INTERPERSONAL RELATIONS AND GROUP PROCESSES Diminishing Self-Disclosure to Maintain Security in Partners'Care UaiverPark UMichee Six studies desire to bond with targ y find ary effe larly.in Study 3 dhiehlevecO lf-di r of tar ive and s or ons (St Keywords:trust.responsiveness.disclosure.bias,motivation Trust that a partner values and cares for the self is an importan observations of target partners'behavior in diagnostic situation uations in which partn tates the dev intimacy (Lauren Barrett.P 0n,201 2007g or the 2007:M 000.n ommo ating respo and desirs of p Michaelis.Kh y eeds or de ation to ld pon to those nee But is trust s so dependent on the information Are per helples d the he onv v.Un ted by National Science Foundation Res ey have disclosed th ir needs and desires to partners,in a manne may. members of h data colle Angel ers'goals ning th Self-Disclosure and Situational Diagnosticity
INTERPERSONAL RELATIONS AND GROUP PROCESSES Diminishing Self-Disclosure to Maintain Security in Partners’ Care Edward P. Lemay Jr. University of Maryland, College Park Michael C. Melville University of New Hampshire Six studies demonstrate that perceivers’ desire to bond with targets motivates perceivers to misconstrue their own self-disclosure in ways that maintain perceivers’ security in targets’ care and commitment. Perceivers who strongly valued relationships with targets reported high levels of global self-disclosure, consistent with many findings suggesting salutary effects of disclosure. However, these same perceivers reported low self-disclosure of needs and desires in hypothetical (Study 1) and actual (Study 2) situations characterized by targets’ unresponsive behavior. Similarly, in daily report (Study 3) and behavioral observation (Study 4) studies, perceivers who valued relationships with targets perceived high levels of self-disclosure when targets were responsive, but they perceived low self-disclosure when targets were unresponsive, and these perceptions seemed partly illusory. In turn, these perceptions of low selfdisclosure in situations characterized by partners’ unresponsive behavior predicted decreased perceptions of diagnosticity of targets’ behavior (Studies 1–3) and buffered the negative affective and interpersonal effects of unresponsive behavior (Study 4). Experimental manipulations (Studies 5 and 6) demonstrated the motivational nature of perceived self-disclosure. Collectively, the results suggest that a desire to bond with targets motivates perceivers to downplay the diagnosticity of targets’ unresponsive behavior through diminishing their self-disclosure, in turn preserving perceivers’ trust in targets’ care and commitment. Keywords: trust, responsiveness, disclosure, bias, motivation Trust that a partner values and cares for the self is an important determinant of relationship quality (Clark & Lemay, 2010; Holmes & Rempel, 1989; Reis, Clark, & Holmes, 2004). This trust facilitates the development of intimacy (Laurenceau, Barrett, & Pietromonaco, 1998; Reis & Shaver, 1988), bolsters relationship satisfaction (Lemay, Clark, & Feeney, 2007; Murray, Holmes, & Griffin, 2000), motivates accommodating responses to interpersonal difficulties (Murray, Bellavia, Rose, & Griffin, 2003; Shallcross & Simpson, 2012; Wieselquist, Rusbult, Foster, & Agnew, 1999), and improves the likelihood of relationship persistence (Arriaga, Reed, Goodfriend, & Agnew, 2006; Downey, Freitas, Michaelis, & Khouri, 1998). Where does this trust come from? Theorizing on interpersonal trust suggests that perceivers’ trust is based on their repeated observations of target partners’ behavior in diagnostic situations— situations in which partners’ behavior is thought to be indicative of their sentiments toward perceivers (Holmes & Rempel, 1989; Kelley et al., 2003; Shallcross & Simpson, 2012; Simpson, 2007a, 2007b). Although the conceptualizations of these situations vary, self-disclosure (i.e., the expression of information about the self) of needs and desires may be an important feature of diagnostic situations. That is, perceivers may view situations as more diagnostic of partners’ relationship motivations when they have disclosed needs or desires to their partners, relative to situations in which they have not disclosed. When perceivers have disclosed needs or desires, they have the ability to gauge their partners’ motivation to respond to those needs and desires, and these observations could impact perceivers’ trust. But is trust so serendipitous, so dependent on the information gleaned from these weighty interactions? Are perceivers helpless victims (or beneficiaries) of diagnostic situations and the messages they convey? Or perhaps is diagnosticity itself in the eye of the beholder? In the current research, we test the prediction that perceivers selectively perceive diagnosticity, including whether they have disclosed their needs and desires to partners, in a manner that supports desired conclusions regarding partners’ care and commitment. In our view, people are not passive observers of diagnostic situations. Rather, the perceived diagnosticity of situations, like most interpersonal cognitions, can be driven by perceivers’ goals. Self-Disclosure and Situational Diagnosticity We propose that perceivers are more likely to perceive partners’ behaviors as diagnostic of partners’ relationship sentiments and This article was published Online First August 12, 2013. Edward P. Lemay, Jr., Department of Psychology, University of Maryland, College Park; Michael C. Melville, Department of Psychology, University of New Hampshire. This research was supported by National Science Foundation Research Grant BCS 1145349 to Edward P. Lemay, Jr. We thank members of the Interpersonal Relationships Lab, University of New Hampshire, for their assistance with data collection and coding, including Angela Neal, Kevin Cannon, Christine Coyne, Heather Balch, Manuel Lopez, Aparecio Peggins, Ryan Keene, Monica Rosskothen, Ashley Scheidegger, Choe Shannon, Kerry Spongberg, and Benjamin Stucker. Correspondence concerning this article should be addressed to Edward P. Lemay, Jr., Department of Psychology, Biology/Psychology Building, University of Maryland, College Park, MD 20742. E-mail: elemay@ umd.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology, 2014, Vol. 106, No. 1, 37–57 © 2013 American Psychological Association 0022-3514/14/$12.00 DOI: 10.1037/a0034161 37
38 LEMAY AND MELVILLE Verette.1999).events that signal partners'lack of care or com and ers"goals to and opportunity to respond in a supportive reflec and heighten That is tang ne. s due to partn lack of a the e feelings of pret s b )For instance not comm and s.Hence people may er a Most relevnt to the.people may thei self-d ern the ap isal of situa and inte upport provision from their partners (Clark&Le 2010 98) Grzelak, ction with the ally from partners who yalue and care for ther On the othe ding the partr nd therefor ependenc (s Murray et al nd loss of er and p 0/ ions of situationa nvolving on th a high risk ofr心je with thei in their partner's ca ors as the partner's true feelings and tion.Of (emyClark.)Hence.peope do sm rust by altering their view er's sentiments toward the self and relationship.whether ositive or negative may 2013) Motivated Perceptions of Lack of Disclosure o clear that the currence of the behavior is un When peonle desire to maintain a close mmunal relationshir with a p ditions welfare (Mills.Clark.Ford.Johnson.2004).they sually wan be partner still cares for the self and relationshin art ners"recin ating care (Clark,Dubash.&Mills,1). gnition Pe ability to reach desired concu y to Thi perceivers' mutual (Atridge.Berscheid.Simpson.9:Drigotas.Rusbult. unresponsive behavior without a seemingly rational justification
motivations when perceivers believe they have disclosed their needs and desires to partners. The perception of this self-disclosure may contribute to their belief that partners were aware of an opportunity to respond in a supportive or otherwise prorelationship manner, which narrows the range of attributions for partners’ unresponsive behavior. That is, when partners behave in a cold, selfish, or neglectful manner, the perception that one disclosed needs and desires reduces the perceived likelihood that partners’ behavior was due to partners’ lack of awareness, increasing the likelihood that perceivers interpret this behavior as the result of partners’ lack of care, lack of commitment, or unwillingness to sacrifice for the self and relationship. In contrast, when perceivers believe they have not communicated their needs and desires to partners, lack of awareness may be a viable explanation of partners’ negative behaviors. Hence, people may often use the clarity of their own self-disclosure of needs and desires as a way of discerning whether situations should be considered diagnostic of partners’ sentiments. A number of findings are consistent with our argument that self-disclosure serves as a source of information regarding situational diagnosticity. When needs for support arise, people tend to express those needs to their partners, and this expression tends to elicit support provision from their partners (Clark & Lemay, 2010; Collins & Feeney, 2000; Laurenceau et al., 1998). Moreover, people often use self-disclosure to build close relationships (Derlega & Grzelak, 1979; Omarzu, 2000). These findings suggest that people know that disclosure sometimes elicits responsiveness, especially from partners who value and care for them. On the other hand, by revealing potentially negative information or vulnerabilities that could be exploited, disclosure also increases the risk of rejection and loss of power, and people also seem to be aware of this risk (Baxter & Montgomery, 1996; Kelly & McKillop, 1996). For instance, people with a high need for approval (Brundage, Derlega, & Cash, 1976) and who perceive a high risk of rejection (Omarzu, 2000) are reluctant to disclose personal information about themselves, and people seem to calibrate their selfdisclosure of emotion with their confidence in their partner’s care, taking the risk of revealing emotional vulnerabilities only to the extent that they are confident that this disclosure will not be met with rejection (Lemay & Clark, 2008). Hence, people do seem to be aware that self-disclosure can create a situation that reveals the partner’s sentiments toward the self and relationship, whether positive or negative. Motivated Perceptions of Lack of Disclosure When people desire to maintain a close, communal relationship with a particular partner, as indicated by strong relationship commitment (Rusbult & Buunk, 1993) and care for the partner’s welfare (Mills, Clark, Ford, & Johnson, 2004), they usually want the partner to reciprocate these sentiments (Holmes & Rempel, 1989). Indeed, care for partners’ welfare coincides with interest in partners’ reciprocating care (Clark, Dubash, & Mills, 1998), and commitment is dependent on partners exhibiting signs of reciprocated commitment (Wieselquist et al., 1999). Perceiving such reciprocation may satisfy evolved needs to forge stable bonds characterized by mutual care (Baumeister & Leary, 1995). As these relationships depend on the motivation of both partners (Attridge, Berscheid, & Simpson, 1995; Drigotas, Rusbult, & Verette, 1999), events that signal partners’ lack of care or commitment may threaten perceivers’ goals to maintain these relationships. Moreover, perceivers’ commitment and care toward a partner reflect dependence on the relationship and heighten their feelings of vulnerability to emotional pain and more tangible losses when they are confronted with the threat of abandonment or mistreatment by the partner (Lemay, Overall, & Clark, 2012; Murray, Holmes, & Collins, 2006; Wieselquist et al., 1999). People engage in a variety of strategies to reduce these feelings of vulnerability (see Murray et al., 2006). For instance, they may engage in prorelationship behaviors that make partners feel similarly dependent on the relationship, which may restore power imbalances and reduce the perceived likelihood of rejection (Murray et al., 2009). Alternatively, they may withdraw their psychological investment in the relationship and devalue their partner as a source of connection, which could also restore power imbalances and reduce the sting of anticipated rejection (Murray et al., 2000). Most relevant to the current research, people may regulate their feelings of vulnerability through the use of motivated cognitive strategies that govern the appraisal of situations and interpretation of partners’ behaviors (see Murray et al., 2006). For instance, people who protect themselves from the threat of rejection by reducing dependence and devaluing the partner may interpret the partner’s behaviors in negative ways that facilitate and justify this distancing decision, whereas people who decide to manage feelings of vulnerability by seeking greater connection with the partner may interpret the same behavior in positive ways that provide reassurance regarding the partner’s trustworthiness and therefore motivate and justify continued dependence (see Murray et al., 2006). These cognitive strategies may target perceptions of situational diagnosticity. Strategies involving reduction of dependence on the partner may involve perceiving cold, neglectful, or selfish behaviors as especially diagnostic of the partner’s underlying sentiments and motivations. Strategies that involve maintaining connection and trust may instead include perceiving such unresponsive behaviors as irrelevant to the partner’s true feelings and relationship motives, which would help justify a decision to maintain connection. Of course, motivated perceivers could also try to regulate trust by altering their views of whether partners behaved in unresponsive ways, and some findings suggest that people do have biased perceptions (Lemay & Clark, 2008; Lemay et al., 2007) and memories (Lemay & Neal, 2013) of partners’ responsiveness. However, the presence of unresponsive behavior may sometimes be so clear that the occurrence of the behavior is undeniable, requiring perceivers to rely on other strategies to facilitate their connection goals, such as redefining the situation. Under these conditions, altering perceptions of situational diagnosticity may be a commonly used method of reaching the desired conclusion that the partner still cares for the self and relationship. However, motivation does not have an unfettered influence over cognitive process (otherwise people would never experience undesired cognitions). Perceivers’ ability to reach desired conclusions is constrained by their ability to construct seemingly reasonable justifications for those conclusions (Kunda, 1990). This constraint may be relevant to perceivers’ attempts at redefining situations. Perceivers would likely have nagging doubts about their own objectivity if they were to deny the diagnosticity of partners’ unresponsive behavior without a seemingly rational justification. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 38 LEMAY AND MELVILLE
DIMINISHING SELF-DISCLOSURE that hance relationship ingly ortive evidence We expect that perceivers ofter s that people are inaccurate a them ro tionship vity (lekes Simpson.2003:Simpson.Ickes. research we exneet that motivated n r (whether one c ar小ycon ds anc nade with less accurac and clarity tha nts of one ers behave (or are perceived to behave)in a cold,selfish,or n.Glas G hat is.pere e dis other situations pasis of asymn ne ato s of perceivers'diselo should suggest that the internal expen may chavior is not diagnos if only itw they judge situati ns as diagnc d by the self-disel of next mo 20120 982 10931 arch o 90 This prediction that targets nsive behavior not diagnostic whic and ma close re Relation to Other The retical Views on Self-Disclosure (Rusbul Van Lange 1996Th Our prediction that most vale relaionships ee lack of se urehas positive interpers of which heir en 1988)model of intima that intir commitme hancing perceptions are trengt ps thr ough a process of disclosing important information about the depth this does noe tre sentiments (Lemay Self-di osure has a consistent p itive as th relation 08 hip evaluations (e.g.,Fin 2011:B their partners in wav rick,Adle 988:La et al..1998;M .He k hat are consistent with their romantic ideals to justify theirwn
To achieve a sense of conviction in the conclusion that unresponsive behaviors are nondiagnostic, they must be able to point to seemingly supportive evidence. We expect that perceivers often underestimate their own self-disclosure of needs and desires as a way of supporting the claim that partners’ behaviors lack diagnosticity. For a number of reasons, this strategy may be especially compelling. First, this strategy depends on a judgment regarding own behavior (whether one clearly communicated needs and desires to partners). Judgments regarding one’s own behavior tend to be made with less accuracy and clarity than judgments of one’s own internal states or others’ behavior (Andersen, Glassman, & Gold, 1998; Vazire, 2010), and this may indicate greater leeway for cognitive distortion of perceivers’ own behavior relative to other strategies, such as those that involve altering perceptions of the behaviors enacted by partners. Second, insufficient disclosure may be argued on the basis of asymmetries in information regarding needs and desires conveyed to partners versus information that is privately experienced. Given that people do not usually express every nuance of their internal experiences, they may easily generate counterfactual thoughts regarding the information that would have elicited partners’ responsiveness, if only it were conveyed (e.g., “I should have expressed how important it was to me”). Third, attributing partners’ unresponsive behavior to perceivers’ own lack of disclosure shifts the responsibility for partners’ unresponsive behavior from partners to perceivers, leaving intact perceivers’ sense of control regarding partners’ responsiveness in future interactions (i.e., next time, if I more clearly disclose my needs, my partner will be responsive) and circumventing potentially destructive feelings of anger (Lemay et al., 2012; Quigley & Tedeschi, 1996; Weiner, Graham, & Chandler, 1982). Perceivers’ chronic relationship motives may determine the use of these strategies. Strong desires to maintain bonds may shift the perceived utility of the available risk-regulation strategies, such that strategies that manage vulnerability by bolstering connection and security, including underestimating disclosure of needs and desires in the face of partners’ unresponsive behavior, are preferred over strategies that manage vulnerability by reducing dependence on the partner. This prediction is consistent with interdependence theoretical perspectives on motivated interpersonal cognition. According to interdependence theory, people who strongly want to maintain close relationships (i.e., people who are high in commitment) are motivated to defend their relationships from threat (Rusbult & Van Lange, 1996; Thibaut & Kelley, 1959). Consistent with this framework, highly committed individuals perceive their relationships as superior to others’ relationships (Rusbult, Lange, Wildschut, Yovetich, & Verette, 2000), and they devalue alternative romantic partners (Johnson & Rusbult, 1989), both of which can maintain their feelings of commitment. Moreover, these commitment-enhancing perceptions are strengthened when perceivers are under threat, suggesting that they reflect motivated processes (Johnson & Rusbult, 1989; Rusbult et al., 2000). Similarly, individuals who care for their partners’ welfare tend to perceive their partners as similarly caring in return, even when this does not reflect partners’ true sentiments (Lemay & Clark, 2008; Lemay et al., 2007). Research examining positive illusions in relationships (Murray, Holmes, & Griffin, 1996a, 1996b) similarly suggests that people view their partners in ways that are consistent with their romantic ideals to justify their own commitment and that these illusions can enhance relationship quality for both partners. Research on motivated inaccuracy in close relationships also suggests that people are inaccurate at reading their partners’ minds when inaccuracy protects them from recognizing threatening information and that this process promotes relationship longevity (Ickes & Simpson, 2003; Simpson, Ickes, & Blackstone, 1995). Hence, in the current research, we expect that motivated perceivers exhibit a bias in which they underestimate their own self-disclosure in unresponsive situations—situations in which partners behave (or are perceived to behave) in a cold, selfish, or neglectful manner. That is, perceivers who strongly value targets should perceive less disclosure of their needs and desires following targets’ unresponsive behavior relative to other situations (i.e., situations in which the partner does not behave in an unresponsive manner), and comparison of disclosure perceptions to external indicators of perceivers’ disclosure should suggest that perceivers underestimate their disclosure in these unresponsive situations. In turn, underestimating disclosure should facilitate the conclusion that partners’ behavior is not diagnostic of their sentiments. In other words, motivated perceivers may exhibit a backward logic (backward relative to the logic implied in most theorizing on trust) in which they judge situations as diagnostic based on their implications for trust. They see situations are diagnostic, and as characterized by the self-disclosure of needs and desires that makes them so, when partners behave in a caring or otherwise prorelationship manner, but they claim inadequate disclosure to see situations as nondiagnostic when partners behave in unresponsive ways that could undermine trust. In addition to being consistent with research on biased judgments of interpersonal relationships, this prediction coincides with research on diagnosticity biases, which suggests that people see events as conveying desired information (Vorauer & Ross, 1993) and with general research on motivated reasoning, which suggests that people construct biased impressions, beliefs, and evaluations that support desired conclusions (Kunda, 1990). Hence, it seems plausible that motivations to maintain relationships with targets can bias perceptions of selfdisclosure, as these perceptions could support the desired conclusion that targets’ unresponsive behavior is not diagnostic, which could protect trust and maintain interdependence. Relation to Other Theoretical Views on Self-Disclosure Our prediction that perceivers who most value relationships wish to see lack of self-disclosure in some situations contradicts most theoretical frameworks involving self-disclosure, which posit that disclosure has positive interpersonal consequences and plays a key role in the maintenance of relationships. For instance, Reis and Shaver’s (1988) model of intimacy proposes that intimacy develops through a process of disclosing important information about the self to a partner and then receiving a responsive response. Similarly, social penetration theory (Altman & Taylor, 1973) proposes that relationship closeness involves increases in the depth of self-disclosure. A number of findings support these views. Self-disclosure has a consistent positive association with relationship evaluations (e.g., Finkenauer, Engels, Branje, & Meeus, 2004; Finkenauer & Righetti, 2011; Hendrick, 1981; Hendrick, Hendrick, & Adler, 1988; Laurenceau et al., 1998; Meeks, Hendrick, & Hendrick, 1998; Pollmann & Finkenauer, 2009). Moreover, metaThis document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. DIMINISHING SELF-DISCLOSURE 39
40 LEMAY AND MELVILLE 19g4.actionbetweend )whoy valuc a relationship with with our as dia hou thes perceivers would see lack of disclosu in situation y.their perception o We do not dispute that self-disclosure often has benefits.How ional facto th shane the m and con tions of di ticity of pa disclos of the ed per .In S by the partner's lack of res Hence.none of ions between omantie parmers.In Study dings e using a 6.v ived sel-dis re that we Ind ed.mo vely.to test predictions regarding the motivational nature self-disclosur that ere based n bia Study 1:Perceptions of Hypothetical Events sured in this way.high self-disclosure should be Study the the ations (thos self-d sure (ReisShaver,1988).These global me ood that perceive define the of elf-d nerceivers should feel trust when rs behave in a and cnhand manner.but lack of trust when they behave in an ion mine he diagnosticity of pa Overview ning relationships with partners.Hence,perceiv ewith their self-disclosureinspecific nzed by the ure when the partne This pr should be their partners should report me self with t views in six studies.In Studies 1 two dill sponsive situs uld b with ntions of self. must in nrtnere'care and comm lack of the globa ositive relationshin disclosure should be associated with positive relationshin percen Partner Valuing Perceived Self. Figure Concepual model guiding the present researh
analyses on self-disclosure suggest that intimate disclosures enhance attraction between disclosers and listeners (Collins & Miller, 1994). From this point of view, people who want relationships should hardly feel threatened by the perception of self-disclosure. To the contrary, their perception of disclosure may even be desired and reassuring. We do not dispute that self-disclosure often has benefits. However, we believe that a more nuanced approach can reveal the situational factors that shape the meaning and consequences of disclosure. None of the findings described above examined perceptions of self-disclosure of needs and desires in situations characterized by the partner’s lack of responsiveness. Hence, none of the findings examined the potentially threatening consequence (i.e., perceived diagnosticity of unresponsive behaviors) of perceived self-disclosure that we have emphasized. Indeed, most of the studies concluding that disclosure enhances relationships involved assessments of self-disclosure that were based on acrosssituation generalizations (i.e., the degree to which one generally discloses to a particular partner). We would agree that, when measured in this way, high self-disclosure should be associated with positive relationship functioning, as it would suggest the building of intimacy and general trust in the partner’s responsiveness to self-disclosure (Reis & Shaver, 1988). These global measures, however, may mask important contextual effects. In specific situations, self-disclosure of needs and desires should increase the likelihood that perceivers define the situation as diagnostic of partners’ sentiments. Therefore, following perceived selfdisclosure, perceivers should feel trust when partners behave in a responsive manner, but lack of trust when they behave in an unresponsive manner. Overview The conceptual model guiding the present research appears in Figure 1. According to this model, perceivers judge their own self-disclosure of needs and desires in accordance with their partner’s responsive behavior, which includes underestimating disclosure when the partner is unresponsive. This process should be stronger when perceivers highly value their relationship with the partner. In turn, given that perceived self-disclosure of needs and desires is thought to serve as a diagnosticity cue, underestimating disclosure should be directly related to perceiving the partner’s behavior as less diagnostic of the partner’s sentiments. We test these views in six studies. In Studies 1 and 2, using two different methodologies, we contrasted global perceptions of self-disclosure with perceptions of self-disclosure in specific situations characterized by target partners’ lack of responsiveness. Consistent with the large body of research suggesting that global self-disclosure has positive relationship consequences, we expected that motivated perceivers (i.e., perceivers who strongly value a relationship with targets) would claim high global self-disclosure. However, consistent with our analysis of disclosure as diagnosticity, we expected that these perceivers would see lack of disclosure in situations characterized by targets’ lack of responsiveness. Furthermore, we expected that this pattern of high global self-disclosure and low self-disclosure in situations characterized by partners’ unresponsive behavior would be associated with the most positive relationship perceptions (i.e., reduced perceptions of diagnosticity of partners’ unresponsive behavior and increased trust). In Study 3, we used a daily report study to test this model with regard to daily interactions between romantic partners. In Study 4, we tested the model using a behavioral observation study. In Studies 5 and 6, we experimentally manipulated desire to bond and reduction of threat, respectively, to test predictions regarding the motivational nature of this bias. Study 1: Perceptions of Hypothetical Events In Study 1, we compared perceivers’ global perceptions of self-disclosure of needs and desires with their perceptions of self-disclosure in unresponsive situations (those characterized by targets’ lack of responsiveness). This comparison is central to the distinction between prior research and theory on self-disclosure and the current model. As we described earlier, prior research reveals positive associations of relationship sentiments with global self-disclosure, suggesting that the existence or perception of global disclosure can maintain and enhance relationships. However, perceiving insufficient self-disclosure in unresponsive situations may undermine the diagnosticity of partners’ unresponsive behavior and therefore maintain trust in partners’ care and commitment, a desirable outcome for perceivers who strongly value maintaining relationships with partners. Hence, perceivers who strongly value their partners should be much more likely to report more global self-disclosure (i.e., that they disclose generally) relative to self-disclosure in specific situations characterized by the partner’s unresponsive behavior. In addition, perceivers who strongly value their partners should report more global selfdisclosure in their relationships relative to perceivers who do not strongly value their partners. However, when with regard to unresponsive situations, perceivers who value partners should perceive low disclosure, perhaps just as low as the disclosure perceived by those who do not value partners. This tendency to see low selfdisclosure in unresponsive situations should be associated with reduced perceptions of diagnosticity in these situations and more trust in partners’ care and commitment. In contrast, and consistent with prior research on the relational benefits of global selfdisclosure in most other contexts, global perceptions of selfdisclosure should be associated with positive relationship percepPartner's Level of + Responsiveness Perceived SelfDisclosure of Needs and Desire Perceived Diagnosticity of Partner's Behavior Partner Valuing + + Figure 1. Conceptual model guiding the present research. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 40 LEMAY AND MELVILLE
DIMINISHING SELF-DISCLOSURE bypotbetical vignettes.which allowed us to capt while Perceived global self-disclosure. suggests that studies using this methodology produce results communicate my desires to [partner name"do not clearly com examining actual events in Study 2. ompleted on 7-point response scales (I strongly disagree;7 Method M =33 low.but they were that higher values indicae local Intemet For each hypothetical se ad Sta participants completed an itema sing pe name]would do this because [partne Turkare more demo olege student sampl that quality is s a as follows:73.5%Caucasian.9.1%African American.11.3 sted or Using the s he Intemnet.We randomly assigned participants to complete the on (e.g. “Partn mel woul thi o this h se I did kedocompltet e I did no d to the devaed nditio ere asked t 77).em to p fou they th higher scores reflected greater situation-specific perceived desc 6 ompleted measures of perceived partner care and perceived par the name]want and the or a long time":Cro 91 global trust in the care and( ed the 94 the testing leted a five- item measure of care for the partner tht was arch tasks w "Crombach'Scale to bou and th Rushult.Martz Agnew.1998:ee"I want our relationshin to I am this our-item measure sire to b alued by the pa (e.g. sed no ant [partner name]to be committed to our relat ship "I wan
tions (i.e., perceiving unresponsive behaviors as nondiagnostic and trusting partners’ care and commitment). We tested these predictions using hypothetical vignettes, which allowed us to capture general explanatory styles while holding constant the specific features of interactions. Prior research on attributions in relationships suggests that studies using this methodology produce results that are parallel to studies examining attributions of actual events (Bradbury & Fincham, 1990). We directly address this issue by examining actual events in Study 2. Method Participants. A sample of 288 participants (M age 33 years; 90 males; 195 females) was recruited using two methods. Advertisements were posted on local Internet bulletin boards across the continental United States inviting participants to complete the questionnaire in exchange for entry in cash raffles. In addition, participants were recruited through Amazon Mechanical Turk, an open website that offers “workers” the ability to complete brief tasks over the Internet in exchange for a small payment. Prior research suggests that samples collected via Amazon Mechanical Turk are more demographically diverse than typical online and college student samples and that data quality is comparable (Buhrmester, Kwang, & Gosling, 2011). The racial distribution was as follows: 73.5% Caucasian, 9.1% African American, 11.3% Asian, and 6.1% other.1 Procedure. Participants completed a questionnaire posted on the Internet. We randomly assigned participants to complete the questionnaire with regard to a valued or devalued partner. Participants who were assigned to the valued partner condition were asked to complete the questionnaire with regard to someone they “see often and care very much about.” Participants who were assigned to the devalued partner condition were asked to complete the questionnaire with regard to someone they “see often but do not care very much about.” Participants then completed the measures of partner valuing and global self-disclosure described below. Next, they read four hypothetical vignettes in which this partner behaved in an unresponsive manner (i.e., the partner did not invite the participant to an event the participant wanted to attend, the partner did not seem interested in talking about the participant’s upsetting experience at school or work, the partner rejected the participant’s request for a ride to the airport, and the partner made a critical remark about the participant regarding a personal quality about which the participant was insecure). After each scenario, participants completed the scenario-specific measures described below. Finally, participants completed the measures of global trust described below. Measures. Partner valuing (manipulation check measures). Participants completed a five-item measure of care for the partner that was adapted from the Communal Strength Scale (Mills et al., 2004; e.g., “Helping [partner name] is a high priority for me”; “I care for [partner name]’s needs”; Cronbach’s .93), a five-item measure of commitment adapted from the Investment Model Scale (Rusbult, Martz, & Agnew, 1998; e.g., “I want our relationship to last for a very long time”; “I am committed to maintaining my relationship with [partner name]”; Cronbach’s .93) and a four-item measure of desire to be valued by the partner (e.g., “I want [partner name] to be committed to our relationship”; “I want [partner name] to have positive views of me”; Cronbach’s .94). Items were completed on 7-point response scales (1 strongly disagree; 7 strongly agree). Perceived global self-disclosure. Participants completed a three-item measure assessing global perceptions of their selfdisclosure of needs, desires, and feelings (e.g., “I do not clearly communicate my desires to [partner name]”; “I do not clearly communicate my needs to [partner name]”; “I do not clearly communicate my feelings to [partner name]”; Cronbach’s .92). Items were completed on 7-point response scales (1 strongly disagree; 7 strongly agree). Items were worded in the negative direction to maximize comparability with the scenario-specific measure described below, but they were reverse-scored so that higher values indicate more self-disclosure. Scenario-specific measures. For each hypothetical scenario, participants completed an item assessing perceived diagnosticity of the partner’s unresponsive behavior (e.g., “[Partner name] would do this because [partner name] did not want to spend time with me”; “[Partner name] would do this because [partner name] did not care about my feelings”; “[Partner name] would do this because [partner name] did not care about me”). Items were completed on 7-point response scales (1 strongly disagree; 7 strongly agree; Cronbach’s .93) and were averaged across the four vignettes. Higher values indicate greater perceived diagnosticity of the unresponsive behavior. Using the same response scales, participants also completed an item assessing disclosure of needs, feelings, and preferences with regard to each situation (e.g., “[Partner name] would do this because I did not clearly express my desires to go”; “[Partner name] would do this because I did not clearly express my needs to [partner name]”; “[Partner name] would do this because I did not clearly express my situation to [partner name]”; Cronbach’s .77). Items were averaged across all four vignettes and scored so that higher scores reflected greater situation-specific perceived disclosure. Global trust in partner’s care and commitment. Participants completed measures of perceived partner care and perceived partner commitment that were analogous to the own care and own commitment measures described above (e.g., “Helping me is a high priority for [partner name]”; “[Partner name] wants our relationship to last for a long time”; Cronbach’s s .88 and .91). Scores on these measures were averaged to create an index of global trust in the partner’s care and commitment (Cronbach’s .94). 1 The challenges of conducting research over the Internet are well documented (Kraut et al., 2004). These challenges include multiple submissions by the same individual, inability to control the testing environment, which introduces noise, high dropout rates, and reductions of participants’ investment of time and energy into the research tasks. We took a number of steps to address these issues. First, we tracked Internet protocol (IP) addresses and eliminated submissions that were identical to a prior submission with regard to both the IP address and the participant’s age, which should address the problem of repeat responders. Second, to address issues of reduced investment of time and energy, we tracked questionnaire completion times and eliminated all responses provided by participants who did not spend more than 5 min on the questionnaire. All analyses were conducted after implementing this rule. Third, to mitigate the issue of increased noise, we collected large samples. These procedures also were implemented in the other online studies. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. DIMINISHING SELF-DISCLOSURE 41
LEMAY AND MELVILLE Results and Discussion in the nresponsive situations were just as low with valued part To be sure that participants selected ers as with valued partners (p neasures of disclosure in unresponsive situation M elative to the devalue condition (M =3.01.2.83.and .however.should well-functionn relation ships and positive relationship perceptions and therefore should b ba spec perceptions function of participantsn of their pan diagnosticity perceptions on both situation-specific and globa pecrcptionsofscfdisclosur Situation-specifc self- ated measures on th neasure type facto The Me 10 20 m 05.P 01,and partner conc saw tar gative bcha d vhen they sav First we compared es to the global alued partne the s.Tha which produced similar result se ed by the lack of responsiveness.F(1.10)-15.14. ved cat and mitment 8-22 t(245) -4372 10.ps 001 global self-dise was positively lobal self-di sure 99 when they laim hig unre ve sure can have opposite implica. ions for trus orted ding d par in hypothetica situations featuring ived low levels o ure in unresponsive situatior pattern of high sponsive behavior and incre care and sng benefits of perceived lack of disclosure in unresponsive vith this
Results and Discussion Manipulation check. To be sure that participants selected partners who varied in intended ways, we compared scores on the measures of partner valuing (care, commitment, and desire to be valued by the partner) across the valued and devalued partner conditions. Participants reported more care for the partner (M 6.03), commitment to the relationship (M 6.22) and desire to be valued by the partner (M 6.25) in the valued partner condition relative to the devalued partner condition (M 3.01, 2.83, and 3.48, respectively), t(285) 20.75, p .001; t(284) 22.59, p .001; and t(284) 17.47, p .001; respectively. Hence, the selected partners varied in intended ways.2 Comparing global and specific perceptions of self-disclosure. We compared global and specific perceptions of self-disclosure as a function of participants’ valuing of their partner using a 2 (measure type: global or specific) 2 (partner condition: valued or devalued partner) mixed analysis of variance with repeated measures on the measure type factor. The Measure Type Partner Condition interaction was significant, F(1, 259) 78.10, 2 .23, p .001, and qualified main effects of measure type, F(1, 259) 12.21, 2 .05, p .01, and partner condition, F(1, 259) 32.84, 2 .11, p .001. Means are presented in Figure 2. First we compared responses to the global and specific measures in the devalued and valued partner conditions. That is, do perceivers perceive less self-disclosure in unresponsive situations relative to their perceptions of global disclosure? Participants in the devalued partner condition reported less global self-disclosure relative to self-disclosure in situations characterized by the partner’s lack of responsiveness, F(1, 130) 15.14, 2 .10, p .001, whereas participants in the valued partner condition reported more global self-disclosure (M 5.27) relative to self-disclosure in these unresponsive situations (M 3.86), F(1, 129) 71.87, 2 .36, p .001, consistent with our prediction that motivated perceivers would claim high global disclosure but would deny disclosure in unresponsive situations. We also compared the devalued and valued partner conditions on both global and specific measures. In other words, do perceivers who value partners report more or less global and specific self-disclosure relative to perceivers who do not value partners? Whereas participants reported more global self-disclosure to valued partners relative to devalued partners, F(1, 262) 91.77, 2 .26, p .001, their self-disclosure in the unresponsive situations were just as low with valued partners as with devalued partners (p .18), supporting our prediction that the association between self-disclosure and positive relationship sentiments applies to global measures of disclosure, but not to measures of disclosure in unresponsive situations. Predicting perceived diagnosticity and trust. We expect that perceivers who strongly value a relationship with partners are motivated to perceive low self-disclosure in situations characterized by partners’ lack of responsiveness because high self-disclosure in these situations engenders perceptions of diagnosticity. Global self-disclosure, however, should reflect well-functioning relationships and positive relationship perceptions, and therefore should be associated with perceiving unresponsive behaviors as less diagnostic. To test these predictions, we regressed situation-specific diagnosticity perceptions on both situation-specific and global perceptions of self-disclosure. Situation-specific self-disclosure (in unresponsive situations) positively predicted diagnosticity perceptions, .29, t(258) 5.66, sr2 .083, p .001, whereas global self-disclosure negatively predicted diagnosticity perceptions, .55, t(258) 10.65, sr2 .295, p .001. Consistent with our predictions, participants saw targets’ negative behavior as nondiagnostic of care—a trust-protective interpretation—when they saw low self-disclosure in unresponsive situations, and when they generally perceived their relationship to be high in self-disclosure. We tested a similar model of global trust in targets’ care and commitment, which produced similar results. Whereas selfdisclosure in unresponsive situations was inversely associated with perceived care and commitment, .22, t(245) 4.37, sr2 .046, p .001, global self-disclosure was positively associated with perceived care and commitment, .64, t(245) 12.90, sr2 .399, p .001. Hence, participants were the most trusting when they perceived lack of disclosure in situations characterized by targets’ unresponsive behavior and perceived high disclosure globally. These results provide strong evidence that global and specific perceptions of self-disclosure can have opposite implications for trust. Summary. Results of this study supported our predictions regarding differences between global perceptions of selfdisclosure and perceptions in hypothetical situations featuring partners’ unresponsive behavior. Participants who strongly valued a relationship with a target person perceived high levels of selfdisclosure in a global sense, but they also perceived low levels of self-disclosure in unresponsive situations. This pattern of high global self-disclosure and low self-disclosure in unresponsive situations was associated with reduced perceptions of diagnosticity of unresponsive behavior and increased trust in partners’ care and commitment, which is consistent with both prior research suggesting benefits of self-disclosure generally, and the current prediction proposing benefits of perceived lack of disclosure in unresponsive 2 We argued in the introduction that people who highly value relationships with partners, as indicated by high care or commitment, tend to desire reciprocation of these sentiments. Consistent with this argument, the correlations of care and commitment with desire to be valued by the partner were very strong, r(286) .84, p .001; and r(286) .89, p .001. 0 1 2 3 4 5 6 Devalued Valued Perceived Self-Disclosure Partner Condition Global Specific Figure 2. Perceived self-disclosure as a function of type of measure (global vs. specific) and partner valuing (Study 1). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 42 LEMAY AND MELVILLE
DIMINISHING SELF-DISCLOSURE "At the time of this event.Idid ot fully for help e more disc e in unresponsive situatic ons than globally municated my stress to partner namel"Cronbach's93). mswere scored so that higher s rflected greater disclo trust-diminishing perceptions. Results and Discussion Study 2:Memories of Prior Events 128) Method ntended wa Participants.Study 2included 130 participants(M age-34 Comparing global and specific perceptions 1.First.we red global and specific percep follows:79%Caucasian,8 African American,,and of se Procedure.Participants completed a questionnare posted on (partner condition d partner)mixed analysis of We randomly assigned part icipants to report ona ures of pa ne 1,127) They the condition,F(1.127)=14.72.n2=.10.p <001.Means ar he- d in which the identifie or work).Particip then pleted the situation-specific mea even tenneynreonsive sitationshe situations m () .7R 58) hether perceivers who value partners (relat wh gree eived global four-itemm tive to devalued partners.F(1.128)=27.36.n2=.18.p<.001 ure of neec gs to the partner (e.g.."I do not ssmy nceds ner namel" ach's 90).Item he partner applied to global self-disclosure.but not to self- Situation-specific measures.Participants completed four disclosure in unresponsive situations Sex did h bomyll-being oine on in my life"Cronbach's 91)Items were completed Hence.sex was using 7-point response scales (1=strongly disagree:7=strongly sing the sam po ts also ompleted four 001: e during the event (e.g
situations.3 In contrast, participants who did not value a relationship with the target person exhibited a trust-diminishing tendency to see more disclosure in unresponsive situations than globally. Both patterns are consistent with the view that levels of relationship desire determine whether people hold trust-enhancing or trust-diminishing perceptions. Study 2: Memories of Prior Events We conducted Study 2 to replicate results of Study 1 using a different methodology. Rather than using hypothetical scenarios, we asked participants to describe an actual prior event in which they were recipients of unresponsive behavior. Method Participants. Study 2 included 130 participants (M age 34 years; 34 males; 96 females) who were recruited using the same two methods described in Study 1. The racial distribution was as follows: 79% Caucasian, 8.9% African American, 8.1% Asian, and 4% other. Procedure. Participants completed a questionnaire posted on the Internet. We randomly assigned participants to report on a valued or devalued partner using the same instructions used in Study 1. Participants then completed the measures of partner valuing and global self-disclosure described below. They then recalled and described a time within the last 12 months in which they experienced a stressful event and in which the identified relationship partner did not provide help or support. To facilitate participants’ recall of such an event, participants first completed a checklist containing some common stressful events (e.g., moving to a new residence, problems in a relationship, a setback at school or work). Participants then completed the situation-specific measures with regard to the selected event. Measures. Partner valuing (manipulation check measures). Participants completed the same measures of partner valuing described in Study 1, including care for the partner (Cronbach’s .92), commitment (Cronbach’s .92), and desire to be valued by the partner (Cronbach’s .90). Items were completed using the same 7-point response scales (1 strongly disagree; 7 strongly agree). Perceived global self-disclosure. Participants completed a four-item measure assessing global perceptions of their selfdisclosure of needs and feelings to the partner (e.g., “I do not clearly communicate my feelings to [partner name]”; “I clearly express my needs to [partner name]”; Cronbach’s .90). Items were completed using the same 7-point response scales and were scored so that higher values indicate more disclosure. Situation-specific measures. Participants completed four items assessing perceived diagnosticity of the partner’s unresponsive behavior (e.g., “He/she was not helpful or supportive because he/she did not care about my well-being”; “He/she was not helpful or supportive because he/she was not concerned about what was going on in my life”; Cronbach’s .91). Items were completed using 7-point response scales (1 strongly disagree; 7 strongly agree). Using the same response scales, participants also completed four items assessing their own self-disclosure during the event (e.g., “At the time of this event, I did not fully express a need for help or support to [partner name]”; “At the time of this event, I clearly communicated my stress to [partner name]”; Cronbach’s .93). Items were scored so that higher scores reflected greater disclosure. Results and Discussion Manipulation check. Participants reported more care for the partner (M 6.13), commitment to the relationship (M 6.33) and desire to be valued by the partner (M 6.26) in the valued partner condition relative to the devalued partner condition (M 3.73, 3.67, and 4.28, respectively), t(128) 12.51, p .001; t(128) 13.48, p .001; and t(128) 9.98, p .001; respectively. Hence, once again, the selected partners varied in intended ways.4 Comparing global and specific perceptions of self-disclosure. We followed the same analysis strategy described in Study 1. First, we compared global and specific perceptions of self-disclosure as a function of participants’ valuing of their partner using a 2 (measure type: global or specific) 2 (partner condition: valued or devalued partner) mixed analysis of variance with repeated measures on the measure type factor. The Measure Type Partner Condition interaction was significant, F(1, 127) 6.21, 2 .05, p .05, and qualified main effects of measure type, F(1, 127) 31.07, 2 .20, p .001, and partner condition, F(1, 127) 14.72, 2 .10, p .001. Means are presented in Figure 3. First we compared responses to the global and specific measures in the devalued and valued partner conditions. That is, do perceivers perceive less self-disclosure in unresponsive situations relative to their perceptions of global disclosure? Consistent with predictions that motivated perceivers are reluctant to perceive self-disclosure in unresponsive situations, the tendency to report greater global self-disclosure relative to selfdisclosure in specific unresponsive situations was much stronger for participants reporting on a valued partner (global M 5.26 vs. specific M 3.70), F(1, 69) 35.01, 2 .34, p .001; than for participants reporting on a devalued partner (global M 3.78 vs. specific M 3.32), F(1, 58) 3.12, 2 .05, p .09. We also examined conditional effects of partner condition, which examine whether perceivers who value partners (relative to perceivers who do not value partners) report more or less self-disclosure. Participants reported more global self-disclosure to valued partners relative to devalued partners, F(1, 128) 27.36, 2 .18, p .001, but perceptions of self-disclosure in specific unresponsive situations did not vary across participants reporting on valued and devalued partners (p .21). Consistent with our predictions, the association between self-disclosure and positive sentiments about the partner applied to global self-disclosure, but not to selfdisclosure in unresponsive situations. 3 Sex did not have a significant main or moderating effect in any of the analyses (ps .18). In addition, sex did not have consistent moderating effects in subsequent studies. Hence, sex was dropped as a moderator. In some cases, sex did have a significant main effect, but in all cases, critical results remained significant when controlling for sex. 4 Once again, care and commitment were highly correlated with desire to be valued by the partner, r(130) .83, p .001; and r(130) .86, p .001; respectively, supporting our argument that people who value relationships with partners desire reciprocation. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. DIMINISHING SELF-DISCLOSURE 43
44 LEMAY AND MELVILLE Patte gra (ieacross the daily reports).This positive sentiments ver we expected a more would judge their Global 2 ■Specific vould disanpear gain.we expecte d that perc ns or lo elt-d iving Devalu thes ns of low anc edby perceivers to ived diag As de uracy benchmarks.W relationships with part rers lack of biased in elf-disclosu these situations she disclosure we-functionn ceptions from analys Study 1 examined attributions of hypothetical events.which may unresp onsive ositivel ciated with diagno .B=42 ve altered the results. The 556,1 .001,global se over.Studies 126) 6.99.2 256.P<001.Hence participants saw tha e ant in light of the curre t model and ceived the situations (rather than iust upresponsive situations)The curren eves of sf-discoure evels behavior. with reduc sticity of the trust-protect Method Participants.Study 3 included a sample of 98 heterosexua romantically involved dyads.Due toa dismiss the unresponsive behavior as nondiagnostic. Study 3:Daily Report Study Sex did not significantly moderate any of the .Th report a high level of commitment and care)would report more
Predicting perceived diagnosticity. As described earlier, we expect that perceivers who strongly value relationships with partners are motivated to perceive low self-disclosure in situations characterized by partners’ lack of responsiveness because high self-disclosure in these situations should be related to perceiving unresponsive behaviors as diagnostic. In contrast, global selfdisclosure should reflect well-functioning relationships and be associated with positive relationship perceptions. Therefore, global self-disclosure should be associated with perceiving unresponsive behaviors as less diagnostic. To test these ideas, we regressed situation-specific diagnosticity perceptions on global perceptions of self-disclosure and perceptions of self-disclosure in specific unresponsive situations. Whereas self-disclosure in specific unresponsive situations was positively associated with diagnosticity perceptions, .42, t(126) 5.56, sr2 .162, p .001, global self-disclosure was negatively associated with diagnosticity perceptions, –.52, t(126) 6.99, sr2 .256, p .001. Hence, participants saw partners’ prior unresponsive behaviors as nondiagnostic—a trustprotective response—when they saw lack of disclosure in that specific situation and when they generally perceived their relationship to be high in self-disclosure. These findings provide replication evidence regarding the divergent associations of global selfdisclosure and self-disclosure in unresponsive situations. Summary. Perceivers who strongly valued a relationship with targets perceived high levels of self-disclosure in a global sense, but they also perceived low levels of self-disclosure in a prior event featuring targets’ lack of responsiveness. This combination was associated with reduced perceptions of diagnosticity of the unresponsive behavior, which is a trust-protective interpretation. These findings are consistent with prior research on the interpersonal benefits of global disclosure as well as with the current prediction that perceiving lack of disclosure in situations characterized by partners’ lack of responsiveness can help perceivers dismiss the unresponsive behavior as nondiagnostic.5 Study 3: Daily Report Study In Study 3, we tested our predictions using a daily report study involving both members of romantic dyads. We expected that perceivers who valued relationships with partners (i.e., those who report a high level of commitment and care) would report more self-disclosure in general (i.e., across the daily reports). This pattern would be consistent with prior research suggesting that self-disclosure usually facilitates closeness and is associated with positive sentiments. However, we expected a more nuanced pattern once we considered the daily context of partners’ behavior. Specifically, we expected that these perceivers would judge their disclosure in accordance with partners’ responsive behavior and that the tendency to report higher self-disclosure to valued partners would disappear on days when partners are unresponsive. Once again, we expected that perceptions of low self-disclosure would have a benefit, in that they would be associated with perceiving the partner’s unresponsive behavior as less diagnostic of the partner’s sentiments. As we stated earlier, these perceptions of low disclosure and diagnosticity may be biased and constructed by perceivers to subjectively fulfill their goals. As a first step in addressing this bias aspect of our predictions, we used target partners’ perceptions as accuracy benchmarks. We assessed target partners’ perceptions of perceivers’ disclosure and the diagnosticity of their own behavior, and we controlled for these perceptions in our analyses. Following prior research on biased interpersonal perceptions (e.g., Kenny & Acitelli, 2001; Lemay et al., 2007; Murray et al., 1996a), this procedure allows us to statistically remove the influence of targets’ perceptions from analyses of perceivers’ perceptions. To the extent that convergence across perceivers and targets reflects perceivers’ accuracy, this procedure statistically removes the portion of perceivers’ perceptions explained by accuracy. This study also addresses some limitations of Studies 1 and 2. Study 1 examined attributions of hypothetical events, which may not reflect attributions of actual events. Study 2 examined attributions of actual events, but these events occurred in the distant past and so memory biases could have altered the results. The daily report method used in Study 3 addresses these issues by examining perceptions of everyday relationship events. Moreover, Studies 1 and 2 compared perceptions of disclosure in specific unresponsive situations to perceptions of global disclosure. Although this comparison is theoretically relevant in light of the current model and the bulk of prior research on global perceptions of disclosure, it did not address disclosure perceptions in specific situations characterized by the partner’s responsive behavior, leaving open the possibility that perceived disclosure is threatening to trust in all specific situations (rather than just unresponsive situations). The current study corrects this issue by examining specific relationship events (i.e., daily interaction) that vary in terms of valence of partner behavior. Method Participants. Study 3 included a sample of 98 heterosexual romantically involved dyads. Due to a computer error, responses from three participants were missing (N 193 participants). Participants were recruited through a variety of methods, including 5 Sex did not significantly moderate any of the reported effects (p .32). However, sex did have a significant main effect on judgments of disclosure, F(1, 125) 3.93, p .05. Women reported more disclosure (M 4.15) than men (M 3.61). The Measure Type Partner Condition interaction remained significant when including sex as a covariate, F(1, 126) 5.92, p .05, and the pattern of the interaction remained the same. Figure 3. Perceived self-disclosure as a function of type of measure (global vs. specific) and partner valuing (Study 2). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 44 LEMAY AND MELVILLE
DIMINISHING SELF-DISCLOSURE Results and Discussion ements(Mage=23e3rs】 Mos cipants described within dyads.Rando n intercepts were estimated for each dyad 1,6.2 an American.6.7 Asia and 7 oth daily a sments of ach member.The models iependence.Given the limited degrees of freedom.slopes aires ven cutive evenings creen re tion(ieone per to n dai 1320 ments)We regr ed per eivers'daily percentions of self disclosure on their to m are not relevant t urrent investigatior that self-disclosure usually facilitate Participa completed two me and is heir study partner.includinga i osure aeross the sampled days (42).(1811)=5 nitment (e.g. hip to of self-disclos for the tested the prediction that perceivers' 80 especially likely to depend arge sponsiv sing 9-poin ponse scales (1 extr ivers should p rceive high self highly gree d in ved in an unn perhaps just a as th re per targ Dail perceptio r partner's their ptions of targets'res nd a prod derate or thoughtful was your p "Toda C.g. tions of self-disclosure to control for accurac .e hich targets agree ivers'chron valuin cre av aged to indicate morere lued are plotted in Figure 4.We xamined e-disc edures recommende Cohen.Cohen.West.an low and high level ted t perceivers deny s e on day item liscl sure(ie."Today,my partner clearly comm nicated to me 9-point agreiy ined in the response scale participants completed a sinele item as thi about me)Pantic the dai ted an analogous item Still.the g the dail regarding the diagnosticity of their own behavior n person me
an undergraduate psychology subject pool, flyers posted on college campuses and various retail locations, and newspaper and Internet advertisements (M age 23 years). Most participants described their relationship as dating (75%), engaged to be married (4%), or married (13%).6 The racial distribution was as follows: 80% Caucasian, 6.2% African American, 6.7% Asian, and 7% other. Procedure. Participants arrived to the laboratory with their study partner and completed the baseline measures described below (in addition to measures that are not relevant to the current investigation). Participants were asked to complete daily questionnaires for seven consecutive evenings, beginning the following evening. Daily questionnaires were completed on a secure website that recorded the date and time of completion, which allowed us to screen responses with regard to timing of completion (i.e., one per evening). Due to missing or invalid (due to timing) daily reports, the total number of daily observations was 1,239. The daily questionnaires included the daily measures described below (in addition to measures that are not relevant to the current investigation). Baseline measures. Partner valuing of targets. Participants completed two measures of the extent to which they valued their relationship with their study partner, including a five-item measure of relationship commitment (e.g., “I want our relationship to last for a very long time”; “I am committed to maintaining my relationship with him/ her”; Cronbach’s .87), and a five-item measure of care for the partner (e.g., “Helping him/her is a high priority for me”; “I care for his/her needs”; Cronbach’s .81). Items were completed using 9-point response scales (1 extremely disagree; 9 extremely agree). In turn, responses on these two measures were highly correlated (Cronbach’s .77) and were averaged to create an index of partner valuing. Daily measures. Daily perception of partner’s responsive behavior. Participants completed an eight-item measure of their perceptions of their partner’s daily responsive behavior (e.g., “Today, how considerate or thoughtful was your partner toward you?”; “Today, to what extent did your partner sacrifice [e.g., his/her time, goals, or personal wishes] to do something for you?”; “Today, how critical or insulting was your partner toward you?”). Items were completed using 9-point response scales (1 not at all; 9 extremely). After reverse-scoring negatively worded items, responses were averaged to create an index of daily perceptions of the partner’s responsive behavior (Cronbach’s .87). Higher values indicate more responsive behavior. Daily perceptions of self-disclosure. Participants completed a single item assessing the extent to which they disclosed their needs and desires to their partner (“Today, I clearly communicated to my partner my needs and preferences”). Participants also completed an analogous item assessing perceptions of their partner’s selfdisclosure (i.e., “Today, my partner clearly communicated to me his/her needs and preferences”). Items were completed using 9-point response scales (1 extremely disagree; 9 extremely agree). Daily perceptions of diagnosticity. Using the same 9-point response scale, participants completed a single item assessing perceptions of the diagnosticity of the partner’s behavior (“Today, my partner’s behavior toward me was a reflection of how he/she feels about me”). Participants also completed an analogous item regarding the diagnosticity of their own behavior. Results and Discussion Analysis strategy. We tested predictions using multilevel models that accounted for the nesting of days and individuals within dyads. Random intercepts were estimated for each dyad member to account for the covariance due to making repeated daily assessments of each member. The models estimated the covariance across the two dyad members of these intercepts and of the day-specific residuals, which accounts for potential dyadic interdependence. Given the limited degrees of freedom, slopes were modeled as fixed. Effects of partner valuing on average levels of self-disclosure. First we sought to replicate prior findings suggesting a positive association between participants’ sentiments toward their partners and self-disclosure in general (i.e., averaged across the daily assessments). We regressed perceivers’ daily perceptions of selfdisclosure on their chronic valuing of targets (care and commitment) as assessed during the baseline session. Consistent with prior research suggesting that self-disclosure usually facilitates relationship quality and is associated with positive sentiments, perceivers’ valuing of target partners predicted increased reports of self-disclosure across the sampled days (b .42), t(181.71) 5, p .001. Responsiveness-driven perceptions of self-disclosure. Next we tested the prediction that perceivers’ perceptions of selfdisclosure are especially likely to depend on targets’ responsive behaviors when perceivers strongly value relationships with targets. That is, highly valuing perceivers should perceive high selfdisclosure when target partners behaved in a responsive manner, but they should also perceive low disclosure when target partners behaved in an unresponsive manner, perhaps just as low as the disclosure perceived by those who do not value targets. To test this prediction, we regressed perceivers’ daily perceptions of selfdisclosure on their daily perceptions of targets’ responsive behavior, perceivers’ chronic valuing of targets, and a product term representing their interaction. We controlled for targets’ perceptions of perceivers’ self-disclosure to control for accuracy of perceivers’ perceptions (i.e., the extent to which targets agreed with perceivers’ judgments of self-disclosure).7 The anticipated interaction between perceivers’ chronic valuing of targets and perceivers’ daily perceptions of targets’ behavior was significant, b .11, t(948.01) 3.25, p .01. Predicted valued are plotted in Figure 4. We examined conditional effects following procedures recommended by Cohen, Cohen, West, and Aiken (2003). First, we examined effects of daily perceived partner responsiveness at low and high levels of chronic valuing, which address whether perceivers deny self-disclosure on days 6 Although we explicitly stated that participants must bring a romantic partner to the lab, some participants (7%) described their relationship with their study partner as a friendship. It is unclear whether this term refers to a platonic friendship or a casual sexual relationship. Responses provided by these participants were retained in the data analysis. 7 We centered the predictors on sample means. We did not center the daily predictors on person means because this would preclude us from comparing low valuing and high valuing perceivers when their partner’s behavior is similarly unresponsive. Centering a daily variable on person means eliminates the ability to equate participants on the daily variable. Still, the predicted interaction emerged even when centering the daily predictors on person means (b .15, t 3.58, p .001). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. DIMINISHING SELF-DISCLOSURE 45
46 LEMAY AND MELVILLE -Low Perceptions of Partner's Responsive Behavior(-1 SD) self-disclo nsive was no High Perceptions of Partner's Responsive Behavior (+1 SD) 8 hich If-disclo was.af st in part.a pe eptual bias 6 tions of high 5 partner's w nsive beh ior is not di 4 ostic of the par -dis 4 8 9 rce ivers'Valuing of Partn ption of self. of dia o ticity on their dail e'cal i daily o b argets(i ccur perceptions) 》P when partners are unresponsive but claim self-disclosure on days 451110m63 16.g3n之001 Percei who sav when partners are responsive. tions of partne If-disclosure on a particular day perc disclosure for who h ts (1 SD abo of pant mean. self-perception Summary. Perceivers who strongly valued relationships with resp ex ive relation hip sentiments.However.these perceivers had per word s,do pe vers who valu ow dis who did not ships with partners.Thi th ure than iow- 38.409.83 eptions of low self-disc sentiments,suggesting that these perceptions of low self-disclosure aved in a r ive r When and o ers'repo s of the f-disclosure as did not chr 19.1=66 This i that i their self-discl 255 ,05.This 137 elf-discl ent is an of se f dis for the dicted bia for th ions of per self-discl s the o place eptic abOrcetve argets'respor ay int 20).Hence.the tendency for highly valuing perceivers to deny
when partners are unresponsive but claim self-disclosure on days when partners are responsive. Daily perceptions of partners’ responsiveness more strongly predicted daily perceptions of selfdisclosure for perceivers who highly valued targets (1 SD above the mean, b .62, t(989.83) 12.79, p .001, than for low-valuing perceivers (1 SD below the mean, b .38, t(1021.39) 6.76, p .001. That is, perceivers who valued targets were more likely to consider targets’ responsiveness when judging their own selfdisclosure. Next, we examined the effects of perceivers’ chronic valuing at low and high levels of daily perceived partner responsiveness. In other words, do perceivers who value targets always see more self-disclosure relative to perceivers who do not value targets, or does this depend on whether partners behaved in a responsive or unresponsive manner that day? Perceivers’ who chronically valued targets perceived more self-disclosure than low-valuing perceivers on days when they perceived that targets were relatively responsive (1 SD above the mean, b .38, t(409.83) 3.88, p .001, but not on days when they thought targets were unresponsive (1 SD below the mean; p .55). These results suggest that the above effect in which partner valuing was associated with more selfdisclosure only pertained to situations in which the partner behaved in a responsive manner. When partners were unresponsive, perceivers who valued partners saw just as little self-disclosure as perceivers who did not chronically value partners. This interaction pattern was found while controlling for target partners’ judgments of perceivers’ self-disclosure, which also predicted perceivers’ judgments of their self-disclosure (b .07), t(1078.61) 2.55, p .05. This unsurprising effect indicates that, in addition to the bias demonstrated above, there was a small degree of agreement between perceivers and targets regarding perceivers’ self-disclosure. Such agreement is an indicator of accuracy. Most important for the present research, we found evidence for the predicted bias while controlling for this agreement between perceivers and targets, suggesting that accuracy cannot explain these results. In addition, we tested another model using partners’ perceptions of perceivers’ self-disclosure as the outcome variable in place of perceivers’ self-perceived disclosure. The interaction reported above was not significant in this model (p .20). Hence, the tendency for highly valuing perceivers to deny self-disclosure on days when partners were unresponsive was not corroborated by their partner’s perceptions of perceivers’ selfdisclosure, which also suggests that motivated perceivers’ denial of self-disclosure was, at least in part, a perceptual bias. Predicting perceived diagnosticity. We expect that motivated perceivers deviate from their typical perceptions of high self-disclosure when partners are unresponsive because the perception of low self-disclosure helps perceivers make the claim that the partner’s unresponsive behavior is not diagnostic of the partner’s feelings or motivations. To test whether low self-disclosure is indeed associated with low perceived diagnosticity, we regressed perceivers’ daily perceptions of diagnosticity on their daily perceptions of self-disclosure. We controlled for target partners’ daily perceptions of perceivers’ self-disclosure and target partners’ daily perceptions of the diagnosticity of their own behavior to be sure that results could not be explained by perceptions of disclosure or diagnosticity that were shared by both perceivers and targets (i.e., accurate perceptions). Perceivers’ daily perceptions of self-disclosure strongly predicted their diagnosticity perceptions (b .45), t(1102.65) 16.93, p .001. Perceivers who saw themselves as low in self-disclosure on a particular day perceived partners’ behavior as less diagnostic of partners’ underlying sentiments on that day, and this effect was independent of partners’ perceptions of perceivers’ disclosure and partners’ self-perceptions of diagnosticity.8 Summary. Perceivers who strongly valued relationships with targets reported more disclosure across the daily reports, consistent with other findings suggesting that disclosure covaries with positive relationship sentiments. However, these perceivers had perceptions of disclosure that were contingent on partners’ daily responsive behavior. They saw high self-disclosure on days when partners behaved in a responsive manner, but they saw low disclosure on days when partners were unresponsive, just as low as perceivers who did not value relationships with partners. This pattern was not corroborated by partners’ reports of perceivers’ self-disclosure, consistent with the hypothesis that perceivers’ perceptions of low self-disclosure are, in part, illusory. Seeing lack of disclosure when partners were unresponsive was related to viewing the unresponsive behavior as less diagnostic of partners’ sentiments, suggesting that these perceptions of low self-disclosure 8 Partners’ reports of perceivers’ disclosure and partners’ reports of the diagnosticity of their behavior also predicted perceivers’ diagnosticity perceptions (b .07, t 2.36, p .05 and b .19, t 6.63, p .001). These effects suggest that, in addition to bias, diagnosticity perceptions were partly accurate. Again, we used variables that were centered on sample means for this analysis. However, the same key effect of perceived self-disclosure on perceived diagnosticity emerged when we instead used daily variables that were centered on person means (b .40, t 13.78, p .001). Our conceptual model indicates that perceivers’ valuing of partners moderates effects of daily perceptions of partners’ responsiveness on their perceptions of self-disclosure. In turn, perceptions of disclosure are closely tied to perceiving partner behaviors as diagnostic, and this link is not moderated. This is why we included different predictors in models of perceived disclosure and perceived diagnosticity. Indeed, in subsequent analyses we found that the link between daily perceived disclosure and daily perceived diagnosticity was not moderated by perceivers’ chronic valuing of targets (p .38), by daily perceptions of targets’ responsiveness, (p .74), or by the combination of perceivers’ chronic valuing and daily perceptions of targets’ responsiveness (i.e., a three-way interaction; p .20). Figure 4. Daily perceptions of self-disclosure as a function of daily perceptions of partner’s responsive behavior and chronic valuing of the partner (Study 3). This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 46 LEMAY AND MELVILLE
