Rejection Sensitivity Moderates the Impact of Rejection on Self-Concept Clarity

Abstract

Self-concept clarity (SCC) refers to the extent to which self-knowledge is clearly and confidently defined, internally consistent, and temporally stable. Research shows that SCC can be undermined by failures in valued goal domains. Because preventing rejection is an important self-relevant goal for people high in rejection sensitivity (RS), it is hypothesized here that failures to attain this goal would cause them to experience diminished SCC. Study 1, an experimental study, showed that high-RS people’s SCC was undermined following rejection but not following an aversive experience unrelated to rejection. Study 2, a daily diary study of couples in relationships, used occurrence of partner conflicts to operationalize rejection. Replicating the findings in Study 1, having a conflict on any given diary day predicted a greater reduction in the SCC of high- compared to low-RS people on the following day. The implications for understanding the conditions under which rejection negatively affects the self-concept are discussed.

It is now widely accepted that the self-concept is relational in nature; its contents, structure, coherence, and associated goals are partly based on the close interpersonal relationships within which it is embedded (Chen, Boucher, & Tapias, 2006). It is assumed that the relational nature of the self develops out of the survival function that close relationships serve for the human infant by providing safety, security, and nurturance (e.g., Hart, Shaver, & Goldenberg, 2005). As such, the needs to secure acceptance and avoid rejection have been deemed among the most central and fundamental motivations to humans (Baumeister & Leary, 1995).

The dependence of the self on acceptance and rejection feedback from significant others implies that the failure to fulfill belongingness needs may have particularly negative consequences for the self-concept. Despite a prolific literature documenting how failure to achieve acceptance lowers self-esteem (Leary, Tambor, Terdal, & Downs, 1995; see Leary, 2006, for review), there has been relatively little emphasis on the potential impact of rejection on other aspects of the self-concept, such as its organization or stability. Therefore, in the present research, we focused on the effect of rejection on self-concept clarity (SCC; J. D. Campbell, 1990; J. D. Campbell et al., 1996), which captures the extent to which self-knowledge is clearly and confidently defined, internally consistent, and temporally stable. SCC has been shown to be an important marker for adaptive functioning. Specifically, relative lack of clarity in the self-concept correlates with global indicators of compromised functioning such as neuroticism, anxiety, and depression (Bigler, Neimeyer, & Brown 2001; J. D. Campbell, 1990; J. D. Campbell, Assanand, & Di Paula, 2003; J. D. Campbell et al., 1996) as well as with aggressive reactions to failure (Stucke & Spore, 2002).

The existing body of work on SCC suggests that the attainment of personally relevant goals contributes to maintaining a sense of consistency and stability of the self (Lavallee & Campbell, 1995). By the same token, failures in valued goal domains seem to be particular vulnerability points for the clarity of the self. For example, in a diary study, Lavallee and Campbell (1995) showed that daily setbacks that impinged on important goals were associated with higher negative affect and lower SCC than setbacks that were not relevant to personally valued goals. Nezlek and Plesko (2001) reported findings demonstrating daily covariations between a wide range of normatively negative events (e.g., rejection, academic failure), elevated negative mood, and lowered SCC over 10 weeks. These studies suggest that rejection may have important implications for SCC. However, the literatures on SCC and on rejection have yet to be systematically linked.

Drawing from these findings, the present research had three primary goals. The first was to focus on rejection as a situational factor contributing to SCC fluctuations. In Nezlek and Plesko’s (2001) study, the occurrence of rejection was one among several daily events (e.g., academic failure) that were combined to assess goal failure. Thus, from these findings it is difficult to isolate the specific relation between experiences of interpersonal rejection and SCC. Second, although previous research has used ecologically meaningful daily diary data to assess covariations between goal failure and SCC, these data were cross-sectional and correlational in nature. Therefore, in the present research we aimed to examine the impact of rejection on SCC using not only experimental methods but also lagged analyses of longitudinal data. Third, in previous research dispositional variables such as self-esteem, depression, negative affectivity, and chronic self-focus did not moderate the relation between goal failure and SCC (Lavallee & Campbell, 1995; Nezlek & Plesko, 2001). However, these studies focused on broad dispositions rather than dispositions particularly relevant for the goal domain at hand. Therefore, there was little correspondence or specificity between the goals that defined the dispositional types under examination and the goal domains in which failures were assessed. Our final goal was to address this limitation by examining the role of rejection sensitivity (RS; Downey & Feldman, 1996)—a personality disposition for which rejection concerns and efforts to avert it are central—in moderating the link between rejection and SCC reductions. Because individual differences in the personal significance of a particular goal can directly determine the intensity of people’s reactions to failures in that domain (e.g., Cantor et al., 1991; Emmons, 1991; Higgins, 1996), we situate our remaining discussion of the rejection–SCC link primarily in the context of RS, first elaborating on the processing dynamics of this disposition and then outlining the rationale for our hypotheses.

PROCESSING DYNAMICS OF RS

The RS model grew out of attachment and attribution accounts of relationship behavior. Its basic tenet is that early experiences of rejection lead people to develop expectations of future rejection, combined with anticipatory anxiety about the possibility of such rejection even before it actualizes (Downey & Feldman, 1996; Downey, Khouri, & Feldman, 1997; Feldman & Downey, 1994). Therefore, anxious expectations of rejection form the core component of high RS both in its conceptualization and measurement (Downey & Feldman, 1996); people who both expect rejection and are highly anxious about its occurrence are referred to as high in RS, whereas those who expect acceptance and are minimally concerned over the possibility of rejection are referred to as low in RS.

Once developed, anxious expectations of rejection affect interpersonal cognition, affect, and behavior in future relationships. More specifically, research shows that high-RS individuals experience heightened activation of the defensive motivational system (DMS) in situations that signal the possibility of rejection (Downey, Mougios, Ayduk, London, & Shoda, 2004). DMS is a domain-general, affectively based physiological system evolved to protect the self against possible threats (e.g., Lang, Bradley, & Cuthbert, 1990; Lang, Davis, & Öhman, 2000) whether these threats are biologically based or socially learned as in the case of high RS. Activation of the DMS is presumed to serve self-protective functions in that it (a) orients the individual to monitor and detect cues that are congruent with the anticipated threat, (b) potentiates action tendencies to avoid or prevent the threat from occurring, and (c) mobilizes resources to enact defensive responses should efforts to prevent threat fail.

Research has documented the operation of each of these three processes in the dynamics of RS. For example, high-RS individuals are more likely to perceive rejection in partners’ ambiguous or somewhat negative behaviors both in their ongoing romantic relationships and in laboratory-based interaction studies (Downey & Feldman, 1996; Downey, Lebolt, Rincon, & Freitas, 1998). Consistent with the notion that activation of DMS motivates people to put effort into preventing the anticipated threat from actualizing, high-RS individuals are also more likely to engage in behaviors aimed at preventing rejection and gaining acceptance. Compared to those low in RS, people high in RS report suppressing their thoughts and feelings to prevent relationship conflicts to a greater degree (Ayduk, May, Downey, & Higgins, 2003) and making more sacrifices to avoid angering or disappointing their partners (Impett, Gable, & Peplau, 2005). Along similar lines, high-RS adolescent girls report greater willingness than their low-RS counterparts to do things that they know are wrong to keep their partners in the relationship (Purdie & Downey, 2000), and high-RS men are more likely to self-ingratiate (i.e., willingness to do tasks no one else wants to do) to gain acceptance from social groups (Romero-Canyas et al., 2009). There is also evidence showing that high-RS individuals engage in defensive behaviors when they perceive rejection to have occurred despite their best efforts to prevent it. When they get into conflicts with romantic partners, high-RS women express greater hostility (both verbally and nonverbally), which reduces their partners’ satisfaction, leading them to consider ending the relationship (Downey, Freitas, Michaelis, & Khouri, 1998). Experiments that deliver rejection in the lab have also demonstrated high-RS participants’ greater tendency to engage in retaliatory, hostile behavior with the intention to hurt when they are led to believe that interaction partners have rejected them (Ayduk, Downey, Testa, Yen, & Shoda, 1999, Study 1; Ayduk, Gyurak, & Luerssen, 2008). Taken together, existing research suggests that rejection is a particularly salient threat for high-RS individuals, and fulfillment of a goal as central as preventing rejection may be critical for maintaining their SCC.

THE CURRENT RESEARCH

What additional processes might be triggered when the efforts of high-RS individuals to prevent rejection fail? Akin to an aspiring Olympic athlete who must suddenly ask herself, “if not an Olympian, who am I?” upon learning she did not make the cut for the national team, high-RS individuals may be especially likely to feel confused and lost about who they are when they experience goal failure—that is, rejection. In addition, goal disruptions increase the accessibility of failed goals (Förster, Liberman, & Higgins, 2005; Liberman, Förster, & Higgins, 2007) and initiate regulatory behaviors aimed at reducing the discrepancies between one’s standards and the actual outcomes experienced (Carver & Scheier, 1981). One way in which high-RS people typically try to prevent rejection is to change themselves, for example, by silencing their thoughts and emotions or subjugating their preferences to those of their partners (Ayduk et al., 2003; Purdie & Downey, 2000). Therefore, when faced with rejection, these regulatory strategies may increase in accessibility and initiate a search for additional ways in which the self can be altered to reduce goal–outcome discrepancy. This process may then further elicit confusion about the self.

Based on this reasoning, and consistent with Person × Situation models of social behavior (e.g., Mischel & Shoda, 1995), we expected the impact of rejection on SCC to depend critically on RS. Specifically, we hypothesized that rejection should be a potent force that undermines SCC among high-RS individuals. Furthermore, we anticipated that because low-RS individuals do not define their self around the goal of preventing rejection, failures in this domain should have a less pronounced or no effect on their SCC.

We tested these hypotheses in two studies. In Study 1 we manipulated the occurrence of rejection in the lab to directly test the notion that rejection leads to dips in SCC, particularly in high-RS individuals. In Study 2, we aimed to replicate these findings in a more ecologically valid setting using a daily diary study of couples that assessed daily fluctuations in SCC following the occurrence of conflicts with romantic partners, which is known to covary with feelings of rejection.

In examining the joint effect of rejection and RS in explaining reduced SCC, we employed several controls to rule out alternative explanations. First, prior research has shown that high-RS people tend to get more depressed following rejection (Ayduk, Mischel, & Downey, 2002). Because SCC also significantly covaries with depression (e.g., Lavallee & Campbell, 1995; Nezlek & Plesko, 2001), we controlled for depressive symptomatology (Studies 1 and 2) and depressive mood (Study 2) to establish the relation between RS and SCC independent of depression. Second, to more unequivocally demonstrate the specificity of rejection as an elicitor of SCC for high-RS people, we also examined the role of noninterpersonal stressors (i.e., academic work overload) in eliciting SCC fluctuations in high-RS people (Study 2).

STUDY 1

Method

Participants and Procedure

This study reports previously unpublished findings from the sample described in Ayduk et al., 2008.¹ One hundred and twenty-nine undergraduates (72.87% women) participated in a two-session study for course credit or money (age: M = 21.66 years, SD = 5.27 years). The sample was 54.1% Asian, 2.46% African American, 3.28% Hispanic, 28.54% White, and 11.62% Other or mixed race/ethnicity.

Participants completed the Rejection Sensitivity Questionnaire (RSQ; Downey & Feldman, 1996) and the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) in an initial survey session, which was followed by the experimental session 1 to 2 weeks later. In the experimental session, two same-sex participants were run at a time to make the cover story more credible as well as to save resources. Upon arriving at the laboratory, participants received the instructions for the study together. The experiment was described as a study on determinants of partner selection during online interactions. Participants were told that a third participant named Alex was also scheduled for the same session and that Alex would choose which of the two participants to interact with after the participants wrote a short biosketch about themselves. Alex was always referred to by the gender pronoun opposite to participants’ sex to prime a dating context. Participants were told that whoever was chosen would do a 15-min online chat with Alex, whereas the person not chosen would complete other tasks. At this point, the participants were taken to different locations and completed the rest of the study individually. When a second participant could not be scheduled, participants were run with a same-sex confederate (40% of the cases; see Ayduk et al., 2008, for details); however, this factor did not affect any of the results.

Subsequently, the experimenter instructed participants to type their biosketch on the computer for about 10 min, including as much or as little personal detail as they wished. At the end of 10 min, the experimenter purportedly sent the participants’ biosketch file to Alex via e-mail and told participants to wait about 5 min for Alex to read both biosketches and make a decision. They were told that Alex would e-mail them back his/her response. In the meantime, the experimenter went to a separate room and randomly assigned each participant either to the rejection or the control condition; thus, the experimental conditions of the two participants were independent of one another. In the rejection condition (n = 64), participants received an e-mail purportedly from Alex (which was in fact sent by the experimenter) informing them that they were not chosen for the interaction. In the control condition (n = 59), after remotely disconnecting the Internet connection from the participants’ computer, the experimenter went back to the participants’ room and told them that the online interaction would not take place because the Internet system in the lab was down. To add credibility to the cover story, when the experimenter tried to open the e-mail account, it failed to upload. Immediately after the manipulation, participants completed the Self-Concept Clarity Scale (SCCS; J. D. Campbell et al., 1996) among other measures. Participants were then taken to another room and completed other tasks unrelated to this study. Finally, they were debriefed, probed for suspicion, thanked, and dismissed. Participants largely indicated that they believed the manipulations, and suspicion ratings were not different as a function of experimental condition (see Ayduk et al., 2008, for details). Six respondents had incomplete data on key variables for this study due to computer problems; therefore, the results reported here are based on 123 participants.

Measures

RSQ

The RSQ measures the degree to which people expect rejection and are anxious about its occurrence (see Downey & Feldman, 1996, for details). The measure includes 18 hypothetical interpersonal situations that afford the possibility of rejection (e.g., you ask someone you do not know well out on a date; you ask your boyfriend/girlfriend if he/she really loves you). For each situation, participants indicate their level of anxiety about the possibility of a negative outcome (1 = not anxious, 6 = very anxious) and perceived likelihood of acceptance (1 = very likely, 6 = very unlikely). The latter is reverse scored to index expectations of rejection then multiplied by level of anxiety for each situation. The multiplicative terms are averaged across the 18 scenarios to index overall levels of anxious expectations of rejection. The possible scores range from 1 to 36. The measure is internally reliable (Downey & Feldman, 1996) and shows good predictive utility (see Pietrzak, Downey, & Ayduk, 2005, for a review).

In this sample the mean RSQ score was 9.41 (SD = 3.30, α = .88). RSQ scores did not differ by experimental group or by participants’ sex (ts < 1).

BDI

This is a widely used 21-item measure in which participants rate their experience of affective, cognitive, and behavioral symptoms of depression in the past week on a 4-point scale (0–3). Ratings were summed to create a composite score (M = 5.10, SD = 4.59) with no significant sex differences (t < 1). RS and BDI scores were significantly correlated, r(121) = .54, p < .001.

SCCS

The SCCS was originally designed as a trait measure of SCC (see J. D. Campbell et al., 1996, for details). Because Study 1 was an experimental study we needed a measure that would be sensitive to capturing temporary fluctuations in SCC in response to situational factors manipulated in the lab. We addressed this issue in two ways. First, in preliminary work, we asked two expert personality psychologists among the faculty at University of California, Berkeley to rate the 12 items of the SCCS on the extent to which they referred to trait-level constructs that were highly stable versus state-level constructs that could be moved around temporarily by situational cues (1 = very trait like, highly stable, 4 = has elements of both, 7 = highly state like, very unstable). The interjudge agreement was high (α = .87) and ratings were averaged. There were 6 items that were rated by the judges to have state-like qualities (with mean ratings of 4 and above): “I spend a lot of time wondering about what kind of person I really am”; “Sometimes I feel that I am not really the person that I appear to be”; “When I think about the kind of person I have been in the past, I’m not sure what I was really like”; “My beliefs about myself seem to change frequently”; “Even if I wanted to, I don’t think I would tell someone what I’m really like”; and “Generally I have a clear sense of who I am and what I am.” Second, prior research (Fleeson, 2001) demonstrates that state fluctuations can be readily captured in stable personality traits when participants are asked to complete relevant questionnaires under “state” instructions (how extraverted were you within the last hour?). Therefore, following the experimental manipulation, participants in this study were asked to rate their agreement (1 = strongly disagree, 5 = strongly agree) with these items with respect to their current thoughts and feelings.

Agreement ratings across the six items were averaged to create a composite SCC score, with higher scores indicating greater SCC. The mean in the present sample was 3.37 (SD = .84, α = .87) with no significant sex differences, F(1, 121) = 1.04, p = .31.

Results

In preliminary analyses neither sex nor ethnicity was a predictor of SCC either alone or in interaction with any of the key predictors; therefore, these variables are not discussed further. The main analyses were conducted using the generalized linear model (GLM) procedure in the SAS statistical package with RS and BDI scores centered on their grand mean (Aiken & West, 1991). RS (continuous), experimental condition (0 = control vs. 1 = rejection), and the interaction between them were entered as between-subjects predictors of SCC with BDI (continuous) entered as a covariate. Following the omnibus analysis, two sets of simple slopes analyses were conducted with one set examining the simple slope of RS in each experimental condition and the other examining the simple slope of experimental condition among low-RS (1 SD below the mean on the RSQ) and high-RS (1 SD above the mean on the RSQ) individuals.

Consistent with previous research, results revealed that BDI was a significant predictor of SCC, F(1, 118) = 11.97, p = .0008, β = −.32. Neither RS nor experimental condition had significant main effects (Fs < 1). As predicted, however, a significant interaction between RS and experimental condition emerged, F(1, 118) = 5.11, p = .026, which is illustrated in Figure 1 based on the parameter estimates obtained from this analysis.

Figure 1.

Self-concept clarity as a function of rejection sensitivity (RS) and experimental manipulation.

NOTE: Higher numbers indicate more self-concept clarity. Predicted values were based on the following parameter estimates:

$$\mathrm{SCC}=3.42-.058(\mathrm{BDI})-.009(\mathrm{RS})-.962(\text{experimental condition})-.0922(\mathrm{RS}\times \text{Experimental Condition})$$

, where SCC = self-concept clarity; BDI = Beck Depression Inventory.

To further understand this interaction, simple slope analyses (Aiken & West, 1991) were conducted as described previously. We first examined the slope of RS in the rejection and in the control conditions separately in the context of the whole sample. Results indicated that RS was negatively related to SCC in the rejection condition, t(119) = −3.48, p < .0007, β = −.39, but not in the control condition, t < 1, β = −.04. Next, we examined the slope of experimental condition among high-and low-RS individuals. These analyses revealed that exposure to rejection (compared to the control) condition was associated with lower SCC among high-RS individuals, t(119) = −2.26, p < .03, β = −.24, but not among low-RS individuals, t(119) = 1, β = .12.

Discussion

The findings from Study 1 showed that high-RS people experienced lower SCC than low-RS people when faced with rejection, whereas no differences in SCC occurred as a function of RS when an opportunity to have a potentially positive interpersonal relationship was removed for reasons unrelated to rejection. These results support a causal relation between social rejection and SCC decreases in vulnerable, high-RS individuals. Because the analysis controlled for BDI scores, the present results also suggest that the effect of rejection in eliciting lower SCC in high-RS individuals is independent of the covariation that occurs among SCC, depression, and RS.

STUDY 2

The overarching aim of Study 2 was to examine our hypotheses in the context of people’s real-world, ongoing personal relationships. By using data from a 3-week daily diary study of dating couples, we aimed not only to provide conceptual replications of the findings from Study 1 but also to establish their ecological validity. Conflicts with partners covary with reduced relationship satisfaction (e.g., L. Campbell, Simpson, Boldry, & Kashy, 2005) and have been found to elicit considerable levels of hurt feelings (Sanford, 2007). These findings suggest that in the day-to-day interactions of couples, conflicts may be important elicitors of rejected feelings. Additionally, behavioral, affective, and physiological reactions during and after conflicts seem to have distinctive utility in predicting important relationship outcomes (e.g., Gottman, 1979, 1993), suggesting that they may also have important implications for SCC. Therefore, in the current study we validated and then used the daily occurrence of conflicts with partners to operationalize rejection.

The longitudinal design afforded by a diary study allowed us to assess whether the within-subjects changes in SCC that occur in response to relationship conflicts were different for low- and high-RS individuals. Because conflicts should represent failures in a domain more central to the self-concept of high- than low-RS people, we predicted that perceived occurrence of conflicts would elicit greater confusion about the self to the degree that people scored high on RS. In testing this hypothesis, we controlled for BDI scores as well as daily depressive mood to demonstrate that the RS–SCC relation is independent of the association that exists between RS and depression.

What relation should one expect between conflict and SCC among low-RS people? In Study 1, low-RS people’s SCC was not affected by the rejection experience they were exposed to in the lab. This was an experimental situation in which participants had neither any prior relationship with the rejector nor an opportunity to establish one in the future. Therefore, low-RS people’s lack of reactivity in Study 1 was not surprising. However, several trends of evidence point to the possibility that conflicts in the context of ongoing, real-life relationships can serve an adaptive purpose for low-RS people. For example, conflicts do not reduce partner satisfaction in low-RS people’s relationships like they do for partners of high-RS individuals (Ayduk et al., 2003; Downey, Feldman, & Ayduk, 2000; Downey, Freitas, et al., 1998). Along similar lines, Murray, Griffin, Rose, and Bellavia (2003) found that individuals who report feeling valued by their partner reinforce relationships ties and actually draw closer to their partners following perceived hurt and conflict. These findings are consistent with Gottman and Krokoff’s (1989) observational studies that found expressing disagreement and even anger may increase relationship satisfaction in the long run when conflicts are not indicative of defensiveness, stubbornness, or withdrawal. Therefore, we speculated that because conflicts occur in the backdrop of relatively well-functioning relationships for low-RS individuals and are characterized by less hostility and withdrawal in general, they may provide contexts in which self-relevant values and goals are articulated and affirmed, leading to greater SCC for those low in RS.

Additionally, Study 2 provided us with an opportunity to examine the specificity of rejection per se, rather than negative events in general, as an elicitor of self-concept confusion in high-RS people. In her work on life tasks, Cantor identified doing well academically as one of the most frequently pursued goals by college students (Cantor et al., 1991). Although not a direct measure of academic failure, the daily diary questionnaire included a question on daily occurrence of academic work overload, which represents potential threats to academic success. Furthermore, chronic work overload has been linked to increased levels of morning cortisol levels, suggesting that academic work overload is likely to be a significant negative stressor in college students’ lives (Schulz, Kirschbaum, Prüsner, & Hellhammer, 1998). We hypothesized that the perceived presence of this stressor would not increase high-RS people’s vulnerability to experience lower SCC because although they may be normatively significant goals for college students in general, they are not central concerns around which high-RS people organize their self-concept.

Method

Participants

Monogamous, nonmarried, English-speaking dating couples who had been together for at least 3 months were recruited through posters posted around the University of California, Berkeley campus to participate for pay in a study on romantic relationships. Fifty-three heterosexual couples participated in the study. Average response rate was 91.70% of diary days with no significant sex differences (t < 1). On average, less than 0.5% of the daily diaries had to be eliminated because responses were submitted outside of the time window specified. The mean age of the participants was 20.58 years (SD = 2.43 years). Couples had been in relationships for an average of 16.52 months (SD = 14.37). Two participants did not provide racial or ethnic information. In the rest of the sample, 46.15% of the participants were Asian, 35.58% Caucasian, 17.31% Other. Ethnically, 9.71% indicated being Hispanic. Same-race relationships accounted for 63.46% of the sample (33.33% Caucasian, 57.58% Asian, 9.09% Hispanic).

Procedure

The sample was recruited from the undergraduate population for a 21-day (3-week) diary and a 2-hr laboratory study. Couples completed background questionnaires online before the laboratory session. During the lab session, couples talked about their relationship and completed various other measures that were not the focus of the current study. Members of the couple were paid $30 each for the completion of the background questionnaires and the laboratory session.

At the end of the lab assessment, participants received a Web link to the structured daily diary questionnaires that they were to complete at the end of each day for 21 days. Participants were asked to complete the diary questionnaires separately from their partner and to refrain from discussing their responses until the study ended. Electronic mail was sent to participants each evening to remind them to complete the online questionnaires. Participants entered their questionnaire responses online every day between 6 p.m. and 3 a.m. the following day. Questionnaire submissions were time stamped electronically, and participants could not go back and modify their responses. Upon completion of the 21 diaries, each member of the couple received payment for the full $55 or a prorated amount.

Background Measures

The background measures included the RSQ, the BDI, global measures of relationship satisfaction and commitment, and several questionnaires unrelated to the purposes of this study. For correlations among key Study 2 variables, see Table 1.

TABLE 1.

Zero-Order Correlations Among Study 2 Variables

	1	2	3	4	5
1. Daily conflict	—
2. Daily work overload	.25**	—
3. Daily SCC	.06	.08	—
4. Daily depressive mood	.21*	.25**	−.42***	—
5. Rejection sensitivity	.13	.14	−.33**	.38***	—
6. Depression (BDI)	.11	.19*	−.34**	.43***	.40***

NOTE: Daily variables have been averaged across the diary period for each individual. SCC = self-concept clarity; BDI = Beck Depression Inventory.

^*p ≤ .05.

^**p ≤ .01.

^***p ≤ .0001.

RSQ

Same as Study 1. In this sample, the mean RSQ score was 9.21 (SD = 2.92, α = .81). There were no significant sex differences in RSQ scores (t < 1). Partners’ RSQ scores were not significantly correlated, r(52) = .21, p = .12.

BDI

Same as Study 1. The mean BDI score in this sample was 6.42 (SD = 5.02) with no significant sex differences (t < 1). Partners’ scores were not significantly correlated, r(52) = .20, p = .13.

Diary Measures

The structured daily diary included questions about the couple’s daily interactions, occurrence of conflict and conflict behavior, reports on partner’s behavior, daily mood, the self-concept, and noninterpersonal stressors and goals. The specific focus of the current study was on the variables described next.

Perceived occurrence of conflicts with partners

Each diary day, participants indicated whether they experienced a conflict with their partners (yes = 1, no = 0). Of the 53 couples, 45 reported having at least one conflict with their partner during the 3-week period. Conflicts occurred on 11.39% of the diary days. Members of a couple agreed about whether a conflict had occurred on 91.1% of the days, with a kappa coefficient of .58.

Consistent with prior findings (Ayduk et al., 2003; Downey, Freitas, et al., 1998), conflict rates over the diary period were not significantly related to RS in the sample (t < 1). This result is expected because in relatively committed relationships, high-RS people are motivated to avoid conflicts that provide contexts for rejection (Ayduk et al., 2003). Higher BDI scores were also not related to conflict frequency across the diary period, t(52) = 1.64, p = .12, b = .002.

To validate our assumption that conflicts indexed rejection, we examined the daily covariation between occurrence of conflicts and a perceived partner rejection index (“My partner did/said things that hurt me and made me feel not cared for” and “My partner acted in a hostile manner towards me”; α = .78, M = 1.30, SD = .68). As expected, participants reported greater perceived partner rejection on days when conflicts occurred than on days when they did not occur, F(1, 52) = 209.07, p < .0001, b = .96. This difference was not significantly moderated by RS (F < 1). Furthermore, in preliminary analyses, using perceived rejection ratings instead of conflict occurrence yielded results highly similar to the findings reported here.²

Perceived occurrence of academic work overload

Each day, participants indicated if they felt overloaded with academic responsibilities and tasks due (yes = 1, no = 0). Work overload occurred on 51.94% of the diary days, and RS was not significantly related to the rate of experiencing this stressor over the diary period, t(52) = 1.29, p = .20, b = .008.

Daily depressive mood

Participants rated their daily mood (1 = not at all, 5 = extremely), and their ratings on the items “depressed” and “happy” (reverse scored) were averaged to index daily depressive mood (α = .60, M = 2.06, SD = .74). Zero-order correlations reported in Table 2 indicate that depressed mood was positively associated with the frequency of both conflict and academic workload across the diary period, supporting the notion that both are negative events in nature.

Daily SCC

Participants indicated their daily level of SCC by rating themselves (1 = not at all, 5 = extremely) on the item “Today I felt like I had a clear sense of who I am and what I want in life” (M = 2.54, SD = 1.13). This item was created by modifying and combining two items from the SCCS (Campbell et al., 1996; items 11 and 12) that had face validity in tapping into having a clear idea of who the person is and the goals the person wants to pursue.

Diary Data Analyses

The diary data involved a hierarchical structure where participants were nested within couples, and days of assessment were nested within participants. For each couple, this structure represented a two-level model and required the simultaneous analysis of within-person and between-person levels that are hierarchically organized (see Laurenceau & Bolger, 2005, for a detailed discussion of a two-level analysis of diary data in couples).² The analyses were conducted using the mixed procedure in the SAS statistical package, which is based on a hierarchical linear model approach and permits the simultaneous analysis of within- and between-person variation (Kenny, Kashy, & Bolger, 1998). For each member of a couple, the lower level within-person analysis generated estimates of the relationship among daily measures (e.g., the association between occurrence of conflicts and SCC). The higher level between-person analyses then examined whether these within-person processes were a function of between-subjects differences in RS.

In addressing whether conflicts or academic workload overload reduce high-RS people’s SCC to a greater degree, we focused on associations with a 1-day lag to make use of the longitudinal nature of the data and to more directly explore causal direction of effects with respect to our hypothesis. That is, our statistical models tested whether the within-subjects relationships between previous day’s conflict or academic work overload and today’s SCC was moderated by our between-subjects predictor, RS, while also controlling for previous day’s SCC. These analyses can be interpreted as testing whether the change in SCC from the conflict day to the next was different for high- and low-RS individuals.

All analyses assumed an error structure allowing for contemporaneous (same-day) dependence between the errors within a couple and a first-order autoregressive structure within a person in a couple. In addition, variances were allowed to differ between males and females. To adopt a conservative approach to significance testing, we used the number of couples to compute degrees of freedom. All continuous predictors were centered on their grand mean (Aiken & West, 1991) with perceived occurrence of conflicts and work overload coded as 0 (absence) versus 1 (presence).

Results

Preliminary Analysis

In preliminary analyses we controlled for partners’ RS as an additional covariate in the main analyses reported next following the recommendation of the actor-partner interdependence model (L. Campbell et al., 2005; Kashy & Kenny, 2000). Partners’ RS did not have a significant effect on participants’ SCC in any of the models, and the main findings reported next remained the same. Preliminary analyses also tested for sex differences (where sex was effect coded as −1, 1). Again, sex did not predict SCC in any of the models, neither did it moderate the interaction between RS and different stressors (i.e., conflict, workload). Therefore, we report results without these variables.

Do High-RS People Report Lower Average Levels of SCC Across the Diary Period?

The Proc mixed analysis was run to estimate the relation between RS and average levels of SCC across the diary days while controlling for BDI and average levels of daily depressive mood across 21 days. These analyses revealed depressive mood, F(1, 51) = 11.56, p < .0001, b = −.35, and RS, F(1, 51) = 3.55, p < .0008, b = −.06, but not BDI (F < 1) to be significant predictors of SCC.

Is High-RS People’s SCC Contingent on Perceived Occurrence of Conflicts?

Mixed analysis was conducted on SCC with previous day’s conflict (lagged value of occurrence of conflicts), RS, and their interaction as predictors. Lagged value of SCC was included as a covariate to control for any effect of previous day’s SCC on today’s SCC. In addition, we controlled for both BDI and same-day depressive mood to isolate the role of RS in regulating daily SCC independent of depression.

We found a significant effect of same-day depressive mood, F(1, 51) = 45.88, p < .0001, and RS, F(1, 51) = 7.48, p < .0001, in predicting SCC. Neither BDI nor previous day’s conflict was a significant predictor of SCC (Fs < 1). However, consistent with our hypothesis, there was a significant RS × Previous Day’s Conflict interaction, F(1, 51) = 3.84, p = .055, b = −.043.³ Figure 2 illustrates the interaction based on the estimates obtained from the mixed analysis.

Figure 2.

Self-concept clarity as a function of rejection sensitivity (RS) and previous day’s conflict.

NOTE: Higher numbers indicate more self-concept clarity. Predicted values were based on the following parameter estimates:

$$\mathrm{SCC}=2.50+.15(\text{previous day’s SCC})-.37(\text{same-day depressive mood})-.003(\mathrm{BDI})-.043(\mathrm{RS})-.001(\text{previous day’s conflict})-.043(\mathrm{RS}\times \text{Previous Day’s Conflict})$$

, where SCC = self-concept clarity; BDI = Beck Depression Inventory.

We further probed this interaction pattern by conducting simple slopes analyses. First, we examined the slope of RS in predicting SCC on days following conflicts versus no conflicts. These analyses revealed that high-RS individuals reported significantly lower SCC than low-RS individuals on days preceded by conflicts, t(51) = −3.94, p < .0002, b = −.086. RS was also negatively related to SCC on days not preceded by conflicts, t(51) = −2.74, p < .0085, b = −.043, but as indicated by the significant interaction term, the link between RS and SCC was twice as strong on days following conflicts than on days not following conflicts.

Next we examined the effect of occurrence of conflicts in reducing SCC for people at 1 SD above and below the mean on the RSQ. Among high-RS participants, occurrence of conflicts was negatively related to SCC, even though this relation was not statistically significant, t(51) = −1.31, p = .19, b = −.13. Among low-RS participants, the relation between occurrence of conflicts and SCC was in the opposite direction but not significant, t(51) = 1.21, p = .23, b = .12.

Is High-RS People’s SCC Contingent on Perceived Presence of Other Stressors?

To further examine the specificity of partner conflicts in eliciting greater SCC reductions in high-RS people, we repeated our initial analysis, replacing perceived occurrence of partner conflicts with perceived occurrence of work overload on the previous day. SCC was predicted by same-day depressive mood, F(1, 51) = 43.65, p < .0001, and RS, F(1, 51) = 10.57, p < .0001, but not by BDI (F < 1), the presence/absence of work overload on the previous day, F(1, 51) = 2.05, p = .16, or the RS × Work Overload interaction (F < 1). Thus, it appears that only relationship conflict interacts with RS in predicting SCC changes.

Discussion

Findings from Study 2 revealed that although high-RS people showed a tendency to experience less SCC than low-RS people even in the absence of conflicts, the differences between low- and high-RS people were more pronounced when conflicts with partners had occurred the day before. Furthermore, by controlling for daily depressive mood and BDI, these results established SCC as an outcome that is distinct from both depressive mood states and more chronic depressive symptoms. The results from Study 2 also underscored the specific role conflicts play in accentuating high-RS people’s confusion with their self-concept since noninterpersonal stressors such as work overload did not differentially affect high- and low-RS individuals’ SCC.

Despite the similarities in the pattern of findings across the two studies, there was a significant relation between RS and SCC even on days that were not preceded by conflicts in Study 2, whereas RS was not related to SCC in the absence of rejection (i.e., control condition) in Study 1. There are several possible explanations. First, it is possible that high-RS participants in the control condition in Study 1 had momentary certainty that rejection was not a possibility because they were explicitly told that the interaction would not take place because of computer problems. This certainty, in turn, may have buffered their SCC. Another possibility is that the RS–SCC relation observed in the diary study reflects the cumulative effect of past and ongoing interpersonal experiences and therefore captures, at least in part, the chronic SCC differences between high- and low-RS individuals. In fact, in the current study, RS significantly predicted average levels of SCC across the diary period, F(1, 51) = 3.55, p < .0008, b = −.06, controlling for BDI and average levels of depressed mood.

Study 2 findings also suggest that conflicts may have a more clarifying effect on low-RS individuals’ self-concept compared to the effect of conflicts on high-RS individuals’ self-concept. Although we had anticipated this possibility based on the literature showing that conflicts and disagreements have beneficial effects on satisfaction when they occur in the backdrop of well-functioning relationships, the exact mechanisms through which this may operate needs further examination. One possibility that we highlighted earlier is that low-RS people use conflicts as an opportunity to articulate their point of view, beliefs, values, and so on, and therefore come out of conflicts feeling more sure about who they are. Low-RS individuals may also receive more affirmation from their partner because they tend to be less destructive in conflicts. However, because the simple slope of conflict among low-RS individuals was not statistically different from zero, these possibilities should be evaluated with caution and investigated further in future research.

GENERAL DISCUSSION

In two studies we provided evidence that interpersonal rejection results in SCC decreases in high-RS people. Using an experimental design, Study 1 demonstrated the role rejection plays in eliciting lower SCC among high-RS people. Study 2 replicated these findings using a daily diary study design. Study 2 showed that in ongoing relationships, occurrence of conflicts with partners, but not occurrence of noninterpersonal stressors, accentuated the relation between high RS and low SCC. Importantly, these effects were not attributable to depression, and they seemed to be specific to failure in the domain of rejection. The combination of the experimental data from Study 1 and the longitudinal data from Study 1 provides evidence in favor of the causal role rejection plays in reducing high-RS people’s SCC. Additionally, given that Study 2 drew daily samples from the participants’ ongoing romantic relationships, our data suggest that these processes generalize outside the laboratory setting.

It is important to underscore the benefits of having examined the rejection–SCC link in the context of RS. We did not find a main effect of rejection on SCC in either study; rather, rejection had a detrimental effect on SCC only among high-RS individuals. As such, these findings draw attention to the utility of examining SCC fluctuations in response to goal disruptions that specifically occur in domains that are self-defining. In addition, the findings presented here inform our general understanding of the conditions under which rejection leads to SCC fluctuations and the conditions under which it does not.

Implications of SCC for Other Negative Outcomes Elicited by Rejection

The present findings also further our understanding of RS dynamics. A long history of research suggests that people high in RS are at greater risk for showing maladaptive affect and behavioral responses in interpersonal relationships including depression, anger, jealousy, and aggressive and controlling behavior (Ayduk, Downey, & Kim, 2001; Ayduk et al., 1999; Ayduk et al., 2003; Downey & Feldman, 1996). In addition to outcomes that occur in relationship contexts, RS has also been linked to outcomes that are not relationship specific per se, for example, academic difficulties and drug use (Ayduk et al., 2000; Downey, Lebolt, et al., 1998).

Before the present research, it was assumed that such outcomes likely came about primarily through interpersonal difficulties, as, for example, when a high-RS student does not do well in a class because he or she is afraid to approach a professor. However, the current research suggests that there may be an alternative mechanism—SCC—that can help explain the occurrence of such outcomes for high-RS people. For example, as a result of not having a clear sense of who one is and where one is going, a high-RS person may lose interest in a class that he or she was initially excited about and/or have difficulty finding internal motivation to keep up with school-work and thus slip academically. It will be important for future research to more directly examine the mediating role of SCC in explaining high-RS individuals’ vulnerability both for relationship outcomes and for conduct problems and personal difficulties.

Understanding the effect of rejection on SCC may shed additional light on some of the mechanisms that underlie violent reactions to rejection. Leary, Kowalski, Smith, and Phillips (2003) note that school shootings of the last decade have often been precipitated by an acute rejection experience (e.g., breakup with a girlfriend) that occurred in the context of more chronic social isolation or bullying. Reasons that have been proposed for this link included self-esteem repair and a gaining of respect. The current findings suggest that attempts to repair SCC might also have been a contributing factor. For example, some of the perpetrators seemed to be motivated by a desire to create an identity that would transcend their physical existence by the atrocity of their crimes. Furthermore, the copycat nature of some incidents (e.g., the Columbine shooters enacting scenes from The Matrix) may attest to a search for solid, readily accessible identities when SCC is compromised as a result of rejection, exclusion, and bullying. These possibilities underscore the importance of well-established, clear identities among youth who are at risk for engaging in extreme violence.

Caveats and Conclusions

We acknowledge several limitations to this research. First, although in Study 2 we compared the specificity of rejection (i.e., relationship conflict) with academic stress in eliciting SCC reductions among high-RS individuals, the results should be replicated with a more direct measure of academic failure (e.g., failing an exam). Second, because of space constraints, we measured daily SCC by a single item in Study 2 that focused on people’s clarity about knowing themselves and the goals they are pursuing. Although psychometric analyses of the 12-item SCCS from which this item was drawn indicate a single-factor structure, it is important to replicate Study 2 findings with additional SCC items to make sure the results generalize to other features of SCC. For example, future research should test whether a sense of continuity between past and present selves changes following rejection (e.g., “When thinking about the kind of person I was in the past, I am not sure what I was really like”). Furthermore, the present research focused exclusively on rejection from actual or potential romantic partners. Establishing the generalizability of the findings to rejection by significant others such as parents or friends awaits future research. Finally, this research assessed RS as a moderator of reactions to interpersonal rejection. However, conceptually similar individual difference variables (e.g., insecure attachment) should be tested in future research to further evaluate the predictive power of RS.

Despite these caveats, the current research demonstrated that setbacks associated with rejection of belongingness strivings are potent elicitors of confusion about the self-concept among vulnerable high-RS people. The results presented here have importance for mental health, particularly in adolescent and young adult populations who go through a critical period of identity development at the same time that the negotiation of interpersonal relationships (independence from parents, establishing romantic relationships) becomes an important life task.