Shifting Dependence

Abstract

High and low self-esteem people typically have divergent responses to interpersonal risk. Highs draw closer to their partner, whereas lows self-protectively distance. However, these responses should be more likely when people are dependent on the rewards their partner offers. Two experiments tested the hypothesis that structural changes in the situation of interdependence lead high and low self-esteem people to reverse their typical responses to risk. When partners were instrumental to a current goal pursuit (and participants were more dependent on the rewards partners could offer), highs drew closer and lows distanced when risk was primed. However, when partners were not instrumental to an active goal (and participants were less dependent on the rewards partners could offer), these responses were reversed. Reducing one’s dependence on a partner to attain one’s personal goals appears to reduce highs’ incentive to connect, whereas it appears to increase lows’ incentive to connect.

Keywords

self-esteem close relationships instrumentality interdependence

Romantic relationships tempt people with the promise of heady rewards. Risking interdependence offers the pleasures of sex (Impett, Strachman, Finkel, & Gable, 2008), the comfort of shared self-disclosure (Reis & Shaver, 1988), and the exhilaration of self-expansion (A. Aron, Paris, & Aron, 1995). Such connections can also boost self-esteem (Murray, Holmes, & Griffin, 2000), alleviate attachment insecurity (Davila, Karney, & Bradbury, 1999; Murray, Holmes, & Collins, 2006), and facilitate the pursuit of one’s personal goals (Collins & Feeney, 2004; Fitzsimons & Bargh, 2003; Fitzsimons & Shah, 2008; Shah, 2003a, 2003b).

Yet the experience of such rewards comes with hidden costs. Depending on the rewards a partner provides makes the prospect of a partner’s rejection more likely, and potentially, all the more painful (Berscheid, 1983; Murray, Holmes, et al., 2006). Imagine that Ron needs a busy and preoccupied Gayle’s assistance in preparing for an upcoming exam. Asking for Gayle’s help could elicit the support Ron needs, but making this request also leaves him vulnerable to her refusal and rejection. Moreover, even if Gayle agrees, routinely relying on her help in such situations only increases how much Ron stands to lose if she later proves to be rejecting.

In this article, we utilize these potential benefits and costs of interdependence to predict when low and high self-esteem people will risk connecting with a partner in the face of acute uncertainty about the partner’s caring. Research on the risk-regulation model reveals that depending on a partner to provide valued rewards, such as advice or comfort, is riskier for low self-esteem people than for high self-esteem people because lows usually expect their partner to reject and disappoint them (Murray et al., 2000, Murray, Holmes, et al., 2006). Hampered by distrusting expectations, low self-esteem people respond to the possibility of rejection by distancing from their partner. But, protected by more trusting expectations, high self-esteem people instead connect (see Cavallo, Murray, & Holmes, 2013 for a review). In this article, the novel hypothesis that subtle shifts in the structure of interdependence can reverse these risk regulation dynamics is tested. We argue that situations that limit dependence on the partner as a source of rewards paradoxically feel safer and more enticing for low self-esteem people because they have less to lose from the possibility of rejection when their partner can offer fewer rewards in the first place. Through this subtle shift in interdependence, we expect situations that limit dependence on the partner to motivate low, but not high, self-esteem people to seek connection with their partner in the face of risk.

Regulating Risk in Romantic Relationships

Risk is a central part of romantic life (Murray & Holmes, 2009). No matter how compatible, couples inevitably encounter specific situations where they have divergent preferences, interests, and goals (Lykken & Tellegen, 1993). In such situations, relying on one’s partner to set aside his or her own interests to give support, provide companionship, or make a sacrifice leaves one vulnerable to the partner’s rejection and nonresponsiveness. Therefore, conflicts of interest routinely put the desire to seek connection and its rewards in conflict with the desire to self-protect against the possibility of loss and rejection.

According to the risk regulation model (Murray, Holmes, et al., 2006), people use their feelings of trust and security in their partner’s caring to navigate such interdependent situations. Being more trusting signals the partner’s likely responsiveness, motivating people to seek greater closeness in the face of risk. In contrast, being less trusting motivates people to self-protect and distance from a potentially nonresponsive partner, thereby minimizing both the likelihood and pain of rejection in advance (Campbell, Simpson, Boldry, & Rubin, 2010; Derrick, Leonard, & Homish, 2012; Murray et al., 2000; Murray, Holmes, et al., 2006; Murray, Rose, Bellavia, Holmes, & Kusche, 2002; Overall & Sibley, 2009).

How Self-Esteem Often Shapes Regulatory Responses to Risk

Because trust regulates responses to risk in this way, people who can more readily sustain trust in their partner’s caring should be more willing to seek connection with their partner in risky situations (Murray, Holmes, et al., 2006). People’s beliefs about how valuable they are to their partner provide an important basis for their feelings of trust or security in their partner’s caring. When people believe that their partners see special and irreplaceable value in them, they report more trust in the partners’ commitment and motivation to be responsive (Murray et al., 2009). However, when people are uncertain of their value to their partner, they tend to doubt that their partners will be responsive to their needs.

People base their estimates of how much their partners value their personal qualities in large part on their own self-views (Murray et al., 2000). People who are high in self-esteem believe that their partners view them as positively as they view themselves. People who are low in self-esteem, on the other hand, mistakenly believe that their partners share their relatively negative self-views. Consequently, low self-esteem people generally report trusting their partner’s caring and responsiveness less than high self-esteem people do (Murray et al., 2000; Murray, Holmes, Griffin, Bellavia, & Rose, 2001). Because self-esteem generally constrains trust in the partner’s responsiveness, high and low self-esteem people typically approach conflict-of-interest situations with opposite interpersonal goals in mind (see Cavallo et al., 2013 for a review).

Being more trusting, high self-esteem people can afford to risk prioritizing connection over self-protection goals. For highs, the risk of a partner behaving nonresponsively usually prompts heightened efforts to seek closeness and depend on the partner even more (Cavallo, Fitzsimons, & Holmes, 2009; Cavallo, Holmes, Fitzsimons, Murray, & Wood, 2012; Murray et al., 2002). For example, high self-esteem people actually perceive their partners to be even more accepting and value their partners more when they are led to believe their partners have an unspoken complaint about their behavior or personality (Murray et al., 2002). Being less trusting, low self-esteem people typically prioritize self-protection over connection goals. For lows, the risk of a partner behaving nonresponsively usually prompts heightened efforts to push the partner away and reduce closeness. For instance, when low self-esteem people are led to believe that their partners perceive a large number of faults in them, they report feeling less close to their partners and self-protectively devalue their partners (Murray et al., 2002). Satisfying the goal to self-protect is such a strong priority for low self-esteem people that even implicitly priming the general goal to approach the social world is itself sufficient to motivate most lows to distance from their partners (Murray, Derrick, Leder, and Holmes, 2008).

Partner Instrumentality: Shifting Incentives to Connect

In prior research on risk regulation, the rewards offered by partners were often abstract, intangible, and wedded to the relationship. For instance, by drawing closer to a partner who seems irritated, people might gain their partner’s renewed approval and affection (Murray et al., 2002). Similarly, by drawing closer to their partner in the face of personal setbacks at work, people stand to gain their partner’s comfort and support (Murray, Griffin, Rose, & Bellavia, 2006). Such rewards are enticing for low and high self-esteem people alike. But, low self-esteem people typically feel less valued by others than they desire, a deprived state that makes them especially dependent on the rewards and acceptance their partner can offer (Leary, Tambor, Terdal, & Downs, 1995). Therefore, any situation that raises new questions or concerns about a partner’s continued willingness to be rewarding, accepting, and responsive is likely to be especially threatening for lows—motivating them to self-protect.

But not all situations highlight the potential rewards attached to depending on a partner to the same degree. In fact, the burgeoning literature on interpersonal goal pursuit suggests that connecting to a partner might be markedly less rewarding in specific situations. Indeed, rather than being stable across situations, the rewards that even a valued and loved partner can offer may be controlled in part by one’s own personal goals pursuits (Fitzsimons & Shah, 2008). Consequently, people’s experienced dependence on the rewards their partner offers might also vary across situations according to their personal goal pursuits.

Mental representations of personal goals include representations of significant others who are associated with those goals (Fitzsimons & Bargh, 2003). Thinking of a significant other automatically activates goals associated with the significant other (Fitzsimons & Bargh, 2003). Similarly, activating a goal automatically brings to mind partners who facilitate the pursuit of that goal (Fitzsimons & Shah, 2008). Because goals are implicit in significant other representations, reminders of significant others can also increase the pursuit of the goals associated with those partners and can improve performance on goal-related tasks, particularly when people are closer to the partners (Fitzsimons & Bargh, 2003; Shah, 2003a). In contexts in which a personal goal is active, people also draw closer to partners who contribute positively to attainment of that goal (Fitzsimons & Fishbach, 2010; Fitzsimons & Shah, 2008). However, people readily distance themselves from partners when they are pursuing a goal the partner does not facilitate. Just as people see less value in other objects that interfere with goal pursuit (Fishbach, Zhang, & Trope, 2010), people tend to view noninstrumental partners less favorably when they are pursuing personal goals that the partner could impede.

These findings suggest that shifts in one’s current goal pursuits might also shift how much people depend on their partner as a source of valued rewards. That is, shifts in one’s current goal pursuits might change the actual structure of interdependence (Kelley, 1979). If Ron needs help studying for an important exam, Gayle’s academic prowess and intellectual support offers him a strong incentive to connect with her. In reaching out to Gayle, Ron can gain her support and companionship, self-regulate more effectively, and perform better academically. But, in depending on Gayle in this way, Ron also risks her behaving nonresponsively and denying him the instrumental support he needs. Thus, situations that highlight Gayle’s instrumentality to Ron’s goals increase his dependence on her; such situations symbolically and practically tie his outcomes to her anticipated responsiveness (Kelley, 1979). Conversely, situations that limit or constrain Gayle’s instrumentality to Ron’s current goal pursuits decrease his dependence on her; such situations separate his outcomes from her anticipated responsiveness because Gayle no longer holds the key to the rewards he values (Fitzsimons & Shah, 2008).

Shifting the Incentives for Low and High Self-Esteem People to Connect

Therefore, when people are pursuing a personally important goal, their personal successes and failures are likely to be more tied to the actions of partners who can facilitate this goal (i.e., instrumental partners) than they are to the actions of partners who cannot (i.e., noninstrumental partners). In such situations, people’s greater dependence on instrumental partners should make the possibility of being rejected or unsupported especially salient (Kelley, 1979). Therefore, in such high-relationship-stakes situations, low and high self-esteem people should evidence their typical regulatory responses to risk (Murray, Holmes, et al., 2006). Being more trusting, high self-esteem people should feel safe enough to risk pursuing stronger connections with instrumental partners when they are uncertain of their responsiveness. In contrast, being less trusting, low self-esteem people should self-protectively distance themselves from instrumental partners when they are uncertain of their responsiveness, and thereby, minimize both the likelihood and potential pain of rejection in advance.

However, when the partner is not instrumental to one’s current goals, being less dependent on the benefits the partner can offer should have the reassuring effect of minimizing what might be lost. Consequently, in such low-relationship-stakes situations, low self-esteem people might feel safe enough to risk seeking connection with a potentially rejecting partner precisely because being less dependent on the partner for the pursuit of their currently active goals makes the prospect of rejection seem less likely. Consistent with this logic, subtle shifts in the situation can motivate low self-esteem people to take the risk of connecting with their partner. For instance, inducing low self-esteem people to perceive a fault in their partner increases their willingness to connect with the partner (Murray et al., 2005). Daily successes at work also lead low self-esteem people to feel more accepted and loved by their partners (Murray, Griffin, et al., 2006). As well, training low self-esteem people to interpret compliments from their partner as evidence of the partner’s caring makes them more willing to seek closeness in risky situations (Marigold, Holmes, & Ross, 2010). Indeed, when others’ acceptance is guaranteed, low self-esteem people are usually quite eager to pursue social connections (Anthony, Wood, & Holmes, 2007).

However, for people high in self-esteem, situational shifts that minimize their need to depend on their partner to accomplish their goals may also reduce the motivational impetus to connect with the partner in the face of risk. High self-esteem people are generally oriented toward seeking reward in risky interpersonal situations (Cavallo et al., 2009; Cavallo et al., 2012). Although pursuing rewards often takes the form of seeking interpersonal connection, it can also involve pursuing nonsocial rewards, such as a risky but lucrative investment strategy (Cavallo et al., 2009). Thus, when connecting with a partner would not be particularly rewarding (i.e., the partner is noninstrumental), highs might seek rewards in other ways, such as by pursuing an active personal goal. Indeed, demonstrating competence via goal pursuit seems to be particularly rewarding for high self-esteem people when they experience self-threats. In such self-threatening situations, high self-esteem people will sometimes seek to demonstrate their prowess even at the expense of seeking social connections (Vohs & Heatherton, 2001).

Summary of Hypotheses

We designed two experiments to examine shifts in risk regulation dynamics as a function of subtle shifts in the structure of interdependence created by people’s symbolic need to depend on their partner to accomplish their current goal pursuits. To shift the rewards that depending on one’s partner offered, we had participants engage in academic goal pursuit—a ubiquitous personal goal for our sample. We expected that engaging in academic goal pursuit would increase dependence when the partner’s responsiveness could facilitate one’s academic goals (i.e., an instrumental partner) and decrease dependence when the partner’s responsiveness could not facilitate one’s goals (i.e., a noninstrumental partner). We also manipulated the general risk of connecting with partners because making the prospect of partner rejection salient triggers risk regulation processes (Murray, Holmes, et al., 2006). Thus, we only expected to see instrumentality-based shifts in risk regulation dynamics among low and high self-esteem participants when they had reason to worry about the security of their partner’s caring and the stability of their relationship.

Prior research on risk regulation processes revealed that people connect and promote their relationship in the face of risk by increasing their sense of psychological closeness to their partner and by enhancing their partner’s value (Murray et al., 2002). Conversely, people self-protect and distance themselves from the prospect of rejection by decreasing their sense of closeness and by derogating their partner’s value. Accordingly, in Study 1, we indexed risk regulation through the tendency to include the partner in one’s own self-concept, an index of psychological closeness. In Study 2, we indexed risk regulation through both the tendency to include the partner in the self and the tendency to ascribe valued traits to the partner.

We expected dependence on the rewards the partner could offer to be amplified when partners were instrumental to the academic goals participants were currently pursuing. In such high-relationship-stakes situations, we expected low and high self-esteem people to evidence prototypic risk regulation dynamics. That is, we expected high self-esteem participants to draw closer to and value instrumental partners more in the face of relationship risk. In contrast, we expected low self-esteem participants to distance and value instrumental partners less in the face of risk.

Conversely, we expected dependence on the rewards the partner could offer to be minimized when partners were not instrumental to the academic goals participants were currently pursuing. We also expected this subtle shift in the relationship stakes of interdependence to change low and high self-esteem people’s responses to risk. When the rewards of relying on a partner are minimized, and dependence is decreased, low self-esteem people should feel safer drawing closer to and valuing noninstrumental partners. Thus, we expected low self-esteem participants to draw closer to and value noninstrumental partners more in response to relationship risk. However, when the partner was not instrumental to the academic goals they were pursuing, high self-esteem people no longer had a strong incentive to connect with their partners. Without this motivational pull to connect, they should prioritize other goals that may offer more opportunity for reward. Because academic achievement is a salient alternative goal in this situation, we expected high self-esteem people to prioritize this goal and distance themselves psychologically from their noninstrumental partners in the face of risk.

Study 1

In Study 1, we first measured people’s perceptions of their partner’s instrumentality to their academic, social, and fitness goals during a pretesting session. In the subsequent laboratory experiment, we manipulated interpersonal risk and had participants engage in an academic-goal-related activity. Participants in the high-risk condition learned that their responses to a relationship behaviors inventory had revealed that their relationships were somewhat more problematic than average. Participants in two control conditions were either given no feedback or were told that their responses indicated their relationships were somewhat less problematic than average. Then all participants engaged in the academic-goal-related activity—completing Graduate Record Examination (GRE) analogy problems. We had participants complete this analogy task to induce academic goal pursuit because prior research suggests that completing such academic tasks activates academic achievement goals. For example, participants engaged in a word-creation task show an implicit positive bias toward goal-related objects (Ferguson & Bargh, 2004). Describing a verbal task as diagnostic of academic ability similarly primes achievement goals, as evidenced by success- and failure-contingent shifts in implicit positivity toward goal-related objects (Moore, Ferguson, & Chartrand, 2011). Next, we measured risk regulation responses through postmanipulation feelings of closeness, which we assessed through participants’ tendency to include the partner into the self (A. Aron, Aron, & Smollan, 1992). When partners were instrumental to academic goals, we expected high self-esteem people to connect and promote the relationship in response to risk (i.e., increase closeness) and low self-esteem people to distance (i.e., decrease closeness). However, when partners were not instrumental to academic goals, we expected low self-esteem people to connect in the face of risk and high self-esteem people to distance.

Method

Participants

One hundred fourteen participants in romantic relationships were recruited in exchange for Introductory Psychology credit. Three participants were dropped because they had been in their relationships for less than 4 months,¹ and 11 participants were dropped for indicating that they were suspicious of the experimental feedback, leaving a sample of 100 participants (54 men). The sample was 19.22 years (SD = 1.78), and participants had been in their relationships for 18.67 months (SD = 16.65).

Procedure

During a mass testing session at the beginning of the academic semester, participants evaluated how instrumental their partner was to their academic, fitness, and social goals (adapted from Fitzsimons & Shah, 2008), how personally important their academic goals were, and their academic performance relative to their partner. Participants also evaluated their own perceived academic success and how close they felt to their partners. On arriving at the lab, participants were told that the experiment examined how experiences in people’s relationships and daily lives affected verbal thinking.

Participants completed the study using MediaLab software. They first completed demographic measures and the Rosenberg (1965) Self-Esteem Scale. All participants then completed the “Interpersonal Behaviors Inventory” (modified from Murray et al., 2002). Participants were asked to indicate how frequently they engaged in a variety of behaviors in their relationships (e.g., “How often do you tease your partner about some aspect of his or her behavior or personality?”). In the high-risk condition, the items were worded in such a way to induce participants to endorse lower values for positive behaviors (e.g., “How often do you do really go out of your way to do something thoughtful or considerate for your partner?”) and higher values for negative behaviors (e.g., “How often are you at all critical of something your partner has said or done?”). Participants in the high-risk condition then received feedback that their relationship was somewhat worse than average. This feedback was intended to generate doubts in participants about the stability of their relationships. In the low-risk condition, the items were worded so that participants would endorse higher values for positive behaviors (e.g., “How often do you do something thoughtful or considerate for your partner?”) and lower values for negative behaviors (e.g., “How often are you critical of something your partner has said or done?”). Participants in the low-risk condition then received feedback that their relationships were somewhat better than average. In both conditions, the endpoints of the scale were also constructed so that participants would select the targeted values, making the feedback they received more believable. Participants in the control condition completed the same questionnaire as participants in the low-risk condition, but did not receive any feedback.

Consistent with the cover story, all participants then completed the measure of verbal thinking, 15 moderately difficult GRE analogy problems.² Each problem consisted of a related pair of words (e.g., HANGAR: AIRCRAFT). Participants were instructed to select from five answer choices the pair of words that expressed the same relationship between words as did the stem pair. This activity was intended to foster engagement in an academic goal pursuit.

Participants then completed the Inclusion of Other in the Self (IOS) Scale (A. Aron et al., 1992), filler items, a manipulation check for the effectiveness of the Interpersonal Behaviors Inventory, and a measure of suspicion. At the end of the study, participants were fully debriefed and thanked for their participation.

Measures

Pretest instrumentality

This three-item measure assessed the extent to which partners facilitated participants’ academic, social, and fitness goals (e.g., “To what extent does your romantic partner facilitate your achievement of your academic goals?” 1 = not at all, 5 = very much).

Inclusion of Other in the Self

The Inclusion of Other in the Self Scale (A. Aron et al., 1992) is a one-item measure that asks participants to indicate which of a series of seven overlapping circles best represents how close (or connected) they feel to their partners.

Results

We conducted hierarchical regression analyses to test our predictions. In the initial step, we entered the centered main effects of self-esteem, pretest partner instrumentality, and risk salience. We used dummy codes to capture the three conditions (each vector compared one of the two control conditions [scored 1] against the high-risk condition [scored 0]). In the initial step, we also included the self-rated importance of one’s academic goals and partners’ relative academic performance as control variables to ensure that any effects of instrumentality did not simply reflect these potentially confounding influences (Fitzsimons & Shah, 2008).³ In subsequent steps, we entered all possible two- and three-way interaction terms between self-esteem, pretest partner instrumentality, and the condition vectors. Table 1 contains the results. We discuss only the significant effects involving the experimental condition.

Table 1.

Summary of Regression Analyses in Study 1.

IOS
	β	t
Goal importance	−.12	−1.12
Partner performance	−.01	−0.11
Pretest instrumentality	.34	1.72^†
Self-esteem	.14	0.63
High-risk vs. Low-risk	.21	1.69^†
High-risk vs. Control	.15	1.21
Self-esteem × Instrumentality	1.05	4.34**
Self-esteem × High-risk vs. Low-risk	.14	0.88
Self-esteem × High-risk vs. Control	−.06	−0.31
Instrumentality × High-risk vs. Low-risk	−.20	−1.36
Instrumentality × High-risk vs. Control	.02	0.10
Self-esteem × Instrumentality × High-risk vs. Low-risk	−.68	−3.34**
Self-esteem × Instrumentality × High-risk vs. Control	−.51	−3.17**

Note. IOS = Inclusion of Other in the Self Scale.

†

p < .10. *p < .05. **p < .01.

Inclusion of Other in the Self

The expected three-way interaction between self-esteem, pretest instrumentality, and the contrast comparing the high-risk and the control condition was significant, b = −1.21, t(86) = −3.17, p = .002.⁴ The expected and opposite self-esteem by risk interactions also emerged for people with more instrumental, b = −1.50, t(86) = −2.87, p = .005, and less instrumental partners, b = 1.29, t(86) = 2.18, p = .032.

Figures 1a and 1b contain the predicted scores. Inspecting the simple effects of risk revealed the hypothesized effects. When people were more dependent on their partner for their academic pursuits (i.e., the partner was more instrumental at pretest), posing a threat to the relationship prompted high self-esteem people to draw closer to their partner and low self-esteem people to distance. That is, high self-esteem people tended to report greater closeness to instrumental partners in the high-risk than control condition, b = −.98, t (86) = −1.83, p = .07. In contrast, low self-esteem people reported less closeness to instrumental partners in the high-risk than control condition, b = 1.79, t(86) = 2.30, p = .024. Conversely, when people were less dependent on their partner for their academic pursuits (i.e., the partner was less instrumental at pretest), posing a relationship threat had the opposite effects. That is, high self-esteem people reported less closeness to noninstrumental partners in the high-risk than control condition, b = 1.53, t(86) = 2.06, p = .042. In contrast, low self-esteem people reported greater closeness to noninstrumental partners in the high-risk than control condition, although this simple effect was not significant, b = −.85, t(86) = −1.42, p = .158.

Figure 1.

Predicted values of Inclusion of Other in the Self for participants with (a) highly instrumental and (b) noninstrumental partners in Study 1.

Inspecting the simple effects of self-esteem also revealed the expected risk regulation effects. For participants with instrumental partners in the high-risk condition, being higher in self-esteem predicted increased closeness, b = 1.70, t(86) = 3.74, p < .001, the prototypic risk regulation effect. But, for participants with noninstrumental partners in the high-risk condition, being lower in self-esteem predicted increased closeness, b = −1.34, t(86) = −2.98, p = .004, the reversed risk regulation effect. The corresponding simple effects of self-esteem were not significant in the control condition.

Manipulation check

As expected, participants in the high-risk condition (M = 4.40) were more likely to report that their responses to the measure indicated that their relationships were somewhat worse than average than were participants in the low-risk condition (M = 2.06), t(63) = 5.14, p < .001.

Discussion

When people’s symbolic dependence on their partner for their academic success was heightened (i.e., when the partner was highly instrumental), low and high self-esteem people evidenced their prototypic risk-regulatory responses. Highs connected and promoted the relationship in response to risk, whereas lows distanced. But, when people’s symbolic dependence on their partner for their academic success was lessened (i.e., the partner was less instrumental), low and high self-esteem people evidenced opposite responses. Low self-esteem people connected to their partner in response to risk, whereas high self-esteem people distanced.

Several potential alternative explanations should be addressed. We believe that the partner’s initial instrumentality to the participant’s academic goals played its moderating role because participants were engaged in a goal-related activity. Once engaged in academic goal pursuit, the partner’s instrumentality for the participant’s current scholarly pursuits then helped define people’s symbolic dependence on their partner to achieve the goal. However, we did not manipulate instrumentality. Therefore, it is possible that the moderating effects of the partner’s instrumentality did not emerge because of the partner’s utility (or lack of utility) to the academic goal context we created. Instead, the moderating effects of partner instrumentality might simply reflect the effects of perceiving one’s partner as generally supportive of or facilitating one’s goals.

If that were the case, the partner’s instrumentality to nonactive goals (e.g., fitness and social goals) might produce similar moderating effects. To evaluate this possibility, we conducted follow-up analyses in which partners’ instrumentality to participants’ fitness and social goals replaced academic instrumentality as a moderator. The three-way interaction between fitness goal instrumentality, self-esteem, and risk did not significantly predict IOS (p = .23). Similarly, the three-way interaction between social goal instrumentality, self-esteem, and risk did not significantly predict IOS (p = .66). Moreover, the predicted three-way academic instrumentality by self-esteem by risk interaction predicting IOS remained robust (p = .007) when we controlled for the interactions between self-esteem, risk, and instrumentality to other goals (and their constituent lower order effects). In this analysis, neither of the latter three-way interactions significantly predicted IOS.

Participants’ perceptions of their partner’s instrumentality could also reflect a broader construct, such as participants’ general feelings of closeness to their partners or participants’ own perceived academic success. To address these possibilities, we conducted follow-up analyses in which pretest closeness and success replaced instrumentality as moderators. The three-way self-esteem by risk by closeness interaction did not significantly predict IOS (p = .94). Moreover, the predicted three-way partner instrumentality by self-esteem by risk interaction remained marginally significant (p = .099) when we controlled for interactions between closeness, self-esteem, and risk, but the three-way interaction between closeness, self-esteem, and risk remained nonsignificant. In addition, the three-way perceived success by self-esteem by risk interaction did not significantly predict IOS (p = .71). The predicted partner instrumentality by self-esteem by risk interaction also remained significant when the interactions between success, self-esteem, and risk were controlled (p = .005), whereas the three-way interaction between perceived success, self-esteem, and risk remained nonsignificant. These analyses provide discriminate validity for our measure of partner instrumentality, distinguishing it from other potentially related constructs.

Finally, it is possible that our findings were driven by social comparison processes. For instance, high self-esteem people derogate downward social comparison targets in response to self-threats (Gibbons & McCoy, 1991). Low self-esteem people, on the other hand, do not derogate downward social comparison targets but instead experience positive outcomes just by having such a standard available. Such findings raise the possibility that the partners’ own academic aptitude, rather than their ability to facilitate participants’ goals, may have influenced how low and high self-esteem participants regulated responses to risk. To address this potential alternative, we conducted analyses in which we replaced pretest instrumentality with partner’s relative academic performance as a moderator. The three-way partner performance by self-esteem by risk interaction did not significantly predict IOS (p = .16). In addition, the predicted partner instrumentality by self-esteem by risk interaction remained robust when the interactions between partner performance, self-esteem, and risk were controlled (p = .006), whereas the partner performance by self-esteem by risk interaction remained nonsignificant. Taken together, these follow-up analyses lend support to our contention that a partner’s perceived instrumentality to one’s currently active goal pursuits can alter people’s symbolic dependence on their partner, leading to shifts in risk regulation processes.

Study 2

Study 1 yielded promising initial support for our hypotheses, but not all of the expected simple effects were significant across measures. We designed Study 2 as a conceptual replication using a different manipulation of risk, a different academic task, and expanded dependent variables. We again measured perceptions of a partner’s academic instrumentality in a pretesting session. In the subsequent experiment, participants in the high-risk condition read a textbook passage suggesting that couples often overestimate the quality of their relationships. Participants in the low-risk condition read a passage suggesting that couples often underestimate the quality of their relationships. To elicit academic goal engagement, participants completed a verbal skills task in which they formed as many words as possible from a string of letters. In Study 1, we only mentioned to participants that we were assessing verbal ability in the cover story. Although we found the predicted effects, it may not have been salient to participants that the GRE analogy problems were assessing academic skills. To address this limitation, we explicitly described the verbal skills task in Study 2 as diagnostic of verbal ability. We then measured feelings of closeness and evaluations of the partner’s interpersonal qualities to gauge different means by which people can regulate risk. When partners were instrumental to academic goals, we expected high self-esteem people to connect and promote their relationship in the face of risk (i.e., increase closeness and partner-valuing) and low self-esteem people to distance (i.e., decrease closeness and partner-valuing). However, when partners were not instrumental to academic goals, we expected low self-esteem people to connect and value their partner more in response to risk and high self-esteem people to distance and devalue their partner.

Method

Participants

Ninety-four participants were enrolled in the second study in exchange for course credit in their Introductory Psychology course. Two participants were dropped because they did not complete the verbal ability task, leaving a final sample of 92 (44 men). Participants had been in their relationships for an average length of 18.27 months (SD = 16.10). The mean age of the participants was 19.26 (SD = 1.77).

Procedure

During a mass testing session at the beginning of the semester, participants completed measures assessing their partners’ instrumentality to their academic goals, the personal importance of their academic goals, and their partners’ relative academic performance. At the lab, participants learned that the experiment would examine how experiences in people’s relationships and daily lives affect their verbal thinking. Using MediaLab software, participants first completed demographic measures as well as the Rosenberg (1965) Self-Esteem Scale.

Next, participants read a passage allegedly from an upcoming social psychology textbook summarizing some research from our lab.⁵ In reality, this passage was fabricated to manipulate relationship risk (Cavallo et al., 2009). Participants were told to read the passage carefully and to try to remember the information because they would be asked to recall details from the passage at the end of the study. In the high-risk condition, the passage indicated that research has found that couples generally overestimate the quality of their relationships. In support of this point, the passage described a series of events in a fictional couple’s week (e.g., cooperating with household tasks). Although these events were fairly commonplace, the passage described the couple’s behaviors as evidence that they were overlooking signs of disregard for one another’s feelings. This passage was intended to create doubts about the perceived quality of participants’ own relationships. In the low-risk condition, the passage indicated that research has found that couples generally underestimate the quality of their relationships. The passage supported this point by describing the same behaviors as evidence of good relationship functioning. This manipulation has been used successfully to activate risk regulation processes in prior research (see Cavallo et al., 2009).

To elicit academic goal pursuit, all participants then completed a measure of verbal ability. Participants were told that the measure “tends to be diagnostic of your overall verbal abilities, and can thus predict success at such tasks as the verbal sections of standardized tests for graduate, law, and medical school admissions, and also vocabulary and essay writing skills” (Fitzsimons & Bargh, 2003). In the task, participants were given seven letters and were instructed to form as many words as possible from the letters. They were given 5 min to complete the task. Next, participants completed the dependent measures. They first completed the IOS Scale to provide a measure of psychological closeness (A. Aron et al., 1992). Then they rated their partner’s interpersonal qualities to provide a measure of partner-valuing (Murray et al., 2002). After some filler items, participants were asked to list as many details as they could remember from the passage, were debriefed, and were thanked for their participation.

Measures

Interpersonal qualities

Participants rated their partners on 20 interpersonal attributes (Murray et al., 2000). The attributes included positive qualities such as “warm” and “responsive” as well as negative qualities such as “lazy” and “critical or judgmental,” reversed. Ratings were made on a 9-point scale (1 = not at all characteristic, 9 = very characteristic).

Results

We conducted hierarchical regression analyses to test our predictions. In the initial step, we entered the main effects of the centered continuous variables and condition (high-risk = 1 and low-risk = −1). In the initial step, we also entered the covariates included in Study 1 (goal importance and participants’ partners’ relative academic performance). In the subsequent steps, we entered all possible two- and three-way interaction terms between self-esteem, pretest partner instrumentality, and condition. Table 2 contains the results.

Table 2.

Summary of Regression Analyses in Study 2.

	IOS		Partner-valuing
	β	t	β	t
Goal importance	.14	1.33	−.01	−0.06
Partner performance	−.21	1.93^†	.05	0.42
Pretest instrumentality	.07	0.63	.12	1.04
Self-esteem	.12	1.18	.17	1.54
Risk	.06	0.57	.02	0.19
Self-esteem × Instrumentality	.23	2.08*	−.01	−0.09
Self-esteem × Risk	.05	0.50	.20	1.87^†
Instrumentality × Risk	−.04	−0.39	.07	0.66
Self-esteem × Instrumentality × Risk	.41	3.77**	.25	2.17*

Note. IOS = Inclusion of Other in the Self Scale.

†

p < .10. *p < .05. **p < .01.

Inclusion of the Other in the Self Scale

The expected interaction between instrumentality, self-esteem, and risk was significant, b = .41, t(79) = 3.77, p < .001. The expected and opposite self-esteem by risk interactions also emerged for people with more instrumental, b = .57, t(79) = 3.22, p = .002, and less instrumental partners, b = −.44, t(79) = −2.23, p = 03.

Figures 2a and 2b present the predicted scores. Inspecting the simple effects of risk generally revealed the hypothesized effects. When people were more dependent on their partner for their academic pursuits (i.e., when the partner was more instrumental at pretest), increasing the risks of connection led high self-esteem participants to draw closer to their partner and low self-esteem participants to distance. High self-esteem participants reported greater closeness to instrumental partners in the high-risk condition than the low-risk condition, b = .60, t(79) = 2.28, p = .025. In contrast, low self-esteem participants reported less closeness to instrumental partners in the high-risk than low-risk condition, b = −.56, t(79) = −2.27, p = .03. Conversely, when people were less dependent on their partner for their academic pursuits (i.e., when the partner was less instrumental at pretest), increasing the risks of connection tended to have the opposite effects. High self-esteem people tended to feel less close to noninstrumental partners in the high-risk condition, although this simple effect was not significant, b = −.32, t(79) = − 1.12, p = .27. In contrast, low self-esteem participants felt significantly closer to noninstrumental partners in the high-risk than the low-risk condition, b = .58, t(79) = 2.00, p = .049.

Figure 2.

Predicted values of Inclusion of Other in the Self Scale for participants with (a) highly instrumental and (b) noninstrumental partners in Study 2.

The simple effects of self-esteem also revealed the predicted risk regulation effects. For participants with instrumental partners in the high-risk condition, being higher in self-esteem predicted increased reports of closeness, b = .98, t(79) = 3.57, p = .001. But, for participants with noninstrumental partners in the high-risk condition, being lower in self-esteem tended to predict increased closeness, b = −.54, t(79) = −1.76, p = .082. The corresponding simple effects of self-esteem were not significant in the control condition.

Partner-valuing

The predicted three-way interaction between pretest instrumentality, self-esteem, and risk predicting partner-valuing was significant, b = .23, t(79) = 2.17, p = .03. The predicted self-esteem by risk interaction emerged for people with more instrumental partners, b = .52, t(79) = 3.03, p = .003, but not for people with less instrumental partners, b = −.04, t (79) = −.23, p = .82.

Figures 3a and 3b contain the predicted scores. When people were more dependent on their partner for their academic pursuits (i.e., when partners were more instrumental at pretest), high self-esteem participants valued their partners more when the risks of connection were increased, whereas low self-esteem participants valued their partners less. That is, high self-esteem participants valued their instrumental partners more in the high-risk than low-risk condition, b = .64, t(79) = 2.50, p = .014. In contrast, low self-esteem participants tended to value their instrumental partners less in the high-risk than low-risk condition, b = −.41, t(79) = −1.74, p = .085. Unexpectedly, when people were less dependent on their partner for their academic pursuits (i.e., when partners were less instrumental at pretest), the simple effects of risk were not significant for either high self-esteem, b = −.11, t(79) = −0.39, p = .695, or low self-esteem participants, b = −.02, t(79) = −0.07, p = .945. Paralleling these effects, the simple effect of self-esteem revealed the predicted risk regulation effect for people with instrumental partners in the high-risk condition; being higher in self-esteem predicted increased partner-valuing, b = .68, t (79) = 2.56, p = .012. The corresponding simple effect was not significant for people with noninstrumental partners in the high-risk condition or in the control condition.

Figure 3.

Predicted values of partner-valuing for participants with (a) highly instrumental and (b) noninstrumental partners in Study 2.

Discussion

These findings partially replicate and extend those from the first study. Namely, when people were more dependent on their partner for the academic goals they were pursuing (i.e., when partners were more instrumental), high self-esteem participants drew closer to their partners and valued their partners more in the high-risk than low-risk condition, whereas low self-esteem participants distanced and devalued their partners. Conversely, when people were less dependent on their partner for the academic goals they were pursuing (i.e., when partners were less instrumental), high self-esteem participants felt less close to their partner in the high-risk than low-risk condition, whereas low self-esteem participants felt closer to their partner in the high-risk than low-risk condition.

Meta-Analysis

The predicted, significant three-way interactions between pretest partner instrumentality, self-esteem, and risk predicting psychological closeness (i.e., the IOS Scale) emerged consistently across studies. But the lower order effects were not perfectly consistent across studies. To examine the robustness of the lower order effects, we conducted a meta-analysis of the effects for the IOS using Winer’s (1971) method of combined t tests. The two-way risk by self-esteem interactions were significant for both participants with highly instrumental partners, z = 4.25 p < .05, and participants with less instrumental partners, z = −3.09, p < .05. Among participants with highly instrumental partners, the predicted simple effects of risk were robust for participants who were high in self-esteem, z = 2.87, p < .05, and for participants who were low in self-esteem, z = −3.23, p < .05. In addition, among participants with less instrumental partners, the predicted effects of risk were robust for participants with high self-esteem, z = −2.22, p < .05, and for participants with low self-esteem, z = 2.42, p < .05. The expected simple effects of self-esteem were also significant. That is, for participants with highly instrumental partners in the high-risk condition, being higher in self-esteem predicted increased closeness, z = 5.11, p < .05. However, for participants with less instrumental partners in the high-risk condition, being lower in self-esteem predicted increased closeness, z = −3.31, p < .05.

General Discussion

The present studies suggest that subtle shifts in the nature of the interdependent situation people confront can alter risk regulation dynamics dramatically. We created a context (i.e., pursuing an academic goal) in which the potential rewards and costs of depending on the partner varied according to the partner’s instrumentality to the participant’s academic goals. When the context heightened dependence on the partner (i.e., when partners were instrumental to the academic goals participants were pursuing), low and high self-esteem people evidenced their typical responses to risk. High self-esteem participants reported feeling closer to their partners and perceived their partners as even more valued when the risks of interdependence were primed (as compared with the control condition). In contrast, low self-esteem participants reported less closeness to instrumental partners in the high-risk than control condition. They also devalued their instrumental partners in the high-risk as compared with the control condition. However, when the context we created minimized people’s dependence on the partner (i.e., when partners were less instrumental to the academic goals participants were pursuing), these well-established risk regulation effects were largely reversed. High self-esteem people reported feeling less close to noninstrumental partners in the risk than control condition. But, low self-esteem people drew closer to noninstrumental partners in the high-risk than control condition, only allowing themselves to rely on their partner when they did not need them quite so much, and thus, the likelihood and pain of rejection was minimized.

In sum, when people were pursuing an active goal that an instrumental partner could facilitate, such heightened dependence seemed to motivate high self-esteem people to connect to their partner in the face of risk and low self-esteem people to distance. But, when people were pursuing a goal that a noninstrumental partner might only impede, such restricted dependence motivated low self-esteem people to connect to their partner in the face of risk and high self-esteem people to distance. The present findings speak to the power of even subtle shifts in interdependence because the partner was not physically present to aid or hinder participants’ academic goals. Instead, simply being more or less symbolically dependent on the partner as a needed resource may be sufficient to direct risk regulation dynamics.

The current findings have several further strengths. We obtained these effects using two different manipulations of risk and two different methods of fostering academic goal pursuit. We also showed that subtle contextual variations in participants’ dependence on the rewards more or less instrumental partners can offer predict reversals in high and low self-esteem people’s tendencies to connect to or distance from their partners in the face of risk. Although high self-esteem people generally trust in their partner’s regard (Murray et al., 2000) and typically risk closeness in the face of risk (Murray et al., 2002), their willingness to connect to their partners appears to be reduced when they have less to gain from depending on their partner. Conversely, although low self-esteem people generally question their partners’ responsiveness and are reluctant to risk closeness in the face of risk, they seem to feel more comfortable connecting to their partners when they have less to gain by depending on them.

These findings also extend previous work demonstrating that high and low self-esteem individuals differ in their broad motivational responses to risk. High self-esteem people become more oriented to both relational and nonsocial rewards when faced with interpersonal risk (Cavallo et al., 2009). Our findings suggest that when the rewards of depending on one’s partner are limited by one’s current goal pursuits, highs may distance from their partners to pursue rewards in other available domains, such as academic achievement. In contrast, low self-esteem people typically respond to interpersonal risk by distancing from their partner. However, our findings suggest that when lows feel relatively safe because the goals they are pursuing make them less dependent on their partner, they then become more willing to connect.

Limitations and Directions for Future Research

Although the present research is the first to show that subtle shifts in the situational structure of interdependence can shift risk regulation dynamics, it has limitations. First, we measured, not manipulated, perceptions of the partner’s instrumentality to one’s academic goals. Consequently, the hypothesized effects of being more or less dependent on the partner might actually reflect the effects of a related construct, such as the partner’s general supportiveness, feelings of closeness to the partner, participants’ own perceived success, or the nature of the academic comparisons one makes with the partner. Study 1 provided evidence against all of these possibilities. When we compared the partner’s instrumentality with one’s academic goals against such likely competitors, we only observed moderating effects for academic instrumentality. The predicted effects did not emerge when we utilized partners’ fitness or social goal instrumentality, pretest closeness, perceived success, or partners’ academic performance as moderators. Nonetheless, the present evidence cannot firmly establish the causal role of instrumentality and dependence in risk regulation processes.

Second, we did not manipulate goal pursuit. We believe that engaging in an academic goal pursuit increased participants’ symbolic dependence on instrumental partners and decreased dependence on noninstrumental partners. But, we cannot prove that academic goal engagement was necessary to elicit the effects because we held academic goal pursuit constant across participants. Thus, our design leaves open the possibility that we would have found exactly the same moderating effects of the partner’s academic instrumentality had participants pursued an alternate, nonacademic goal. We think this is unlikely. If the match between the partner’s instrumentality and current goal pursuits was immaterial to the effects, then any variable that captured the partner’s instrumentality to the self should have similarly moderated risk regulation dynamics. But, in Study 1, neither the partner’s instrumentality to one’s social goals, nor the partner’s instrumentality to one’s fitness goals, nor one’s initial feelings of closeness to the partner significantly moderated the effects. This suggests that dependence—that is, the match between the partner’s instrumentality and one’s current goal pursuits—is crucial.

Nevertheless, it is still possible that completing the academic task did not just engage participants in a goal pursuit, but instead primed other constructs, such as success or failure. For instance, perhaps completing the academic task primed success in people with instrumental partners and failure in people with noninstrumental partners. However, performance on the academic task did not vary as a function of partner instrumentality in either study. We also reanalyzed our data controlling for participants’ performance on the verbal skills tasks. The predicted three-way self-esteem by instrumentality by risk interactions predicting our dependent variables remained significant in both studies. Moreover, performance on the verbal skills tasks did not significantly moderate the predicted interaction in either study. Furthermore, academic success and failure both trigger relational insecurities in low self-esteem people (Murray, Holmes, MacDonald, & Ellsworth, 1998). If completing the verbal skills task had primed academic success in people with instrumental partners and failure in people with noninstrumental partners, low self-esteem people should have distanced from their partners in response to risk regardless of whether their partners were instrumental or not.

Prior research has utilized engaging in an academic goal to manipulate academic goal pursuits using procedures very similar to the ones we used (Ferguson & Bargh, 2004; Moore et al., 2011). But, the specific verbal task we utilized as an induction of goal pursuit has been used as a dependent measure indexing goal pursuit in other work (Fitzsimons & Bargh, 2003). In using this verbal task to induce rather than index goal pursuit, we are not trying to challenge the arguments advanced by Fitzsimons and Bargh (2003). Instead, despite utilizing goal-related tasks in different ways, both sets of findings illuminate the dynamic interplay between people’s mental representations of their partners and their ongoing goal pursuits.

Third, Study 2 did not reveal the expected risk regulation effects on partner-valuing among participants with noninstrumental partners. Although this null effect bears replication, it might be the case that reducing psychological closeness to noninstrumental partners may be more relevant to effective goal pursuit than devaluing noninstrumental partners. That is, minimizing psychological closeness might reduce the possibility that a noninstrumental partner’s physical or psychological presence will interfere with goal pursuit, whereas derogating the partner may not have such self-regulatory benefits. In fact, effective self-regulators reduce closeness to noninstrumental others when a relevant goal is primed, whereas ineffective self-regulators do not make these goal-dependent shifts in closeness (Fitzsimons & Shah, 2008).

The strengths and weaknesses of the studies also help point to questions for future research. Further research might utilize manipulations of the participant’s ongoing goal pursuits, the partner’s instrumentality, or the partner’s physical presence to shift the bases of dependence. Future research might also explore the relational implications of the present findings. High self-esteem people might experience self-regulatory benefits from distancing from their partners when their partner might hinder their current goal pursuits (Fitzsimons & Shah, 2008), but distancing may have detrimental effects on their relationships across time, such as decreasing the noninstrumental partner’s commitment or trust. Conversely, low self-esteem people may not reap self-regulatory benefits from drawing closer to noninstrumental partners, but drawing closer in response to risk might benefit their relationships. For example, lows who draw closer to noninstrumental partners might respond less destructively to partner transgressions (Marigold et al., 2010) or experience increases in their state of self-esteem (Murray et al., 2005).

Conclusion

The present studies revealed new and intriguing facets of the interplay between the rewarding and costly aspects of interdependence. The rewards and costs that come from being more or less dependent on a romantic partner seem to pull low and high self-esteem people in different directions. Amplifying dependence seems to increase high self-esteem people’s willingness to connect with their partner in the face of risk and decrease low self-esteem people’s willingness to connect. Conversely, minimizing dependence seems to increase low self-esteem people’s willingness to connect with their partner in the face of risk and decrease high self-esteem people’s willingness to connect. Needing a partner more thus seems to motivate high self-esteem people to connect in the face of risk and lows to distance, whereas needing the partner less instead motivates low self-esteem people to connect and highs to distance.

Footnotes

Acknowledgements

We would like to thank John Holmes, Mark Seery, and Shira Gabriel for their insightful comments on earlier drafts of this manuscript, and our undergraduate research assistants for their role in collecting data.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Anthony

D. B.

Wood

J. V.

Holmes

J. G.

(2007). Testing sociometer theory: Self-esteem and the importance of acceptance for social decision-making. Journal of Experimental Social Psychology, 43, 425-432.

Aron

E. N.

Smollan

(1992). Inclusion of Other in the Self Scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63, 596-612.

Aron

Paris

Aron

E. N.

(1995). Falling in love: Prospective studies of self-concept change. Journal of Personality and Social Psychology, 69, 1102-1112.

Berscheid

(1983). Emotion. In Kelley

H. H.

Berscheid

Christensen

Harvey

J. H.

Huston

T. L.

Levinger

Peterson

D. R.

(Eds.), Close relationships (pp. 110-168). New York, NY: Freeman.

Campbell

Simpson

J. A.

Boldry

J. G.

Rubin

(2010). Trust, variability in relationship evaluations and relationship processes. Journal of Personality and Social Psychology, 99, 14-31.

Cavallo

J. V.

Fitzsimons

G. M.

Holmes

J. G.

(2009). Taking chances in the face of threat: Romantic risk regulation and approach motivation. Personality and Social Psychology Bulletin, 35, 737-751.

Cavallo

J. V.

Holmes

J. G.

Fitzsimons

G. M.

Murray

S. L.

Wood

J. V.

(2012). Managing motivational conflict: How self-esteem and executive resources influence self-regulatory responses to risk. Journal of Personality and Social Psychology, 103, 430-451.

Cavallo

J. V.

Murray

S. L.

Holmes

J. G.

(2013). Regulating interpersonal risk. In Simpson

J. A.

Campbell

(Eds.), Oxford handbook of close relationships (pp. 116-136). New York, NY: Oxford University Press.

Collins

N. L.

Feeney

B. C.

(2004). Working models of attachment shape perceptions of social support: Evidence from experimental and observational studies. Journal of Personality and Social Psychology, 87, 363-383.

10.

Davila

Karney

B. R.

Bradbury

T. N.

(1999). Attachment change processes in the early years of marriage. Journal of Personality and Social Psychology, 76, 783-802.

11.

Derrick

J. L.

Leonard

K. E.

Homish

G. G.

(2012). Dependence regulation in newlywed couples: A prospective examination. Personal Relationships, 19, 644-662.

12.

Ferguson

M. J.

Bargh

J. A.

(2004). Liking is for doing: The effects of goal pursuit on automatic evaluation. Journal of Personality and Social Psychology, 87, 557-572.

13.

Fishbach

Zhang

Trope

(2010). Counteractive evaluation: Asymmetric shifts in the implicit value of conflicting motivations. Journal of Experimental Social Psychology, 46, 29-38.

14.

Fitzsimons

G. M.

Bargh

J. A.

(2003). Thinking of you: Nonconscious pursuit of interpersonal goals associated with relationship partners. Journal of Personality and Social Psychology, 84, 148-164.

15.

Fitzsimons

G. M.

Fishbach

(2010). Shifting closeness: Interpersonal effects of personal goal progress. Journal of Personality and Social Psychology, 98, 535-549.

16.

Fitzsimons

G. M.

Shah

J. Y.

(2008). How goal instrumentality shapes relationship evaluations. Journal of Personality and Social Psychology, 95, 319-337.

17.

Gibbons

F. X.

McCoy

S. B.

(1991). Self-esteem, similarity, and reactions to active versus passive downward social comparisons. Journal of Personality and Social Psychology, 60, 414-424.

18.

Impett

E. A.

Strachman

Finkel

E. J.

Gable

S. L.

(2008). Maintaining sexual desire in intimate relationships: The importance of approach goals. Journal of Personality and Social Psychology, 94, 808-823.

19.

Kelley

H. H.

(1979). Personal relationships: Their structures and processes. Hillsdale, NJ: Erlbaum.

20.

Leary

M. R.

Tambor

E. S.

Terdal

S. K.

Downs

D. L.

(1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68, 518-530.

21.

Lykken

D. T.

Tellegen

(1993). Is human mating adventitious or the result of lawful choice? A twin study of mate selection. Journal of Personality and Social Psychology, 65, 56-68.

22.

Marigold

D. C.

Holmes

J. G.

Ross

(2010). Fostering relationship resilience: An intervention for low self-esteem individuals. Journal of Experimental Social Psychology, 46, 624-630.

23.

Moore

S. G.

Ferguson

M. J.

Chartrand

T. L.

(2011). Affect in the aftermath: How goal pursuit influences implicit evaluations. Cognition & Emotion, 25, 453-465.

24.

Murray

S. L.

Derrick

J. L.

Leder

Holmes

J. G.

(2008). Balancing connectedness and self-protection goals in close relationships: A levels-of-processing perspective on risk regulation. Journal of Personality and Social Psychology, 94, 429-459.

25.

Murray

S. L.

Griffin

D. W.

Rose

Bellavia

(2006). For better or worse? Self-esteem and the contingencies of acceptance in marriage. Personality and Social Psychology Bulletin, 32, 866-882.

26.

Murray

S. L.

Holmes

J. G.

(2009). The architecture of interdependent minds: A motivation-management theory of mutual responsiveness. Psychological Review, 116, 908-928.

27.

Murray

S. L.

Holmes

J. G.

Collins

N. L.

(2006). Optimizing assurance: The risk regulation system in relationships. Psychological Bulletin, 132, 641-666.

28.

Murray

S. L.

Holmes

J. G.

Griffin

D. W.

(2000). Self-esteem and the quest for felt security: How perceived regard regulates attachment processes. Journal of Personality and Social Psychology, 78, 478-498.

29.

Murray

S. L.

Holmes

J. G.

Griffin

D. W.

Bellavia

Rose

(2001). The mismeasure of love: How self-doubt contaminates relationship beliefs. Personality and Social Psychology Bulletin, 27, 423-436.

30.

Murray

S. L.

Holmes

J. G.

MacDonald

Ellsworth

(1998). Through the looking glass darkly? When self-doubts turn into relationship insecurities. Journal of Personality and Social Psychology, 75, 1459-1480.

31.

Murray

S. L.

Leder

MacGregor

J. C. D.

Holmes

J. G.

Pinkus

Harris

(2009). Becoming irreplaceable: How comparisons to the partner’s alternatives differentially affect low and high self-esteem people. Journal of Experimental Social Psychology, 45, 1180-1191.

32.

Murray

S. L.

Rose

Bellavia

Holmes

Kusche

(2002). When rejection stings: How self-esteem constrains relationship-enhancement processes. Journal of Personality and Social Psychology, 83, 556-573.

33.

Murray

S. L.

Rose

Holmes

Derrick

Podchaski

E. J.

Bellavia

Griffin

(2005). Putting the partner within reach: A dyadic perspective on felt security in close relationships. Journal of Personality and Social Psychology, 88, 327-347.

34.

Overall

N. C.

Sibley

C. G.

(2009). Attachment and dependence regulation within daily interactions with romantic partners. Personal Relationships, 16, 239-262.

35.

Reis

H. T.

Shaver

P. R.

(1988). Intimacy as an interpersonal process. In Duck

(Ed.), Handbook of research in personal relationships (pp. 367-389). London, England: Wiley.

36.

Rosenberg

(1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.

37.

Shah

J. Y.

(2003a). Automatic for the people: How representations of significant others implicitly affect goal pursuit. Journal of Personality and Social Psychology, 84, 661-681.

38.

Shah

J. Y.

(2003b). The motivational looking glass: How significant others implicitly affect goal appraisals. Journal of Personality and Social Psychology, 85, 424-439.

39.

Vohs

K. D.

Heatherton

T. F.

(2001). Self-esteem and threats to self: Implications for self-construals and interpersonal perceptions. Journal of Personality and Social Psychology, 81, 1103-1118.

40.

Winer

B. J.

(1971). Statistical principals in experimental design (2nd ed.). New York, NY: McGraw-Hill.