytic form. Finally, all measures in this study are self-report measures. and cognitive—behavioral theories of depression. Psychoanalytic Psy- Future research on the interpersonal influence of noninterper- chology, 9, 157-190. sonal personality traits should include transactional or observa- Browne, M. W. (1992). Circumplex models for correlation matrices. tional data to comprehensively assess these relationships. Psychometrika, 57, 469-497. Browne, M. W, & DuToit, S. H. C. (1992). Automated fitting of non- standard models. Multivariate Behavioral Research, 27, 269-300. References Buss, D. M. (1996). Social adaptation and the five major factors of personality. In J. S. Wiggins (Ed.), The five-factor model of personal- Alden, L. E., Wiggins, J. S., & Pincus, A. L. (1990). Construction of ity: Theoretical perspectives (pp. 180-208). New %rk: Guilford circumplex scales for the Inventory of Interpersonal Problems. Jour- Press. nal of Personality Assessment, 55, 521—536. Carson, R. C. (1969). Interaction concepts of personality. Chicago: Alpher, V. S. (1996). Identity and introject in dissociative disorders. Aldine. Journal of Consulting and Clinical Psychology, 64, 1238-1244. American Psychiatric Association. (1980). Diagnostic and statistical Carson, R. C. (1991). The social-interactional viewpoint. In M. Hersen, manual of mental disorders (3rd ed.). Washington, DC: Author. A. Kazdin, & A. Bellack (Eds.), The clinical psychology handbook Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among (2nd ed., pp. 185-199). New York: Pergamon Press. young adults: A test of a four-category model. Journal of Personality Cassidy, J. (1995). Attachment and generalized anxiety disorder. In D. C. and Social Psychology, 61, 226-244. Toth (Ed.), Rochester symposium on developmental psychopathology: Bellak, L. (1993). as a science. New "fork: Allyn & Emotion, cognition, and representation (pp. 343-370). Rochester, Bacon. NY: University of Rochester Press. Benjamin, L. S. (1974). Structural analysis of social behavior. Psycho- Costa, P. T, Jr. (1994). Can the SASB bridge personality theory and logical Review, 81, 392-425. clinical psychology? A view from the O, C, E, A, N. Psychological Benjamin, L. S. (1983). Intrex users' manual: Part one. Madison, WI: Inquiry, 5, 319-322. Intrex Institute. Costa, P. T, Jr., & McCrae, R. R. (1992). NEO-P1-R/NEO-FFIprofes- Benjamin, L. S. (1984). Principles of prediction using structural analy- sional manual. Odessa, FL: Psychological Assessment Resources. sis of social behavior. In A. Zucker, J. Aronoff, & J. Rubin (Eds.), Costa, P. T, Jr., & McCrae, R. R. (1995). Solid ground in the wetlands Personality and the prediction of behavior (pp. 121 -173). New \brk: of personality: A reply to Block. Psychological Bulletin, 117, 216- Academic Press. 220. Benjamin, L. S. (1987). Use of the SASB dimensional model to develop Costa, P. T, Jr., & Widiger, T. A. (1994). Personality disorders and the treatment plans for personality disorders: I. Narcissism. Journal of five-factor model of personality. Washington, DC: American Psycho- Personality Disorders, 1, 43—70. logical Association. Benjamin, L. S. (1988). Intrex users' manual. Madison, WI: Intrex Davison, M. L. (1985). Multidimensional scaling versus components Institute. analysis of test intercorrelations. Psychological Bulletin, 97, 94-105. Benjamin, L. S. (1993a). Every psychopathology is a gift of love. Psy- Davison, M. L. (1994). Multidimensional scaling of personality re- chotherapy Research, 3, 1—24. sponding. In S. Strack & M. Lorr (Eds.), Differentiating normal and Benjamin, L. S. (1993b). Interpersonal diagnosis and treatment of per- abnormal personality (pp. 196-215). New 'fork: Springer. sonality disorders. New "fork: . Dickinson, K. A., Pincus, A. L., & Ruiz, M. A. (1996, August). Attach- Benjamin, L. S. (1994). SASB: A bridge between personality theory ment styles and parental representations: A test of Benjamin's inter- and clinical psychology. Psychological Inquiry, 5, 273-316. personal theory. Paper presented at the 104th Annual Convention of Benjamin, L. S. (1995). Good defenses make good neighbors. In H. the American Psychological Association, Toronto, Ontario, Canada. Conte & R. Plutchik (Eds.), Ego defenses: Theory and measurement Florsheim, P., Henry, W. P., & Benjamin, L. S. (1996). Integrating indi- (pp. 38-78). New York: Wiley. vidual and interpersonal approaches to diagnosis: The Structural Anal- Benjamin, L. S. (1996a). A clinician-friendly version of the Interper- ysis of Social Behavior and attachment theory. In F. Kaslow (Ed.), sonal Circumplex: Structural Analysis of Social Behavior (SASB). Handbook of relational diagnosis (pp. 81-101). New 'fork: Wiley. Journal of Personality Assessment, 248-266. Benjamin, L. S. (1996b). Introduction to the special section on Struc- Freedman, M. B., Leary, T, Ossorio, A. G., & Coffey, H. S. (1951). tural Analysis of Social Behavior. Journal of Consulting and Clinical The interpersonal dimension of personality. Journal of Personality, Psychology, 64, 1203-1212. 20, 143-161. Benjamin, L. S., & Wonderlich, S. A. (1994). Social perceptions and Gifford, R. (1991). Mapping nonverbal behavior on the Interpersonal borderline : The relation to mood disorders. Jour- Circle. Journal of Personality and Social Psychology, 61, 279-288. nal of Abnormal Psychology, 103, 610-624. Gifford, R., & O'Connor, B. (1987). The interpersonal circumplex as Bentler, P.M. (1990). Comparative fit indexes in structural models. a behavioral map. Journal of Personality and Social Psychology, 52, Psychological Bulletin, 107, 238-246. 1019-1026. Blackburn, R., & Renwick, S. S. (1996). Rating scales for measuring Guisinger, S., & Blatt, S. J. (1994). Individuality and relatedness: Evolu- the Interpersonal Circle in forensic psychiatric patients. Psychological tion of a fundamental dialectic. American , 49, 104-111. Assessment, 8, 76-84. Gurtman, M. B. (1994). The circumplex as a tool for studying normal Blatt, S. J. (1990). Interpersonal relatedness and self-definition: Two and abnormal personality: A methodological primer. In S. Strack & 1644 PINCUS, GURTMAN, AND RUIZ

M. Lorr (Eds.), Differentiating normal and abnormal personality (pp. inventory of impact messages: An operational bridge between emotion 243-263). New York: Springer. and interpersonal behavior. In R. Plutchik & H. R. Conte (Eds.), Guttman, L. (1954). A new approach to factor analysis: The radex. In Circumplex models of personality and emotion (pp. 221 -244). Wash- P. Lazarsfeld (Ed.), Mathematical thinking in the social sciences (pp. ington, DC: American Psychological Association. 258-348). Glencoe, IL: Free Press. Kiesler, D. J., & Watkins, L. M. (1989). Interpersonal complementarity Henry, W. P. (1992, June). Categorical patterns of introject change and the therapeutic alliance: A study of relationship in psychotherapy. during short-term dynamic psychotherapy. Paper presented at the Psychotherapy, 26, 183-194. meeting of the Society for Psychotherapy Research, Berkeley, LaForge, R., & Suczek, R. F. (1955). The interpersonal dimensions of California. personality: III. An interpersonal check list. Journal of Personality, Henry, W. P. (1994). Differentiating normal and abnormal personality: 24, 94-112. An interpersonal approach using the structural analysis of social be- Leary, T. (1957). Interpersonal diagnosis of personality. New York: havior. In S. Strack & M. Lorr (Eds.), Differentiating normal and Ronald. abnormal personality (pp. 316-340). New York: Springer. Lichtenstein, J., & Cassidy, J. (1991, March). The Inventory of Adult Henry, W. P. (1996). Structural Analysis of Social Behavior as a com- Attachment: Validation of a new measure. Paper presented at the mon metric for programmatic psychopathology and psychotherapy meeting of the Society for Research in Child Development, Seattle, research. Journal of Consulting and Clinical Psychology, 64, 1263- WA. 1275. Loehlin, J. C. (1992). Latent variable models (2nd ed.). Hillsdale, NJ: Henry, W. P., Schacht, T. E., & Strupp, H. H. (1986). Structural Analysis Erlbaum. of Social Behavior: Application to a study of interpersonal process Lorr, M. (1991). A redefinition of dominance. Personality and Individ- in differential psychotherapy outcome. Journal of Consulting and ual Differences, 12, 877-879. Clinical Psychology, 54, 27-31. Mahler, M. S., Pine, F., & Bergman, A. (1975). The psychological birth Henry, W. P., Schacht, T. E., & Strupp, H. H. (1990). Patient and thera- of the infant. New \fork: Basic Books. pist introject, interpersonal process, and differential psychotherapy Maiti, S. S., & Mukherjee, B. N. (1990). A note on the distributional outcome. Journal of Consulting and Clinical Psychology, 58, 768- properties of the Joreskog-Sorbom fit indices. Psychometrika, 55, 774. 721-726. Hogan, R. (1996). A socioanalytic perspective on the five-factor model. McCrae, R. R. (1996). Social consequences of experiential openness. In J. S. Wiggins (Ed.), The five-factor model of personality: Theoreti- Psychological Bulletin, 120, 323-337. cal perspectives (pp. 163-179). New \brk: Guilford Press. McCrae, R. R., & Costa, P. T., Jr. (1989). The structure of interpersonal Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Urefio, G., & Villasenor, traits: Wiggins' circumplex and the five-factor model. Journal of V. S. (1988). The Inventory of Interpersonal Problems: Psychometric Personality and Social Psychology, 56, 586—595. properties and clinical applications. Journal of Consulting and Clini- McCrae, R. R., & Costa, P. T, Jr. (1997). Personality trait structure as cal Psychology, 56, 885-892. a human universal. American Psychologist, 52, 509-516. Horowitz, L. M., Rosenberg, S. E., & Bartholomew, K. (1993). Interper- Mead, G. H. (1932). Mind, self, and society. Chicago: University of sonal problems, attachment styles, and outcome in brief dynamic psy- Chicago Press. chotherapy. Journal of Consulting and Clinical Psychology, 61, 549- Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational measure- 560. ment (3rd ed., pp. 13-103). New \brk: Macmillan. Horowitz, L. M., Rosenberg, S. E., & Kalehzan, B. M. (1992). The Millon, T. (1987). Manual for the MCMI-1I (2nd ed.). Minneapolis, capacity to describe other people clearly: A predictor of interpersonal MN: National Computer Systems. problems and outcome in brief dynamic psychotherapy. Psychotherapy Paddock, J. R., & Nowicki, S. (1986). The circumplexity of Leary's Research, 2, 37-51. interpersonal circle: A multidimensional scaling perspective. Journal Jackson, D. J. (1974). Personality Research Form manual. Goshen, NY: of Personality Assessment, 50, 279-289. Research Press. Paivio, S. C, & Greenberg, L. S. (1995). Resolving "unfinished busi- Joreskog, K. G., & Sorbom, D. (1986). Analysis of linear structural ness": Efficacy of experiential therapy using empty-chair dialogue. relationships by the method of maximum likelihood. Mooresville, IN: Journal of Consulting and Clinical Psychology, 63, 419-425. Scientific Software. Parker, G., Tupling, H., & Brown, L. B. (1979). A parental bonding Kiesler, D. J. (1983). The 1982 Interpersonal Circle: A taxonomy for instrument. British Journal of Medical Psychology, 52, 1-10. complementarity in human transactions. Psychological Review, 90, Pincus, A. L. (1994). The interpersonal circumplex and the interpersonal 185-214. theory: Perspectives on personality and its pathology. In S. Strack & Kiesler, D. J. (1985). The 1982 Interpersonal Circle: Acts version. Un- M. Lorr (Eds.), Differentiating normal and abnormal personality (pp. published manuscript, Virginia Commonwealth University, Richmond. 114-136). New York: Springer. Kiesler, D. J. (1987). Check list of psychotherapy transactions-revised Pincus, A. L., & Gurtman, M. B. (1995). The three faces of interper- and check list of interpersonal transactions-revised. Unpublished sonal dependency: Structural analyses of self-report dependency mea- manuscript, Virginia Commonwealth University. sures. Journal of Personality and Social Psychology, 69, 744-758. Kiesler, D. J. (1988). Therapeutic metacommunication. Palo Alto, CA: Pincus, A. L., Newes, S. L., Dickinson, K. A., & Ruiz, M. A. (1998). Consulting Psychologists Press. A comparison of three indices to assess the dimensions of SASB. Kiesler, D. J. (1991). Interpersonal methods of assessment and diagno- Journal of Personality Assessment, 70, 145-170. sis. In C. R. Snyder & D. R. Fbrsythe (Eds.), Handbook of social and Pincus, A. L., & Ruiz, M. A. (1997). Parental representations and di- clinical psychology (pp. 438-468). New \ork: Pergamon Press. mensions of personality: Empirical relations and assessment implica- Kiesler, D. J. (1996). Contemporary interpersonal theory and research. tions. Journal of Personality Assessment, 68, 436-454. New York: Wiley. Pincus, A. L., & Wiggins, J. S. (1992). An expanded perspective on Kiesler, D. J., & Goldston, C. S. (1988). Client-therapist complemen- interpersonal assessment. Journal of Counseling and Development, tarity: An analysis of the Gloria Films. Journal of Counseling Psychol- 71, 91-94. ogy, 35, 127-133. Plutchik, R., & Conte, H. R. (1997). Circumplex models of personality Kiesler, D. J., Schmidt, J. A., & Wagner, C. C. (1997). A circumplex and emotion. Washington, DC: American Psychological Association. STRUCTURE OF SASB 1645

Quintana, S. M., & Meara, N. M. (1990). Intemalization of therapeutic (1994). Not a complete bridge, but several solid spans. Psychological relationships in short-term psychotherapy. Journal of Counseling Psy- Inquiry, 5, 326-329. chology, 37, 123-130. Wagner, C. C, Kiesler, D. J., & Schmidt, J. A. (1995). Assessing the Russell, J. A. (1980). A circumplex model of affect. Journal of Person- interpersonal transaction cycle: Convergence of action and reaction ality and Social Psychology, 39, 1161-1178. interpersonal circumplex measures. Journal of Personality and Social Schacht, T. E. (1994). SASB and clinical psychology: Further hypothe- Psychology, 69, 938-949. ses. Psychological Inquiry, 5, 324-326. Wiggins, J. S. (1979). A psychological taxonomy of trait-descriptive Schaefer, E. S. (1959). A circumplex model for maternal behavior. Jour- terms: The interpersonal domain. Journal of Personality and Social nal of Abnormal and Social Psychology, 59, 226-235. Psychology, 37, 395-412. Schaefer, E. S. (1961). Converging conceptual models for maternal be- Wiggins, J. S. (1982). Circumplex models of interpersonal behavior in havior and for child behavior. In J. C. Glidwell (Ed.), Parental atti- clinical psychology. In P. Kendall & J. Butcher (Eds.), Handbook of tudes and child behavior (pp. 124-146). Springfield, IL: Charles C. research methods in clinical psychology (pp. 183—221). New %rk: Thomas. Wiley. Schaefer, E. S. (1997). Integration of configurational and factorial mod- Wiggins, J. S. (1991a). Agency and communion as conceptual coordi- els for family relationships and child behavior. In R. Plutchik & H. R. nates for the understanding and measurement of interpersonal behavior. Conte (Eds.), Circumplex models of personality and emotion (pp. In W. Grove & D. Cicchetti (Eds.), Thinking clearly about psychol- 133-153). Washington, DC: American Psychological Association. ogy: Essays in honor of Paul E. Meehl (Vol. 2, pp. 89-113). Minne- Schaefer, E. S., & Burnett, C. K. (1987). Stability and predictability of apolis, MN: University of Minnesota Press. women's marital relationships and demoralization. Journal of Person- Wiggins, J. S. (1991b). The Interpersonal Adjective Scales (IAS) pro- ality and Social Psychology, 53, 1129-1136. fessional manual. Odessa, FL: Psychological Assessment Resources. Schaefer, E. S., Edgerton, M., & Burnett, C. K. (1992). Revised marital Wiggins, J. S. (1994). Shoring up the SASB bridge between personality relatedness and autonomy inventory. Unpublished manuscript, Univer- theory and clinical psychology. Psychological Inquiry, 5, 333-335. sity of North Carolina, Chapel Hill. Wiggins, J. S. (1996a). The five-factor model of personality: Theoretical Siegelman, M., & Roe, A. (1979). Manual: The Parent-Child Relations perspectives. New "York: Guilford Press. Questionnaire II. New York: Author. Wiggins, J. S. (1996b). An informal history of the interpersonal circum- Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New plex tradition. Journal of Personality Assessment, 66, 217-233. "Vbrk: Norton. Wiggins, J. S., & Broughton, R. (1985). The interpersonal circle: A Tracey, T. J. G., Gurtman, M. B., & Rounds, J. (1996). Examination of structural model for the integration of personality research. In R. the general factor with the interpersonal circumplex structure: Appli- Hogan & W. Jones (Eds.), Perspectives in personality (Vol. 1, pp. cation to the Inventory of Interpersonal Problems. Multivariate Behav- 1-47). Greenwich, CT: JAI Press. ioral Research, 31, 441-466. Wiggins, J. S., & Broughton, R. (1991). A geometric taxonomy of Tracey, T. J. G., & Rounds, J. (1995). The arbitrary nature of Holland's personality scales. European Journal of Personality, 5, 343-365. RIASEC types: A concentric-circles structure. Journal of Counseling Wiggins, J. S., Phillips, N., & Trapnell, P. D. (1989). Circular reasoning Psychology, 42, 431-439. about interpersonal behavior: Evidence concerning some untested as- Tracey, T. J. G., & Rounds, J. (1997). Circular structure of vocational sumptions underlying diagnostic classification. Journal of Personality interests. In R. Plutchik & H. R. Conte (Eds.), Circumplex models of and Social Psychology, 56, 296-305. personality and emotion (pp. 183-201). Washington, DC: American Wiggins, J. S., Steiger, J. H., & Gaelick, L. (1981). Evaluating circum- Psychological Association. plexity in personality data. Multivariate Behavioral Research, 16, Tracey, T. J. G., & Schneider, P. L. (1995). An evaluation of the circular 263-289. structure of the Checklist of Interpersonal Transactions and the Check- Wiggins, J. S., Trapnell, P. D., & Phillips, N. (1988). Psychometric and list of Psychotherapy Transactions. Journal of , geometric characteristics of the revised Interpersonal Adjective Scales 42, 496-507. (IAS-R). Multivariate Behavioral Research, 23, 17-30. Trapnell, P. D., & Wiggins, J. S. (1990). Extension of the Interpersonal Wilkinson, L. (1990). SYSTAT: The system for statistics. Evanston, IL: Adjective Scales to include the Big Five dimensions of personality. SYSTAT. Journal of Personality and Social Psychology, 59, 781-790. Upton, G. J. G., & Fingleton, B. (1989). Spatial data analysis by exam- Received October 10, 1996 ple: Vol. 2. Categorical and directional data. New "fork: Wiley. Revision received July 28, 1997 Van Denburg, T. E, Kiesler, D. J., Wagner, C. C, & Schmidt, J. A. Accepted July 29, 1997 •

View publication stats