Health Communication, 24: 270–283, 2009 Copyright © Taylor & Francis Group, LLC ISSN: 1041-0236 print / 1532-7027 online DOI: 10.1080/10410230902806070 HHTH HIV Disclosure Messages Do Message Features Influence Reactions to HIV Disclosures? A Multiple-Goals Perspective John P. Caughlin Department of Communication University of Illinois at Urbana-Champaign Jennifer J. Bute School of Communication Studies Ohio University Erin Donovan-Kicken Department of Communication Studies University of Texas at Austin Kami A. Kosenko Department of Communication North Carolina University Mary E. Ramey and Dale E. Brashers Department of Communication University of Illinois at Urbana-Champaign People who are HIV-positive must make decisions about disclosing their status to others but do so in the context of stigma and social isolation reported by many with the disease. Disclosing an HIV-positive diagnosis is necessary to seek social support, to manage health care, and to negotiate sexual encounters, but fear of how others will respond is a strong barrier to revealing that information. This investigation focuses on various ways that HIV can be disclosed. Using a multiple-goals perspective, 24 disclosure messages (representing 6 different types) were created. Participants (N = 548) were asked to imagine one of their siblings revealing an HIV-positive diagnosis, using 1 of the 24 messages. Participants’ reactions to the disclosures differed substantially across the various message types. The discussion focuses on theoretical explanations for the variations in responses and the utility of these findings for practical interventions concerning HIV disclosures. Despite significant biomedical advances that have lengthened the life spans of people who acquire HIV (Brashers et al., 1999; Siegel & Lekas, 2002), those living with the virus still face difficult psychosocial challenges. People Correspondence should be addressed to John P. Caughlin, Department of Communication, University of Illinois at Urbana-Champaign, 702 S. Wright St., No. 244 Lincoln Hall, Urbana, Illinois 61801. E-mail: caughlin@ illinois.edu with HIV typically must cope with uncertainty; gain access to social support; navigate personal relationships; and decide who, when, and how to tell about their HIV-positive status (Brashers et al., 2003; Goldsmith, Brashers, Kosenko, & O’Keefe, 2008). These psychosocial issues can have important physical health implications. Disclosing an HIV-positive status, for instance, is important for accessing support and for maintaining adherence to the complex medication regimens of HAART (highly active anti-retroviral therapy), HIV DISCLOSURE MESSAGES both of which can lead to better health outcomes for the infected (Serovich, 2001). There are also broader social, legal, and health implications of HIV disclosures. Secondary prevention efforts hinge on status disclosure to sexual partners, and medical ethicists, health advocates, and the legal community contend that persons living with HIV have a social and legal responsibility to divulge their status to potential sexual partners (e.g., Stein et al., 1998). Congruent with the obvious importance of HIV disclosures, a large research literature has emerged examining issues pertaining to revelations of HIV status (for a review, see Greene, Derlega, Yep, & Petronio, 2003). Much of the HIV disclosure research has examined how people living with the disease make decisions about divulging their status (e.g., Derlega, Winstead, & Folk-Barron, 2000; Derlega, Winstead, Greene, Serovich, & Elwood, 2004). This literature indicates that people living with HIV frequently choose not to disclose their status because of the stigma associated with the illness and concerns about others’ reactions (e.g., Alonzo & Reynolds, 1995). When concerns about stigma prevent people from gaining benefits from disclosure (e.g., gaining support, acquiring medical care), this becomes an important health issue, warranting counseling programs to help these individuals manage the social, psychological, and health aspects of their illness (Rintamaki, Kosenko, Scott, Jensen, & Jordan, 2005). In addition to examining individuals’ decisions about revealing their HIV status, there has been some crucial work on disclosers’ perceptions of the responses they receive when telling others about their HIV-positive diagnosis (Greene & Faulkner, 2002; Serovich, Kimberly, & Greene, 1998). Understanding how people respond to disclosures about HIV-positive status is extremely important. People living with HIV often experience uncertainty about how others will react (Brashers et al., 2003). If this uncertainty contributes to anxiety about revealing their status, research describing common reactions to HIV disclosures can be useful by illuminating “the issues and dilemmas that may be encountered when disclosing” (Serovich et al., 1998, p. 15). In short, there is a rich and growing literature examining how people with HIV make decisions about disclosing and how people who receive those disclosures tend to react. One related issue that has not received sufficient attention, however, is whether (and how) the characteristics of particular disclosure messages influence people’s reactions to revelations of HIVpositive status. Research on how people manage information suggests that some ways of revealing an HIV-positive test result are likely to be viewed more favorably than others (Brashers, Goldsmith, & Hsieh, 2002), yet little is known about what specific features of HIV-disclosure messages lead to more (or less) positive reactions. A better understanding of how features of HIV-disclosure messages influence people’s reactions could have enormous practical implications. Given the challenges involved in HIV disclosures, it would be useful for researchers to be able to offer guidance about the types of messages that have the greatest 271 likelihood of allowing disclosure while minimizing some of the negative reactions to it. This investigation represents a first step toward that end. Specifically, this article takes a multiplegoals perspective to examine how HIV-disclosure messages may vary in the reactions they tend to elicit. MULTIPLE-GOALS PERSPECTIVE A useful framework for conceptualizing the diversity in disclosure messages is a multiple-goals perspective (e.g., Goldsmith, 2004; Wilson, 2002). There are numerous specific theories within this perspective, but our primary focus here is on some common assumptions. Most obviously, “it is a truism that individuals generally pursue multiple goals simultaneously during their interactions with others” (Berger, 2005, p. 422). Goals can be described at different levels of abstraction (e.g., Wilson, 2002). At a broad level, three types of goals that are relevant to a wide variety of conversations are (a) instrumental goals, which are the primary focus of a communicator’s task; (b) identity goals, which relate to portraying or protecting a desired image of the self or conversational partner; and (c) relational goals, which are associated with the development and maintenance of relationships (Clark & Delia, 1979). Another common distinction among goals is between primary and secondary ones (e.g., Wilson, 2002) or dominant and subsidiary ones (O’Keefe & Shepherd, 1987). A key point of this distinction is that even if a person has a main goal in a conversation, other goals are still relevant to the situation and may shape or constrain the pursuit of the primary goal. That is, message production often is not simply a matter of choosing a singular course of action (e.g., pursuing the goal of revealing information); instead, message producers frequently attend to multiple objectives simultaneously. A person who decides to reveal an HIVpositive status, for instance, may have a primary (instrumental) goal of conveying that information; however, research indicates that people considering HIV status disclosure frequently worry about stigma (e.g., Alonzo & Reynolds, 1995; Derlega et al., 2000; Derlega et al., 2004). An individual’s decision to disclose an HIV-positive status does not necessarily indicate that his or her concerns about others forming negative impressions have disappeared. Such identity concerns instead are likely to remain salient, and a multiple-goals perspective would suggest that individuals might attempt to craft disclosure messages intended to ameliorate the chances of others having a negative reaction. In addition to providing a theoretical account of how message producers address multiple objectives, multiplegoals perspectives are pertinent to understanding how recipients respond to messages (Goldsmith, 2004; Wilson, 2002). Message recipients are influenced by whether (and how) particular goals appear to be pursued in conversations. A message recipient will judge others as inconsiderate, for 272 CAUGHLIN ET AL. instance, if they fail to recognize and attend to the relevant identity goals of the recipient (O’Keefe, 1988). In many cases, there are goals that are inherently relevant to the situation. A person giving advice, for instance, risks creating the impression of being nosy (Goldsmith & Fitch, 1997). That is, people who receive advice make judgments about the identity management objective of not appearing nosy, and they do so irrespective of whether the person giving the advice is actually concerned with appearing nosy. Put more broadly, message recipients’ reactions are influenced by the goals addressed in the messages, irrespective of whether the message producer endorses the importance of particular goals. In short, a multiple-goals perspective provides an account for how individuals manage various objectives to produce messages, and it also suggests that variations in whether particular goals are addressed influence recipients’ reactions to messages. This theoretical framework can be applied to any communication scenario, but the particular goals that are likely to be salient are shaped by the specific situation (Goldsmith, Miller, & Caughlin, 2008; O’Keefe & Shepherd, 1987). That is, although the general classes of instrumental, identity, and relationship goals are broadly relevant, the goals that are conventionally relevant to HIVdisclosure scenarios are particular to that circumstance. It is thus important to consider commonly salient goals for people disclosing that they have HIV. RELEVANT GOALS WHEN DISCLOSING HIV-POSITIVE STATUS Theoretical models of how individuals decide whether to reveal an HIV-positive status (e.g., Derlega et al., 2004; Serovich, 2001) do not explicitly examine the goals that people have for that disclosure. Nevertheless, research supporting these models suggests some common concerns or objectives for people making HIV disclosures. Derlega, Lovejoy, and Winstead (1998), for instance, found 11 different reasons that people with HIV had for disclosing and 7 reasons they had for not disclosing. Such reasons for revealing (or not) can be conceptualized as objectives that would be pursued when disclosing, and these objectives ought to shape the particular disclosure message (Goldsmith, Miller et al., 2008). A person who is disclosing to secure help (e.g., Derlega et al., 1998), for instance, might ask for assistance. The literature on HIV-disclosure decisions suggests that there are many possible objectives that people might have for revealing their status. Some of these goals may be idiosyncratic or at least uncommon. For example, in Derlega et al.’s (1998) study of reasons for disclosure, 1 respondent (out of 42) reported disclosing to a sister because she “was so filled with hatred toward me and no compassion toward anyone else. It was almost like I have got to do something to make her feel something” (p. 155). Although there are many specific objectives that people may have when revealing an HIV-positive status, the research on the reasons why people do and do not tell (e.g., Derlega et al., 2004; Serovich, 2001) implies a number of common goals. Rather than attempting to provide an exhaustive list of the goals that might be pertinent in specific situations, our focus is on some of the more salient goals in the literature. An examination of this literature indicates that all three of the canonical types of goals—instrumental, identity, and relational—are relevant. The most obvious instrumental goal in HIV-disclosure scenarios is disclosing the information. By definition, this goal would be evident in any revelation of HIV-positive status. The literature also suggests another common instrumental goal: seeking support (e.g., Brashers et al., 2003; Goldsmith, Brashers, et al., 2008). Sometimes this support involves emotional support: One of the most commonly cited reasons for revealing that one has HIV is a belief that “the other person would be understanding, consoling, concerned, and supportive” (Derlega et al., 1998, p. 153). People with HIV also may seek more tangible support, as in cases when a person asks for assistance or advice from a person who is knowledgeable about HIV (Derlega et al., 1998). As suggested earlier, people who reveal having HIV often are concerned about the associated stigma (e.g., Alonzo & Reynolds, 1995; Derlega et al., 2004). Sometimes they expect that others will have negative reactions (Greene & Faulkner, 2002), and they may even feel shame about their status (Derlega et al., 1998; Derlega et al., 2004). That is, people who have HIV frequently are concerned about maintaining a positive identity, and these concerns would be evident in identity goals when revealing their status. There are at least two distinct issues involved. First, as evidenced by concerns that the person being told may not understand or may judge them harshly (Derlega et al., 1998; Greene & Faulkner, 2002), disclosers may attempt to avoid negative evaluation by the recipient. The identity concerns also extend beyond the particular individual being told. People who disclose having HIV often are worried that “people have big mouths and they might go running around telling other people” (Derlega et al., 2000, p. 68). Consequently, managing their identity means trying to prevent their status from becoming a topic of gossip; thus, they try to limit who knows the information. That is, even when revealing their HIV-positive status to one person, people may be concerned with preventing subsequent disclosure to a third party (Greene & Faulkner, 2002). In addition to identity concerns, relational goals are probably salient for many people revealing that they have HIV. Derlega et al. (1998) found that the most frequently cited reason for not telling others about having HIV was a fear of rejection, including the concern that the other person would no longer like them once the HIV-positive status was revealed. Thus, people who do choose to disclose having HIV DISCLOSURE MESSAGES HIV may be concerned with the goal of maintaining the relational bond with the recipient of the disclosure. In addition, people who disclose having HIV often do so because they feel a duty to do so (Derlega et al., 2000). Often this sense of duty is tied to the nature of the relationship with the recipient; for example, people may feel that family members simply have a right to know the information (Derlega et al., 1998). Thus, one relational goal in revealing an HIV-positive status is recognizing or honoring the other person’s right to know. THIS STUDY Research on multiple goals in communication indicates that the various goals that are pertinent to HIV disclosures will shape the content of the disclosure messages. This does not imply that all disclosers will attend to all of the same goals; indeed, some people who decide to reveal may produce messages that appear to select only the goal of revelation (e.g., O’Keefe & Shepherd, 1987). Thus, possible disclosure messages will vary in the goals that are (conventionally understood to be) addressed. As noted earlier, the extent to which goals are addressed shapes the interpretation of messages; for instance, being inconsiderate is the result of inadequate attention to identity issues (O’Keefe, 1988). Based on this general principle, we expected that HIV-disclosure messages with varying attention to the goals summarized above would elicit varying reactions from recipients; however, given the lack of research on the effects of various HIV-disclosure messages, it was not precisely clear how the responses would be influenced. A person who seeks to prevent subsequent disclosures by asking the recipient not to repeat the information, for instance, might receive an assurance of confidentiality or might make the recipient feel special to be told the information. On the other hand, asking the recipient not to repeat the information also could elicit negative reactions, such as disappointment or anger that the individual wants to keep this information secret. Because it was not possible to make specific hypotheses about how the reactions would be influenced by the various messages, we posed a general research question: RQ: Do HIV disclosure messages that vary in the extent to which they address various goals elicit different reactions? To examine this general idea, we focused on disclosures to a sibling. The choice of the sibling relationship was based on several reasons. First, HIV disclosures to family members can be particularly complicated. People living with HIV are often particularly hesitant to reveal their status to family members (Greene, 2000) and are more likely to regret telling family members than friends or partners (Serovich, Mason, Bautista, & Toviessi, 2006). 273 Such findings suggest that concerns about being negatively evaluated are quite salient in disclosures to family members. Moreover, people with HIV may feel obligated to tell family members despite not wanting to (Derlega et al., 2004). In short, people who reveal to family members that they have HIV often have a complex and competing set of objectives. Such complex situations are more likely than simple ones to reveal varying effects of different messages (e.g., O’Keefe, 1988). Second, disclosing to siblings is often important to people living with HIV, but HIV disclosures “to siblings are not well understood in the literature” (Greene et al., 2003, p. 71). Third, because the purpose of this study was a systematic assessment of whether various message types were evaluated differently, it was necessary to pick a relational context that would be salient to most people. METHOD Participants We recruited 581 participants from introductory communication classes at a large Midwestern university. Because our focus was on disclosures from siblings, participants were first asked if they had at least one sibling. In some cases, participants asked whether stepbrothers or stepsisters counted as siblings; they were told to decide whether they considered the stepsibling to be a sibling. Thirty-three participants reported having no siblings. They were asked to complete a separate questionnaire about a cousin. Because the number of people reporting on cousins was not large enough to allow for systematic examination, the surveys pertaining to cousins were excluded, leaving an effective sample size of 548. The final sample included 239 (43.6%) men, 299 (54.6%) women, and 10 (1.8%) who did not indicate their sex. The majority of the participants reported being European American (392, 71.5%), and the other participants reported being African American (56, 10.2%), Asian American (40, 7.3%), Latino/Latina (30, 5.5%), and other (17, 3.1%). Thirteen (2.4%) individuals did not provide information about ethnic background. The average age of the respondents was 19.72 years (SD = 1.85). Procedures Each participant was given a short questionnaire on a single sheet of paper and an envelope that initially concealed a longer questionnaire containing a variety of closed-ended items. The first sheet began, “Imagine you are alone talking with your sibling. (If you have more than one sibling, pick your oldest sibling.) After a few minutes of small-talk, your sibling says, ‘Well, I have something to tell you . . .’” This introductory passage was followed by 1 of 24 disclosure 274 CAUGHLIN ET AL. messages, all of which indicated that the sibling had HIV (see Appendix). After the HIV disclosure, participants were asked, “If this were to occur, what would you do or say in response? If you would say something, write down exactly the words you would use.” Respondents were given the majority of the page (and back if needed) to respond. Prior to beginning the study, participants were told that they should wait until they had finished the first sheet before removing the second questionnaire from the envelope. After completing the first page, participants enclosed it in the envelope before beginning the second part of the questionnaire. The primary purpose of the envelope was to prevent participants’ responses on the first part from being influenced by the second part, which contained items about possible responses. Because even individuals who cannot produce sophisticated responses may be able to select them (Burleson et al., 1988), it was crucial that participants not see the second part before answering the first part. There was no identifying information on any of the questionnaires, so participant’s responses could not be identified. Questionnaires Message manipulations. Based on the aforementioned discussion of relevant goals, there were six different types of disclosure messages, hereafter referred to as conditions. The Appendix lists the conditions and the specific messages for each condition. The first condition focused on the instrumental goal of revealing the HIV-positive status. The four messages in the first condition also served as the kernel messages (Jackson, 1992) in the other conditions. That is, these messages were crossed with the other conditions, with each appearing as part of one of the larger messages in the other conditions. In addition to the kernels, the messages in the second condition explicitly referred to the instrumental goal of seeking support. The third and fourth conditions concerned identity issues, specifically avoiding negative evaluation and preventing subsequent disclosure (i.e., asking the recipient of the message not to tell others). Conditions five and six, respectively, referred to the relationship goals of maintaining the relational bond with the recipient and honoring the other’s right to know the information. Participants’ responses to the disclosure. There were both closed-ended and open-ended assessments of the responses. For the closed-ended measures, participants completed a series of 7-step Likert-type items about how they would react. Based on past research on reactions to disclosure of HIV-positive status (e.g., Greene & Faulkner, 2002; Serovich et al., 1998), we constructed items intended to assess five types of responses: seeking information about transmission (e.g., “I would ask my sibling how he/she got it”), social support (e.g., “I would offer to help in any way I could”), relational assurance (e.g., “I would tell my sibling that HIV would not affect the bond that we share”), negative evaluation (e.g., “I would think less of him/her”), and upset (e.g., “This information would really upset me”). All 14 items were submitted to a principal-components analysis with varimax rotation. Both the eigenvalues and the scree plot suggested a four-factor solution rather than the expected five-factor solution. The factor loadings for the rotated solution were examined using the 60/40 rule, which is a “very conservative” (McCroskey & Young, 1979, p. 380) criterion for assigning items to factors. The results indicated that social support and relational assurance items loaded strongly on the same factor, with the lowest factor loading among these items being .69 (and no secondary loadings on other factors as high as .20). Given the lack of evidence for an empirical distinction between social support and relational assurance, and the fact that it was theoretically reasonable to combine them, a single index called comforting was created. Using the 60/40 criterion, all of the other items loaded exactly as expected, so we retained the four-factor solution suggested by the principal-components analysis. These four factors (eigenvalues: 4.68, 2.50, 1.64, and 1.34) accounted for 72.64% of the variance. Before using the factors as a basis for subscales, we also examined the Cronbach’s alphas to determine whether any particular item substantially lowered reliability. For the index of upset, dropping one item increased the reliability from .74 to .80, so this item was removed. Our final scales for the reactions measure were: seeking information about transmission (2 items, a = .89, M = 5.77, SD = 1.63), comforting (6 items, a = .89, M = 6.56, SD = 0.83), negative evaluation (3 items, a = .82, M = 2.57, SD = 1.43), and upset (2 items, a = .80, M = 6.77, SD = 1.63). Perceived competence and relationship change. In addition to the items developed based on the HIV-disclosure literature, we asked participants about some general reactions to the messages. Specifically, respondents rated the competence of the sibling’s disclosure message and the extent to which their relationship with the sibling would be changed by such a conversation. To assess competence, we used four semantic-differential items (e.g., competentincompetent, effective–ineffective) from Goldsmith and MacGeorge’s (2000) Message Effectiveness Scale. All items were recoded so that higher scores indicated higher competence on a scale ranging from 1 to 7. After performing a factor analysis and checking reliability (a = .85), all four items were retained (M = 5.31, SD = 1.27). Participants responded to four items from Vangelisti and Young’s (2000) Relational Distancing Scale. These 7-point items assessed how much respondents believed their relationship with the sibling would change due to the conversation involving the revelation of HIV. Based on reliability analysis and examination of the items, we determined that two items (e.g., “more tense” vs. “more relaxed”) likely HIV DISCLOSURE MESSAGES confounded relational change with the particular situation of HIV disclosure (e.g., one might experience tension without relational distancing in this circumstance). These two items were dropped, and the remaining items were coded so that low scores indicated that the relationship would become more distant or remote, the scale midpoint (4) indicated no change, and high scores indicated that the relationships would become closer or more intimate. We labeled this variable more closeness (a = .85, M = 5.16, SD = 1.22). Coding Open-Ended Responses Five hundred thirty-six participants wrote responses to the disclosures. The responses were first unitized into thought units (Hatfield & Weider-Hatfield, 1978). After initial training, two coders unitized a subset of the data, and then met with the first author to clarify the unitization rules. Next, the two coders independently unitized data from 130 participants. The resulting reliability was quite high (U = .99, see Guetzkow, 1950), and each coder unitized half of the remaining data. After unitization, these responses amounted to 2,930 distinct thought units. Four of the authors each next took approximately onefourth of the data and independently determined which message functions were present. The four met to discuss their categorization schemes and found that all four coders had identified similar functions. The main differences involved labels and level of differentiation (e.g., negative emotion vs. sadness, fear, and anger). Such discrepancies were resolved through discussion. Then two authors each coded 308 units independently to assess intercoder reliability (Guetzkow’s 275 P = .78). This reliability was consistent with other studies using similar techniques (e.g., Lambert & Gillespie, 1994), and the coders split the remaining units and assigned a function code to each. The responses ranged in number of functions from 1 to 10. Most messages included multiple functions, and the majority of response messages included 2 to 4 functions. Realism of Disclosure Messages After examining the first group of responses, we decided it would be important know whether participants believed the messages were realistic. Thus, all subsequent participants (n = 325) rated the realism of the disclosure message. Realism was assessed with two items (“Think about how your sibling disclosed in the hypothetical situation described above. To what extent was this similar to how somebody with HIV might disclose this to a sibling?”), each of which was rated on a scale that ranged from 1 to 7, with higher scores being more realistic (a = .77). On average, participants reported that the messages were fairly realistic, with a mean of 4.65 (SD = 1.38), which was significantly greater than the scale midpoint, t(324) = 8.50, p < .01. RESULTS Function Descriptions A total of 17 message functions emerged from the data. A complete list of the function categories, along with descriptions and examples of each, can be found in Table 1. TABLE 1 Functions Expressed in Responses to Hypothetical HIV-Status Disclosures Function Description Sadness Anger Fear Surprise Concern Emotional inhibition Blame Expressing sadness, sorrow, grief; crying Being angry, frustrated, or disappointed Expressing fear Demonstrating disbelief, shock, confusion, denial; silence Mentioning concerns about the sibling or other people Inhibiting any expressions of affect, or minimizing emotions Stating or implying that the sibling is at fault for acquiring HIV Information seeking about transmission General information seeking Emotional support Explicit advice Implicit advice Instrumental support Relationship affirmation Identity affirmation Asking questions about how the sibling acquired HIV Gratitude Privacy assurance Asking questions not pertaining to transmission, including those about living with HIV, the sibling’s history, current status, and future Comforting; expressing affection; showing empathy Providing, or offering to provide, information or advice Using off-record comments or questions to imply a course of action Providing or offering tangible assistance Emphasizing a strong and enduring relational bond Assuring the sibling that he or she is still perceived positively or viewed in the same way Stating that the participant appreciates something that the sibling did Assuring that the information will be kept in confidence Example(s) “I’m sad to hear this.” “I’m really angry.” “I’m so scared for you.” “Oh my god!”; “I don’t know what to say.” “How is your wife dealing with this?” “I would try to remain calm.” “Why didn’t you put on a condom?”; “You’re an idiot.” “How did you get it?” “Who have you told?”; “Are you going to die?” “I’m here for you.” “You need to tell mom and dad.” “Have you seen a doctor yet?” “I’ll help you get the best medical care.” “You’re my brother, no matter what.” “You are still the same person.” “Thank you for telling me.” “Of course I won’t tell anyone.” 276 CAUGHLIN ET AL. Several of the functions involved emotional expressions, including sadness (n = 154 thought units, 5.26% of all utterances), anger (n = 27, 0.92%), fear (n = 11, 0.38%), surprise (n = 437, 14.91%), and concern (n = 42, 1.43%). Also, some messages described emotional inhibition (n = 8, 0.27%) or attempts to reduce participants’ expression of emotions. Another function was blame (n = 106, 3.62%), which included overtly critical remarks and comments with a clearly accusatory tone, insinuating that the sibling was at fault for his or her HIV status. Two functions involved participants seeking information. Because questions about how the virus was acquired were so common, this type of message content was designated as its own category, information seeking about transmission (n = 330, 11.37%). Another category, general information seeking (n = 286, 9.76%), comprised all other types of information seeking. Another general area of functions corresponded to types of social support. Emotional support (n = 600, 20.48%) included comments that demonstrated comforting behaviors. Informational support was articulated through the functions explicit advice (n = 146, 4.98%) and implicit advice (n = 225, 7.68%). The instrumental support (n = 158, 5.39%) category contained units that were offers of tangible assistance. The remaining categories involved various forms of reassurance. Relationship affirmation (n = 61, 2.08%) was characterized by statements acknowledging a close relational bond between the participant and the sibling. The identity affirmation (n = 40, 1.37%) category emphasized that the sibling would not be perceived in a fundamentally different or negative way. Statements of privacy assurance (n = 35, 1.19%) expressed promises to keep the sibling’s disclosure a private matter, and, finally, gratitude (n = 27, 0.92%) primarily included statements thanking the sibling for sharing this information. The remaining units (n = 237, 8.09%) were either uncodable or represented some other function. Many of the responses not represented in Table 1 involved statements directed toward the researchers rather than descriptions of the participants’ reactions (e.g., “I am not very good at being supportive or reassuring”). Research Question The open-ended data pertinent to RQ1 are summarized in Table 2, which lists the proportion of responses that included each function for each experimental condition. The proportions for each condition were compared with tests of independent proportions, which are distributed as z (Brunig & Kintz, 1997). The basic disclosure (kernel) messages prompted more expressions of sadness than any other condition except for the support-seeking condition. The kernel-only condition also evoked more fear than either requests to prevent subsequent disclosure or emphasizing maintenance of the relational bond, both of which were marked by no expressions of fear at all. Surprise was a fairly common element of responses in all conditions but was less common in the maintain-relationship condition than in the basic kernel, avoid negative evaluation, and prevent-subsequent-disclosure conditions. Concern was more commonly expressed in response to support seeking than in response to the kernel-only condition disclosure messages. Expressions of emotional inhibition did not differ significantly by condition. TABLE 2 Proportion of Responses Containing Each Function, by Disclosure Condition Condition Function Kernel Seeking support Avoid negative evaluation Prevent subsequent disclosure Maintain relationship Right to know Sadness Anger Fear Surprise Concern Emotional inhibition Blame Information seeking about transmission General information seeking Emotional support Explicit advice Implicit advice Instrumental support Relationship affirmation Identity affirmation Gratitude Privacy assurance .37a,b,c,d .03 .06a,b .60a .02a .03 .10 .49 .50a,b,c,d .69a .24a .40a,b .32a,b .08 .07 .08a .09 .24 .02 .01 .50 .09a .00 .07a,b .40 .40e .60 .12a,b,c .29 .18a .13a .11a .04 .05 .23a .03 .02 .58b .06 .02 .08 .51 .35a .58 .23b .28 .11b,c,d,e .16b,c,d .06 .04 .06 .20b .06 .00a .60c .06 .00 .17a .48 .30b .49a .16d .34c .24c .04a,b .03a .06 .03 .17c .06 .00b .43a,b,c .05 .01 .13 .48 .26c,e .57 .30c,d .26a .27d .06c .06 .05 .07 .22d .04 .02 .57 .06 .02 .18b .41 .31d .59 .18 .19b,c .27e .05d .05 .01a .06 Note. Proportions in the same row that share subscripts differ significantly at p < .05. HIV DISCLOSURE MESSAGES Participants made more statements of blame when in the prevent-subsequent-disclosure and the right-to-know conditions than in the support-seeking condition. Questions about how HIV was transmitted (information seeking about transmission) were equally common across conditions. Information seeking more generally (general information seeking) appeared with different frequency, however, depending on the condition: Basic disclosures tended to be met with more questions than conditions involving negative evaluation, preventing subsequent disclosure, maintaining relational bonds, or the other’s right to know. Also, general information seeking was more frequent when disclosures involved support seeking than when they emphasized maintaining the relationship. Emotional support was offered less frequently when disclosures contained requests for no subsequent disclosures to others than when they were the kernel-only messages. In terms of other types of support, the different types of advice appeared to operate somewhat differently. Explicit advice was least common in the seeking-support condition, significantly less than in the kernel-only, avoid-negativeevaluation, and maintain-relationship conditions. The maintain-relationship messages also elicited more explicit advice than did the prevent-subsequent-disclosure messages. On the other hand, implicit advice appeared less frequently in response to right-to-know messages than either the kernel-only messages or prevent-subsequentdisclosure messages. Also, implicit advice was more common in the kernel-only condition than in the maintain-relationship condition. Relationship affirmation was expressed more in response to concern about negative evaluation than to requests for preventing subsequent disclosure, emphasis on relational maintenance, or honoring one’s right to know. Relationship affirmation also occurred more after support seeking than after the prevent-subsequent-disclosure messages. In the identity affirmation function category, support seeking produced more identity affirmation than did the prevent-subsequentdisclosure condition. Gratitude was expressed more commonly as a response to the kernel messages than to those 277 honoring the other’s right to know. There were no differences across conditions for privacy assurance. RQ1 also was examined with a series of analyses on the closed-ended questions. Each of the six reaction variables (seeking information about transmission, comforting, negative evaluation, upset, perceived competence, and more closeness in the relationship) was examined with the six message conditions treated as fixed factors and the four kernel messages treated as random factors. Treating the messages as random factors enhances generalizability to the broader population of HIV-disclosure messages and takes into account random variation that might be attributable to differences in messages (Jackson, 1992). Throughout all the analyses, there were no statistically significant effects for the kernel messages or for the condition by messages interactions. This indicates that all the effects of the conditions can be interpreted in a straightforward manner. Because the nonsignificant effects involving the various kernel messages are not of substantive interest, the specific values of these tests are not presented; instead, the results focus on the main purpose of the analyses— examining differences in the conditions. There was not significant evidence that participants’ seeking information about the transmission was influenced by the conditions, F(5, 15) = 1.34, ns. There was, however, a significant effect on comforting, F(5, 15) = 4.36, p = .01, partial h2 = .58. Post hoc analyses indicated that the seekingsupport condition was significantly higher on comforting than were the prevent-subsequent-disclosure and honoringthe-right-to-know conditions (see Table 3). In addition, the maintaining-relationship condition was significantly higher on comforting than right-to-know messages. Participants’ reports that they would negatively evaluate the sibling who disclosed were also associated with the message conditions, F(5, 15) = 3.52, p < .05, partial h2 = .52. Post hoc analyses indicated that messages citing the participants’ right to know evoked the highest levels of negative evaluation, and the right-to-know condition was significantly higher than the kernel-only and seeking-support conditions. Messages citing the goal of maintaining the relationship TABLE 3 Summary of Closed-Ended Reaction Reports by Message Condition Condition Reaction Seeking information about transmission Comforting Negative evaluation Upset Perceived competence More closeness Note. Kernel Seeking support Avoid negative evaluation Prevent subsequent disclosure Maintain relationship Right to know 5.89 (1.34) 6.57 (0.75) 2.30ab (1.14) 6.19 (1.19) 2.43 (1.18) 5.51abc (1.06) 6.02 (1.35) 6.73ab (0.61) 2.43c (1.31) 6.14 (1.05) 2.65 (1.27) 5.25d (1.21) 5.61 (1.96) 6.53 (0.99) 2.61 (1.46) 6.32 (1.27) 2.68 (1.25) 5.26e (1.25) 5.87 (1.67) 6.43a (0.82) 2.50 (1.49) 6.21 (1.15) 2.88 (1.34) 4.86ade (1.22) 5.64 (1.66) 6.67c (0.56) 2.72a (1.41) 6.22 (0.99) 2.66 (1.24) 4.99b (1.15) 5.61 (1.71) 6.40bc (1.09) 2.88bc (1.70) 6.19 (1.11) 2.86 (1.31) 5.02c (1.36) Values given are means (standard deviations). Means in the same row that share subscripts differ significantly at p < .05. 278 CAUGHLIN ET AL. elicited the second-highest level of negative evaluation, and this was significantly higher than that in the kernel condition. There was not a significant difference in levels of upset reported by participants, F(5, 15) = 0.16, ns, and perceived competence of the disclosure was not significantly influenced by the conditions, F(5, 15) = 1.44, ns. However, perceptions that the disclosure would lead to more closeness in the sibling relationship were related to the message conditions, F(5, 15) = 3.15, p < .05, partial h2 = .51. In particular, the condition aimed at preventing subsequent disclosure of the information to third parties produced the lowest scores on the closeness index and was significantly lower than three conditions: kernel-only, seeking-support, and avoid-negative-evaluation. Also, the maintainingrelationship conditions and the right-to-know condition were significantly lower than the kernel messages in terms of educing more closeness. Despite such differences among the groups in leading to more closeness, all of the groups were significantly higher than the scale midpoint, which was defined as “no change” in the relationship. Even the condition with the lowest scores, preventing subsequent disclosure, was significantly higher on average than the midpoint, t(80) = 6.36, p < .01. That is, participants in each condition reported that the relationship would become significantly closer rather than more distant. DISCUSSION This study is an important first step toward understanding how features of HIV-disclosure messages influence the impact of those disclosures. Like any initial study, this one has limitations, one of which is that the sample was composed of undergraduate students. Although undergraduates certainly can have siblings with HIV, the findings from this study may not generalize to other populations. This study also focused specifically on sibling relationships; HIVdisclosure messages may operate differently in different kinds of relationships. For example, if the disclosure is to a sexual or needle-sharing partner, concerns about the other person’s health would be more relevant (Derlega et al., 2000). Another limitation is the use of imagined encounters. Although this concern warrants replicating these findings, it does not undermine the value of this investigation. If the scenarios were not meaningful to some participants, this would introduce random error into the study. Clearly, such randomness could not explain why the various conditions were related systematically to the tenor of the responses. The data also showed little evidence that participants produced only socially desirable responses. If social desirability were excessive, one would expect uniformly positive responses. Instead, there was great variety, with many of the responses being far from socially desirable (e.g., “What the hell did you do? If it was your own fault, then I have no pity. Be careful when you are around me.”). Although the design of this study has limits, many research programs have begun with designs similar to that utilized in this study, and such programs point to the applicability of findings like those presented here. Much of the seminal research on comforting messages, for instance, involved having participants imagine attempting to comfort someone (e.g., Burleson, 1984), and this research has held up when applied more broadly (for a review, see Burleson & MacGeorge, 2002). The broader applicability of imagined interactions also appears in the HIV-disclosure literature. One study that asked individuals to imagine disclosing illnesses found that people are more likely to reveal having HIV to a partner than to family members (Greene, 2000), and the same pattern has been found in reports of actual disclosures (Derlega et al., 2004). In short, like much research in the social sciences, this study could benefit from a more naturalistic replication, but there are also good reasons to believe that the findings presented here provide a good beginning to understanding how variations in HIV messages influence the responses to those messages. More specifically, the results have a number of theoretical implications that pertain to two general areas: (a) the impact of variations in disclosure messages on reactions and (b) the utility of multiple-goals perspectives for studying disclosure processes. Variations in Disclosure Messages The primary goal of this investigation was to examine whether the messages used to reveal HIV-positive status would influence recipients’ reactions to that information. These findings indicate that message features matter. These results provide a helpful complement to previous research on HIV disclosure, which has documented much about when and why people reveal their status, but has focused much less on variations in disclosure messages (Greene et al., 2003). Moreover, this study is an important addition to the larger disclosure literature because that literature also includes relatively few studies examining the importance of message characteristics (Afifi, Caughlin, & Afifi, 2007). Because we examined a number of specific features of participants’ reactions, it is useful to summarize the results pertaining to each condition. Rather than repeating all of the findings, our focus here is on the salient aspects of the various conditions. For some conditions, the findings were mixed, with some fairly positive outcomes and some relatively negative ones. Consider the kernel-only condition, for instance: These messages were particularly likely to elicit negative emotional expressions (e.g., sadness and fear) in participants. Respondents receiving a kernel message were also relatively unlikely to report that they would evaluate their sibling negatively and fairly likely to report that the disclosure would make their relationship with their sibling closer. They also were relatively likely to seek a lot of information (other than about the means of transmission), to give HIV DISCLOSURE MESSAGES implicit and explicit advice, and to offer instrumental support. This pattern of findings, in which kernel messages tend to elicit informational support (i.e., advice) but are not particularly likely to elicit emotional support, is consistent with research showing that people interacting with a distressed person often give advice if they do not know what else to say (Burleson & Goldsmith, 1998). Not surprisingly, the closed-ended data indicated that respondents who received messages seeking support were more likely than several other groups to offer comforting. Respondents’ reports of concern were comparatively high in the seeking-support condition, and their statements blaming their sibling were relatively infrequent. The support-seeking messages were also less likely than several other types of messages to evoke explicit advice. Taken together, these findings suggest that alluding to one’s need for social support may be an effective way to receive comforting while simultaneously suppressing unsolicited advice, which often is unwanted advice (see Goldsmith & Fitch, 1997). The avoiding-negative-evaluation condition had relatively few salient reactions. Across all the closed-ended questions, it was only significantly different from one other condition (prevent subsequent disclosure) on one outcome measure (more closeness). However, the comparatively negative reactions to the prevent-subsequent-disclosure messages seemed to be driving this effect (rather than salient responses to the avoiding-negative-evaluation condition). In terms of the functions of the response messages (see Table 2), avoiding-negative-evaluation disclosures differed from multiple conditions in only two categories: It was lower than several others on instrumental support and higher than several others on relational affirmation. In sum, messages that overtly sought to avoid negative evaluation typically evoked responses that were about average in most respects, but there is some evidence that these messages can be successful at getting reassurance about one’s relationship. Several types of responses distinguished the preventsubsequent-disclosure condition. Notably, there was no evidence that requests to prevent subsequent disclosure to third parties resulted in assurances of such privacy; however, the prevent-subsequent-disclosure condition was the lowest in terms of eliciting emotional support in the open-ended measures, and it was fairly low in the closed-ended comforting measure as well. Responses to the prevent-subsequentdisclosure messages were the least likely to include relational assurance, and participants who received a prevent-subsequentdisclosure message were less likely than several other groups to indicate that their relationship would become closer. The findings associated with the condition concerning maintaining the relationship were a theoretically interesting mix of positive and negative outcomes. In terms of the functions, it was lowest in educing surprise and the highest in eliciting explicit advice. The maintain-relationship 279 condition was second highest (and significantly higher than one other condition) on the quantitative measures of both comforting and negative evaluation. It also was significantly lower than the kernel-only condition in terms of leading to more relational closeness. Together, these findings suggest that asking for relational assurances frequently leads to overt responses that are consistent with that request but simultaneously leads to disappointment and a relatively diminished sense that the relationship will become closer. Last, the condition in which messages referred to the participants’ right to know was not particularly salient in terms of the response functions. The condition honoring the recipient’s right to know was only significantly different from more than one other condition on one function: It was lower than two categories in terms of implicit advice responses. In contrast to the open-ended measures, the closed-ended measures suggested that the right-to-know condition was fairly low on comforting, comparatively high on negative evaluation, and lower than the kernel-only condition in terms of beliefs that the relationship would become closer. This investigation cannot establish exactly why certain conditions fared relatively poorly. Consider the messages that included requests to prevent subsequent disclosure, which generally were rated less positively than basic (kernel) messages that disclosed the information simply and directly. One possible explanation comes from politeness theory (Brown & Levinson, 1987), which suggests that relatively direct messages are expected and often preferred in close relationships. Given that the disclosures in this study involved sibling relationships, it is likely that most of the relationships were fairly close (see Knapp, Ellis, & Williams, 1980). In such a context, pleading to not divulge the information to others may be seen as incongruent with the relationship. Such a plea also could be threatening to the recipient’s identity because it implies that he or she might not guard the information otherwise (i.e., might not be trustworthy). It is also possible that raising the issue of privacy heightens sensitivity to issues of stigma; that is, it reminds the recipient that HIV is a highly stigmatized illness (Alonzo & Reynolds, 1995). This might exacerbate any underlying sense among recipients that an HIV-positive status indicates contagiousness or immorality (e.g., sexual behavior or injection drug use). Telling the sibling not to tell others also might imply that the individual with HIV is not telling his or her status to people who should be told (e.g., sexual partners). By asking a sibling not to tell others, it could raise questions about the infected sibling’s responsible handling of the information. Along similar lines, the messages noting that participants had a right to know the information were viewed more negatively than the plain disclosure messages. Again, identity issues might be relevant: Stating that a sibling is being told because of a right to know might imply that the person with HIV does not really want to share the information, which could be a threat to the recipient’s identity (e.g., by implying that the recipient is only being included due to obligation). It also is plausible that referring to a right to know may 280 CAUGHLIN ET AL. suggest that the information is personally relevant to the sibling, which could heighten any fears the sibling might have about being exposed to HIV (Serovich et al., 1998). Certainly, there are other possible explanations for the various findings, but the scope of this project does not provide evidence regarding the most useful explanations. Also, this investigation focused on disclosures to siblings, and the findings may not be generalizable beyond that particular relationship. Indeed, the types of messages examined here may operate differently in different kinds of relationships. For instance, people who disclose their HIV-positive status to an acquaintance may be viewed as justified in explicitly asking that person not to tell others the information. Multiple-Goals Perspective on Disclosure To inform the creation of various disclosure messages, this investigation took a multiple-goals perspective. The findings have some theoretically interesting implications for such a perspective. First, the open-ended responses revealed little evidence that recipients’ responses matched the goals cited in the disclosure messages (see Table 2). Avoidingnegative-evaluation messages were no more likely to elicit identity affirmation than were other messages, and there was no evidence that the prevent-subsequent-disclosure condition was particularly likely to elicit privacy assurances. The maintain-relationship condition was actually less likely to evoke relationship affirmation than were some message types. The seeking-support condition was not significantly higher (and was sometimes significantly lower) than any other condition on functions like emotional support, advice, and instrumental support. Seeking-support messages did elicit comparatively high levels of relationship affirmation, and were higher than several other conditions in the closed-ended measure of comforting. In short, although requests for support may be met with relational affirmation and comforting, the overall pattern of results provided little evidence that responses addressed the overt goals in the disclosure messages. This pattern, coupled with the findings suggesting that some conditions elicited a mix of positive and negative responses, suggests that appealing to multiple goals while disclosing HIV is more complicated than simply disclosing the information while divulging other concerns. One possible explanation for why the various conditions did not elicit responses that closely matched the disclosure messages is that the various conditions would be relevant to recipients’ goals as well. As suggested earlier, the prevent-subsequentdisclosure condition may have threatened participants’ goal of maintaining an identity as a good and trustworthy sibling, or the right-to-know messages may portray HIV as something particularly relevant to the receiver’s identity, which could heighten anxiety. Thus, although this study demonstrated that various kinds of messages elicit different types of responses, future research could benefit from a more systematic examination of how features of HIV disclosure messages are relevant to recipients’ goals. The most sophisticated and effective disclosure messages might be based on perceiving and addressing recipients’ concerns (rather than focusing on disclosers’ goals alone). Although speculative, this possibility is congruent with abundant research emphasizing the utility of addressing others’ perspectives (e.g., in the social support literature, see Burleson & MacGeorge, 2002). Another set of theoretically interesting findings involved the fact that the prevent-subsequent-disclosure and right-toknow conditions were evaluated more harshly than the simple kernel messages. At first glance, these results appear to run counter to much writing from multiple-goals perspectives, which often suggests that messages that pursue a primary goal (e.g., influence) while also accommodating a secondary goal (e.g., considerateness) are likely to be more appropriate and effective than ones that ignore secondary goals (O’Keefe & Shepherd, 1987; Wilson, 2002). It is important to point out, however, that the number of goals addressed is not the only factor that influences how a message is perceived; the particular goals addressed also matter. As suggested earlier, the prevent-subsequent-disclosure and rightto-know messages overtly addressed possible goals of a discloser but may have been threatening in various ways to the recipients. If the additional goals addressed were instead tailored to recipients’ goals, multiple-goal messages probably would be perceived positively. Moreover, future investigations should expand the range of message features (e.g., direct vs. indirect messages) examined. This study used messages that contained explicit expressions of goals (e.g., avoiding negative evaluation by saying “please don’t judge me” or maintaining a relational bond by saying “I always want to maintain our trusting relationship”). These same goals could be pursued less explicitly. For instance, Cohen-Silver, Wortman, and Crofton (1990) found that discussing one’s good coping efforts when disclosing an illness creates a positive impression and elicits support from others, suggesting that mentioning coping efforts may serve as an indirect request for support. Such findings imply that the manner in which goals are pursued (not just which goals are addressed) may be important. This possibility is consistent with O’Keefe’s (1988) theory of message design logics, which describes three fundamentally different ways to address multiple goals. In expressive messages, one goal is selected and emphasized; in conventional messages, the most important goal is pursued overtly, but additional information may be added to attend to secondary goals; in rhetorical messages, the various goals are integrated by redefining the various goals to be congruent. In this investigation, the messages with more than one goal can probably best be thought of as conventional: They clearly and explicitly addressed the primary goal of disclosure and added separate information designed to address other concerns. It is possible that if the various goals were integrated (i.e., if the HIV-disclosure messages HIV DISCLOSURE MESSAGES had rhetorical design features), messages with multiple goals would have been rated at least as favorably as the kernel messages. It is also conceivable that more complex messages are simply less effective in this situation, and future research ought to examine such possibilities. Of course, not all of the messages with more than one goal fared worse than the simple and direct (kernel) messages. The conditions involving seeking support, avoiding negative evaluation, and maintaining the relationship all received both more positive responses and more negative responses than the kernel condition on at least some dimensions. This suggests that these message categories represent different tradeoffs. For instance, the seeking-support messages seem well suited to eliciting comforting but appeared to reduce offers for instrumental support. To the extent that one wants or needs comforting more than instrumental support (or is willing to ask for the instrumental support in a subsequent message), this type of message may be superior to the plain kernel message. If the person most needs verbal assurances of instrumental support, however, this might not be the best message strategy, or the message may need to be very specific about what forms of support are needed. 281 benefits compared to most of the other message types. One potential drawback of the kernel messages is that they were particularly likely to elicit instrumental support and advice. This is consistent with evidence that people who do not know what to say when conversing with a distressed individual often give advice or attempt to solve the problem (Burleson & Goldsmith, 1998). By not mentioning something else for participants to discuss, the kernel messages may have evoked unsolicited advice, which is often unwelcome advice (Goldsmith, 2004). Despite this potential drawback, in some situations, keeping the message simple may be the most competent means of disclosure. ACKNOWLEDGMENTS The order of the second through fifth authors is alphabetical. We are grateful to Amanda Howald for her assistance with unitizing and coding. REFERENCES Final Thoughts Knowing that different kinds of messages influence reactions to HIV disclosures has enormous practical significance. The ability to shape reactions means that it should be possible to develop a model about normatively effective disclosures (see Goldsmith, 2004). Such a model could be the basis for interventions aimed at helping people living with HIV to disclose in ways that maximize certain benefits (e.g., gaining support, achieving a closer relationship with the recipient) while minimizing the costs (e.g., disappointing the recipient). The specifics of such a program are beyond the scope of this investigation, but the findings provide useful leads. Overall, the messages that explicitly sought support appeared to evoke the most empathetic and positive responses, suggesting that people initially disclosing HIV-positive test results might be well advised to ask for support if they need it. These findings also hint that some types of messages may be ineffective. Messages that referred to the participant’s right to know, for instance, elicited more negative evaluations than did several of the other conditions. Given that concerns about eliciting negative evaluations can be salient for people disclosing that they have HIV (Derlega et al., 2004; Greene & Faulkner, 2002), it is useful to know that mentioning that one believes the sibling had a right to know may amplify the potential for losing the sibling’s positive regard. In addition, the fairly positive reactions to the plain and simple (kernel) messages could be useful. An intervention might note, for example, that if a person has doubts about conveying a more complex message (e.g., if the person is nervous about telling), a simple message has several Afifi, T. D., Caughlin, J. P., & Afifi, W. A. (2007). The dark side (and light side) of avoidance and secrets. In B. H. Spitzberg & W. R. Cupach (Eds.), The dark side of interpersonal communication (2nd ed., pp. 61– 92). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Alonzo, A. A., & Reynolds, N. R. (1995). Stigma, HIV, and AIDS: An exploration and elaboration of a stigma trajectory. Social Science and Medicine, 41, 303–315. Berger, C. R. (2005). Interpersonal communication: Theoretical perspectives, future prospects. Journal of Communication, 55, 415–447. Brashers, D. E., Goldsmith, D. J., & Hsieh, E. (2002). Information seeking and avoiding in health contexts. Human Communication Research, 28, 258–271. Brashers, D. E., Neidig, J. L., Cardillo, L. W., Dobbs, L. K., Russell, J. A., & Haas, S. M. (1999). “In an important way, I did die.” Uncertainty and revival among persons living with HIV or AIDS. AIDS Care, 11, 201– 219. Brashers, D. E., Neidig, J. L., Russell, J. A., Cardillo, L. W., Haas, S. M., Dobbs, L. K., et al. (2003). The medical, personal, and social causes of uncertainty in HIV illness. Issues in Mental Health Nursing, 24, 497– 522. Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. New York: Cambridge University Press. Brunig, J. L., & Kintz, B. L. (1997). Computational handbook of statistics (4th. ed.) New York: Longman. Burleson, B. R. (1984). Age, social–cognitive development, and the use of comforting strategies. Communication Monographs, 51, 140–153. Burleson, B. R., & Goldsmith, D. J. (1998). How the comforting process works: Alleviating emotional distress through conversationally induced reappraisals. In P. A. Andersen & L. K. Guerrero (Eds.), The handbook of communication and emotion (pp. 245–280). Thousand Oaks, CA: Sage. Burleson, B. R., & MacGeorge, E. L. (2002). Supportive communication. In M. L. Knapp & J. A. Daly (Eds.), Handbook of interpersonal communication (3rd ed., pp. 374–424). Thousand Oaks, CA: Sage. Burleson, B. R., Wilson, S. R., Waltman, M. S., Goering, E. M., Ely, T. S., & Whaley, B. R. (1988). Item-desirability effects in compliance-gaining research: Seven studies documenting artifacts in the selection procedure. Human Communication Research, 14, 429–486. 282 CAUGHLIN ET AL. Clark, R. A., & Delia, J. G. (1979). Topoi and rhetorical competence. Quarterly Journal of Speech, 65, 187–206. Cohen-Silver, R., Wortman, C. B., & Crofton, C. (1990). The role of coping in support provision: The self-presentational dilemma of victims of life crises. In B. R. Sarason, I. G. Sarason, & G. R. Pierce (Eds.), Social support: An interactional view (pp. 397–426). New York: Wiley. Derlaga, V. J., Lovejoy, D., & Winstead, B. A. (1998). Personal accounts of disclosing and concealing HIV-positive test results: Weighing the benefits and risks. In V. J. Derlega & A. P. Barbee (Eds.), HIV and social interaction (pp. 147–164). Thousand Oaks, CA: Sage. Derlega, V. J., Winstead, B. A., & Folk-Barron, L. (2000). Reasons for and against disclosing HIV-seropositive test results to an intimate partner: A functional perspective. In S. Petronio (Ed.), Balancing the secrets of private disclosures (pp. 53–69). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Derlega, V. J., Winstead, B. A., Greene, K., Serovich, J., & Elwood, W. N. (2004). Reasons for HIV disclosure/nondisclosure in close relationships: Testing a model of HIV-disclosure decision making. Journal of Social and Clinical Psychology, 23, 747–767. Goldsmith, D. J. (2004). Communicating social support. New York: Cambridge University Press. Goldsmith, D. J., Brashers, D. E., Kosenko, K. A., & O’Keefe, D. (2007). Social support and living with HIV: Findings from qualitative studies. In T. Edgar, S. M. Noar & V. S. Freimuth (Eds.), Communication perspectives on HIV/AIDS for the 21st century (pp. 101—136). New York: Taylor & Francis. Goldsmith, D. J., & Fitch, K. (1997). The normative context of advice as social support. Human Communication Research, 23, 454–476. Goldsmith, D. J., & MacGeorge, E. L. (2000). The impact of politeness and relationship on perceived quality of advice about a problem. Human Communication Research, 26, 234–263. Goldsmith, D. J., Miller, L. E., & Caughlin, J. P. (2008). Openness and avoidance in couples communicating about cancer. Communication Yearbook, 31, 62–115. Greene, K. (2000). Disclosure of chronic illness varies by topic and target: The role of stigma and boundaries in willingness to disclose. In S. Petronio (Ed.), Balancing the secrets of private disclosures (pp. 123–135). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Greene, K., Derlega, V. J., Yep, G. A., & Petronio, S. (2003). Privacy and disclosure of HIV in interpersonal relationships. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Greene, K., & Faulkner, S. L. (2002). Expected versus actual responses to disclosure in relationships of HIV-positive African American adolescent females. Communication Studies, 53, 297–317. Guetzkow, H. (1950). Unitizing and categorizing problems in coding qualitative data. Journal of Clinical Psychology, 6, 47–58. Hatfield, J. D., & Weider-Hatfield, D. (1978). The comparative utility of three types of behavioral units for interaction analysis. Communication Monographs, 45, 44–50. Jackson, S. (1992). Message effects research: Principles of design and analysis. New York: Guilford Press. Knapp, M. L., Ellis, D. G., & Williams, B. A. (1980). Perceptions of communication behavior associated with relationship terms. Communication Monographs, 47, 262–278. Lambert, B. L., & Gillespie, J. L. (1994). Patient perceptions of pharmacy students’ hypertension compliance-gaining messages: Effects of message design logic and content themes. Health Communication, 6, 311–325. McCroskey, J. C., & Young, T. J. (1979). The use and abuse of factor analysis in communication research. Human Communication Research, 5, 375–382. O’Keefe, B. J. (1988). The logic of message design: Individual differences in reasoning about communication. Communication Monographs, 55, 80–103. O’Keefe, B. J., & Shepherd, G. J. (1987). The pursuit of multiple objectives in face-to-face persuasive interaction: Effects of construct differentiation on message organization. Communication Monographs, 54, 396–419. Rintamaki, L. S., Kosenko, K., Scott, A. M., Jensen, R., & Jordan, C. (2005, June). HIV status disclosure to past, present, and prospective sexual partners. Paper presented at the meeting of the National HIV Prevention Conference, Centers for Disease Control and Prevention, Atlanta, GA. Serovich, J. M. (2001). A test of two HIV disclosure theories. AIDS Education and Prevention, 13, 355–364. Serovich, J. M., Kimberly, J. A., & Greene, K. (1998). Perceived family member reaction to women’s disclosure of HIV-positive information. Family Relations, 47, 15–22. Serovich, J. M., Mason, T. L., Bautista, D., & Toviessi, P. (2006). Gay men’s report of regret of HIV disclosure to family, friends and sex partners. AIDS Education and Prevention, 18, 132–138. Siegel, K., & Lekas, M. H. (2002). AIDS as a chronic illness: Psychosocial implications. AIDS, 16 (Suppl. 4), S69–S76. Stein, M. D., Freedberg, K. A., Sullivan, L. M., Savetsky, J., Levenson, S. M., Hingson, R., et al. (1998). Sexual ethics: Disclosure of HIV-positive status to partners. Archives of Internal Medicine, 158, 253–257. Vangelisti, A. L., & Young, S. L. (2000). When words hurt: The effects of perceived intentionality on interpersonal relationships. Journal of Social and Personal Relationships, 17, 393–424. Wilson, S. R. (2002). Seeking and resisting compliance: Why people say what they do when trying to influence others. Thousand Oaks, CA: Sage. APPENDIX Condition 1: Instrumental Goal was Disclosing the Information (Kernel Messages) • I just got back the results of some medical tests, and it turns out that I have HIV. • I tested positive for HIV. • I found out that I am HIV positive. • I recently found out that I have HIV. Condition 2: Additional Instrumental Goal was Seeking Support • I just got back the results of some medical tests, and it turns out that I have HIV. I’m not sure what lies ahead, but I’ll probably need to rely on you for help. • I wanted to tell you this because I think I am going to need your help. I tested positive for HIV, and I am going to need assistance with certain things. • I found out that I am HIV positive. The reason that I am telling you this is because you have always given me valuable advice about whatever it is that I am going through. • You are always so supportive of me, so I knew that you would be a good person to confide in. I recently found out that I have HIV. HIV DISCLOSURE MESSAGES Condition 3: Identity Goal of Avoiding Negative Evaluation • But I want to ask you, please don’t judge me. I just got back the results of some medical tests, and it turns out that I have HIV, which is bad enough, but I don’t think I could take it if you ended up thinking less of me. • I’m afraid that this would change the way people see me, and I don’t want that to happen with us. You see, I tested positive for HIV. • I don’t want you to be disappointed in me, but I found out that I am HIV positive. • I hope that this doesn’t make people think negatively of me, but I recently I recently found out that I have HIV. Condition 4: Identity Goal of Preventing Subsequent Disclosure to a Third Party • I just got back the results of some medical tests, and it turns out that I have HIV. I am trying to keep this quiet because not everybody will be understanding about this. So, please keep this information to yourself, okay? • I have not really told many people about this yet, and I would hope that you will keep this just between us. I tested positive for HIV. • But it’s important to me that this information not get spread around. I found out that I am HIV positive. • I recently found out that I have HIV. I would really appreciate you not telling anybody else because there are some people I don’t want to know. 283 Condition 5: Relational Goal of Maintaining Relational Bond • I just got back the results of some medical tests, and it turns out that I have HIV. I’m letting you know this because my relationship with you has always been so important to me, and I want that to continue no matter what happens. • You know I trust you completely, and I hope you feel the same way. Because I always want to maintain our trusting relationship, I think it’s important for me to tell you that I tested positive for HIV. • I found out that I am HIV positive. I am telling you about this because I appreciate how good you have always been to me, and how we have always been able to share the good and bad things in our lives with each other. I hope we always stay so close. • We have always been so close and you are very important to me and that’s why I think that I can tell you that I recently found out that I have HIV. Condition 6: Relational Goal of Honoring Other’s Right to Know • I think it’s important for me to tell the people who are close to me. I just got back the results of some medical tests, and it turns out that I have HIV. • I tested positive for HIV. I thought you deserved to know that. • This is not good news, but it’s something I think you should know. I found out that I am HIV positive. • I recently found out that I have HIV. I wanted to tell you right away because I think you have a right to know. Health Communication, 27: 356–368, 2012 Copyright © Taylor & Francis Group, LLC ISSN: 1041-0236 print / 1532-7027 online DOI: 10.1080/10410236.2011.586988 Assessing Health Diagnosis Disclosure Decisions in Relationships: Testing the Disclosure Decision-Making Model Kathryn Greene, Kate Magsamen-Conrad, and Maria K. Venetis Department of Communication Rutgers University Maria G. Checton Department of Health Care Management College of Saint Elizabeth Zhanna Bagdasarov Department of Communication Rutgers University Smita C. Banerjee Memorial Sloan Kettering Cancer Center, New York Illness affects millions of Americans each year, and the disclosure of health conditions can facilitate access to social support, in addition to other physical and physiological benefits. This article tests the Disclosure Decision-Making Model (DD-MM; Greene, 2009) to predict factors that influence the likelihood of disclosing (and past disclosure of) nonvisible physical or mental health-related information. One hundred eighty-seven (n = 187) people were recruited for a study to report on both disclosing and not disclosing a nonvisible health condition. Measured variables included information assessment, relational quality, anticipated reactions (support, relational consequences), confidence in response, disclosure efficacy, and disclosure (likelihood of disclosure and depth of disclosure). Structural equation modeling results supported many of the proposed hypotheses, with a great deal of similarity across models. Specifically, assessing information predicted efficacy, and to some extent relational outcomes. Closeness was related to response overall and to efficacy in one model. Response predicted outcome overall and likelihood of disclosure in one model. Finally, efficacy predicted likelihood of disclosure and depth of disclosure. The article discusses the implications of the findings for understanding information, relationship assessments, and efficacy in disclosing health diagnoses. People with nonvisible health conditions such as early-stage cancer must make decisions about sharing their health diagnosis with others. Such sharing has been linked to social support and physical and mental well-being (see Frattaroli, 2006). Individuals face a number of challenges (e.g., physical or emotional) when they are initially diagnosed with an illness and as Correspondence should be addressed to Kate Magsamen-Conrad, 700 Roseto Avenue, Bangor, PA 18013. E-mail: Kmagsame@rutgers.edu they continue to manage that illness. People may experience a range of emotions, including uncertainty, fear, helplessness, and/or anxiety (see Braitman et al., 2008; Epstein & Street, 2007). A prominent issue includes the decision of whether or not to disclose this information to others. Individuals manage personal or private health information through a variety of strategies (see Derlega, Metts, Petronio, & Margulis, 1993; Petronio, 2002). One strategy that has received considerable research attention is disclosure. Scholars have recently focused attention on HEALTH DIAGNOSIS DISCLOSURE DECISIONS developing models that identify factors influencing disclosure decisions. These models (e.g., Cycle of Concealment Model, Afifi & Steuber, 2010; Disclosure Decision-Making Model, Greene, 2009; Revelation Risk Model, Afifi & Steuber, 2009) endeavor to outline the process of coming to the decision to disclose or conceal private information or secrets to particular recipients. Much of the research on privacy and disclosure is grounded in dialectical approaches that have emerged in relationship research (see Baxter & Braithwaite, 2008). The specific application of information management as self-disclosure often emphasizes boundaries or the tension between sharing and withholding. This boundary concept was initially described by Derlega and Chaikin (1977), and Derlega has developed a line of quantitative research emphasizing, among other concepts, the self, other, and relationship reasons for and against disclosure in the context of HIV. Petronio (2002) developed a framework (Communication Privacy Management, CPM) that integrates dialectics and boundaries and describes a number of constructs that operate in privacy management processes. The CPM framework has been utilized extensively in numerous qualitative studies to frame research findings and has significant heuristic value grounded in prior quantitative research (e.g., gender and disclosure). Other researchers have provided model testing based on related yet narrower phenomena including secrets (see Kelly, 2002; or Vangelisti & Caughlin, 1997; Vangelisti, Caughlin, & Timmerman, 2001) and self-disclosure (see Afifi and coworkers already cited, or Caughlin & Afifi, 2004; Caughlin, Afifi, Carpenter-Theune, & Miller, 2005), including health disclosure (see Greene, 2009, and Greene and coworkers, cited later; or Derlega and coworkers’ work, already cited). The DD-MM emerged from this general area as a model that is narrow in scope and focuses on the health disclosure decision process, a subprocess of the more global information management area. The DD-MM (Greene, 2009) is particularly relevant to health diagnosis disclosures because it fully explicates the assessment of the health information component of decision making. Disclosure research has moved beyond theorizing to testing relations between and among variables. In general, health disclosure decision making involves coping with dialectical dilemmas of balancing risks with rewards (e.g., Greene, Derlega, & Mathews, 2006; Petronio, 2002). The assessment of these risks may also involve a compromise between individuals’ needs and their concerns for self-, other-, and relationship-protection issues (see Afifi & Steuber, 2009; Greene, Derlega, Yep, & Petronio, 2003). Because disclosure can contribute to disclosers’ feelings of vulnerability (Afifi & Olson, 2005; see also, Petronio, 2002), individuals make deliberate choices about how, when, and with whom they choose to share their diagnoses (e.g., Petronio, Reeder, Hecht, & Mon’t Ros-Mendoza, 1996). The DD-MM (Greene, 2009) is located at the crux of health communication and interpersonal communication 357 research in secrets, privacy, avoidance, uncertainty, and information management. What is similar about this research is recognition of the dialectical nature of the process of sharing information, yet the DD-MM model specifically tested seeks to examine what factors are quantitatively weighed in this disclosure decision process (where potential disclosers assess information and recipients for possible sharing, as well as their own efficacy for disclosing the information). The DD-MM further separates itself as a model with uncertainty at the core, uncertainty related to specific predictors of disclosure decisions. This article tests the DD-MM in two different ways. Study I tests the DD-MM as conceptualized within the process of sharing information not disclosed to a specific person. Study II assesses the same diagnosis where participants report retrospectively about disclosing the information to a different person. DISCLOSURE DECISION-MAKING MODEL (DD-MM) The DD-MM (Greene, 2009) explicates the process of coming to a disclosure enactment based on three assessments, including both direct and indirect effects. The model is grounded in uncertainty (cf. Babrow, 2001; Brashers, 2001) and how people balance potential risks related to different aspects of the disclosure decision (see Derlega et al., 1993; Petronio, 2002). As one component, individuals assess their diagnosis or the information under consideration for disclosure. People also evaluate several aspects related to the person to whom they might disclose and their perceived disclosure efficacy to share the information. The DD-MM argues that people base their decision of whether or not to disclose the information on evaluation of these three factors. To date, there is one study testing the DD-MM with general disclosure but not with health information (see also Checton & Greene, in press). Greene et al. (2009) surveyed 283 couples about information they had not yet shared with the person they brought to the study. Greene et al. (2009) tested four key predictions of the DD-MM and found basic support for the model. We review model components next. Assessing information (the diagnosis). One aspect in the process of coming to a disclosure decision is an assessment of the health diagnosis or the information; that is, what is disclosed matters, consistent with many prior disclosure/privacy theories and frameworks (e.g., Derlega et al., 1993; Kelly, 2002; Petronio, 2002) but often limited in prior measurement to information valence. In the DD-MM this assessment was developed specifically for health information and consists of consideration of five potentially overlapping factors: stigma, prognosis, symptoms, preparation, and relevance. Other disclosure models assess information 358 GREENE ET AL. as key (e.g., CPM, RRM) but do not measure beyond valence (and also are not health specific). The present study is the first to measure the five information assessment factors, and this is crucial to test the model and assess the significance of the model contribution. The information is conceptualized as a single construct with five subfactors, but this structure is unconfirmed to date; the factors could plausibly form a second-order factor or two (or more) correlated latent constructs. In previous tests of the DD-MM (Greene et al., 2009), more general information valence positively predicted receivers’ anticipated reactions. Negative diagnosis assessment is expected to decrease perceived efficacy and intention to disclose, but these relations will be mediated by assessments of receiver reactions. The first of the five information components is the stigma associated with the diagnosis, widely discussed within the context of illness and the most studied of the information components (see Derlega, Winstead, Greene, Serovich, & Elwood, 2004; Herek, Capitanio, & Widaman, 2002). Perceptions of stigma likely decrease intentions to disclose, although these effects may be mediated by anticipated reactions and efficacy. Additionally, the discloser evaluates the disease prognosis, including consideration of the outcome of the disease (e.g., treatable, chronic vs. terminal) and prognosis uncertainty. Symptoms (especially visibility of symptoms and disease progression) of the disease are also evaluated as part of information assessment. Another consideration is disclosers’ preparation for the diagnosis. Illness diagnoses may be somewhat anticipated (e.g., if there is a family history of illness, such as breast cancer) or may be completely unexpected (the result of a routine medical exam). Finally, information assessment involves consideration of the relevance of the diagnosis to others (e.g., whether others are directly or indirectly affected by the diagnosis). When people believe that the diagnosis is relevant to others they are more likely to disclose their health diagnosis, especially if the disease can be transmitted or is genetically linked (see Greene, 2009). These five components of the information form a more complex and robust conceptualization of health information, and this article presents both data and measures to contribute to the field. Besides the information, disclosers also consider aspects of the receiver in deciding to disclose. Assessing the Receiver Another part of the disclosure decision-making process is analyzing the potential receiver. The discloser evaluates the quality of the relationship with and anticipated reactions of a specific disclosure target. In general, better relational quality is associated with more positive perceptions of anticipated response (see Afifi & Olson, 2005; Greene et al., 2009; Petronio, 2002; Vangelisti, Caughlin, & Timmerman, 2001). Receiver assessment is a component of several disclosure decision-making models; however, few models operationalize anticipated reactions (specifically, not separating anticipated response and outcome). Relationship quality. People generally choose to disclose to those with whom they feel “close” and whom they can trust (see Greene, 2009; Petronio, 2002). Relational quality has been a component of most disclosure and privacy theorizing. Greene et al. (2009) determined that people feel more confident in their abilities to disclose personal/private information to people when they feel close and expect more positive responses from these people. Finally, better relational quality and more positive anticipated responses are related to increased disclosure intentions or willingness to disclose (e.g., Afifi & Steuber, 2009; Caughlin & Afifi, 2004; Vangelisti & Caughlin, 1997). Anticipated reaction. Another aspect of receiver assessment is anticipated reactions or consideration of what would happen if someone did disclose a diagnosis (see Caughlin, Afifi, Carpenter-Theune, & Miller, 2005; Greene & Faulkner, 2002; Vangelisti et al., 2001; see also disclosure ramifications in Petronio, 2002). Anticipated reaction has been operationalized in a number of ways, for example, in terms of valence, but also through a variety of motivations (or goals) for disclosure (see Derlega et al., 2004; Greene et al., 2006). Greene et al. (2009) introduced a conceptualization separating anticipated reactions in two dimensions that are reported in prior literature: anticipated response (e.g., provision of support) and anticipated outcome (e.g., relational consequences). One primary distinction between response and outcome is temporal (see Magsamen-Conrad, 2010). That is, after the personal/private information is shared, there is a more immediate response from the recipient (e.g., emotional reaction). This response may be defined as an immediate reply to the disclosure/discloser as communicated in words and/or actions. Anticipated response may be comprised of a number of subtypes (Magsamen-Conrad, 2010). For the present study of health disclosure, we initially focus on anticipated supportive responses because of the prevalence in the illness literature. Outcome, compared to response, is conceptualized as the end result or consequence of the disclosure and may have an aspect of finality (e.g., relationship dissolution). Because response and outcome are related to the same information (between the same discloser and recipient), it is expected that perceptions of anticipated response will influence perceptions of anticipated outcomes (and not the reverse). Confidence in response. The DD-MM identifies an additional receiver-oriented variable that may affect an individual’s decision to disclose. The variable confidence in response (see Greene, 2009, p. 239) reflects the degree to which the discloser is certain that the intended target (the receiver) will respond to the disclosed information in the way that the discloser anticipates. This variable may HEALTH DIAGNOSIS DISCLOSURE DECISIONS Hypothesized Model for Undisclosed Information be related to anticipated reactions (e.g., if the anticipated response or outcome is expected to be “negative,” how sure must disclosers be before refusing to disclose?). At this time, it is unclear how confidence in response affects the disclosure process because this variable has not been included in prior disclosure or health models. However, we propose that after individuals consider how the receiver might react (both response and outcome) they also consider how confident or certain they are in that reaction. Being able to accurately anticipate another’s response (either good or bad) with some degree of certainty may result in people feeling more prepared (less uncertain) about the other’s reaction. When confident in response, individuals may feel they are better able to prepare themselves and thus feel more efficacious about disclosing. Based on the preceding rationale, two models are hypothesized (undisclosed and disclosed). The first model (see Figure 1) examines health information that has not yet been shared, the undisclosed model. When people negatively assess the health condition, they will anticipate more negative responses (H1a) and outcomes (H1b) and have less disclosure efficacy (H1c). Additionally, when people perceive themselves in a close relationship, they will also anticipate more positive responses (H2a), have more confidence in receiver reactions (H2b), and have more disclosure efficacy (H2c). A person’s perceptions that the receiver will respond positively to the disclosure (e.g., offer support) should result in perceptions of more positive relationship outcomes (e.g., “closer” relationship; H3a) as well as an increased likelihood that the discloser will disclose this diagnosis to this receiver (H3b). When individuals anticipate more positive relational outcomes, they will be more confident in their ability to anticipate disclosure target’s responses (H4). Finally, increased disclosure efficacy will predict disclosure intentions (H6) such that when individuals have more confidence in their ability to disclose the information they are more likely to share. Disclosure Efficacy The third DD-MM assessment is efficacy for sharing diagnosis (disclosure efficacy rather than communication efficacy). Other disclosure models have included various forms of efficacy (e.g., Afifi, Olsen, & Armstrong, 2005; Afifi & Steuber, 2009), and the DD-MM describes how both confidence and skills are necessary to disclose health information. People may also resort to “alternative” methods of disclosure (e.g., through computer-mediated communication [CMC] or using a third party) if they do not feel that they have the skills necessary to successfully disclose the information and produce the desired result. Disclosure models have confirmed that people who felt that they had more confidence in their ability to share the diagnosis were also more likely to disclose that information (Afifi & Steuber, 2009; Greene et al., 2009). Information Assessment Hypothesized Model for Disclosed Information In order to replicate and test model stability across health decisions, this article also assesses predictors in the DD-MM based on retrospective reports of health diagnoses already disclosed. Despite this retrospective limitation, health diagnosis disclosure decisions are expected to function in similar ways, with the exceptions that (a) some variables must be operationalized differently, (b) some variables are no H1c H1b H1a – – – Anticipated Response H3a Anticipated Outcome + H3b H4 + H2a Relational Quality + H2b + Confidence in Response H5 + + Disclosure Efficacy H6 + + H2c Likelihood of Disclosure FIGURE 1 359 Theoretical model—undisclosed. 360 GREENE ET AL. H1b Information Assessment – H1a – Antic. Response (Retrospective) + H3a + Disclosure Efficacy H2a H4 + + H3b H2b + Relational Quality Depth of Disclosure FIGURE 2 Theoretical model—disclosed. longer applicable (e.g., confidence in response), and (c) one variable was not assessed to simplify recall instructions (retrospective reports of anticipated outcomes). Based on the preceding rationale, the following disclosed model is hypothesized (see Figure 2). First, the health information assessment directly predicts both retrospective report of anticipated response (H1a) and disclosure efficacy (H1b). Relational quality predicts retrospective report of anticipated response (H2a) and disclosure efficacy (H2b). Retrospective report of anticipated response positively predicts efficacy (H3a) and disclosure depth (H3b). Finally, increased disclosure efficacy predicts disclosure depth (H4) such that when individuals have more disclosure efficacy they will report deeper disclosure of their health information. METHOD Procedure Participants provided self-report data about a nonvisible health condition. For minimal extra credit, students from communication courses at a large university in the northeastern United States recruited individuals who met study criteria. Conceptually, researchers sampled for serious/significant “nonvisible” illnesses for which patients were currently under treatment. Researchers distributed an announcement listing qualifying health conditions.1 1 Examples of conditions listed on the announcement as qualifying included STIs, eating disorders, cancers (except skin cancer), and lupus. Examples of conditions that were listed on the announcement as not qualifying included allergies, high blood pressure/hypertension, migraines, broken bones, and ulcers. To qualify, a person must have a current, diagnosed, qualifying condition and be under treatment (and/or in recovery in the case of addiction). Participants were screened privately by a researcher to ensure that they met all criteria. Participants were individually screened by a researcher upon arrival and, if the condition qualified, completed a survey about the health diagnosis and managing that information. Participants reported on one person to whom they had shared and another to whom they had not disclosed the health diagnosis. Participants The final sample consisted of 183 (n = 183) male (n = 65) and female (n = 118) individuals ranging in age from 18 to 82 years of age (M = 23.48 years, SD = 10.88 years). Approximately three-quarters of the participants were Caucasian (74%); others were Asian (7%), bi-/multiracial (4%), African-American (4%), Hispanic (3%), South Asian (2%), Middle Eastern/Arab (2%), and other (4%). Participants reported knowing the person to whom they had disclosed their health condition for an average of 9.5 years (SD = 10.00, range = less than 1 month to 62 years), 9.4 years (SD = 12.23, range = less than 1 month to 63 years) for the person not told. Participants reported the status of these relationships as friend (51%), dating partner/spouse (27%), family member (18%), and other (4%). Measures Variables measured for both the disclosed and the undisclosed scenario included information assessment, relational quality, and efficacy. Variables measured in the undisclosed scenario included anticipated reaction (response and outcome), confidence in response, and likelihood of disclosure. Disclosed scenario variables included retrospective report of anticipated response and disclosure depth. Data were screened for normality and outliers, and there were no transformations necessary or multivariate outliers removed. Confirmatory factor analyses (CFAs) were HEALTH DIAGNOSIS DISCLOSURE DECISIONS conducted on multi-item scales to ensure that they met the criteria of face validity, internal consistency, and parallelism. After confirming the dimensionality of scales,2 composite scores were created by averaging responses to individual items. Due to the limited prior measurement for most constructs in this area, we conducted extensive analyses to ensure adequate psychometrics. Information assessment. Five information assessment subfactors were each measured with five Likert-type items developed based on prior research, with responses ranging from 1 (strongly disagree) to 5 (strongly agree). Items were averaged to form scales with a higher score indicating more of that particular subscale (e.g., more stigma, preparation). Some items were removed due to low/cross loadings to improve model fit statistics and reliability. Stigma was ultimately measured with five items (e.g., “Some people think my health condition is my fault,” M = 3.11, SD = 1.00), prognosis with three items (e.g., “My prognosis is good with my health condition” (R), M = 2.28, SD = .66), symptoms with three items (e.g., “It would be difficult for others to notice my health condition” (R), M = 2.05, SD = 1.02), preparation by three items (e.g., “I had a sense that I was going to be diagnosed with my health condition,” M = 3.04, SD = 1.12), and relevance with four items (e.g., “I worry about spreading my health condition to others,” M = 1.75, SD = 1.08). The best overall information assessment fit was obtained from two correlated latent factors (r = .20). Stigma and progno...
Purchase answer to see full attachment