Copyright 2000 by the American Psychological Association, Inc. 0022-3514/00/$5.00 DOI: 10.1O37//0O22-3514.78.4791
Journal of Personality and Social Psychology 2000, Vol. 78, No. 4, 791-808
The Relational-Interdependent Self-Construal and Relationships Susan E. Cross, Pamela L. Bacon, and Michael L. Morris Iowa State University Three studies describe the development and validation of a measure of the relational-interdependent self-construal, which is defined as the tendency to think of oneself in terms of relationships with close others. Study 1 reports the development, psychometric properties, and tests of validity of this new measure. Individuals who scored high on the Relational-Interdependent Self-Construal (RISC) Scale characterized their important relationships as closer and more committed than did individuals who scored low on diis measure (Study 1) and were more likely to take into account the needs and wishes of others when making decisions (Study 2). In Study 3, using a dyadic interaction paradigm with previously unacquainted participants, the partners of persons who scored high on the RISC scale viewed them as open and responsive to their needs and concerns; these perceptions were related to positive evaluations of the relationship.
Cross-cultural psychologists have recently identified a basic dimension that differentiates how people think about themselves— the degree to which elements of the social world (such as close relationships, contexts for behavior, or important group memberships) are included in the self (Markus & Kitayama, 1991; Shweder & Bourne, 1984; Triandis, 1989). They argue that members of collectivist cultures, such as Japan or India, tend to think of themselves as interdependent with close others and as defined by important roles and situations. In contrast, members of Western cultures, such as the United States, tend to think of themselves as independent of relationships and as autonomous or separated from others.
(we explain this further below). Although this difference in selfstructure has been identified by other researchers (e.g., Lykes, 1985; Surrey, 1991), a measure of the interdependent selfconstrual appropriate for Western populations has not been available for interested researchers. In this article we describe the development and validation of a new measure of the interdependent self-construal. Individual Differences in the Self-Construal
There may be considerable variation within North American culture with respect to the self-construal, however. Members of many ethnic and religious groups tend to think of themselves as interdependent or relational (Allen, Dawson, & Brown, 1989; Marin & Triandis, 1985; McCombs, 1985; Oved, 1988). Women are more likely than men in American society to construct an interdependent or relational self-view (Cross & Madson, 1997; Markus & Oyserman, 1989). The nature of this Western version of the interdependent self-view is likely to be somewhat different from the group-oriented interdependence that characterizes the self-views of members of collectivist societies; North Americans and Western Europeans are more likely to include representations of close relationships rather than group membership into the self
For years, Western researchers have assumed a single model of the self—what has been termed the independent self-construal (Markus & Kitayama, 1991). This understanding of the self is based on Euro-American beliefs about individualism, personal rights, and the autonomy of the individual from social groups (Bellah, Madsen, Sullivan, Swidler, & Tipton, 1985; Guisinger & Blatt, 1994; Markus & Kitayama, 1994; Sampson, 1985). The underlying principle that shapes the independent self-construal is the premise that the person is essentially separate from others. The primary components of the independent self-construal are one's unique traits, abilities, preferences, interests, goals, and experiences, and these are differentiated from social contexts, interpersonal relationships, and group memberships. To maintain and enhance this independent view of the self one must maintain a sense of autonomy from others and "be true to one's own internal structures of preferences, rights, convictions, and goals" (Markus & Kitayama, 1994, p. 459).
Susan E. Cross, Pamela L. Bacon, and Michael L. Morris, Department of Psychology, Iowa State University. This research was supported by National Institute of Mental Health Grant MH 54540-01. We wish to thank Dan Russell for his help with the analyses and Dan Russell and Carolyn Cutrona for their helpful comments on the manuscript. We also wish to thank Angela Anderson, Paige Boland, Summer Brunscheen, Frindee Daly, John Egan, Angela Fontanini, Amy McGregor, Alyssa Puffer, Carrie Rosentrater, Becky Schmidt, Kelly Southwell, and Laura Wright for their help with the data collection. Correspondence concerning this article should be addressed to Susan E. Cross, Department of Psychology, W112 Lagomarcino Hall, Iowa State University, Ames, Iowa 50011. Electronic mail may be sent to scross® iastate.edu.
In contrast, the underlying principle that shapes the interdependent self-construal is the premise that the person is connected to others, so that the self is defined, at least in part, by important roles, group memberships, or relationships. For individuals with this self-construal, representations of important relationships and roles share the self-space with abstract traits, abilities, and preferences. To maintain and enhance this interdependent view of the self, individuals will tend to think and behave in ways that emphasize their connectedness to others and that strengthen existing relationships. These differences in self-construals raise many issues with regard to self-related processes. For example, cognitive processes 791
792
CROSS, BACON, AND MORRIS
may vary depending on one's self-construal. Specifically, information for close others that is closely linked to information about the self may function much like self-relevant information in cognitive processes. As a result, individuals with an interdependent self-construal may attend more closely to information about close others than will individuals with an independent self-construal. Similarly, individuals with an interdependent self-construal may remember information about close others better than individuals with an independent self-construal. Motivational processes may also vary for individuals with divergent self-construals. It is generally assumed that the desire to promote or enhance the self is a central motivator of behavior. However, self-promotion or self-enhancement will depend on the nature of the self-construal. For individuals with an independent self-construal, standing out, being better than others on selfdefining domains, and maintaining a positive ratio of successesto-aspirations contribute to a positive view of the self and to enhanced self-esteem (Blaine & Crocker, 1993; Harter, 1993; James, 1890/1983; Tesser, 1988). In contrast, for the person with an interdependent self-construal, positive feelings about the self will in some part derive from developing and maintaining close relationships with others. Indeed, a variety of evidence suggests that positive feedback from others and harmonious relationships are more important for women's self-esteem than for men's (e.g., Moran & Eckenrode, 1991; Roberts & Nolen-Hoeksema, 1989, 1994; Shwalbe & Staples, 1991; Zuckerman, 1989). Individuals with interdependent self-construals may also approach relationships differently than those with independent selfconstruals. If maintaining close relationships is central to maintaining a stable sense of self and self-esteem for people with interdependent self-construals, then these individuals should strive to develop skills and abilities that foster close relationships. For example, these persons may be more likely to self-disclose on intimate topics or to consider the implications of their decisions for close others, relative to individuals with an independent selfconstrual. In addition, the needs and wishes of close others may strongly influence the thoughts and behaviors of these individuals. These and other implications of differences in self-construals for cognition, emotion, motivation, and social relationships are reviewed by Cross and Madson (1997). Two Forms of Interdependence Due to cultural differences between collectivist and individualist cultures, the specific form of the interdependent self-construal developed by members of these different types of cultures will vary. In collectivism-based interdependence, the individual's position in the group or situation dictates behavior; therefore, knowing one's place, behaving according to one's role, and putting the needs of the group before one's own needs are central dictums that shape the self-construal. Measures designed to tap this selfconstrual, such as Yamaguchi's Collectivism scale (Yamaguchi, 1994; Yamaguchi, Kuhlman, & Sugimori, 1995), include items such as, "I am prepared to do things for my group at any time, even though I have to sacrifice my own interest" and "I respect decisions made by my group" (see also Matsumoto, Weissman, Preston, Brown, & Kupperbusch, 1997). This group-oriented notion of interdependence, however, does not adequately describe the relationship-centered conception of interdependence that charac-
terizes North Americans. Group memberships are relatively unimportant to U.S. adults as compared with members of East Asian cultures (Triandis, 1989). Americans treat group memberships rather casually and have relatively little loyalty to in-groups. American in-groups are larger, place relatively fewer demands on members, and are more voluntary than in collectivist cultures (Triandis, 1989). Rather than in-groups, Americans are more likely to include individual relationships (e.g., a spouse, mother, best friend, or colleague) in their self-representations.1 Indeed, Kashima et al. (1995) contrasted collectivist-oriented self-construals with the concept of relatedness to argue that the two concepts are not the same. They created a 4-item measure of relatedness that focused on the emotional relatedness of the self with other individuals (a sample item is, "I feel like doing something for people in trouble because I can almost feel their pains."). Comparing samples from Australia, the United States, Japan, and Korea, they found that individualism, collectivism, and relatedness were empirically separable. Cultures were most likely to be differentiated on the individualism/collectivism dimension, but men and women were best differentiated on the relatedness dimension. It is necessary, therefore, to differentiate between the collectivistoriented interdependence that characterizes members of cultures like Japan and China from the relational-interdependence that characterizes many North Americans. In addition, some measures of interdependence that have been based on East Asian collectivist ideology are multi-dimensional.2 For example, Kashima et al. (1995) report that a factor analysis of a collectivism scale revealed three factors: a factor they called "collectivism," which can be described as the willingness to sacrifice self-interest for the sake of the group; "agency," which is characterized by items indicating a willingness to be independent of the group; and "assertiveness," which is the willingness to speak up in opposition to the group. Other scales, such as the Singelis (1994) interdependent self-construal measure, include additional dimensions of collectivism, such as respect for authority figures, that are conceptually independent of the concept of the self as connected or related to specific others (see also Realo, Allik, & Vadi, 1997). A measure that taps a relational form of interdependence is needed to further explore the variation of the selfconstrual and its function among Western populations. Currently, no measure of relational interdependence operationalizes this construct in terms of the self-concept.3 Other measures, such as the one used by Kashima et al. (1995), tap emotional consequences of investment in close relationships. Although this emotional sensitivity to others may be a characteristic of individuals with a relational-interdependent self-construal, these items do not directly assess the extent to which others, especially close relationships, are a part of one's personal identity or self-concept. Similarly, the Communal Orientation Scale (Clark, Ouellette, Powell, & Milberg, 1987) focuses on the individual's desire for 1 Brewer & Gardiner (1996), however, argued that a collectivist or group-oriented interdependence is characteristic of men in American culture. 2 See Triandis (1995) for an overview of the development and use of various measures of individualism and collectivism. 3 We do know of one related yet unpublished measure by Rude, Welch, and Sandere (1998).
RELATIONAL-INTERDEPENDENT SELF-CON STRUAL reciprocity and expectations of equity in close relationships, but does not assess the person's self-views as relational or connected to others. If the relational-interdependent self-construal is conceptualized in terms of the self-concept, then an adequate measure must include this cognitive component of thinking of oneself in terms of close others or identifying with close others. In a related line of research, Aron and his colleagues have argued that individuals often include representations of a particularly close relationship into their mental representations of themselves (Aron, Aron, & Smo 11 an. 1992; Aron, Aron, Tudor, & Nelson, 1991). In their work, individuals who include a specific partner in the self are more likely to view themselves as sharing perspectives, resources, and characteristics of the other (Aron et al., 1991). We agree that most people will include specific relationships (e.g., one's mother or spouse) as part of the self, but individuals differ markedly in the degree to which other relationships are self-defining—such as friendships or relationships with siblings. The Aron et al. (1991) measure, the Inclusion of Others in the Self Scale, focuses on a single relationship, whereas our emphasis is on interdependence as a general orientation toward representing oneself in terms of close relationships. In addition, our goal is a better understanding of the structure and function of the self in contrast to Aron's focus on understanding the consequences of close relationships. Although our discussion thus far has represented relationalinterdependent and independent self-construals as two types, we recognize that these constructs represent two continuous dimensions. Our assumption of two dimensions (rather than one bipolar dimension) is based on previous research with other measures, such as the Singelis (1994) measures of the interdependent and independent self-construals. Singelis reports that these two scales are orthogonal. Other studies also find a similar lack of correlation between measures of these constructs (Cross, 1995). The primary focus of the research reported here was the development of a measure of the relational-interdependent self-construal; therefore, we focus on people who think of themselves in very relational or interdependent terms compared to those who do not. We do not assume that persons who do not think of themselves as relational or interdependent have necessarily constructed very independent self-views. Additional theorizing and research is necessary to spell out the implications of these two dimensions for self-related processes. Research Aims The first step in examining the role of the relationalinterdependent self-construal in psychological phenomena is to devise a reliable and valid measurement tool. In much earlier work, interdependence has not been measured directly; instead, gender or cultural background have been used as proxies for the interdependent self-construal. More careful tests of the hypotheses related to individual differences in self-construal can be conducted if this construct can be reliably and validly measured. Our first goal was to develop a new measure that directly conceptualizes the relational-interdependent self-construal in terms of the degree to which individuals include close relationships in their self-concepts. In Study 1 we describe the development of this measure, its psychometric properties, and its relations to other personality constructs.
793
Our second goal was to assess the relation between this operationalization of the interdependent self-construal and individuals' experiences with relationships. If the self is defined, at least in part, in terms of one's close relationships, then the individual should be motivated to develop and nurture close relationships. Consequently, individuals who view themselves as interdependent with others may be more likely to self-disclose and to permit close others to influence how they think or behave. They are expected to be more likely to include close others in the self, in Aron's terms, and to be strongly committed to important relationships. In Studies 1, 2, and 3 we examine these hypothesized associations between interdependence and orientations toward relationships. Study 1: Scale Development and Validation In Study 1 we report the development and psychometric properties of the Relational-Interdependent Self-Construal Scale (RISC). Eight samples of students were used to examine the properties of the measure, its reliability, and its convergent and discriminant validity. The convergent validity of the RelationalInterdependent Self-Construal scale was examined by assessing its association with measures related to thinking of oneself as interdependent with others, such as the Communal Orientation Scale (Clark et al., 1987) and the Interdependent Self-Construal measure (Singelis, 1994). Because other research indicates that measures of the independent self-construal and collectivism are not correlated, we expected the RISC scale to be unrelated to measures of independence or individualism. Women are more likely to include close relationships in their self-concepts than are men, so we examined its association with a measure of traditional sex roles, the Spence, Helmreich, and Stapp (1974) Personal Attributes Questionnaire (PAQ). The PAQ taps self-views related to feminine sex roles (termed "expressivity") and masculine sex roles (termed "instrumentality"). According to Helgeson (1994), the expressivity subscale is a useful measure of a communal orientation, whereas the instrumentality subscale can be viewed as a measure of the independent self-construal. Thus we expected the RISC scale to be positively correlated with the expressivity scale and uncorrelated with the instrumentality scale. We also examined the relations between the RISC scale and several personality measures. Theorists have argued that a critical mark of the relational-interdependent self-construal is empathy for others (Surry, 1991), so we examined the relations between the RISC scale and measures of empathy and perspective taking. The tendency to think of oneself in terms of close relationships may be a consequence of underlying personality structure; therefore, measures of the five-factor model of personality were included in this research (Costa & McCrae, 1992). We expected the RISC scale to be moderately related to the interpersonal traits, agreeableness and extroversion, but to be unrelated to neuroticism and openness to experience. (We had no hypothesis concerning conscientiousness.) We also examined the association of the RISC scale with the tendency toward responding in a socially desirable fashion. Some researchers have argued that very relational or interdependent persons may be vulnerable to depression or to reduced well-being due to their sensitivity to the distress of others (e.g., Belle, 1982; McGrath, Keita, Strickland, & Russo, 1990). Therefore, several measures of well-being, such as depression, selfesteem, and life satisfaction, were included in this research.
794
CROSS, BACON, AND MORRIS Table 1
Descriptive Statistics and Gender Differences for the Relational Interdependent Self-Construal Scale Men
Total
Difference
Women
Sample
M
SD
N
M
SD
n
U
SD
n
1 2 3 4 5 6 7
54.10 56.12 54.91 54.01 56.84 54.12 58.10 56.08
9.29 10.25 10.39 10.11 ,9.27 9.92 10.68 9,58
271 299 940 609 735 535 621 273
52.89 54.27 53.90 52.20 54.27 50.85 54.73 54.48
8.07 9.56 10.25 10.03 9.17 10.34 10.17 9.38
111 128 406 253 341 221 248 111
55.11 57.44 55.66 55.72 59.13 56.48 60.49 57.78
10.03 10.56 L0.39 9.93 8.76 8.87 10.47 9.50
152 165 525 315 383 285 362 143
8
d -1.92* -2.65** -2.59** -4.19*** —7.29*** -6.59*** -6.76*** -2.76**
-.24 .32 -.17 -.35
-.52 -.57 -.56 -.35
Note. All t tests were one-tailed. Because of participants who failed to indicate their gender, the total number of participants may be greater than the number of women and men listed. *p < .05. **p < .01. ***p < .001.
Finally, we used a version of the known groups validation technique (i.e., examining gender differences) to assess criterion validity. We followed this with regression analyses aimed at examining the unique predictive power of scores on the RISC scale. We began to examine the relations between scores on the RISC scale and individuals' descriptions of their close relationships in this study. When relationships are self-defining, the individual should desire to develop and maintain close and supportive ties with others. Thus, it is likely that they will develop a greater number of important relationships than will individuals who have not constructed an interdependent self-construal. Individuals with an interdependent view of themselves may also describe their important relationships as closer than do others. Measuring closeness, however, is not entirely straightforward. Some researchers have defined closeness in terms of the impact of a relationship on the person's life. For example, Berscheid, Snyder, and Omoto (1989) draw upon Kelley et al.'s (1983) definition of a close relationship as one in which the partners are interdependent. In their view, the three most important aspects of a close or interdependent relationship are frequency of contact, involvement in a diverse array of activities together, and strong impact on the individual's decisions, plans, and behaviors. Other researchers have focused on intimacy of discussion and self-revelation as central to a close relationship (Reis & Shaver, 1988). For example, Airman and Taylor's (1973) theory of social penetration posits that close and rewarding relationships are characterized by extensive and intimate self-disclosure. A newer conceptualization of closeness has been suggested by Aron and his colleagues, who argue that closeness can be understood as "including the other in the self (Aron et al., 1991). As described earlier, Aron and his colleagues have investigated the cognitive consequences of incorporating specific close relationships into the self, and they have developed a simple, Venndiagram-like measure in which the self and other are represented as overlapping circles (Aron et al., 1991). We included measures of each of these representations of closeness in this study in order to assess the concept broadly. In addition, if a relationship is self-defining, the individual should desire to protect and maintain it. The loss of a close relationship may be especially threatening or difficult for the
person who has staked some of his or her identity in it. Thus we expected individuals who scored high on the RISC scale to be more committed to their close relationships than low scorers. In summary, we hypothesized that individuals who scored high on the RISC scale would view more relationships as important to them than would others. We also hypothesized that they would seek out frequent and diverse contact with close relationship partners, would be more likely to permit those relationships to influence their lives, and would be more likely to open up and reveal personal or intimate information about themselves in order to promote the relationship. We also expected that these individuals with a very interdependent self-construal would tend to incorporate their close relationships into the serf and would be more committed to their close relationships than other individuals. One consequence of this commitment to relationships may be that individuals who think of themselves as interdependent with others may also perceive higher levels of social support than do others.
Method Participants Eight samples of undergraduate students from introductory psychology courses completed the RISC scale along with a packet of questionnaires, as part of large data collection efforts. Students received extra course credit for their participation. Due to time constraints, participants in each sample completed a subset of the measures described below. Sample sizes ranged from 267 to 956; the sample sizes and numbers of men and women in each sample are presented in Table 1. To reduce the possible influence of cultural differences in self-construal, non-American participants and participants who did not indicate their citizenship were excluded from the analyses (except for Samples 5, 6, and 7, where citizenship information was not obtained). This resulted in the exclusion of 22 to 40 participants per sample.4
4
A comparison of the American students' scores (n = 2,374) on the RISC scale with the noncitizens' scores (n = 109) across all the samples revealed that the Americans tended to score higher than the noncitizens, AfA^oan, = 54.89, SD = 10.11, M oonddzcn9 = 50.85, SD = 9.52, f(2481) = 4 . 0 8 , p < .001.
795
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL
Measures For all the measures described below (except where noted), participants responded using a 7-point Likert-type scale. Indexes were created so that high scores indicate high levels of the construct (e.g., high scores on the RISC scale indicate higher levels of the interdependent self-construal). Relational-Interdependent Self-Construal Scale (RISC). Many of the initial items were generated by the authors, based on the conceptualization of the relational form of the interdependent self-construal as including close others in the self-concept. Additional items were created by modifying items from measures of conceptually related constructs, such as cross-cultural measures of the interdependent self-construal. The modification typically included changing the focus of the question from one's relationship with people in general or a group to a close friend or family members. During the development phase, some items were changed, others were dropped, and new items were constructed. Using data from the first sample, we examined the individual questions to identify items that (a) had low item-total correlations, (b) were strongly associated with social desirability, or (c) were worded poorly. When an item consistently violated these criteria, it was eliminated. Using this system, we narrowed the number of items from 28 to 11 (see Table 2). Of these 11 items, 2 were phrased negatively and are reverse-scored in the data analyses. When completing the RISC scale, participants were instructed to "indicate the extent to which you agree or disagree with each of these statements." Participants responded using a Likert-type scale, ranging from 1 (strongly disagree) to 7 (strongly agree). The possible range of scores was from 11 to 77. Measures related to interdependence and independence. The Communal Orientation Scale (Clark et al., 1987) consists of 14 items; individuals high in communal orientation feel responsible for the welfare of close others and are responsive to their needs. An example of an item from the Communal Orientation scale is, "I believe people should go out of their way to be helpful." The reliability of this scale was .78 (Clark et al., 1987). Singelis (1994) developed orthogonal, 12-item measures of the interdependent and independent self-construals based on. the Markus and Kitayama (1991) conceptualization of cultural differences in the self. Examples of these items are, "It is important for me to maintain harmony within my group" (Interdependent Self-Construal subscale) and "I prefer to be direct and forthright when dealing with people I've just met" (Indepen-
dent Self-Construal subscale). The reh'abilities of the Interdependent and Independent subscales were .73 and .69, respectively (Singelis, 1994). The Collective Self-Esteem Scale (CSB; Luhtanen & Crocker, 1992) is a 16-item measure that assesses collective self-esteem along four dimensions: Private Collective Self-Esteem, Public Collective Self-Esteem, Importance to Identity, and Membership Esteem (each subscale is composed of 4 items). The general positivity of one's social identity is assessed by totaling all 16 items. Examples of items from the CSE are, "I feel good about the social groups I belong to" (Private subscale; Cronbach's a = .74); "In general, others respect the social groups that I am a member o f (Public subscale; a = .80); "The social groups I belong to are an important reflection of who I am" (Importance to Identity subscale; a = 76); and "I am a worthy member of the social groups I belong to" (Membership subscale; a = .73). Luhtanen and Crocker (1992) also report that the reliability for the total CSE scale was .85. The Personal Attributes Questionnaire (PAQ; Spence et al., 1974) is composed of eight traits related to instrumentality (or a communal orientation) and eight traits related to expressivity (or an independence orientation), which are thought to be independent constructs. The items are presented in a 5-point bipolar format (e.g., 1 = not at all aggressive; 5 = very aggressive). Helmreich, Spence, and Wilhelm (1981) report Cronbach alphas ranging from .67 to .77 for the Instrumentality scale and .72 to .80 for die Expressivity scale. Two additional items were included to assess the validity of the RISC scale. First, participants were asked, "When you think about yourself, how important are your relationships with others for your self-concept?" Second, participants were asked, "Some people tend to see themselves as very independent and separate from others, whereas others are more likely to think of themselves as interdependent and connected in important ways to their close friends. Which type of person do you think you are?" Participants responded using a 5-point Likert-type scale that ranged from 1 (very independent) to 5 (very interdependent). These questions allowed us to assess the associations between the RISC scale and these alternative approaches to tapping the relational-interdependent self-construal. Personality measures. The 7-item Empathic Concern subscale of the Interpersonal Reactivity Index (Davis, 19S0) assesses feelings of concern and sympathy for others, whereas the 7-item Perspective Taking subscale
Table 2 Final Relational-Interdependent Self-Construal Scale Items With Corrected Item-Total Correlations Item 1. My close relationships are an important reflection of who I am. 2. When I feel very close to someone, it often feels to me like that person is an important part of who I am. 3. I usually feel a strong sense of pride when someone close to me has an important accomplishment. 4. I think one of the most important parts of who I am can be captured by looking at my close friends and understanding who they are. 5. When I think of myself, I often think of my close friends or family also. 6. If a person hurts someone close to me, I feel personally hurt as well. 7. In general, my close relationships are an important part of my self-image. 8. Overall, my close relationships have very little to do with how I feel about myself.3 9. My close relationships are unimportant to my sense of what kind of person I am.a 10. My sense of pride comes from knowing who I have as close friends. 11. When I establish a close friendship with someone, I usually develop a strong sense of identification with that person. Note. N = 4,288. Response scale ranges from 1 (strongly disagree) to 7 (strongly agree). a Reverse-keyed item.
Corrected itemtotal r .68 .69 .54 .64 .63 .53 .69 .54 .52 .56 .60
796
CROSS, BACON, AND MORRIS
measures respondents' ability to see things from another person's viewpoint. Cronbach's alpha ranged from .70 to .78 (Davis, 1980). The NEO-FFI (Costa & McCrae, 1992) was used to assess the dimensions of the five-factor model: Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Costa and McCrae (1992) report Cronbach's alphas of .89 to .95 for the scales, which have been widely validated and have shown good convergent and discriminant validity. Socially desirable responding was assessed using the 33-item MarloweCrowne Social Desirability Scale (Crowne & Marlowe, 1960). Participants responded 'True" or "False" to each item. The Kuder-Richardson 20 (KR-20) reliability of the Social Desirability Scale was .88 (Crowne & Marlowe, 1960). Measures of well-being. The 20-item CES-D depression scale (Radloff, 1977) assesses the frequency of depressive symptoms, such as depressed mood, feelings of guilt and worthlessness, problems sleeping, and loss of appetite. The CES-D correlates strongly with the number of negative life events and other measures of depression, such as the Beck Depression Inventory (Beck, 1967) and the Self-Rating Depression Scale (Zung, 1965; see Shaver & Brennan [1991] for further information on the scale properties). Participants responded using a 4-point scale (1 = rarely or none of the time; 4 = most of the time). Cronbach's alpha for this sample was .91. The Satisfaction With Life Scale (Diener, Emmons, Larsen, & Griffin, 1985) is a 5-item scale that has been used extensively and has good psychometric properties. Participants responded to the items using a 5-point scale (1 = very slightly or not at all; 5 = extremely). Cronbach's alpha in this sample was .84. The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965), is a commonly used 10-item measure of global self-esteem. Reliabilities of the RSES were .77 and .88 (reported in Blascovich & Tomaka, 1991). The Life Orientation Test (LOT; Scheier & Carver, 1985) is an 8-item measure of dispositional optimism. Scheier and Carver (1985) report a Cronbach's alpha of .76 for the LOT. Evaluations of relationships. The social network diagram (Antonucci, 1986) was used to assess the number of important relationships in participants' social networks. This measure consists of four concentric circles, the innermost one marked "YOU." Participants were asked to write in the names of people with whom they had personal relationships according to the following instructions. In the inner circle, they were to write the names of people so close and important that "it is hard to imagine life without them"; in the middle circle, they were to write the names of people who were not quite as close but still important to them; and in the outermost circle, they were to write the names of people who were important enough that they should be placed somewhere in the network. This diagram has been used with a variety of samples that varied across age, class, and culture and has been found to support theory in the social support literature (Antonucci, 1986). Participants selected the most important person in their social network and completed the following measures with that relationship in mind. First, they described their closeness to this person using the Relationship Closeness Inventory (RCI). The RCI (Berscheid et al., 1989) was developed as a multidimensional measure of closeness for specific relationships. The participant was asked to categorize the type of relationship into one of the following categories: work, family, romantic, friend, or other. Thirty-three percent of the most important relationships were categorized as close friends; 32% were romantic partners; 24% were family members; and 11% were categorized in the "other" category. Participants rated the average amount of time they had spent with the target in the mornings, afternoons, and evenings of the past week. Participants were then presented with a list of activities and were asked to check all those that they had initiated with the target in the past week. This was a change from the original RCI, which asks participants to check activities that they had done alone with the target. Both the total amount of time
spent with the target and the activities data were converted to 1 to 10 scales as in Berscheid et al. (1989). Next, participants responded to a series of 27 items concerning the target's influence on the participant's thoughts, feelings, and behavior. Examples of influence items include "X influences important things in my life" and "X influences how I spend my free time." Cronbach's alpha was .89 in this sample. Participants also responded to 7 items concerning the degree to which the target influenced their future goals and plans. Responses were made on a scale ranging from 1 (not at all) to 7 (a great extent), and items included "my marriage plans" and "my plans for achieving a particular standard of living." Cronbach's alpha was .89. The 10-item self-disclosure scale (Miller, Berg, & Archer, 1983) asked participants to indicate the extent to which they talked with this person - about several topics. Examples of items included "My personal habits" and "My deepest feelings." Responses were made on a 4-point scale ranging from 1 (discuss not at all) to 4 {discuss fully and completely). Due to low correlations with other items, the "personal habits" item was dropped. With this 9-item scale, Cronbach's alpha was .91. The Inclusion of Other in the Self scale (IOS) is a single-item pictorial measure of closeness (Aron et al., 1992). Participants selected a picture from a set of Venn-like diagrams that present two circles with varying degrees of overlap; the circles represent the self and the relationship partner. The IOS scale has good test-retest and alternate form reliability, and has been shown to predict how long romantic couples stay together, with no relation to social desirability. Rusbult's (1983) 4-item scale was used to measure commitment. Examples of items included "How likely is it that you will end your relationship with person X in the near future?" and "To what extent are you 'attached' to person X?" Responses were made on a 7-point scale ranging from 1 {not at all likely) to 7 {extremely likely). Due to low correlations with other items, the "end your relationship" item was dropped, leaving a total of 3 items. Cronbach's alpha was .65. The Social Provisions Scale (SPS; Cutrona & Russell, 1987) was used to assess participants' overall level of social support. This 24-item measure taps 6 dimensions of social support identified by Weiss (1974): guidance, reliable alliance, attachment, social integration, reassurance of worth, and opportunity to provide nurturance. Evidence of the scale's validity is provided by Cutrona (1989) and Cutrona and Russell (1987). A 4-point scale was provided for the participants' responses (1 = strongly disagree; 4 = strongly agree). Cronbach's alpha for the combined scale in this sample was .92.
Results Basic Psychometric Properties Factor structure. Principal components analysis was used to conduct an exploratory factor analysis on the pooled data from the eight samples. Negatively phrased items were recoded, resulting in positive factor loadings. Only one factor emerged with an eigenvalue greater than 1.0. All 11 items loaded between .59 and .77 on the first factor. The scree plot for the factor analysis also supported a single factor, with the elbow of the plot occurring after the first factor. The first factor accounted for 47% of the total variance. Descriptive statistics. To create an index, the negative items were reversed and responses to the 11 items were summed. Table 1 shows the mean RISC scale scores across the eight samples. Skewness ranged from —.805 to —.289 across these samples, and kurtosis ranged from —.017 to 1.41.
Reliability Internal consistency. Evaluating the samples individually, the coefficient alphas ranged from .85 to .90, with a mean of .88. The
797
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL
means of the inter-item correlations ranged from .35 to .46. The item-total correlations from the eight samples ranged from .54 to .73. When the samples were combined the coefficient alpha was .88 (N = 4,288), and the inter-item correlations ranged from .25 to .66, with a mean of .41. Table 2 presents the corrected item-total correlations for the pooled sample. Test-retest reliability. Stability over time was assessed by administering the RISC scale to participants in Samples 3 and 5 twice in either a one- or two-month interval. The test-retest reliabilities over the 2-month interval in the two samples were .73, n = 67, p < .001 and .63, n = 317, p < .001. The test-retest reliability over one month was .74, n = 405, p < .001 and .76, n = 46, p < .001. Convergent
Validity
Other measures of collectivism and interdependence. The RISC scale is intended to provide a better measure of the relational-interdependent self-construal for Americans than current measures of collectivism or the interdependent self-construal, which tend to focus on group identity. It should, however, be correlated with these other measures. As shown in Table 3, the RISC scale correlated moderately with the Communal Orientation Scale (r = .41), and Singelis's (1994) group-oriented Interdependent Self-Construal Scale (r = .41). The RISC scale was moderately related to the total Collective Self-Esteem score (r = .37). It was most strongly related to the Identity subscale (r = .47) and the Private subscale (r = .22). Similarly, the correlation of the RISC scale with the PAQ measure of expressivity (which has also been described as a measure of communal orientation) was .32. These moderate correlations suggest that although the RISC is related to other measures of collectivism or communal orientation, it is clearly not identical to those measures. Because individuals with a very interdependent self-construal should find relationships more important to how they think of themselves than do others, we expected a positive correlation between the RISC scale and the item that asked "When you think about yourself, how important are your relationships with others for your self-concept?" The correlation was .56, indicating that the RISC taps this relational dimension of the self-concept. In addition, the correlation between RISC scores and degree of selfreported categorization as independent or interdependent was .31. Measures of independence. The interdependent self-construal and the independent self-construal have been described as two orthogonal constructs (Singelis, 1994); therefore, we predicted that our measure would not be related to a measure of the independent self-construal. Supporting our prediction, the RISC was not related to Singelis's (1994) Independent Self-Construal measure (r = .08). Hotelling tests showed that the relation between the RISC and Singelis's measure of the interdependent self-construal (r ~ .41) was stronger than the relation between the RISC and his measure of the independent self-construal (r - .08), r(601) = 6.67, p = .001. The RISC scale was also unrelated to the instrumentality dimension of the PAQ, which has been described as indicating an orientation toward independence (r = —.06). A Hotelling test showed that the relation between the RISC and the expressivity dimension of the PAQ was stronger than the relation between the RISC and the instrumentality dimension, r(601) = -6.79, p < .001.
Table 3 Correlations of the Relational-Interdependent Self-Construal (RISC) Scale With Other Scales Measure Measures of collectivism and interdependence Communal Orientation Scale Interdependent Self-Construal Scale (Singelis) Collective Self-Esteem Scale Private Public Identity Membership Total Personal Attributes Questionnaire—Expressivity Importance of relationships item Self-categorization item Measures of independence or individualism Independent Self-Construal Scale (Singelis) Personal Attributes Questionnaire—Instrumentality Personality measures Interpersonal Reactivity Index Empathic Concern Perspective Taking NEO-FFI Agreeableness Extraversion Conscientiousness Neuroticism Openness to Experience Marlowe-Crowne Social Desirability Scale Measures of well-being Depression Satisfaction With Life Rosenberg Self-Esteem Scale Life Orientation Test (Optimism) Evaluation of closest relationship RCI time spent together RCI activities RCI influence RCI future plans Self-disclosure 1OS Commitment Social Provisions Scale (Social Support)
Correlation with RISC
N
.41 .41
598 1,219
.22 .20 .47 .16 .37 .32 .56 .31
1,218 1,218 1,217 1,219 1,219 604 1,216 1,212
.08 -.06
604 604
.34 .13
2,726 2,986
.35 .28 .23 .08 .09 -.05
628 628 628 628 628 1,785
.03 .07
235 235 3,555 749
.01 .17 .12
.11 .27 .15 .26 .26 .22 .26
165" 247 247 246 245 243 247 235
Note. Eight samples are represented in this table. None of the measures was administered to every sample, which accounts for the differences in sample size for each correlation. RCI = Relationship Closeness Inventory; IOS - Inclusion of Other in the Self; NEO-FFI = NEO Five-Factor Inventory. a This correlation excludes participants whose closest relationship partner lived in another city.
Personality measures. In order to maintain self-defining relationships, individuals with highly interdependent self-construals should be aware of the feelings and thoughts of close others. Thus, we predicted that the RISC scale would be related to empathic concern and perspective-taking. As predicted, the RISC was positively related to empathic concern (r = .34), but had a weak relation to perspective taking (r = .13). Hotelling tests revealed that the relation between the RISC scale and empathic concern was significantly stronger than the relation with perspective taking, *(2723) = \i.92,p< .001. The RISC scale was also correlated with each of the five subscales of the NEO-FFI measure of the Five Factor Model. As
798
CROSS, BACON, AND MORRIS
expected, the RISC scale was positively correlated with Agreeableness (r = .35) and Extraversion (r = .28). It was also positively correlated with Conscientiousness (f = ,23); but it was not correlated with Neuroticism (r = .08) or Openness to Experience (r = .09). Hotelling tests revealed that the correlations of the RISC scale with agreeableness, extraversion, and conscientiousness were significantly stronger than the relations with neuroticism and openness to experience, rs(625) > 2.44, ps < .02. There was not a significant relationship between the RISC and the Marlowe-Crowne Social Desirability scale (r = —.05). These results suggest that respondents are not attempting to present themselves in socially approved ways when they answer the RISC scale, thus demonstrating discriminant validity. Measures of well-being. The RISC scale was unrelated to the measures of depression (r = .03), life satisfaction (r = .07), and self-esteem (r = .01). There was a weak correlation between the RISC scale and the measure of optimism (the LOT; r = .16). Thus, individuals with a very interdependent self-construal appear no more at risk of reduced well-being than are others. Group differences. The relational-interdependent self-construal is thought to be more characteristic of Western women than of men (Cross & Madson, 1997). In all eight samples, women scored significantly higher than men on the RISC scale (see Table 1). The size of these differences ranged from small (d = —.17) to moderate (d — -.57). In the pooled sample, d = - . 4 1 .
Examination of Incremental Utility Although the zero-order correlations presented in Table 3 provide some evidence that the Relational-Interdependent SelfConstrual Scale is related to other measures as predicted, it is unclear from these analyses the extent to which this measure taps a unique construct. We expected that the RISC scale would account for additional variance in relevant dependent variables beyond that accounted for by related constructs, such as expressivity, empathic concern, and a measure of the group-oriented interdependent self-construal. In these analyses, our criterion variables were the relatively global measures of collective self-esteem and communal orientation. We selected these measures because they tap constructs that are fairly broad or inclusive in scope (i.e., attitudes toward relationships and evaluations of one's in-groups). In these analyses we examined the incremental predictive power of the RISC scale in predicting these constructs, controlling for the expressivity subscale of the PAQ, empathic concern, and the Singelis Interdependent Self-Construal Scale. In separate hierarchical regression equations, the criterion variable was regressed on the covariates in Step 1, and scores on the RISC scale were added at Step 2. The correlations among the variables are presented in Table 4, and the results of the hierarchical regressions are presented in Table 5. As shown in Table 5, the RISC scale is a significant predictor of these criterion measures after controlling closely related constructs.
Evaluations of Relationships In the final phase of this study, we examined the relations between the RISC scale and participants' perceptions of their social network and closest relationship. The social network diagram, consisting of a set of concentric circles with "YOU" in the
Table 4 Correlations Among the Variables in the Hierarchical Regressions (Stuiiy 1) Variable
1
2
3
4
RISC scale PAQ Expressivity Empathic Concern Singelis Interdependent Self-Construal 5. Collective Self-Esteem 6. Communal Orientation
_ .32 .28 .40
— .61 .34
— .33
—
.30 .41
.24 .61
.33 .67
.30 .40
1. 2. 3. 4.
5
6
— .43
—
Note. N = 582. RISC = Relational-Interdependent Self-Construal; PAQ = Personal Attributes Questionnaire. All correlations are significant at p < .001.
center, was used to assess the hypothesis that individuals who scored high on the RISC scale would view a greater number of relationships as very important to them than would others. As expected, scores on the RISC scale and the number of people listed in the inner circle (i.e., people it is hard to imagine life without) were positively associated (r = .24, p < .05). There were no associations between the RI^C scale and the number of people in the middle and outer circles (r = .12, p > .05 for the middle circle, r = .09, p > .05 for the outer circle). This suggests that high scorers on the RISC scale tend to have more close and important relationships than others, but have similar numbers of more distant relationships. Based on their responses to the social network diagram, participants picked their most important relationship. We predicted that people who scored high on the RISC scale would view their most important relationship as closer than would others (as indicated by the RCI subscales, the self-disclosure scale, and the IOS scale) and as more committed than would others. As shown in Table 3, the RISC scale was positively related to the degree to which the individual reported being influenced by their relationship partner (r = .27), their self-disclosure in the relationship (r = .26), the degree to which the other is included in the self (the IOS scale; r .26), and commitment to the relationship (r = .22). There were not, however, strong relations between the RISC scale and the time spent with the relationship partner, the number of activities initiated with this person, or the degree to which this person influenced the participant's future plans. If individuals with an interdependent self-construal are more committed to and close to their relationship partners, they may also perceive higher levels of social support than others. As predicted, individuals who described themselves as high on the RISC scale tended also to perceive higher levels of social support from others, r = .26. The correlations with the individual subscales of the Social Provisions Scale ranged from a low of r = .13 with the reassurance of worth subscale to a high of r = .30 with the opportunity for nurturance subscale.
Discussion The results of Study 1 indicate that the relational-interdependent self-construal is a relatively stable individual difference construct that can be measured with self-report items. Factor analyses indicated that a single factor underlies the RISC scale. The scale also
799
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL
Table 5 Hierarchical Regression of Collective Self-Esteem and Communal Orientation on PAQ Expressivity, Empatkic Concern, Singelis Interdependent Self-Construal Scale, and RISC Scores: Beta Weights and R2 (Study 1) Collective Self-Esteem Predictor Step 1 PAQ Expressivity Empathic Concern Singelis Interdependent Self-Construal Step 2 RISC scale Total/? 2
j3
Aft2
Communal Orientation /3
AR2 .54***
.03 24***
.28*** ,45***
.21***
.15** .02***
17***
.02*** .17**
27***
In the studies that follow, we examine the implications of the relational-interdependent self-construal for relationship processes. A key assumption of this perspective is that when others are included in one's self-representations, attention to these relationships and efforts to maintain and enhance them will serve to maintain and enhance the self. In Study 2, we examine the process of making important decisions in terms of the relationalinterdependent self-construal.
.56***
Note. N = 582. PAQ = Personal Attributes Questionnaire; RISC = Relational-Independent Self-Construal. **/> < .01. ***p < ,001.
exhibited both high internal consistency and good test-retest stability. In addition, the results provided evidence of the scale's convergent and discriminant validity. As expected, the RISC scale was moderately associated with other measures of conununalism and interdependence and was unassociated with measures of independence. The scale related as expected to other personality constructs: Scores on the RISC scale were moderately related to measures of empathy, agreeableness, and extroversion, but were unrelated to neuroticism and social desirability. In addition, RISC scale scores did not correlate with measures of well-being, such as depression or satisfaction with life. Criterion validity was also shown in the pattern of sex differences: Women consistently scored higher on the RISC scale than did men. When controlling for closely related constructs, the RISC scale predicted a significant proportion of incremental variance in relatively global measures of relatedness (collective self-esteem and communal orientation), thus demonstrating its discriminant validity. Scores on the RISC scale were also related to the size of participants' social networks and to their evaluations of their closest relationship. Persons with a very interdependent selfconstrual reported a greater number of relationships that were very important to them and higher levels of social support than did others. High scores on the RISC scale were also associated with high levels of self-disclosure, perceived closeness, and commitment in the relationship. Unexpectedly, scores on the RISC scale were not related to the amount of time participants spent with their relationship partner or to the number of activities they initiated with this person. These results suggest that whereas most people enjoy being with close friends and seek out opportunities to do things together regardless of the degree of interdependence of their self-construal, the quality of that time together may differ depending on individuals' self-construals. Individuals with very interdependent self-construals may be more likely than others to spend their time with significant relationship partners disclosing important or sensitive information and may be more responsive to their partner's needs and desires (as evidenced by their willingness to let their partner influence their choices and behaviors).
Study 2: The Role of Relationship Considerations in Decision Making For individuals with an interdependent self-construal, the wishes and opinions of close relationship partners are likely to be taken into account when the individual is making important decisions. For example, the student who is considering going out of state for a summer job may think of the implications of the decision for a romantic relationship. The student making a decision about taking a summer job versus going to summer school may consider his or her family's wishes in the matter. The person who is low in interdependence may not be as likely to consider other people's wishes or reactions or to consult other people for information or advice. Consequently, in this study we asked students to describe an important decision they were in the process of making and to elaborate the factors that were influencing their decision. After describing their decision, they were asked several follow-up questions about it. Embedded in the list of items were questions designed to elicit directly the extent to which close others influenced their decisions. We hypothesized that individuals who scored high on the Relational-Interdependent Self-Construal scale would tend to list relational factors that had influenced their decision making and to indicate on the close-ended items that other people and close relationships were taken into account in thendecision process.
Method Participants Participants were introductory psychology students who participated in exchange for extra course credit. Two hundred sixty-six participants (120 men, 145 women, and 1 unstated) provided complete data for all portions of the study. The open-ended descriptions of 18 participants could not be coded either due to poor handwriting or failure to follow directions.
Procedure and Materials The RISC scale and the decision questionnaire were administered in large group sessions of 30 to 200 as part of a mass-testing process. The first packet that the students completed included the RISC scale and several other instruments, many of which were designed for other studies. Cronbach's alpha for the RISC scale was .89. The participants were given the decision questionnaire last. On the first page, they were asked to "Describe a decision that you are in the process of making. Please describe a decision that has a fairly significant impact on your life." The instructions went on to give several examples of possible decisions one might be making (e.g., how to spend one's summer or with whom to room next year). The participants were asked to describe as thoroughly as possible the factors that they were considering in making their decision—both major factors and relatively minor factors. No men-
800
CROSS, BACON, AND MORRIS
tion was made of the potential influence of others in this decision. The participants were encouraged to be explicit and to write in essay style, not to simply list pros and cons. Students who could not think of a decision they were in the process of making were encouraged to describe a recent decision they had already made. Most of the first page was empty for students to write out their thoughts. On the second page, participants answered several questions about their decision. They were enjoined not to add anything to their written description after beginning these items. First they were asked if this was a decision they were still making or had already made. Then they were asked to categorize the decision as to whether it had to do with personal relationships, academics, family, jobs, housing, or other. They were allowed to select as many categories as were appropriate. The remainder of the questionnaire consisted of items that assessed their decision. Crucial to this investigation were three items that addressed the role of others in the decision. They were "To what extent do you take into account the needs and wishes of others when thinking about this decision?" "How important in your decision process are the opinions of your close friends?" and "How important in your decision process are the opinions of your family members?" These items were embedded with other filler items (e.g., "How important are financial concerns in your decision making?").
Coding the Open-Ended Descriptions Each of the factors mentioned by participants as an influence on their decision was coded into one of two categories: a relational category or a nonrelational category. (Statements that were not factors in the decision were not coded.) Examples of relational statements were "I'd like to be close to my family" and "My roommates wanted the first apartment." Examples of nonrelational statements were "I could get a better job here" and "The second apartment was cheaper." Two trained judges coded the descriptions; 25 percent of the protocols were coded by both judges to establish interrater reliability. (The coders were blind to participants' scores on the RISC scale.) The number of relational and nonrelational factors in each protocol was determined independently by the two judges; interjudge reliability was .92 for the relational factors and .89 for the nonrelational factors.
Results Participants generated an average of 3.3 relational factors that influenced their decisions (SD = 3.96, range = 0 to 23) and an average of 5.3 nonrelational factors (SD = 3.71, range = 0 to 19). As hypothesized, the RISC scale was correlated with the number of relational factors that were described (r = .18, p < .01, two-tailed), but it was not related to the number of nonrelational factors included in the descriptions (r = .01). The RISC scale was also related to the extent to which the participants took into account the needs and wishes of others when making the decision (r = .24, p < .01) and the importance of the opinions of family (r = .27, p < .01) and friends (r = .21, p < .01) in the decision. We also reasoned that the relation between the RISC and the extent to which others are considered when making decisions would be stronger for decisions that did not directly concern a relationship (e.g., where to work for the summer or whether to take summer school classes). When making decisions about relationships or family issues, virtually everyone is likely to mention the influence of the other persons involved in the decision. When making decisions that are not explicitly relational, however, individuals with more interdependent self-construals may be more likely to think about and take into account the opinions or desires of their friends or family members. To look at these types of
decisions separately, decisions that were coded by the participants into either the relational or family category were included together in a single relational category (n = 126); all other decisions were included in a nonrelational category (n = 135). There were no gender differences in the type of decision described, x2 = .00. As expected, the correlations between the RISC scale and the items indicating the influence of others were generally stronger for nonrelational decisions than for the relational decisions. When describing nonrelational decisions, the RISC scale was positively related to the number of relational factors mentioned (r = .21 versus r = .11 for relational decisions) and to the importance of family members' opinions (r = .37 compared with r = .15 for relational decisions). Although the relation between the RISC scale and reports of the extent to which the needs and wishes of others were taken into account was larger for the nonrelational decisions (r = .24) than for the relational decisions (r = . 18), this difference was small. The exception to this pattern was the item assessing the importance of friends' opinions. For this item, the relation with interdependence was stronger for the relational decisions (r = .25) than for the nonrelational decisions (r = .11), indicating that students who scored high on the RISC scale were more likely to turn to their friends for advice when making decisions about their relationships than when making other sorts of decisions.5
Discussion This study supported the hypothesis that individuals who tended to think of themselves in relational, interdependent terms would be more likely to consider the consequences of their decisions for other people or to take into account the opinions or needs of close others. When describing an important decision, individuals who scored high on the RISC scale were more likely than those who scored low to freely generate relationship-oriented factors in their decision. The instructions for the decision description task were carefully designed to permit the generation of a wide variety of factors that could influence decisions; relational factors were not specifically prompted in the instructions to the participants. Thus, any mention of the role of others in this decision was selfgenerated by the respondents and not likely to have been a consequence of subtle demand factors. In addition, participants who 5
We also examined the extent to which the relational-interdependent self-construal mediated sex differences in the ratings of the influences on nonrelational decisions. Independent sample t tests revealed sex differences in ratings of the extent to which others were taken into account, Wwt>men = 3.07, SD = 1.09, M men = 2.66, SD = 1.12, «133) = -2.15,p < .04, d = — .37; in the importance of family members' opinions, M womeo = 3.58, SD = 1.01, M men = 3.11, SD = 1.00, /(133) = 2.68, p < .01, d = -.47; and on the RISC scale, Mwomen = 55.19, SD = 8.11, Wmcn = 48.36, SD = 10.16, f(133) = -4.35, p < .001, d = -.15. Sex alone accounted for 3% of the variance in the extent to which the needs and wishes of others were considered in the decision. When the RISC was added to the equation, sex was reduced to accounting for only an additional 1% of the variance in the extent variable. TTius, scores on the RISC scale accounted for 71% of the relation between sex and the criterion variable. For the item assessing the importance of family members' opinions, sex alone accounted for 5% of the variance in the dependent variable. When the RISC was added to the equation, sex was reduced to accounting for only 1% of the variance. Thus, scores on the RISC scale accounted for 78% of the relation between sex and the criterion variable.
801
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL
Two key components of the process of relationship building are self-disclosure and responsiveness to one's partner's needs and concerns (Reis & Patrick, 1996; Reis & Shaver, 1988). Cross and Madson (1997) hypothesized that individuals with an interdependent self-construal would be more likely to self-disclose to others, to be sensitive to others, and to describe their relationships as close, and the studies described above provide self-report evidence in support of these claims. But do other people concur with these self-descriptions? In this study of dyadic interactions, we examined the degree to which individuals' self-construals were associated with their partners' evaluations of their openness and responsiveness in the interaction and to their general evaluation of the relationship.
We hypothesized that the partners of individuals who scored high on the RISC scale would perceive them as disclosing more about themselves and as more responsive than would the partners of individuals who scored low on the self-construal measure. Given that self-disclosure and sensitive responding are central ingredients in the development of a satisfying or close relationship, we reasoned that individuals paired with a high scorer on the RISC scale would also tend to evaluate the relationship and the interaction more positively than would others. Therefore, we examined the direct relationship between individuals' RISC scale scores and their partners' satisfaction with the interaction, their liking for the individual, and their perceptions of closeness to their partner. In addition, we examined a mediational model, in which this expected relation between one individual's self-construal and her partner's evaluations of the relationship are mediated by the individual's self-disclosure and her partner's perceptions of her responsiveness. Previous research has shown that perceptions that one's partner is responsive to one's needs mediate the effects of self-disclosure on the closeness of a relationship (Laurenceau, Barrett, & Pietromonaco, 1998). Thus, we hypothesized that individuals who score high on the RISC scale will be more likely to self-disclose and will consequently be viewed as responsive by their partners in a get-acquainted task. As a result, their partners will evaluate their relationship more positively than will partners of low scorers.
To address this issue, one might look at long-term relationships, asking the partners of people who are high or low in the RISC scale to describe them. Although close friends or romantic partners can provide helpful information on these dimensions, their responses may be biased. Individuals tend to choose friends and relationship partners who are similar to themselves (Byrne, 1971) and who see them as they see themselves (Swann, De La Ronde, & Hixon, 1994). As a result, a close friend of a person with a very interdependent self-construal may also have a very interdependent self-construal and may tend to project his or her attitudes and beliefs onto the person. The friend also may be aware of the target's self-views, and respond as he or she thinks the target would.
This study, therefore, goes beyond self-reports to examine informants' ratings of an interaction designed to promote closeness in a controlled setting. It permits us to test the question posed earlier: Do the partners of individuals with a very interdependent self-construal concur with their self-reports of open self-disclosure and sensitivity to their relationship partners? In addition, this study allows us to examine the processes whereby individuals with an interdependent self-construal begin building a relationship. We expect that these individuals will view this situation much like any other opportunity to get to know a new acquaintance and so will open up about themselves and will be perceived as responsive by their partners, resulting in positive evaluations of the relationship by their partners.
scored high on the RISC scale also indicated on the experimenterprovided items that they were more likely to take others into account when making their decision. This study supported the hypothesized role of the relationalinterdependent self-construal in decision-making. Study 3 examines the association between the self-construal and relationshipdevelopment strategies. Study 3: The Relational-Interdependent Self-Construal and Relationship Development
One way around this problem is to provide an experimental setting in which randomly paired individuals take time to get to know each other, much like two acquaintances might begin to form a friendship. Using this approach, researchers can simulate the process of developing a friendship but reduce some of the biases inherent in the perceptions of close friends or partners. Fortunately, Aron, Melinat, Aron, Vallone, and Bator (1997) have shown that a temporary feeling of closeness among two previously unacquainted individuals can be developed in an experimental context. Drawing on research emphasizing self-disclosure as the defining feature of a close relationship, they describe a structured set of activities that asks participants to describe their thoughts, feelings, and experiences during a 45-min period. They found that the participants in the experimental group described their relationships as closer than did members of a control group, in which participants engaged in relatively superficial small talk for the same amount of time. Certainly the closeness generated in the laboratory is not the same as that created in the context of an ongoing relationship, but this method permits one to model the processes involved in the development of close relationships and to reduce some of the biases inherent in ongoing relationships.
Method Participants and Procedure Participants were 181 women from introductory psychology classes who participated in exchange for course credit. They signed up for the study in groups of four, and when they arrived at the lab, they were told that the study addressed problem solving by pairs. Specifically, they were told that the study addressed whether or not knowing one's partner had any impact on how effectively pairs solved problems. Participants then provided informed consent. The experimenter checked at this point to determine if any of the participants knew each other. Participants were randomly assigned to partners (except for instances in which two participants knew each other previously; in this case, they were randomly assigned to one of the two remaining participants). After completing the RISC scale in individual cubicles, the participants were seated in cubicles with their randomly assigned partner. (When an odd number of participants showed up for the study, one person was paired with a confederate. These participants' responses were dropped from all analyses.) Participants were told that they had been assigned to the condition in which the partners spent some time getting to know each other before the problem solving task (in fact, all participants got to know either another participant or a confederate).
802
CROSS, BACON, AND MORRIS
The procedure for the interaction task was adapted from the Aron et al. (1997) closeness task. Each partner was given a set of cards, and each card had a different question on it, such as "Given the choice of anyone in the world, whom would you want as a dinner guest?" The Aron et al. (1997) questions range in the sensitivity of the information elicited; after piloting these questions On a separate group of participants, we selected a set of 15 questions that varied in the degree of disclosure elicited. (The cards were arranged with the least revealing questions first.) Members of the pairs were asked to alternate asking and answering the questions, with each person responding to every question. Participants were instructed that they could take as long as they wanted on any single question and should feel free to skip any questions mat they preferred not to answer. They were given a maximum of 15 min to work through the set of questions. After the interaction, participants returned to separate cubicles and completed several questionnaires about their interactions. They were assured that these evaluations would not be seen by their partner. Following completion of these questionnaires, participants were carefully and thoroughly debriefed and dismissed.
Measures For all instruments except one of the selection variables (see below) responses were provided on a 5-point Likert-type scale, with responses ranging from 1 (not at all/strongly disagree) to 5 (very much/strongly agree). Relational-Interdependent Self-Construal (RISC). Responses to the 11-item RISC scale demonstrated a good internal consistency with a Cronbach's alpha of .84. Disclosure. Disclosure was measured with eight items from the SelfDisclosure Index (Miller et al., 1983). (Two items that were judged to be inappropriate for this situation were dropped from the scale.) Participants were asked to complete this scale twice, once for themselves and once for how much their partner disclosed. Cronbach's alpha was .76 for own disclosure and .83 for perceptions of one's partner's disclosure. Partner's responsiveness. In this study, responsiveness was operationalized as the demonstration that one understands, cares for, and values one's partner (Reis & Shaver, 1988). Perceptions of the partner's responsiveness were measured with six items adapted from a scale by Cutrona, Hessling, & Suhr (1997), In addition, the authors created six items designed to tap the perceptions that one has been understood, cared for, and validated (Reis & Patrick, 1996). The items are listed in the Appendix. Cronbach's alpha was .92. Liking. Liking was assessed with three items from Miller et al. (1983) and one additional item constructed by the current researchers. The additional item was "How much would you like to work with your partner on the upcoming problem solving task?" Cronbach's alpha was .88. Satisfaction. Satisfaction was measured with five commonly used items. Examples of the items included "How much did you enjoy the interaction with your partner?" and "How satisfied were you with your interaction with this partner?" Cronbach's alpha was .88. Closeness. The Subjective Closeness Index (Berscheid et al., 1989) was used to measure perceived closeness. This measure includes the items "Relative to all your other relationships, how would you characterize your relationship with this person?" and "Relative to what you know about other people's close relationships, how would you characterize your relationship with this person?" A third item added by the current authors was "Right now, how close do you feel to your partner?" Cronbach's alpha of the 3-item scale was .84. Selection variables. As a check on participants' attitudes toward the interaction task, they were asked to rate the extent to which they and their partners took the discussion time seriously. If either member of a pair indicated that they did not take the discussion time seriously, the couple was excluded from all analyses. A total of three couples were excluded for this reason. Participants were also asked how well they knew their partners
before the interaction. Responses were made on a 6-point Likert-type scale ranging from 1 (not at all) to 6 (we knew each other pretty well). No participants indicated that they knew their partners pretty well. Five couples were dropped for failing to follow directions appropriately. The final sample was comprised of 68 pairs.
Results The first issue addressed in these analyses was whether an actor's RISC scale score was related to her partner's perceptions of the actor's degree of self-disclosure and responsiveness. Kenny (1996) recommends that when analyzing data from interchangeable dyad members, researchers should first examine the correlations between the two dyad members1 scores on the variables of interest to test for nonindependence. As one would expect, pairwise intraclass correlations (Griffin & Gonzalez, 1995) revealed significant correlations between partners' scores for self-disclosure (r = .31), perceptions of one's partner's disclosure (r = .34), perceptions of one's partner's responsiveness (r = .44), satisfaction with the interaction (r = .37), liking for the partner (r - .39), and closeness to the partner (r — .19) (all ps < .05). To examine dyadic effects, Kenny (1996) recommends the use of pooled regressions, which allow for the computation of the effect of an actor's characteristics on her own evaluations of the interaction (termed the "actor effect") and the effect of her partner's characteristics on the actor's evaluations (termed the "partner effect"). The values computed in this analysis represent unstandardized regression coefficients. As the first two lines of Table 6 show, participants whose partners scored high on the RISC scale perceived their partners as disclosing more about themselves and as being more responsive to their needs and concerns than did participants whose partners scored low on the RISC scale (summarized in the "partner effect" column). In addition, an individual's own RISC score was related to how she perceived her partner (summarized in the "actor effect" column).
Table 6 Estimation of Actor and Partner Effects of the RISC Scale on Evaluations of the Dyadic Interaction (Study 3)
Dependent variable Perceptions of partner's disclosure Perceptions of partner's responsiveness Self-disclosure Satisfaction Liking of partner Closeness to partner
Actor effect*
Partner effect*
t
df
.54
7 22***
.19
2.55**
70
.69
13.26***
.36
6.97***
71
.45 .55 .55 .40
6.74***
.25 .44 .42 .17
3.75*** 6.17*** 4.98*** 1.90*
70 70 70 69
7 JI***
6.54*** 4.45***
Note. N = 68 pairs. RISC = Relational-Interdependent Self-Construal. The values in the actor effect and partner effect columns represent unstandardized regression coefficients. The formula suggested by Kenny (1996) to approximate the degrees of freedom results in different values for each analysis. 11 The actor effect is the effect of an actor's RISC scale score on her own evaluations of the interaction. b The partner effect is the effect of an actor's partner's RISC scale score on the actor's evaluations of the interaction. *p < .05. **p < .01. ***/? < .001, one-tailed.
803
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL
Participants' scores on the RISC scale were positively related to their own ratings of their self-disclosure (see Table 6 under "actor effect"), replicating the findings of Study 1. There was also a significant partner effect on ratings of self-disclosure: Individuals paired with a person who scored high on the RISC scale described themselves as self-disclosing more than did individuals paired with a low-scoring partner. Table 6 also shows that participants' ratings of their satisfaction with the discussion, their liking of their partner, and feelings of closeness to their partner were significantly related to their own RISC scale scores and to their partner's scores. As hypothesized, individuals paired with a person who scored high on the RISC scale were more satisfied with the interaction, liked their partner more, and felt closer to their partner than did individuals paired with a person who scored low on the RISC scale. We hypothesized that an actor's self-disclosure and her partner' s ratings of the actor's responsiveness would mediate the direct effects of the actor's self-construal on her partner's overall evaluations of the interaction. To create a single index of participants' overall evaluations of the interaction, the measures of satisfaction, liking, and closeness were standardized and summed together. The reliability for this composite satisfaction score was .95. The hypothesized relations among partners' RISC scale scores, selfdisclosure, perceptions of their partner's disclosure, and overall satisfaction with the relationship are shown in Figure 1. This model includes the direct effects of an actor's RISC scale scores on her partner's overall satisfaction (paths a), as well as mediators of this relation. The most important relations in the model are captured by paths b, c, and d: An actor's RISC scale scores are expected to be related to her own self-disclosure (paths b), which will in turn predict her partner's perceptions of the actor's responsiveness (paths c). Perceptions of one's partner's responsiveness
were expected to predict the individual's overall satisfaction (paths d). In addition, we examined the possible direct effects of an actor's RISC scale scores on her perceptions of her partner's responsiveness (paths e) and on her partner's perceptions of the actor's responsiveness (paths f). We also included possible direct effects of an actor's self-disclosure on her perceptions of her partner's responsiveness (paths g), on her own overall satisfaction with the relationship (paths h), and on her partner's overall satisfaction (paths i). To test this model, structural equation analyses were conducted using maximum likelihood estimation from the LISREL 8 program. Members of the dyads were randomly assigned to be person 1 (PI) or person 2 (P2); Table 7 presents the correlation matrix and the standard deviations for the variables used in this model. In these analyses, the path coefficients for the two participants were constrained to be equivalent (e.g., the path from Pi's RISC score to Pi's self-disclosure score was set to be equivalent to the path from P2's RISC score to P2's self-disclosure score). In addition, the model included the correlations between the parallel measures from the two participants (e.g., the correlation between P i ' s selfdisclosure and P2's self-disclosure was included in the model). Although this initial model presented a relatively good fit to the data, the modification indices and the path coefficients indicated that the paths from RISC scores to perceptions of the partner's responsiveness (paths f) and the paths from self-disclosure to one's partner's overall satisfaction score (paths i) were nonsignificant and should be dropped from the model. This revised model was found to fit the data well, ^ ( 1 6 , JV = 68) = 26.83, p < .05, GFI = .92, CFI = .93. The standardized path coefficients from this model are presented in Figure 2. Participants' RISC scale scores were directly related to their partner's
Pis perception of her partner's responsiveness
Pi's overall satisfaction with the relationship
P2's overall satisfaction with the relationship P2 s perception of her partner's responsiveness
Figure 1. Initial path model. The bold arrows represent the paths in the hypothesized mediation model. RISC = Relational-Interdependent Self-Construal scale; PI = Partner 1; P2 = Partner 2.
804
CROSS, BACON, AND MORRIS
Table 7 Correlations and Standard Deviations of Partner 1 (PI) and Partner 2 (P2) Variables Used in the Mediation Model Predicting Overall Satisfaction With the Interaction (Study 3) Variable
1
1. PI RISC scale 2. PI self-disclosure 3. PI perceptions of partner's responsiveness 4. PI overall satisfaction 5. P2 RISC scale 6. P2 self-disclosure 7. P2 perceptions of partner's responsiveness 8. P2 overall satisfaction
(-56) .23* .13
(.58) .32**
.17 .19 -.08 -.03
.31** .26* .25*
.05
(-55)
51**
.77** -.04 .04 31**
(2.23) .13 .10 .40**
,38**
.27*
.24*
(.47) .22* .40** .24*
(-60) .45** .54*
(-47)
.64*
(2.21)
Note. RISC = Relational-Interdependent Self-Construal. Standard deviations of each variable are on the diagonal. *p < .05. **p < .01, one-tailed.
overall satisfaction with the relationship such that individuals paired with a person who had a very interdependent self-construal were more satisfied and liked their partner more than did individuals paired with a partner who scored low on the RISC scale. In addition, there were significant indirect effects of RISC scale scores on participants' partners' composite satisfaction scores, Ppi RISC-P2 evaluation
=
- ^ 0P2 RISC-PI evaluation
=
-03> P» < .05,
one tailed. As shown by the bold arrows in Figure 2, individuals who scored high on the RISC scale tended to self-disclose more than did others (path b); the self-disclosure scores were signifi-
cantly related to their partner's perceptions of their responsiveness (path c), which were strongly related to their partner's satisfaction with the relationship (path d).
Discussion Disclosing personal information about oneself and being sensitive and responsive to one's partner's disclosures are central processes in the development of relationships (Laurenceau et al., 1998; Reis & Shaver, 1988). We hypothesized that individuals
Pi's perception of her partner's responsiveness
Pi's overall satisfaction with the relationship
P2's overall satisfaction with the relationship P2 s perception of her partner's responsiveness
Figure 2. Final path model. All paths are significant atp < .05, one-tailed. The bold arrows represent the paths in the hypothesized mediation model. RISC = Relational-Interdependent Self-Construal scale; PI = Partner 1; P2 = Partner 2. Correlations among the parallel measures for PI and P2 were as follows: RISC scale, r = .18, ns; self-disclosure, r = .20, ns; perceptions of partner responsiveness, r = .21, p < .05; overall evaluation, r = .07, ns.
RELATIONAL-INTERDEPENDENT SELF-CON STRUAL who scored high on the RISC scale would be more likely than others to use these strategies to develop new relationships. To avoid the biases inherent in ongoing relationships, we asked randomly assigned strangers to take some time becoming acquainted and then to indicate how open and responsive their partners had been. As expected, individuals' RISC scale scores were positively related to their partner's ratings of the person's self-disclosure and responsiveness. In addition, participants who were paired with a person who scored high on the RISC scale evaluated their relationship with their partner more positively than did participants who were paired with a person who scored low on the RISC scale. Structural equation analyses showed that the relation between individuals' RISC scale scores and their partners' overall satisfaction with the interaction and liking of their partner were mediated by the individual's degree of self-disclosure and the partner's perceptions of the individual's responsiveness. This model suggests that persons with a very interdependent self-construal are more likely to self-disclose than are others; individuals who are very disclosing are in turn evaluated as responsive by their partners. Finally, participants who view their partners as responsive and sensitive to their concerns evaluate the relationship more positively. In short, these results suggest that individuals with a very interdependent self-construal develop and nurture new relationships by being open about themselves and by showing sensitivity and concern for their relationship partners, even when these partners are randomly assigned strangers.
General Discussion We report the development of a self-report measure that focuses on individuals' self-representations as relational or interdependent with close others. In Study 1, we found strong evidence of the scale's reliability and validity. The 11-item measure taps a single, general factor and evidences good internal and test-retest reliability. An examination of the RISC scale's associations with other measures of interdependence, independence, and personality attests to its convergent and discriminant validity. The RISC scale predicted unique variance in global measures of collective selfesteem and communal orientation when other related constructs were controlled. Thus, we have good evidence that the RISC scale measures a construct that is reasonably distinct from other similar measures. These results revealed no zero-order correlations between the RISC scale and measures of well-being or self-esteem. Other research has found that individuals who are very needy or dependent on others tend to be vulnerable to depression (Rude & Burnham, 1995). Our results suggest that defining oneself in terms of close relationships need not necessarily lead to these negative consequences. Further research is necessary, however, to explore the conditional relationships between the interdependence of the self-concept and well-being. For the person with a very interdependent self-construal, overall well-being may depend at least in part on the well-being of close relationships. If close relationships are threatened or undergoing periods of difficulty, the person with a very interdependent self-construal may suffer more than the person whose self-concept is not based on close relationships. These studies also supported the hypothesis that women would be more likely to define themselves in terms of their close relationships than would men. The effect sizes for the RISC scale in
805
these studies ranged from fairly small (d = -.17) to moderate (d - -.57). These effect sizes are similar in magnitude to many other gender differences in behavior (Hyde, 1996). We have mentioned throughout the article that individuals who have a very interdependent self-construal should be motivated to develop and maintain close relationships, but we unfortunately overlooked measures of related motives, such as the intimacy motive or the need for affiliation, in our examination of the convergent and discriminant validity of the RISC scale. Me Adams describes the intimacy motive as "the desire for warm, close, and communicative relations with others" (1989, p. 53), and we would expect the RISC scale to correlate fairly highly with a measure of this motivation. This relation should not be symmetric, however: A person who has constructed an interdependent self-construal should also score high on the measure of intimacy motivation, but the reverse may not be true. As McAdams (1989) explains, many people who score high on intimacy motivation may not have identified this tendency or have incorporated it into their self-views. We focused the remainder of the research on the association between the relational-interdependent self-construal and relationship-oriented thoughts and behaviors. Consistent with the hypothesis that individuals with a very interdependent selfconstrual will have a larger social network, we found that individuals who scored high on the RISC scale listed more people who were very important to them and perceived higher levels of social support than did lows (Study 1). Individuals who scored high on the RISC scale also described their most important relationships as closer and more committed than did others. These results provide criterion validity for the RISC scale in that they support the hypothesis that individuals with a very interdependent selfconstrual should be especially motivated to develop and maintain close relationships. Additional research is necessary to uncover more about the processes and motivations underlying these associations. The findings of Study 1, however, suggest that quality (i.e., self-disclosure and mutual influence), not quantity, of time spent with close others differentiates the approach to relationships adopted by very interdependent persons from that adopted by those who do not think of themselves as interdependent with others. If close relationships are part of the self, then individuals should take care to promote and protect these relationships. In Study 2, highly interdependent persons were more likely than others to take into account the opinions or needs of friends and family when making important decisions. When close relationships are selfdefining, then the needs and wishes of close others may be nearly as important or salient as one's own needs and wishes. In addition, decisions made without heeding their implications for close others may cause conflict or threaten relationships. Study 3 supported the findings of Studies 1 and 2, showing that when individuals with a very interdependent self-construal were asked to get to know a new person, they were more likely to reveal information about themselves and to be viewed as responding sensitively to their relationship partners than were others. These conclusions were based not only on self-reports, but also on the reports of their partners in these interactions. Thus, this study provides important criterion validity for the RISC scale. In addition, Study 3 begins to answer the question of how individuals with an interdependent self-construal create and maintain close relationships. The structural equation analyses of these
806
CROSS, BACON, AND MORRIS
dyadic interactions revealed that individuals with a very interdependent self-construal tended to self-disclose more, which was related to their partners' perceptions that they were concerned and caring; these perceptions that one's partner was responsive were strongly related to positive evaluations of the relationship. Thus, individuals' RISC scale scores are not only related to their own perceptions and evaluations of themselves and their relationships, but also to others' perceptions of them. These findings are crosssectional and based on a sample of college women; replication in longitudinal studies with a more diverse sample will promote further understanding of the role of the self-construal in relationship processes. Future studies should also examine the relationship of the self-construal to other relationship-enhancing processes in ongoing close relationships.
Research Applications When a relational orientation is conceptualized at the level of the self-concept, researchers can draw upon the extensive research on the self to more clearly specify the mechanisms and processes that underlie the relations between interdependence and other phenomena. There has been much speculation about how viewing oneself as relational or interdependent may influence behavior, but empirical tests of these hypotheses have been thwarted by the lack of an appropriate measure. For example, Cross and Madson (1997) suggest that many theories and research findings in the area of self-enhancement processes have assumed an independent selfconstrual. They argue that the strategies used to enhance the self may vary, depending on the structure of the self. With this new measure, researchers are better equipped to test these and other hypotheses specifically, thus better understanding the sources of social behavior. Similarly, some theorists have attributed gender differences in behavior to differences in the self-construal (Cross & Madson, 1997; Markus & Oyserman, 1989; Surrey, 1991), but others contend that most gender differences in behavior are due to differences in status or power (Ridgeway, 1988; Snodgrass, 1992). With this measure, researchers can more carefully identify the independent and joint effects of the self and the situation in social behavior. A critical next step is to examine the relation between this construct and individuals' self-views as independent and autonomous. As Trafimow, Triandis, and Goto (1991) have suggested, these two types of self-representations may be stored separately from each other in memory and accessed with different frequencies. The effects of the interdependent self-construal on relationship-oriented behavior may depend on the extent to which an individual has developed independence-related self-views. For example, the person with a very interdependent self-construal who also has well-elaborated self-views as independent and self-reliant may be best able to balance the needs of relationship partners with his or her own needs (see Helgeson, 1994, for a review). A consideration of the relational-interdependent self-construal may open the doors to new understanding of the role of the self in cognition, motivation, well-being, and social interaction. References Allen, R. L., Dawson, M. C , & Brown, R. E. (1989). A schema based approach to modeling an African American racial belief system. American Political Science Review, 83, 421-442,
Altman, I., & Taylor, D. A. (1973). Social penetration: The development of interpersonal relationships. New York: Holt, Rinehart & Winston. Antonucci, T. C. (1986). Hierarchical mapping technique. Generations, 10, 10-12. Aron, A., Aron, E. N., & Smollen, D. (1992). Inclusion of other in the self scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63, 596-612. Aron, A., Aron, E. N., Tudor, M , & Nelson, G. (1991). Close relationships as including other in the self. Journal of Personality and Social Psychology, 60, 241-253. Aron, A., Melinat, E., Aron, E. N., Vallone, R. D., & Bator, R. J. (1997). The experimental generation of interpersonal closeness: A procedure and some preliminary findings. Personality and Social Psychology Bulletin, 23, 363-377. Beck, A. T. (1967). Depression: Causes and treatment. Philadelphia: University of Pennsylvania Press. Bellah, R. R , Madsen, R., Sullivan, W. M., Swidler, A., & Tipton, S. M. (1985). Habits of the heart: Individualism and commitment in American life. Berkeley, CA: University of California Press. Belle, D. (Ed.). (1982). Lives in stress: Women and depression. Beverly Hills, CA: Sage. Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The relationship closeness inventory: Assessing the closeness of interpersonal relationships. Journal of Personality and Social Psychology, 57, 792-807. Blaine, B., & Crocker, J. (1993). Self-esteem and self-serving biases in reactions to positive and negative events: An integrative review. In R. F. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 5 5 85). New York: Plenum Press. Blascovich, J., & Tomaka, J. (1991). Measures of self-esteem. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 115-160). San Diego, CA: Academic Press. Brewer, M. B., & Gardner, W. (1996). Who is this "we"? Levels of collective identity and self-representations. Journal of Personality and Social Psychology, 71, 83-93. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Clark, M. S., Ouellette, R., Powell, M. C , & Milberg, S. (1987). Recipient's mood, relationship type, and helping. Journal of Personality and Social Psychology, 53, 94-103. Costa, P. T., & McCrae, R. R. (1992). Professional manual: Revised NEO Personality Inventory (NEO PI-R) and NEO five factor inventory (NEOFFI). Odessa, FL: Psychological Assessment Resources. Cross, S. E. (1995). Self-construals, coping, and stress in cross-cultural adaptation. Journal of Cross-Cultural Psychology, 26, 673-697. Cross, S. E., & Madson, L. (1997). Models of the self: Self-construals and gender. Psychological Bulletin, 122, 5-37. Crowne, D. P., & Marlowe, D. (I960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. Cutrona, C. E. (1989). Ratings of social support by adolescents and adult informants: Degree of correspondence and prediction of depressive symptoms. Journal of Personality and Social Psychology, 9, 3-14. Cutrona, C. E., Hessling, R. M., & Suhr, J. (1997). The influence of husband and wife personality on marital social support interactions. Personal Relationships, 4. 379-393. Cutrona, C. E., & Russell, D. (1987). The provisions of social relationships and adaptation to stress. In W. H. Jones and D. Perlman (Eds.), Advances in personal relationships (Vol. 1, pp. 37-68). Greenwich, CT: JAI Press. Davis, M. H. (1980). A multidimensional approach to individual differences in empathy. JSAS: Catalog of Selected Documents in Psychology, 10, 85. Deiner, EM Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction with Life Scale. Journal of Personality Assessment, 49,
71-75.
RELATIONAL-INTERDEPENDENT SELF-CONSTRUAL Griffin, D., & Gonzalez, R. (1995). Correlational analysis of dyad-level data in the exchangeable case. Psychological Bulletin, 118, 430-439. Guisinger, S., & Blatt, S. J. (1994). Individuality and relatedness: Evolution of a fundamental dialectic. American Psychologist, 49, 104-111. Harter, S. (1993). Causes and consequences of low self-esteem in children and adolescents. In R. F. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 87-116). New York: Plenum Press. Helgeson, V. S. (1994). Relation of agency and communion to well-being: Evidence and potential explanations. Psychological Bulletin, 116, 412428. Helmreich, R. L., Spence, J. T., & Wilhelm, J. A. (1981). A psychometric analysis of the Personal Attributes Questionnaire. Sex Roles, 7. 10971108. Hyde, J. (1996). Half the human experience. Lexington, MA: Heath. James, W. (1890/1983). The principles of psychology. Cambridge, MA: Harvard University Press. Kashima, Y., Yamaguchi, S., Kim, U., Choi, S. C , Gelfand, M. J., & Yuki, M. (1995). Culture, gender, and self: A perspective from individualismcollectivism research. Journal of Personality and Social Psychology, 69, 925-937. Kelley, H. H., Berscheid, E., Christensen, A., Harvey, J. H., Huston, T. L., Levinger, G., McClintock, E., Peplau, L. A., & Peterson, D. R. (1983). Close relationships. New York: Freeman. Kenny, D. A. (1996). Models of non-independence in dyadic research. Journal of Social and Personal Relationships, 13, 279-294. Lauxenceau, J.-P., Barrett, L. F., & Pietromonaco, P. R. (1998). Intimacy as ,an interpersonal process: The importance of self-disclosure, partner disclosure, and perceived partner responsiveness in interpersonal exchanges. Journal of Personality and Social Psychology, 74, 1238-1251. Luhtanen, R., &. Crocker, J. (1992). A collective self-esteem scale: Selfevaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302-318. Lykes, M. B. (1985). Gender and individualistic vs. collectivistic bases for notions about the self. In A. J. Stewart & M. B. Lykes (Eds.), Gender and personality: Current perspectives on theory and research (pp. 268-295). Durham, NC: Duke University Press. Marin, G., & Triandis, H. C. (1985). Allocentrism as an important characteristic of the behavior of Latin Americans and Hispanics. In R. Diaz-Guerrero (Ed.), Cross-cultural and national studies in social psychology (pp. 69-80). Amsterdam: North Holland. Markus, H., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224253. Markus, H. R., & Kitayama, S. (1994). A collective fear of the collective: Implications for selves and theories of selves. Personality and Social Psychology Bulletin, 20, 568-579. Markus, H., & Oyserman, D. (1989). Gender and thought: The role of the self-concept. In M. Crawford & M. Hamilton (Eds.), Gender and thought (pp. 100-127). New York: Springer-Verlag. Matsumoto, D., Weissman, M. D., Preston, K., Brown, B. R., & Kupperbusch, C. (1997). Context-specific measurement of individualismcollectivism on the individual level: The Individualism-Collectivism Interpersonal Assessment Inventory. Journal of Cross-Cultural Psychology, 28, 743-767. McAdams, D. P. (1989). Intimacy: The need to be close. New York: Doubleday. McCombs, H. (1985). Black self-concept: An individual/collective analysis. International Journal of Intercultural Relations, 9, 1-18. McGrath, E., Keita, G. P., Strickland, B. R., & Russo, N. F. (1990). Women and depression: Risk factors and treatment issues. Washington, DC: American Psychological Association. Miller, L. C , Berg, i. H., & Archer, R. L. (1983). Openers: Individuals who elicit intimate self-disclosure. Journal of Personality and Social Psychology, 44, 1234-1244.
807
Moran, P. B., & Eckenrode, J. (1991). Gender differences in the costs and benefits of peer relationships during adolescence. Journal of Adolescent Research 6, 396-409. Oved, Y. (1988). Two hundred years of American communes. New Brunswick, NJ: Transaction Press. Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401. Realo, A., Allik, J., & Vadi, M. (1997). The hierarchical structure of collectivism. Journal of Research in Personality, 31, 93-116. Reis, H. T., & Patrick, B. C. (1996). Attachment and intimacy: Component processes. In E. T. Higgins and A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 523-563). New York: Guilford Press. Reis, H. T., & Shaver, P. (1988). Studying social interaction with the Rochester Interaction Record, In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 269-318). San Diego, CA: Academic Press. Ridgeway, C. L. (1988). Gender differences in task groups: A status and legitimacy account. In M. Webster and M. Foschi (Eds.), Status generalization: New theory and research (pp. 188-206). Stanford, CA: Stanford University Press. Roberts, T., &. Nolen-Hoeksema, S. (1989). Sex differences in reactions to evaluative feedback. Sex Roles, 21, 725-747. Roberts, T., & Nolen-Hoeksema, S. (1994). Gender comparisons in responsiveness to others' evaluations in achievement settings. Psychology of Women Quarterly, 18, 221-240. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rude, S. S., & Burnham, B. L. (1995). Connectedness, neediness: Factors of the DEQ and SAS dependency scales. Cognitive Therapy and Research, 19, 323-340. Rude, S. S., Welch, N., & Sanders, B. (1998). The development and validation of a scale measuring the construct of connectedness. Unpublished manuscript, University of Texas. Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and deterioration) of satisfaction and commitment in heterosexual involvement. Journal of Personality and Social Psychology, 45, 101-117. Sampson, E. E. (1985). The decentralization of identity: Toward a revised concept of personal and social order. American Psychologist, 40, 12031211. Scheier, M. F., & Carver, C. S. (1985). Optimism, coping, and health: Assessment and implications of generalized outcome expectancies. Health Psychology, 4, 219-247. Schwalbe, M. L., & Staples, C. L. (1991). Gender differences in sources of self-esteem. Social Psychology Quarterly, 54, 158-168. Shaver, P. R., & Brennan, K. A. (1991). Measures of depression and loneliness. In J. P. Robinson, P. R. Shaver, & L. W. Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 195290). San Diego, CA: Academic Press. Shweder, R. A., & Bourne, E. J. (1984). Does the concept of the person vary cross-culturally? In R. A. Shweder & R. A. LeVine (Eds.), Culture theory: Essays on mind, self, and emotion (pp. 158-199). Cambridge, England: Cambridge University Press. Singelis, T. M. (1994). The measurement of independent and interdependent self-construals. Personality and Social Psychology Bulletin, 20, 580-591. Snodgrass, S. E. (1992), Further effects of role versus gender on interpersonal sensitivity. Journal of Personality and Social Psychology, 62, 154-158. Spence, J. T., Helmreich, R. L., & Stapp, J. (1974). The Personal Attributes Questionnaire: A measure of sex-role stereotypes and masculinity-
808
CROSS, BACON, AND MORRIS
femininity. Journal Supplement Abstract Service Catalog of Selected Documents in Psychology, 4, 43-44. Surrey, J. (1991). The "Self In-relation": A theory of women's development. In J. V. Jordan, A. G. Kaplan, J. B. Miller, I. P. Stiver, & J. L. Suirey (Eds.), Women's Growth in Connection (pp. 51-66). New York: Guilford Press. Swann, W. B., Jr., De La Ronde, C , & Hixon, J. G. (1994). Authenticity and positivity strivings in marriage and courtship. Journal of Personality and Social Psychology, 66, 857-869. Tesser, A. (1988). Toward a self-evaluation maintenance model of social behavior. In L. Berfcowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 181-227). New York: Academic Press. Trafimow, D., Triandis, H. C , & Goto, S. G. (1991). Some tests of the distinction between the private self and the collective self. Journal of Personality and Social Psychology, 60, 649-655. Triandis, H. C. (1989). The self and social behavior in differing cultural contexts. Psychological Review, 96, 506-520.
Triandis, H. C. (1995). Individualism and collectivism. Boulder, CO: Westview Press. Weiss, R. S. (1974). The provisions of social relationships. In Z. Rubin (Ed.), Doing unto others (pp. 17-26). Englewood Cliffs, NJ: Prentice Hall. Yamaguchi, S. (1994). Collectivism among the Japanese: A perspective from the self. In U. Kim, H. C. Triandis, C. Kagitcibasi, S. C. Choi, and G. Yoon (Eds.), Individualism and collectivism (pp. 175-188). Newbury Park, CA: Sage. Yamaguchi, S., KuhJman, D. M., & Sugimori, S. (1995). Personality correlates of allocentric tendencies in individualist and collectivist cultures. Journal of Cross-Cultural Psychology, 26, 658-672. Zuckerman, D. M. (1989). Stress, self-esteem, and mental health: How does gender make a difference? Sex Roles, 20, 429-444. Zung, W. W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70.
Appendix Items Used to Assess Perceptions of One's Partner's Supportiveness (Study 3) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 1
I felt as if my partner really cared about me." My partner behaved warmly toward me.a My partner listened carefully when it was my turn to talk. My partner tried to see things from my point of view. My partner made me feel comfortable about myself and my feelings.3 My partner seemed sensitive to my feelings.8 My partner seemed uncaring. My partner showed respect for my capabilities and talents.0 My partner did not seem to take my concerns seriously.3 My partner seemed sincere during our interaction. My partner made me feel valued as a person. My partner seemed to understand my concerns.
From the Interaction Supportiveness Scale. Copyright 1997 by C. E. Cutrona. Printed with permission.
Received June 28, 1999 Revision received September 2, 1999 Accepted November 2, 1999
