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 predic￾tions using hypothetical vignettes, which allowed us to capture general explanatory styles while holding constant the specific features of interactions. Prior research on attributions in relation￾ships 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 com￾plete 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. Partic￾ipants 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 mea￾sures of partner valuing and global self-disclosure described be￾low. 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 mea￾sures described below. Finally, participants completed the mea￾sures 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 mea￾sure 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 self￾disclosure of needs, desires, and feelings (e.g., “I do not clearly communicate my desires to [partner name]”; “I do not clearly com￾municate 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 diagnos￾ticity 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 part￾ner 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 sub￾missions by the same individual, inability to control the testing environ￾ment, which introduces noise, high dropout rates, and reductions of par￾ticipants’ 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