“You’re Just Saying That.” Contingencies of Self-Worth, Suspicion, and Authenticity in the Interpersonal Affirmation Process

Abstract

A model of the role and costs of contingent self-worth in the partner-affirmation process was tested. Actors whose self-worth was contingent on appearance or intelligence claimed to have expressed their particular heightened sensitivity to their romantic partners. Suggesting a cost to these reactions, actors’ beliefs about having expressed heightened sensitivity, in turn, predicted their doubts about the authenticity of partners’ positive feedback in the domain of contingency, independently of whether partners claimed to deliver inauthentic feedback. Suggesting a cost for partners, partners of contingent actors appeared to detect actors’ expressions of sensitivity in the domain of contingency and respond by delivering inauthentic feedback to actors in the domain, which in turn predicted partners’ increased relationship anxiety and decreased satisfaction. Results suggest that contingent self-worth may undermine the functioning of the partner-affirmation process through actors discrediting partners’ positive feedback and partners behaving in an inauthentic and controlled manner.

People depend on their romantic partners to meet their needs (Clark & Mills, 1993; Reis, Clark, & Holmes, 2004). Among these needs is the need to feel good about the self; people want their partners to see them in a manner that affirms their feelings of self-worth (Murray, Holmes, & Griffin, 1996). In many cases, the process by which a partner affirms our feelings of self-worth happens easily, as when a partner regularly delivers positive feedback or dispels doubts about our inadequacies. In turn, partners can feel good about themselves and their relationships through providing this support (Clark & Grote, 1998).

However, this process does not work well for everyone. Some people, even those with loving and admiring partners, do not feel adequately valued by their partners and often react to their own failures with relationship-damaging behaviors (Crocker & Park, 2004; Murray, Holmes, & Collins, 2006). How does the partner affirmation process go awry? In the current research, we test the role of contingencies of self-worth as psychological vulnerabilities that may undermine the partner-affirmation process. In particular, by expressing heightened emotional vulnerabilities and sensitivities in a particular domain to partners, contingent individuals may come to doubt the authenticity of partners’ positive feedback in that domain and have partners who feel compelled to provide inauthentic feedback relevant to performance in that domain. Our model is illustrated in Figure 1. We discuss each path in detail below.

Figure 1.

Model of Contingencies of Self-Worth, Partner Authenticity, and Authenticity Doubts

Path A: Contingent Actors Believe They Expressed Self-Worth Sensitivity to Partners

For people who have self-worth that is contingent on performance in a particular domain, successes and failures in that domain generalize to their felt worth as a person (Crocker & Wolfe, 2001). They feel valuable when they succeed in the domain and worthless when they fail (Crocker, Karpinski, Quinn, & Chase, 2003). Contingent self-worth also motivates people to strive for success and avoid failure in the domain as a means of validating their worth (Crocker & Park, 2004; Crocker & Wolfe, 2001).

Due to this emotional vulnerability and pursuit of self-esteem, those who have highly contingent self-worth (termed “actors” in our model) may frequently express their particular self-esteem sensitivity to their romantic partners. They are likely to do so because people generally depend on romantic partners to meet their needs, including self-esteem needs (Drigotas & Rusbult, 1992; Murray, Holmes, & Griffin, 2000; Reis et al., 2004). This expression of sensitivity may include a variety of specific behaviors, such as seeking reassurance about their standing in the relevant domain (Joiner, Katz, & Lew, 1999) and responding to threats in the domain with hurt feelings or antagonism (Crocker & Park, 2004; Murray et al., 2006; Park & Crocker, 2005). For example, Sarah, having a sense of self-worth that is contingent on her physical appearance, may frequently ask Derek, her partner, whether she looks fat, and she may become angry and defensive when he suggests sharing a low calorie meal.

These responses may have provided some emotional gratification at the time they occurred. For instance, having one’s request for reassurance be met with a partner’s reassuring response likely provides a boost to feelings of self-worth. Reacting to threat with hostility might provide a temporary sense of safety (Murray et al., 2006) or vindication (Leary, Twenge, & Quinlivan, 2006). However, later, reflecting on these prior reactions might cause actors to come to the conclusion that they have expressed their heightened sensitivity to partners and, as described later, this belief may come at a cost. That is, our model posits a particular means by which these responses might entail a trade-off between short-term emotional gratification and long-term well-being (see also Baumeister & Scher, 1988; Crocker & Park, 2004).

Path B: Partners Detect Actors’ Sensitivity

Contingent actors’ beliefs about expressing sensitivities should, to some extent, reflect their actual behavior. As a result, partners should, to some extent, agree with actors’ claims that they expressed sensitivity. Partners may be especially likely to detect this information because it conveys information about actors’ needs, and people typically monitor the other’s needs in close relationships (Clark, Mills, & Powell, 1986). Although we do not know of evidence that people are aware of others’ specific contingencies of self-worth, people do seem to detect their partners’ expressions of general sensitivity, including reassurance seeking (Shaver, Schachner, & Mikulincer, 2005) and hostile reactions to negative evaluation (Downey, Freitas, Michaelis, & Khouri, 1998; Murray, Bellavia, Rose, & Griffin, 2003). Continuing with our example, Derek likely detects Sarah’s need for reassurance and her emotional vulnerability regarding attractiveness.

Path C: Partners Who Detect Actors’ Sensitivity Deliver Inauthentic Feedback in the Domain

Partners who perceive actors as especially emotionally dependent on feedback in a particular domain are likely to respond by “walking on eggshells” when providing such feedback, including cautiously concealing negative evaluations and exaggerating positive evaluations. This is likely to be the case for both other-oriented and selfish reasons. In terms of other-oriented reasons, people usually care about their partners’ needs (Reis et al., 2004), including their emotional well-being. As such, they often tell altruistic lies to benefit close partners (DePaulo & Kashy, 1998), especially when partners appear emotionally invested in the performance domain (DePaulo & Bell, 1996). In terms of selfish reasons, people want happy relationships and they wish to avoid being the target of a partner’s anger. The consequences of delivering less-than-positive feedback to a highly contingent partner may elicit reactions that interfere with these goals. Indeed, one’s own happiness may be dependent on the partner’s happiness (Kelley & Thibaut, 1978), and ingratiating deception is common when people are dependent on targets (Jones, 1964). Hence, when actors’ have communicated heightened vulnerability to feedback in particular domains, partners may sacrifice their own authenticity to care for actors’ psychological welfare or avoid the negative consequences of actors’ threat. Continuing with our example, Derek likely responds to his observations of Sarah’s emotional vulnerability regarding her attractiveness by cautiously providing overly positive feedback, perhaps telling Sarah that she looks great even when she does not, and avoiding any behavior that might suggest negative views of her appearance.

Path D. Partners’ Inauthentic Feedback Predicts Actors’ Authenticity Doubts

Partners may communicate their lack of authenticity to actors. This might occur, for instance, before partners learn of actors’ sensitivities, when partners unintentionally communicate less positive evaluations through nonverbal channels than what is expressed verbally (DePaulo, Stone, & Lassiter, 1985), when partners expresses negative evaluations during a moment of anger, or when partners succumb to actors’ pursuit of more veridical evaluations (Swann, 1987). Indeed, lies are more often detected in close relationships than in distant relationships (DePaulo & Kashy, 1998) and people are generally able to detect others’ ingratiation attempts (DePaulo et al., 1985; Jones, Stires, Shaver, & Harris, 1968) and exaggerated positive emotions (Pataki & Clark, 2004). That is, Sarah may realize Derek’s lack of authenticity when, after repeatedly inquiring as to whether she looks fat, he finally concedes that she could lose a few pounds or when his avoidance of eye contact and suggestions for light meals betray his explicit positive feedback.

Path E. Actors’ Expression of Sensitivity Predicts Authenticity Doubts

For contingent actors, suspicion about partners’ authenticity may occur even when partners are authentic. Once actors believe they have expressed their emotional vulnerability, social scripts regarding how others react to this vulnerability may influence their interpretation of partners’ feedback. Indeed, people expect feedback recipients to receive exaggerated evaluations from others when recipients are invested in the feedback (DePaulo & Bell, 1996; Lemay & Clark, in press) or when evaluators are dependent on recipients’ good will (Jones et al., 1968; Vonk, 1998). Hence, contingent actors may look back on their expressions of sensitivity and begin to doubt the authenticity of their partner’s positive feedback, even if the partner remained authentic. Continuing with our example, Sarah’s reflection on her own prior neediness and volatility likely causes her to believe that Derek is “sugar-coating” his feedback regarding her attractiveness, even when his positive feedback is genuine.

Path F. Partners’ Inauthentic Feedback Predicts Increased Anxiety and Reduced Satisfaction

Partners who provide inauthentic feedback may experience increased anxiety and reduced relationship satisfaction. Intimacy depends on each partner disclosing aspects of the self to the other (Reis & Shaver, 1988), which may include views of the other. Not expressing one’s views, even if they are about the other, may signal to the self that the other cannot be trusted (Sagarin, Rhoads, & Cialdini, 1998) and that one is uncomfortable in the relationship. Moreover, consistently monitoring and suppressing one’s reactions so as to not upset the other may be draining and anxiety-provoking. Consistent with these arguments are findings that people experience distress when telling lies to close relationship partners (DePaulo & Kashy, 1998) and to those who are invested in the feedback (DePaulo & Bell, 1996), and that suppression of thoughts and feelings can result in heightened blood pressure in the suppressor and disrupted communication in the relationship (Butler, Egloff, Wlhelm, Smith, Erickson, & Gross, 2003). Hence, Derek’s felt need to “walk on eggshells” around Sarah likely causes him to feel more anxious and less satisfied in the relationship.

Summary

Our model posits that contingent actors express their particular sensitivity to their partners. Consequently, they become suspicious of the authenticity of partners’ positive feedback. Partners also suffer a cost by detecting the heightened sensitivity, behaving in an inauthentic manner and, in turn, feeling more anxious and less satisfied.

We test this model in appearance and intelligence domains. We expect to find evidence of domain-specificity, such that the model describes the process in each domain while controlling for the process in the other domain. Given that level of self-esteem tends to be correlated with contingency of self-worth (Crocker, Luhtanen, Cooper, & Bouvrette, 2003), we controlled for self-esteem in all analyses. We also test whether actors’ self-evaluations, actors’ perceived partner evaluations, or partners’ actual evaluations of actors explain our findings.

Method

Participants

Romantic couples were recruited via advertisements on electronic bulletin boards to complete an electronic survey in exchange for payment of $5 per individual. Data from one dyad of the initial 102 dyads were eliminated from analyses due to missing data. The sample included 96 heterosexual, two same-sex male, and three same-sex female couples (69 married couples or domestic partnerships and 32 dating couples; M age = 33.41 years).

Procedure and Measures

Participants completed the following measures in the order presented. For domain-specific measures, the appearance and intelligence items were mixed and presented in a random order.

Self-esteem

Participants completed the Rosenberg (1965) Self-Esteem Scale using 6-point response scales (1: strongly disagree; 6: strongly agree) (Cronbach’s alpha = .90).

Contingency of self-worth (CSW)

Participants completed the contingency of self-worth (CSW) on appearance subscale of the Contingencies of Self-Worth Scale (Crocker, Luhtanen et al., 2003). Participants also completed a revised version of this subscale to measure contingency of self-worth on intelligence (e.g., “My self-esteem does not depend on whether or not I feel intelligent”). Items were completed on 7-point response scales (1: strongly disagree; 4: neutral; 7: strongly agree) (appearance alpha = .83; intelligence alpha = .85). After reverse-scoring, higher values reflected greater contingency of self-worth.

Appearance and intelligence self-evaluations

Using the same 7-point response scales, participants completed 4-item measures of their self-perceived attractiveness (e.g., “My facial features are attractive;” “My body is unattractive”) and intelligence (e.g., “I am intelligent;” “I am incompetent”). After reverse-scoring, higher scores reflect more positive self-evaluations (appearance alpha = .87; intelligence alpha = .81).

Expression of sensitivity

Using 6-point response scales (1: strongly disagree; 6: strongly agree), participants completed two analogous 3-item scales assessing their beliefs that they have expressed appearance and intelligence sensitivities to their romantic partners. Items assessed feedback seeking (e.g., “I often seek feedback from my partner regarding my attractiveness”), expression of negative emotion (e.g., “I have expressed hurt feelings or anger to my partner because he or she didn’t find me attractive”), and expression of sensitivity generally (e.g., “When with my partner, I have acted insecure about my intelligence”) (appearance and intelligence alphas = .66).

Authenticity doubts

Participants completed two analogous 3-item scales assessing their doubts about the authenticity of their partner’s feedback using the same 6-point response scales. Items assessed perceiving exaggerated positive evaluation (e.g., “My partner expresses more positive views about my intelligence than he or she really feels”), suppressed negative evaluation (e.g., “My partner hides his or her negative thoughts about my physical appearance”), and general caution (e.g., “My partner is overly cautious when it comes to giving me feedback about my appearance”) (appearance alpha = .77; intelligence alpha = .78).

Perception of partner’s expression of sensitivity

Using the same response scales, participants completed revised expression of sensitivity items to assess perceptions of the partner’s expression of sensitivity (e.g., “My partner often seeks feedback from me regarding his or her attractiveness”) (3-item appearance alpha = .79; 3-item intelligence alpha = .81).

Own inauthentic feedback

Using the same response scales, participants completed revised authenticity doubts items to assess their own inauthentic responding to their partners (e.g., “I express more positive views about my partner’s appearance than I really feel”) (3-item appearance and intelligence alphas = .82).

Relationship anxiety

With the same response scales, participants indicated their agreement with statements regarding the emotions they experience when interacting with their partner (e.g., “I feel calm when interacting with my partner”). The emotions included calm, frustrated, nervous, worried, fearful, tense, and relaxed. After reverse-scoring responses to the two positive emotions, higher scores indicate greater relationship anxiety (alpha = .91).

Perceived partner evaluation

Participants completed two 4-item measures, analogous to the self-evaluation measures and using the same response scales, to assess perceptions of the partner’s evaluation (e.g., “This person thinks my facial features are attractive”) (appearance alpha = .87; intelligence alpha = .86). After reverse-scoring, higher scores indicate beliefs of being evaluated more positively by the partner.

Relationship satisfaction

Using 9-point response scales (1: strongly disagree; 5: neither agree or disagree; 9: strongly agree), participants answered three questions assessing their relationship satisfaction (i.e., “This relationship is close to ideal;” “I am satisfied with this relationship;” “This relationship makes me very happy”) (alpha = .95).

Partner’s evaluation of appearance and intelligence

Participants completed two 4-item measures, analogous to the self-evaluation measures and using the same response scales, to assess their evaluation of the partner (i.e., “This person's facial features are attractive”) (appearance and intelligence alphas = .84).

Results

Analysis Strategy

Hypotheses were tested using multilevel models (with the SAS MIXED procedure) that modeled two individuals as nested within couples and intercepts as randomly varying across couples, which accounts for the dyadic interdependence of outcomes (cf. Kenny, Kashy, & Cook, 2006). Indirect effects were tested using the Baron and Kenny (1986) approach. The significance of the indirect effect (independent variable → mediator → dependent variable) was also formally tested using Sobel tests (the version including the s_a²s_b² term) (see Baron & Kenny, 1986). Each participant provided data relevant to testing both actor and partner roles. With the exception of relationship satisfaction, all of the variables varied more at the person level than at the relationship level, suggesting the utility of person-level, relative to couple-level, predictors.

Predicting Actors’ Expressions of Sensitivity

According to our model, CSW predicts actors’ expression of sensitivity in the contingent domain (Path A in Figure 1). We regressed actors’ expressions of sensitivity on same-domain CSW. We also controlled for level of self-esteem and CSW in the other domain. CSW predicted expression of sensitivity in the same domain, appearance b = .44, t = 7.05, p < .001; intelligence b = .31, t = 5.25, p < .001, but not in the other domain, ps > .66. These results support our predictions of a link between CSW and expressions of sensitivity in the contingent domain. These effects were independent of level of self-esteem, which predicted reduced expressions of appearance sensitivity, b = −.35, t = −4.08, p < .001 and intelligence sensitivity, b = −.44, t = − 5.06, p < .001.

Predicting Partners’ Detection of Actors’ Sensitivity

Our model anticipates an indirect effect of actors’ CSW on partners’ detection of actors’ sensitivity via actors’ reported expression of sensitivity (Paths A and B in Figure 1). Two models were used to test these predictions. In Model 1, we regressed partners’ detection of actors’ sensitivity on actors’ same-domain CSW. In Model 2, we included actors’ self-reported expression of sensitivity in the domain as an additional predictor. In all analyses, self-esteem and variables pertinent to the other domain (CSW, expression of sensitivity) were included as controls.

Results of analyses predicting partners’ perception of actors’ appearance and intelligence sensitivity appear in the upper and lower portions of Table 1, respectively. In Model 1, actors’ CSW predicted partners’ detection of actors’ sensitivity in the domain. The effect was substantially reduced (appearance domain) or eliminated (intelligence domain) in Model 2, in which actors’ self-reported expression of sensitivity was controlled. Sobel tests of the indirect effects (actors’ CSW → actors’ expression of sensitivity → partners’ detection of actors’ sensitivity) were significant, appearance z = 4.82, p < .001; intelligence z = 3.64, p < .001. These results are consistent with predictions that high CSW actors would have partners who perceived them as sensitive in the contingent domain because those actors expressed sensitivity in the past.

Table 1.

Predicting Partners’ Perception of Actors’ Sensitivity

Predictor	Model 1	Model 2
Predicting Partners’ Perception of Actors’ Appearance Sensitivity
Actors’ Self-esteem	−.20 (−1.74)†	−.10 (−1.01)
Actors’ Appearance CSW	.40 (5.18)***	.16 (2.04)*
Actors’ Expression of Appearance Sensitivity (Path B)	-	.55 (6.38)***
Predicting Partners’ Perception of Actors’ Intelligence Sensitivity
Actors’ Self-esteem	−.30 (−2.81)**	−.17 (−1.60)
Actors’ Intelligence CSW	.18 (2.59)*	.05 (.72)
Actors’ Expression of Intelligence Sensitivity (Path B)	-	.42 (4.90)***

Note. Coefficients are unstandardized. Values in parentheses are t values. Predictors in the other domain (i.e., CSW, expression of sensitivity) were included as controls.

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

There was some evidence of spillover across domains. Actors’ expression of appearance sensitivity independently predicted partners’ perceptions of actors’ intelligence sensitivity, b = .19, p < .05. However, same-domain effects were stronger and more consistent than cross-domain effects.

Predicting Partners’ Inauthentic Responding

Partners who perceived actors’ sensitivity should report providing more inauthentic feedback (Path C in Figure 1). Three models in each domain of contingency tested the processes linking actors’ contingent self-worth to partners’ authenticity. In Model 1, partners’ inauthentic feedback was regressed on actors’ same-domain CSW. In Model 2, actors’ expression of sensitivity was added as an additional predictor. In Model 3, partners’ detection of sensitivity in the same domain was also added. In all models, self-esteem and variables pertinent to the other domain (CSW, actors’ expression of sensitivity, partners’ detection of sensitivity) were included as controls.

Results are displayed in Table 2. In Model 1, actors’ appearance CSW predicted partners’ inauthentic appearance feedback. Contrary to predictions, actors’ intelligence CSW did not directly predict partners’ intelligence feedback authenticity. This may be due to the somewhat weaker Path A and Path B in the intelligence domain.

Table 2.

Predicting Partners’ Inauthentic Feedback

Predictor	Model 1	Model 2	Model 3
Predicting Partners’ Inauthentic Appearance Feedback
Actors’ Self-esteem	−.04 (−.35)	.02 (.14)	.04 (.40)
Actors’ Appearance CSW	.19 (2.52)*	.07 (.85)	.01 (.17)
Actors’ Expressions of Appearance Sensitivity	-	.28 (3.05)**	.16 (1.65)
Partners’ Perception of Actors’ Appearance Sensitivity (Path C)	-	-	.28 (3.7)***
Predicting Partners’ Inauthentic Intelligence Feedback
Actors’ Self-esteem	−.01 (−.07)	.03 (.29)	.07 (.72)
Actors’ Intelligence CSW	.08 (1.24)	.04 (.61)	.02 (.28)
Actors’ Expressions of Intelligence Sensitivity	-	.13 (1.71)†	−.01 (−.14)
Partners’ Perception of Actors’ Intelligence Sensitivity (Path C)	-	-	.38 (5.7)***

Note. Coefficients are unstandardized. Values in parentheses are t values. Predictors in the other domain (i.e., CSW, expression of sensitivity, partners’ perception of sensitivity) were included as controls.

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

Results of Model 2 revealed that actors’ expressions of sensitivity predicted increases in partners’ inauthentic feedback in each domain. Moreover, actors’ appearance CSW no longer predicted partners’ appearance authenticity in Model 2. The indirect effect (actors’ appearance CSW → actors’ expression of appearance sensitivity → partners’ inauthentic feedback) was significant, z = 2.79, p < .01. That is, high appearance CSW actors had partners who delivered less authentic appearance feedback largely because such actors expressed their appearance sensitivities.

For Model 3, partners’ detection of actors’ same-domain sensitivity was the only significant predictor of partners’ authenticity. The effects of actors’ expressions of sensitivity were eliminated and Sobel tests of the indirect effects (actors’ expression of sensitivity → partners’ detection of actors’ sensitivity → partners’ authenticity) were significant, appearance z = 3.12, p < .01; intelligence z = 3.86, p < .001. Actors who expressed sensitivity had partners who delivered less authentic feedback apparently because those partners viewed actors as especially sensitive in that domain. Evidence of domain-specificity was observed; in Model 3, once all predictors were included, the only significant predictor of partners’ inauthentic feedback in the domain was partners’ detection of actors’ sensitivity in that domain.

Predicting Actors’ Authenticity Doubts

We expected that actors’ doubts about the authenticity of partners’ appearance or intelligence feedback would be predicted by the partners’ reported inauthentic feedback (Path D in Figure 1) and by actors’ beliefs of expressing sensitivity to partners (Path E in Figure 1). We tested four models in each domain of contingency to test the processes linking CSW to authenticity doubts. The predictors of interest are displayed in Table 3 and Table 4. Analogous variables in the other performance domain were included as controls.

Table 3.

Predicting Actors’ Doubts Regarding Authenticity of Partners’ Appearance Feedback

Predictor	Model 1	Model 2	Model 3	Model 4
Actors’ Self-esteem	−.41 (−4.19)***	−.35 (−3.83)***	−.35 (−3.80)***	−.34 (−3.72)***
Actors’ Appearance CSW	.21 (3.19)**	.05 (.76)	.05 (.64)	.03 (.48)
Actors’ Expression of Appearance Sensitivity (Path E)	-	.37 (4.83)***	.35 (4.14)***	.32 (3.84)***
Partners’ Perception of Actors’ Appearance Sensitivity	-	-	.05 (.74)	−.01 (−.10)
Partners’ Inauthentic Appearance Feedback (Path D)	-	-	-	.21 (2.79)**

Note. Coefficients are unstandardized. Values in parentheses are t values. Predictors specific to the intelligence domain (i.e., CSW, expression of sensitivity, partners’ perception of sensitivity, partners’ inauthentic feedback) were included as controls.

^**p < .01.

^***p < .001.

Table 4.

Predicting Actors’ Doubts Regarding Authenticity of Partners’ Intelligence Feedback

Predictor	Model 1	Model 2	Model 3	Model 4
Actors’ Self-esteem	−.40 (−4.98)***	−.28 (−3.68)***	−.27 (−3.57)***	−.29 (−3.8)***
Actors’ Intelligence CSW	.08 (1.57)	−.02 (−.48)	−.03 (−.56)	−.03 (−.57)
Actors’ Expression of Intelligence Sensitivity (Path E)	-	.35 (5.56)***	.32 (4.77)***	.30 (4.57)***
Partners’ Perception of Actors’ Intelligence Sensitivity	-	-	.07 (1.18)	0 (−.02)
Partners’ Inauthentic Intelligence Feedback (Path D)	-	-	-	.22 (2.97)**

Note. Coefficients are unstandardized. Values in parentheses are t values. Predictors specific to the appearance domain (i.e., CSW, expression of sensitivity, partners’ perception of sensitivity, partners’ inauthentic feedback) were included as controls.

^**p < .01.

^***p < .001.

Appearance CSW predicted appearance authenticity doubts in Model 1, although the effect of intelligence CSW on intelligence authenticity doubts did not quite reach conventional significance levels (p = .12; one-tailed p = .06). (The two-tailed effect was marginal only when the appearance domain control variables were excluded from the analysis, b = .10, t = 1.85, p = .07. Note also that the effect of intelligence CSW changed sign in Models 2 through 4, suggesting that requiring a direct effect of intelligence CSW on intelligence authenticity doubts for evidence of mediation is too conservative. See MacKinnon, Krull, and Lockwood, [2000].)

The effect of appearance CSW was eliminated after controlling for actors’ expressions of sensitivity in Model 2. Instead, actors’ expressions of sensitivity predicted actors’ authenticity doubts. Sobel tests of the indirect effects (actors’ CSW → actors’ expression of sensitivity → actors’ authenticity doubts) were significant, appearance z = 4, p < .001; intelligence z = 3.89, p < .001. These findings support the predicted indirect effect of CSW on actors’ authenticity doubts via their self-perceived expression of sensitivity (combination of Paths A and E in Figure 1).

This effect of actors’ expression of sensitivity on their authenticity doubts remained significant in Model 3 (controlling for partners’ perception of actors’ sensitivity) and in Model 4 (controlling for partners’ reported inauthentic feedback). Partners’ reports of providing inauthentic feedback in the domain also significantly predicted actors’ authenticity doubts in Model 4, suggesting some kernel of truth in actors’ doubts. Evidence for domain-specificity was found once all predictors in the same domain of performance were controlled in Model 4; effects of predictors in the other domain of performance were not significant.

These results support our predictions that actors’ authenticity doubts are somewhat accurately based on partners’ inauthentic responding (Path D) and are somewhat biased by actors’ self-perceived expression of sensitivity (Path E).

Effects of Partners’ Authenticity on their Relationship Anxiety and Satisfaction

We predicted that partners who deliver inauthentic feedback would report more anxiety and less satisfaction. We tested the link between partners’ authenticity and these outcomes while controlling for all same-domain variables listed in Table 3 and Table 4. As predicted, partners’ inauthentic appearance and intelligence feedback predicted their anxiety, b = .27, t = 4.36, p < .001 and b = .25, t = 3.53, p < .001, respectively, and satisfaction, b = −.59, t = −5.09, p < .001 and b = −.61, t = −4.44, p < .001, respectively.¹

Alternative Explanations

Perhaps actors express sensitivity to partners because they have negative self-views or believe partners view them negatively, rather than because of contingent self-worth. Indeed, other research suggests that people with self-doubts or doubts about a partner’s regard tend both to express sensitivities to partners (Murray et al., 2006) and question the validity of positive feedback (Marigold, Holmes, & Ross, 2007; Stroebe, Eagly, & Stroebe, 1977). Hence, we retested all models described above after also controlling for actors’ self-evaluations and perceived partner evaluations in the domain. The paths displayed in Figure 1 were still evident (Path D ps < .07; all other ps < .001).

Perhaps partners evaluated contingent actors negatively in the domain, and this negative evaluation, rather than perception of actors’ sensitivity, predicted their inauthentic responding. Such negative evaluation of actors also may predict partners’ satisfaction or anxiety (see Murray et al., 1996). Hence, we tested paths involving partner variables after also controlling for partners’ evaluation of actors in the domain. With one exception (the effect of partners’ inauthentic intelligence feedback on their anxiety; two-tailed p = .18; one-tailed p = .09), the effects predicted by our model remained significant, ps < .05.

Finally, a variety of other mediation models could be formed by rearranging our model variables. Perhaps CSW, as it resembles insecurity, directly predicts doubts about partners’ authenticity, and this lack of trust then predicts their expressions of sensitivity to the partner (switching the order of expressions of sensitivity and authenticity doubts in our model). Alternatively, perhaps partners of contingent actors are dissatisfied or anxious, which causes their inauthentic responding (placing partners’ satisfaction or anxiety between actors’ expression of sensitivity and partners’ inauthentic responding). To test these and all other alternative mediation models, we conducted additional analyses in which every model variable was predicted by all 12 of the other primary model variables, whether theoretically upstream or downstream (actors’ self-esteem, the 5 domain-specific measures in both domains of contingency, and partners’ satisfaction and anxiety). Support was still observed for the paths predicted by our model, ps < .01 Hence, only the predicted mediation effects were found.

Discussion

We have tested a model of how the partner-affirmation process may go awry. As predicted, individuals whose feelings of self-worth are highly contingent on performance in a particular domain reported frequently seeking their partner’s feedback, expressing negative emotion in response to threats, or otherwise expressing emotional sensitivities to partners in the domain of contingency. In turn, having believed that they revealed their heightened emotional vulnerabilities, contingent actors had reason to suspect the authenticity of their partner’s positive feedback. Sadly and ironically, these doubts occurred precisely when actors most wanted to believe positive feedback – when their self-worth was highly contingent on performance. This suspicion may be a cost that contingent actors incur as a result of their own emotional reactions, which may undermine partners’ ability to affirm actors’ self-worth (see Leary & Baumeister, 2000).

Our results suggest that the actors’ expressions of vulnerability also affect the partners’ behavior. Partners of contingent actors appeared to detect actors’ expressions of vulnerability which, in turn, predicted their providing inauthentic feedback to such actors (as indicated by partners’ self-reports). That is, partners seemed to “walk on eggshells’ when providing feedback relevant to actors’ specific vulnerability. Partners who reported responding in this manner reported reduced relationship satisfaction and increased relationship anxiety, suggesting that actors’ reactions to their contingent self-worth were costly to partners.

Of course, the current research is not experimental. Causal inferences must be considered tentative until experimental evidence is available. Experimental manipulations may reveal downstream effects predicted by our model. For example, a manipulation that convinces people that they have expressed a heightened degree of sensitivity to performance in a particular domain may cause them to become suspicious of subsequent positive feedback. Likewise, false information regarding a partner’s contingent self-worth may cause people to “walk on eggshells” around the partner, which may undermine comfort and satisfaction in the interaction. Despite the lack of experimental manipulation, our analyses do rule out several alternative explanations, including the possibility that low self-esteem, actors’ self-evaluations or perceptions of partner evaluations in the domain, or partners’ actual evaluations of actors in the domain operated as third variables.

Theoretical Implications

Crocker and Park (2004) argued that people with contingent self-worth become chronically oriented toward achieving success and avoiding failure in the relevant performance domain as a means of validating their worth, which comes at costs to the self – including controlled motivation and resistance to negative information that could facilitate growth – and costs to others – including reduced consideration of others’ welfare and antagonistic responses.

The current findings are consistent with and extend this reasoning. By expressing their heightened emotional vulnerability, contingent actors seem to incur as a cost to themselves suspicion and doubt about partners’ positive feedback. Whereas prior research suggests that a chronic orientation toward regulating contingent self-worth is related to dismissing negative feedback (Crocker & Park, 2004), the current research suggests that, through the process of communicating heightened emotional vulnerability to partners, this orientation also may result in distrusting positive feedback. Hence, by believing that one’s pursuit of self-esteem has been communicated to one’s partner, the pursuit may be undermined.

Our findings also extend the argument that contingent self-worth is costly for partners. Prior research suggests that contingent individuals are less likely to experience autonomous motivation – a sense that behavior is self-determined (Crocker & Park, 2004; Deci & Ryan, 2000). The present research suggests a similar result for partners; by expressing heightened emotional vulnerability to partners, partners are put in a position in which they must choose between assisting actors in regulating their feelings of worth and retaining their own authenticity. Either decision is likely to have costs for partners, whether it takes the form of their own reduced authenticity or their suffering the wrath of hostile or depressed actors. Hence, contingent actors who rely on partners to regulate their feelings of worth may have partners who, like actors themselves, are in a no-win situation. Moreover, as our results suggest that partners of contingent actors ultimately do behave in an inauthentic manner, it may be the case that partners unwittingly assist actors in resisting negative performance feedback that could facilitate actors’ growth.

At first glance, the current findings might seem to contradict the general perspective that well-functioning close relationships involve each partner expressing needs and vulnerabilities to the other (Clark & Finkel, 2005; Reis & Shaver, 1988). However, the process studied in the current research involves actors being especially and incessantly focused on their own self-evaluative needs, which may blind them to the consequences of their insecurities for their partner’s welfare. Indeed, those with contingent self-worth frequently expressed negative emotions in response to partners’ feedback, which may make partners anxious rather than draw them closer. The functional aspects of disclosing vulnerabilities may require that people do not react negatively to their partner’s feedback. Seeking effective care may require just as much consideration and skill as providing it.