Development and validation of HIV-related dyadic measures for men who have sex with men

Abstract

The purpose of this study was to develop and assess the psychometric properties of several scales that measure dyadic constructs derived from an integrated model of interdependence theory and communal coping perspectives. The scales are specific to HIV risk, and men who have sex with men (MSM) who are also in an intimate relationship. These scales are newly developed measures of perceived severity of HIV, preferences for sexual health outcomes, outcome and couple efficacy to avoid HIV and communal coping strategies. Scale items were created based on theoretical definitions and results from six focus groups with MSM. Face and content validity of the scales were assessed with a panel of six experts in the field of HIV prevention. Revised scales were subsequently administered to an online sample of 638 MSM, who indicated being in a relationship for at least 3 months. All scales showed adequate reliability and evidence for construct validity was obtained for all scales except for perceived severity of HIV. The results indicate that these dyadic scales are psychometrically sound and can be used in future HIV prevention research and practice with MSM couples.

Men who have sex with men (MSM) remain disproportionately affected by HIV/AIDS in the United States (Sullivan & Wolitski, 2007). Current research indicates that more MSM acquire HIV within the context of a same-sex relationship with a main partner rather than through causal relationships (Sullivan, Salazar, Buchbinder, & Sanchz, 2009). Unfortunately, there has been less success in designing effective interventions for MSM than for other risk groups (Centers for Disease Control and Prevention [CDC], 2009). Thus, new intervention strategies for MSM are warranted and should acknowledge the dynamics of MSM’s intimate relationships and determine how those dynamics may influence HIV risk.

There has been an increasing interest in dyad-directed HIV prevention programs. These types of programs are based on the understanding that ‘relationship factors and dynamics may be important determinants of condom use behavior, including communication with sexual partners, and that any self-efficacious behavior is influenced by the behavioral intentions of both partners (Karney et al., 2010). Dyadic models specific to HIV risk related behavior are increasingly being used to illustrate the reciprocal influence that occurs between individuals within couples and may offer new insights into the factors that regulate sustained behavior change (Albarracin et al., 2005). Although dyadic interventions suggest a promising perspective to combat HIV for MSM couples, few dyadic interventions have been designed for MSM couples. In fact, the few dyadic interventions in the published literature pertain to providing behavioral couples therapy for gay couples with alcohol disorders (Fals-Stewart, O’Farrell, & Lam, 2009) or cognitive behavioral couples therapy for gay men in an AIDS setting (Ussher, 1990).

Couples voluntary HIV counseling and testing (CVCT) is one example of a dyadically delivered approach that has been used as an HIV prevention intervention in Africa for over 20 years. CVCT is considered by CDC to be a “high leverage HIV prevention intervention” in that setting (Painter, 2001). In fact, CVCT has been described as the most effective behavioral intervention to prevent HIV transmission among urban African men and women (Allen et al., 2003; Dunkle et al., 2008). CVCT in Africa entails couples participating in the whole cycle of VCT together: they receive pretest information together, receive pretest counseling and risk ascertainment as a couple and receive the results of HIV testing and posttest counseling as a couple. Concordant negative couples are advised to remain monogamous with each other and to use condoms with any outside partners. Concordant positive couples are advised to use condoms with each other to prevent exposure to different strains of HIV as well as with outside partners to prevent transmission. For HIV discordant couples, where one partner is HIV positive and the other is HIV negative, correct and consistent condom use becomes the primary prevention strategy to protect the HIV negative partner from infection.

Extensive research comparing relationship dynamics and adjustment among gay, lesbian and heterosexual couples has found that gay couples do not differ significantly from heterosexual couples (Holmberg & Blair, 2009; Kurdek, 1994; 2005; 2006; 2007; Means-Christensen, Snyder, & Negy, 2003). Thus, adapting a dyadic intervention such as CVCT, which requires communication and cooperation of the partners, to MSM couples residing in the United States may be a feasible and effective intervention strategy.

Interdependence theory as described by Lewis et al. (2006) (see Figure 1) is an integrative model incorporating both interdependence theory and communal coping perspectives and can be used to guide adaptation of CVCT for MSM. As illustrated, this model suggests that predisposing factors such as perceived threat of HIV, shared mutual preferences for corresponding outcomes (i.e., positive or negative consequences experienced from interacting) and more constructive communication styles may influence MSM couples’ responses to HIV and may be modified by CVCT. Dyadic characteristics such as outcome efficacy (expectation that engaging in safer sex is beneficial) and couple efficacy (confidence in the ability of the couple to engage in safer sex behaviors) may also be affected by CVCT, which in turn could affect communal coping strategies. Thus, changes in key dyadic characteristics may be the causal pathway between CVCT and the adoption of safer behaviors.

Figure 1.

Conceptual model of couples’ interdependence theory (adapted from Lewis et al., 2006)

Source: Adapted from Lewis, M. A., McBride, C. M., Pollak, K. I., Puleo, E., Butterfield, R. M., & Emmons, K. M. (2006), with permission.

Although this model may be useful for adapting CVCT for MSM couples, unfortunately, scales are not available to assess all of these theoretical constructs. Few scales have been developed for MSM couples or used with MSM couples. Some of these scales measure condom attitudes and responsibility (Ross et al., 2007), relationship quality (Hoff et al., 2006; Hunsley, Best, Lefebvre & Vito, 2001), investment in the relationship (i.e., the degree to which gay men persist within their sexual relationship with their main partner) (Mitchell, Harvey, Champeau & Seal, 2011; Rusbult, Martz & Agnew, 1998), HIV-related social support (Darbes & Lewis, 2005), or marital satisfaction (Means-Christensen, Snyder, & Negy, 2003). Although these constructs may fall within “Predisposing Factors” in Lewis’s model, to the best of our knowledge, measures of these remaining constructs have not been developed specifically for HIV-related behaviors and in particular, have not been developed specifically for MSM couples. Accordingly, the purpose of this study was to develop several key dyadic characteristic scale measures from this framework specific to MSM and to assess the psychometric properties of the scales. These measures, if shown to be reliable and valid, can be used in other dyad-level interventions with MSM couples or in observational studies involving MSM couples to further our understanding of the interpersonal processes related to safer sex behavior.

Method

Scale Development

Several dyadic constructs derived from both interdependence theory and communal coping perspectives were developed in the present study: couple member’s perceived severity of HIV, preferences for general lifestyle outcomes, preferences for sexual health outcomes, couple efficacy and outcome efficacy to reduce HIV risk and use of communal coping strategies. Six scales were created to measure these theoretical characteristics.

Scale items were informed by a review of the literature and qualitative data collected with gay men, who self-identified as being in a relationship with another man. Six focus group discussions were conducted in Atlanta, Chicago and Pittsburgh (two in each city). We asked men about their current relationships, how they determined relationship satisfaction, their perceived support for their relationships within their community, the methods used by men in same-sex relationships to cope with health or financial threats, their decision-making process with their partner and their degree of confidence that they could work with their partner towards a shared goal.

Data from the focus groups were used to inform the scale items by providing specific language and concepts about how couples communicate and manage risk. A six-step process called a concept analysis was used to guide the scale development and entailed identifying: the definitions and uses of these constructs, the critical attributes, similar and different constructs, dimensions of the construct, antecedents and consequences of the constructs, and writing a model case (DiIorio, 2005).

Once the scale items were developed, six experts in the field of HIV/AIDS prevention examined the scales to determine face and content validity. We provided each scale along with its theoretical definition. For face validity, we asked each expert to indicate (yes or no) whether the scales appeared to be a valid measure of the construct. For those who responded “no”, we asked them to give us specific feedback on how to revise the scale. For content validity, we asked them to assess each item and determine whether it was “essential”, “useful, but not essential”, or “not necessary” to the performance of the construct. Revisions were made to many of the scale items based on expert feedback. Revised scales were subsequently administered to an online convenience sample of MSM to determine the reliability and validity of the scales.

Measures of Predisposing Factors

Perceived Severity of HIV Scale

This construct involves the perception of the personal, psychosocial and physical consequences of a particular health threat. For the present study, we generated items that referred to perceived severity of HIV infection. We developed a total of 13 items that crossed the three pertinent domains: personal, psychosocial and physical. The 13 items were: 1) Contracting HIV would be very serious to me; 2) The thought of contracting HIV scares me; 3) When I think about contracting HIV I feel nauseous; 4) If I contracted HIV my career would be endangered; 5) When I think about contracting HIV it makes me very anxious; 6) Contracting HIV would jeopardize my relationship with my partner; 7) Contracting HIV would jeopardize my relationships with my family; 8) My views of myself would change dramatically if I contracted HIV; 9) My financial security would be greatly endangered if I contracted HIV; 10) Contracting HIV would be more serious than other diseases; 11) If I contracted HIV, my whole life would change; 12) Thinking about contracting HIV stops me from sleeping; and 13) Thinking about contracting HIV stops me from enjoying sex with my partner. The response format for these items was a 5-point bipolar Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). Thus, this scale had a potential range of 13 to 65 with higher scores indicating more severity of HIV infection. All of the expert reviewers indicated that this scale had face and content validity.

Preferences for General Lifestyle Outcomes Scale

This construct is defined as the degree to which interacting partners agree about the shared or joint outcomes in their relationship. The degree to which they can agree can range from fully cooperative (i.e., set goals together) or noncorrespondent (i.e., in complete conflict). We developed two subscales for this construct: Preferences for General Lifestyle scale and Preferences for Sexual Health Outcomes scale.

For the Preferences for General Lifestyle scale, the stem was initially “Please indicate to what extent do you view the following decisions as your decision versus our decision?” with seven outcomes, which were created based on themes from the focus groups: 1) diet and nutrition, 2) exercising, 3) where to go to dinner, 4) financial decisions, 5) preventive health, 6) social activities and 7) career decision. The response format was a 5-point bipolar Likert-type scale ranging from 1 (My decision) to 5 (Our decision). Thus, the scale had a potential range of 7 to 35 and higher scores indicated more cooperation within the couple.

Four of the six expert reviewers indicated the scale had face validity. The two experts who disagreed provided feedback. The expert reviewers also indicated that three of the seven outcomes were not essential and offered their feedback on those as well. Based on this feedback and discussion among the research team, we revised the stem to: “Please indicate the extent to which you and your partner ‘are on the same page’ (agree with each other) when it comes to the following topics.” We eliminated one item (“where to go to dinner”) and revised the other items: 1) what we like to eat, 2) whether to exercise, 3) money matters, 4) preventive health (e.g., flu shots, general check-up, dental), 5) our social activities, and 6) our sexual relationship. The response format was a 5-point Likert-type scale ranging from 1 (We don’t agree at all) to 5 (We completely agree).

Preferences for Sexual Health Outcomes scale

Similar to the Preferences for General Lifestyle Outcomes scale, this scale assessed the extent the participant and his male partner are concordant when it comes to sexual health topics. We used the same initial stem and 5-point Likert response format used for the Preferences for General Lifestyle Outcomes scale, but the outcomes were related to sexual health: 1) reducing your risk for HIV, 2) using condoms, 3) having sex with someone other than your partner, 4) using condoms when having sex with someone other than your partner, 5) getting an HIV test, and 6) being the top or the bottom.

One expert reviewer indicated the scale did not have face validity; thus, we revised the stem so that it matched with Preferences for General Lifestyle Outcomes scale and revised the following items and added an item: 1) using condoms when we have sex with each other, 2) limiting the number of other sex partners, 3) either of us having sex “outside” our relationship, 4) using condoms when either of us has sex outside our relationship, 5) getting tested regularly for STDs and/or HIV, 6) being the top or bottom when we have sex with each other, and 7) being sexually faithful to each other. Thus, the potential range of this scale was 7 to 35 with higher scores reflecting more concordance.

Communal Coping

Outcome Efficacy to Reduce HIV Threat Scale

Communal coping refers to mutual joint influence in couples and involves two constructs, outcome efficacy and couple efficacy, which in turn influence the use of communal coping strategies (Lewis et al., 2006). Communal coping involves couples working together, communicating, and making decisions together to reduce the health threat.

Outcome efficacy is similar to the construct of outcome expectancies in other value-expectancy theories and is defined as the couple members’ view about the effectiveness of communal coping for initiating and maintaining behavioral change. In this study, the health threat is HIV infection and we created three subscales to capture the full range of outcome efficacy related to these three processes of communal coping. All three subscales used the same items to capture the issues related to safer sex, and all used the same 5-point bipolar Likert-type scale (strongly disagree to strongly agree), but had different stems.

For the first subscale, Joint Effort, we used the stem, “My partner and I believe that ‘working together’ versus on our own is an effective strategy;” for the second subscale, Communication, we used the stem, “Communicating with my partner is an effective strategy for;” and for the third subscale, Planning and Decision-making, we used the stem, “My partner and I making decisions together rather than separately is an effective strategy.”

The items for each of the three subscales were the same as the items used for the Preference for Sexual Health Outcomes scale: 1) to reduce the risk of HIV, 2) for using condoms, 3) when considering whether to have sex with someone other than each other, 4) when considering whether to use condoms when having sex with someone other than each other, 5) when considering getting an HIV test, and 6) when considering who is the top or bottom.

One expert reviewer indicated this scale did not have face validity and provided feedback. Based on the expert feedback and discussions among the research team, we made the following revisions to the scales: we revised the stem in the Joint Effort subscale to, “Working together with my partner is an effective strategy for.” We revised the stem in the Planning and Decision-making subscale to, “Making decisions together is an effective strategy for.” The stem for the Communication subscale was not changed. We also matched the items to be parallel to the Preferences for Sexual Health Outcomes scale. Thus, each subscale had a potential range of 7 to 35, with higher scores indicating higher levels of outcome efficacy to reduce HIV threat.

Couple Efficacy to Reduce HIV Threat Scale

Couple efficacy is defined as a couple’s confidence that together they can engage in communal coping efforts and is similar to Bandura’s self-efficacy construct. We developed three subscales in parallel to Outcome Efficacy to Reduce HIV Threat Scale: Joint Effort, Communication, and Planning and Decision-making. All three subscales used the same six items from the Outcome Efficacy scale to capture the issues related to safer sex; however, the response format used was “not at all confident” to “very confident” and slightly different stems were used.

For the first subscale, Joint Effort, we used the stem, “How confident are you that you and your partner can work together;” for the second subscale, Communication, we used the stem, “How confident are you that you and your partner can communicate about;” and for the third subscale, Planning and Decision-making, we used the stem, “How confident are you that you and your partner can make decisions together.” Based on expert feedback, we did not revise the stems to these subscales; however, the items were revised to match the revised items in the Outcome Efficacy scales. Thus, each subscale had a potential range of 7 to 35, with higher scores indicating higher levels of outcome efficacy to reduce HIV threat.

Communal Coping to Reduce HIV Threat Scale

As previously mentioned, this construct refers to the “utilization of strategies, which are characterized as communal in nature such as couple communication about behavior change, joint decision-making and planning regarding the behavior or working together to engage in the behavior” (Lewis et al., 2006, p.1374). Thus, the use of communal coping is a construct that measures how often people engage in joint efforts to manage the health threat.

We developed three subscales in parallel to the Outcome Efficacy Scale and the Couple Efficacy Scale: Joint Effort, Communication, and Planning and Decision-making. All three subscales used the same six items from these other scales to capture the managing of reducing the health threat of HIV, but the stems were different. For the first subscale, Joint Effort, we used the stem, “To what extent do you and your partner work together;” for the second subscale, Communication, we used the stem, “To what extent do you and your partner communicate about;” and for the third subscale, Planning and Decision-making, we used the stem, “To what extent do you and your partner make decisions together about.” Based on expert feedback, we did not revise the stems to these subscales; however, the items were revised to match the revised items in the Preference for Sexual Health Outcomes scale, the Outcome Efficacy Scale and the Couple Efficacy scale. Thus, for each of these subscales, the potential range was 7 to 35 with higher scores indicating greater frequency in engaging in these communal coping strategies.

Psychometric Assessment of Scales

Participants

To assess the psychometric properties of the revised scales, we conducted an online study. Recruitment was conducted through Facebook banner advertising. When Facebook members access their personal Facebook home page, they are routinely shown banner advertisements customized to their self-reported demographic characteristics stored in their profiles (i.e., ‘targeted advertising’). During a five-day recruitment period (May 18, 2010- May 23, 2010) ads were displayed to Facebook members whose profiles matched our study eligibility criteria: males living in the United States, at least 18 years old, “interested in men” and relationship status was “in a relationship, married or engaged.” The banner advertisements were displayed at random times of day. All Facebook members whose profiles fit the criteria had an equal chance of being shown the banner advertisements.

Web analytics regarding the ads showed 2.4 million impressions with 3,155 clicks. Prospective participants who clicked on the ads (81.5%) were directed to the survey. Once participants entered the survey, they were screened for eligibility. If prospective participants met eligibility requirements, they were then directed to a consent form. Once electronic consent was provided, they were directed to the questionnaire. Those who did not qualify based upon eligibility requirements were taken to another screen that thanked them for their interest and provided them with links about HIV prevention for MSM. The Emory University Institutional Review Board (IRB) approved the study protocol prior to implementation.

We recruited 1,366 men through this method; 634 completed the survey in its entirety. The mean age of the sample was 32.77 years (SD = 11.54, Range = 18–75). A majority (87%) of men were White, 1.6% were African American, 5.6% were multi-racial and 5.7% were of other racial backgrounds; 9.6% reported themselves to be Hispanic or Latino. The sample was relatively educated: 41% had at least a college degree; 43% had some college or two-year degree; 15% graduated high school; only 1% had less than high school. Ninety-nine percent self-identified as gay or homosexual.

Due to screening, 100% of the men reported that they currently had a main male sex partner (i.e., “someone you feel committed to above all others, someone you would call a boyfriend, life partner or significant other or husband). The length of this relationship ranged from 3–4 months (6.3%), 4–6 months (6.9%) or more than 6 months (86.8%). Regarding their agreements, only 8.8% indicated that they did not have an agreement, whereas 62.2% (n=395) agreed not to have sex outside the relationship, 25.8% (n=164) indicated that they could have sex outside the relationship, but with conditions, and 3.1% (n=20) agreed they could have sex outside the relationship, with no conditions. Data were missing for .5% (n=3) participants. Regarding self-report HIV status, 86% were HIV negative, 10% HIV positive, and 4% reported their status was unknown.

Online Survey

The online survey was hosted by an online survey service provider called Survey Gizmo. No personally identifying information including computer IP addresses were collected from participants. All questions of a sensitive nature included the option not to answer. Data from the online surveys were stored on a secure server at the Survey Gizmo servers in Boulder, Colorado and were monitored for enrollment through a secure access portal. Access credentials were available only to the PI, study coordinator, and to Survey Gizmo managers.

The 30-minute online survey consisted of the newly developed scales and included several domains of personal information including: demographics (i.e., age, race/ethnicity, education level, sexual orientation), relationship status, relationship length, whether or not they had an agreement with their partner, unprotected anal intercourse with their partner, unprotected anal intercourse with someone other than their partner and self-reported HIV infection. Several additional scales were used to assess validity of the scales.

Validation Measures

Relationship Satisfaction

Hendrick’s (1988) seven-item Relationship Satisfaction Scale was used to measure relationship characteristics. Items included: “How much do you love your partner” with the response 1 (not very much) to 5 (very much); and “In general, how satisfied are you with your relationship” with the response 1 (very unsatisfied) to 5 (very satisfied). Items were averaged with higher scores representing more relational satisfaction. In the present study, Cronbach’s alpha (α) was .91.

Degree of happiness in relationship

This construct was measured with one item. We asked men to “please indicate the degree of happiness, all things considered, of your relationship” with a 7-point bipolar response scale ranging from 1 (Extremely unhappy) to 7 (Perfect).

Communal Confidence

This construct was measured by three items that assessed participants’ level of agreement with their partner on the following broad topics related to their relationship: “Philosophy of life”, “Aims, goals, and things believed important”, and “the amount of time spent together.” A 7-point bipolar response scale was used that ranged from 1 (Always disagree) to 7 (Always agree). Cronbach’s alpha (α) was .86.

Conflict Style

To determine how respondents typically handled conflict in their relationships, we administered the 30-item Conflict Style Inventory (CSI; Levinger & Pietromonaco, 1989). The CSI includes subscales for collaborating (“I share the problem with my partner so that we can work it through together”), compromising (“I try to find a compromise solution”), and contending (“I try to get my way whenever I can”). Items for the contending subscale were recoded and all three subscales were combined into a single scale as an indicator of constructive conflict resolution style. Higher scores indicate constructive strategies (i.e., more collaborating, more compromising, and less contending). Cronbach’s alpha (α) was .78.

Communication Style

The style of communication was measured with the Communication Patterns Questionnaire Constructive Communication subscale (Heavey, Larson, Zumtogel, & Christensen, 1996). This 7-item subscale assesses the overall quality of a couple’s communication by considering both positive and negative behaviors. The seven items assess the frequency of both constructive and destructive communication behaviors; however, the destructive behaviors are subtracted from the constructive behaviors. Cronbach’s alpha (α) was .81.

Agreements

We asked participants “What is the current agreement that you and your partner have about sexual encounters outside of the relationship? (Please choose only one.)” Responses were: 1) Both of us cannot have any sex with an outside partner; 2) We can have sex with outside partners, without any conditions or restrictions; 3) We can have sex with outside partners, but with some conditions or restrictions; or 4) We do not have an agreement. This item was developed by Neilands, Chakravarty, Darbes, Beougher, and Hoff (2010). For the present study, we dichotomized the item so that 0 = No Agreement and 1= Agreement.

Broke their agreement

We asked participants, “Have you ever broken your current agreement about sex with outside partners?” Responses were yes/no (Neilands et al., 2010).

Attitudes toward agreements

This 13-item scale assessed attitudes about the participants’ current agreement about sex outside the relationship (Neilands et al., 2010). Sample items were, “How committed are you to having your current agreement?” “How dedicated are you to your current agreement?” “How obligated do you feel toward your current agreement?” with a 5-point bipolar response scale ranging from 1 (not at all) to 5 (very much). Higher scores indicate a more positive attitude toward having an agreement. Cronbach’s alpha (α) was .93.

HIV-specific Social Support Scale

This measure was assessed with a modified version of Cutrona and Russell’s (1987) Social Provisions Scale. Darbes and Lewis (2005) adapted each of the 24 items to reflect perceived partner support for HIV-preventive behavior. They created two versions: a HIV-negative version and a HIV-positive version. A sample item from the HIV-negative version was, “I can depend on my partner to help me practice safer sex if I need help.” A sample item from the HIV-positive version was, “If I experienced a complication related to being HIV positive or having AIDS, I could count on my partner to help me handle the situation.” Response options ranged from 1 (strongly disagree) to 4 (strongly agree) and higher scores indicate higher perceived partner support with regard to HIV prevention. Both versions were administered in the current study and showed adequate reliability. Cronbach’s alpha (α) was .81 for the HIV-negative version (n=504) and .85 for the HIV-positive version (n=58). Data were missing for some participants (n=76) on this measure.

Perceived Personal Gay-Related Stigma Scale

This 6-item scale was created by the second author for previous studies and was used in the present study for validation of the dyadic scales. The scale assesses participants’ personal perceptions of how much stigma there is for being gay. The stem used was, “Compared to heterosexual couples you know …‥” with the following items: 1) how much respect do you and your partner get? 2) how much do people see you as a real couple? 3) how much do you and your partner, as a couple, face harassment from others? 4) how much is your relationship with your partner respected by your employer? 5) how much is your relationship with your partner recognized by your local or state government? And 6) how much do health care providers respect your relationship with your partner? Response format was a 5-point bipolar Likert scale ranging from 1 (much less) to 5 (much more). Cronbach’s alpha (α) was .73.

Perceptions of Local Stigma Scale

This 7-item scale measures perceptions of stigma related to being gay in participants’ communities (Herek & Glunt, 1995). The stem used was, “Most people in my city or town …” with the following items: 1) believe that a gay/bisexual man is just as trustworthy as the average heterosexual citizen; 2) feel that homosexuality is a sign of personal failure; 3) would not hire a gay/bisexual man to take care of their children; 4) think less of a person who is gay/bisexual; 5) would treat a gay/bisexual man just as they would treat anyone; 6) will willingly accept a gay/bisexual man as a close friend; and 7) will hire a gay/bisexual man if he is qualified for the job. Response options ranged from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicate more perceived stigma. Cronbach’s alpha (α) was .88.

Unprotected anal intercourse with main partner

We asked participants “whether they had unprotected anal intercourse with their partner in the past 12 months? This means that you or your partner did not use a condom at all during anal sex.” Response option was yes or no.

Unprotected anal intercourse with casual partner

We asked participants “whether they had unprotected anal sex with a man other than their partner in the past 12 months, during the time while you were having sex with your partner?” Response option was yes or no.

HIV Status

We asked participants to indicate whether they were: HIV positive, HIV negative, or didn’t know their status.

Data Analysis

Reliability of the scales was assessed using two procedures: internal consistency and the split-half method. We calculated Cronbach’s alpha (α) to assess the internal consistency of the items and then we randomly divided the scale items into two sets of scores and calculated the Spearman-Brown correlation between these two total scores. Adequate reliability was indicated if Cronbach’s alpha and the split-half correlation coefficient were > .70.

As there is no single measure of construct validity (rather, construct validity is based on the accumulation of knowledge about the test and its relationship to other tests and behaviors), we approximated the construct validity of the newly-developed scales in several ways. First, a factor analysis was conducted on two of the scales (Perceived Severity of HIV Infection and Preferences for Outcomes) to determine whether the loadings of measured (indicator) variables conformed to what was expected. Indicator variables were the newly developed items based on the theoretical framework used. The factor analysis was used to see if the indicator variables loaded as predicted on the expected number of factors. A two factor solution for Preferences for Outcomes was extracted given there were two subscales. In addition, the scales were assessed in terms of their convergent validity, meaning correlations were calculated between the newly developed scale scores and the other theoretical constructs and behaviors as specified by the theory (e.g., for couple efficacy: communication style and relationship satisfaction). Divergent validity was assessed by calculating correlations between the new scales and education level, with which the scales should not be related. Finally, scales were correlated with sexual risk behaviors and HIV status to determine criterion-related validity.

Results

Reliability of the Scales

Results for the internal reliability of the developed scales are presented in Table 1. As indicated, all of the scales showed adequate reliability except for the Communal Coping Joint Effort subscale.

Table 1.

Reliability and Descriptive Statistics for Dyadic Scales for Men Who Have Sex with Men Couples

Scale Measure	No. of Items	α	Split-Half Spearman–Brown	Range	M (SD)
Measures of predisposing factors
Perceived Severity of HIV Infection scale	13	.83	.81	18–65	43.00 (9.62)
Preferences for Outcomes scale
General lifestyle	7	.76	.68	7–35	27.39 (4.72)
Sexual health	7	.82	.83	7–35	31.82 (4.60)
Measures of communal coping processes
Outcome Efficacy to Reduce HIV Threat scale
Joint effort	7	.68	.67	7–35	31.30 (5.83)
Communication	7	.86	.83	7–35	30.92 (5.86)
Planning and decision-making	7	.87	.84	7–35	31.15 (5.86)
Couple Efficacy to Reduce HIV Threat scale
Joint effort	7	.91	.90	7–35	31.30 (5.83)
Communication	7	.89	.89	7–35	32.50 (5.03)
Planning and decision-making	7	.91	.90	7–35	32.54 (5.08)
Communal Coping to Reduce HIV Threat scale
Joint effort	7	.68	.67	11–35	28.56 (5.66)
Communication	7	.86	.83	7–35	27.37 (7.78)
Planning and decision-making	7	.87	.84	7–35	29.85 (7.12)|

Descriptive Statistics

Descriptive statistics for each of the developed scales are presented in Table 1.

Factor Analyses

Perceived Severity of HIV Scale

The factor structure of the scale was determined using principal-components analysis with oblique rotation using a Promax solution. The solution for the Perceived Severity of HIV Scale yielded three factors with eigenvalues > 1, accounting for 56% of the variance. Those factors were named Personal Consequences (Factor 1), Psychosocial Consequences (Factor 2), and Physical Consequences (Factor 3). An examination of the pattern matrix confirmed a three-factor solution. The items and factor loadings for the factors are presented in Table 2. All items had factor loadings > .40.

Table 2.

Factor Analysis of Perceived Threat of HIV Infection Scale

Item	Factor Loading
Factor 1: Personal Consequences (34% of variance)
Contracting HIV would be very serious to me	0.72
The thought of contracting HIV scares me	0.88
When I think about contracting HIV I feel nauseous	0.69
When I think about contracting HIV it makes me very anxious	0.71
Contracting HIV would be more serious than other diseases	0.47
If I contracted HIV my whole life would change	0.49
Factor 2: Psychosocial Consequences (11% of the variance)
If I contracted HIV my career would be endangered	0.65
Contracting HIV would jeopardize my relationship with my partner	0.67
Contracting HIV would jeopardize my relationships with my family	0.65
My views of myself would change dramatically if I contracted HIV	0.54
My financial security would be greatly endangered if I contracted HIV	0.75
Factor 3: Physical Consequences (9.8% of the variance)
Thinking about contracting HIV stops me from sleeping	0.78
Thinking about HIV stops me from enjoying sex with my partner	0.83

Preferences for Outcomes Scale

The factor structure of the scale was determined using principal components analysis with oblique rotation using a Promax solution. We forced a two-factor solution because this scale comprised two subscales, Preferences for General Lifestyle Outcomes scale (Factor 1) and Preferences for Sexual Health Outcomes scale (Factor 2). The two factors accounted for 47% of the variance. This confirmatory analysis showed that the seven items associated with the Preferences for General Lifestyle Outcomes scale all loaded onto the first factor (loadings > .40) and the seven items associated with the Preferences for Sexual Health Outcomes scale all loaded on the second factor (loadings > .40). The pattern matrix confirmed a two-factor solution. The items and factor loadings are presented in Table 3.

Table 3.

Factor Analysis of Preferences for Outcomes Scales: Preferences for General Lifestyle Outcomes Scale and Preferences for Sexual Health Outcomes Scale

Item	Factor Loading
Factor 1: General Lifestyle Outcomes scale (13% of the variance)
What we like to eat	0.44
Exercise	0.65
Finances or money matters	0.72
Preventive health	0.67
Our social activities	0.69
Our sexual relationship	0.62
How we resolve our conflicts	0.65
Factor 2: Sexual Health Outcomes scale (34% of the variance)
Using condoms when you have sex with each other	0.46
Limiting the number of other sex partners	0.87
Deciding whether to have sex outside the relationship	0.84
Using condoms when having sex outside the relationship	0.71
Getting tested regularly for STDs and or HIV	0.55
Deciding who will be the top and who will be the bottom
when you have sex with each other	0.54
Being sexually faithful to each other	0.82

Construct validity

Perceived Severity of HIV Scale

The correlations between this scale and the validation measures are presented in Table 4. The results indicate small effects with only a few measures. Moreover, the scale was not related to any of the other dyadic characteristic scales nor was it related to sexual risk behaviors.

Table 4.

Correlations among Perceived Severity of HIV and Preferences for Outcomes Scales (Preferences for General Lifestyle Outcomes Scale and Preferences for Sexual Health Outcomes Scale) and Validation Measures

Validation Measure	Perceived Severity of HIV	Preferences-General Lifestyle	Preferences-Sexual Health
Perceived Severity of HIV	---	−.08	−.03
Preferences—General	−.08	---	.46***
Preferences—Sexual	−.03	.46***	---
Outcome efficacy—joint	−.01	.21***	.45***
Outcome efficacy—comm	−.01	.25***	.52***
Outcome efficacy—plan	−.01	.21***	.49***
Couple efficacy—joint	−.05	.41***	.70***
Couple efficacy—comm	−.08	.36***	.65***
Couple efficacy—plan	−.05	.40***	.67***
Communal coping—joint	.00	.16***	.42***
Communal coping--comm	−.02	.26***	.46***
Communal coping—plan	−.04	.24***	.51***
Relationship Satisfaction	−.08	.64***	.50***
Relationship Happiness	−.05	.54***	.42***
Communal Confidence	−.06	.24***	.16***
Conflict Style	.13**	.04	−.03
Communication Style	−.11*	.55***	.43***
Agreement	−.02	.18***	.29***
Broke agreement	.03	−.19**	−.26***
Attitudes Agreements	.04	.30***	.41***
HIV-Social Support (−)	.04	.39***	.47***
HIV-Social Support (+)	---	.45***	.53***
Personal stigma	.14**	−.17***	.03
Local stigma	.14**	−.08	−.02
UAI-main partner	.03	.05	−.00
UAI-casual	−.02	.06	.25***
HIV status	.00	−.01	−.10*
Education level	−.11**	−.02	−.05
Age	−.33**	.06	−.05

Note: Values in boldface type denote >.40. Joint=joint effort; Comm.=communication; Plan=planning and decision-making; UAI=unprotected anal intercourse.

^*p<05

^**p<01

^***p<.001

Preferences for General Lifestyle Outcomes scale

Correlations for this scale are presented in Table 4. Convergent validity was supported by positive correlation with the dyadic scales and other relationship markers. The scale was not related to sexual risk behavior or education level suggesting discriminant validity.

Preferences for Sexual Health Outcomes scale

Correlations for this scale are presented in Table 4. This scale was related positively to the newly developed dyadic scales, other relationship markers and with HIV-related social support suggesting convergent validity. Criterion-related validity was suggested by the negative significant correlation with unprotected anal intercourse with a casual partner and with being HIV negative. It was not related to education level.

Communal Coping Process

The correlations between three subscales of Outcome Efficacy to Reduce HIV Threat and the other corresponding subscale measures of Couple Efficacy to Reduce HIV Threat and Communal Coping Strategies are presented in Table 5. As shown, the subscales were all positively inter-correlated, indicating convergent validity.

Table 5.

Correlations among Dyadic Characteristic Scale Measures of the Communal Coping Process.

Scale	Outcome Efficacy: Joint	Outcome Efficacy: Comm	Outcome Efficacy: Plan	Couple Efficacy: Joint	Couple Efficacy: Comm	Couple Efficacy: Plan	Communal Coping: Joint	Communal Coping: Comm
Outcome Efficacy:
Comm	.90***
Outcome Efficacy:
Plan	.92***	.94***
Couple Efficacy:
Joint	.56***	.61***	.59***
Couple Efficacy:
Comm	.47***	.54***	.52***	.85***
Couple Efficacy:
Plan	.52***	.57***	.56***	.91***	.94***
Communal Coping:
Joint	.42***	.46***	.44***	.43***	.38***	.40***
Communal Coping:
Comm	.45***	.48***	.46***	.47***	.43***	.43***	.60***
Communal Coping:
Plan	.52***	.53***	.51***	.53***	.50***	.50***	.61***	.75***

Note: Outcome Efficacy=Outcome Efficacy to Reduce HIV Threat scale; Couple Efficacy=Couple Efficacy to Reduce HIV Threat scale; Communal Coping=Communal Coping to Reduce HIV Threat scale; Joint=joint effort; Comm.=communication; Plan=planning and decision-making.

^***p<.001

Outcome Efficacy to Reduce HIV Threat Scale

Correlations with the validation measures are presented in Table 6. The three subscales of this measure were all positively correlated with other relationship markers and HIV-related social support. All three subscales were also significantly related to unprotected anal intercourse with a casual partner in the negative direction. Also, they were related negatively to breaking an agreement and to HIV status, suggesting criterion-related validity. None of the subscales were related to education level.

Table 6.

Correlations among Outcome Efficacy Subscales and Validation Measures

Validation Measure	Outcome efficacy-Joint	Outcome efficacy-Comm	Outcome efficacy Plan
Relationship Satisfaction	.26***	.27***	.25***
Relationship Happiness	.40***	.44***	.43***
Communal Confidence	.07	.09*	.08
Conflict Style	−.04	−.06	−.06
Communication Style	.20***	.23***	.22***
Agreement	.19***	.18***	.18***
Broke agreement	−.23***	−.17***	−.19***
Attitudes Agreements	.14**	.23***	.22***
HIV-Social Support (−)	.31***	.32***	.30***
HIV-Social Support (+)a	.36**	.31*	.31*
Personal stigma	.05	.07	.08
Local stigma	.07	.04	.06
UAI-main partner	−.03	−.01	−.02
UAI-casual	−.24***	−.19***	−.23***
HIV status	−.10*	−.12**	−.15***
Education level	.05	.04	.05

Note: Values in boldface type denote >.40. Outcome Efficacy=Outcome Efficacy to Reduce HIV Threat scale; Joint=joint effort; Comm.=communication; Plan=planning and decision-making; UAI=unprotected anal intercourse.

^an=56

^*p<.05

^**p<.01

^***p<.001

Couple Efficacy to Reduce HIV Threat Scale

Correlations with the validation measures are presented in Table 7. The three subscales of this measure were all significantly correlated in a positive direction with additional measures of relationship characteristics, indicating convergent validity. The subscales were negatively related with having had unprotected anal intercourse with a casual partner, HIV status, and breaking an agreement suggesting strong criterion-related validity. Finally, the three subscales were all positively related to HIV-related social support. None were related to education level.

Table 7.

Correlations among Couple Efficacy Subscales and Validation Measures

Validation Measure	Couple efficacy-Joint	Couple efficacy-Comm	Couple efficacy-Plan
Relationship Satisfaction	.48***	.46***	.50***
Relationship Happiness	.40***	.44***	.43***
Communal Confidence	.17***	.22***	.21***
Conflict Style	−.05	−.14**	−.08
Communication Style	.38***	.37***	.40***
Agreement	.30***	.24***	.24***
Broke agreement	−.23***	−.17***	−.19***
Attitudes Agreements	.34***	.37***	.35***
HIV-Social Support (−)	.46***	.40***	.42***
HIV-Social Support (+)a	.59***	.47***	.55***
Personal stigma	.02	.06	.05
Local stigma	.01	.00	.00
UAI-main partner	−.03	−.01	−.02
UAI-casual	−.24***	−.19***	−.23***
HIV status	−.15***	−.10*	−.12**
Education level	−.03	−.07	−.06

Note: Values in boldface type denote >.40. Couple Efficacy= Couple Efficacy to Reduce HIV Threat scale; Joint=joint effort; Comm.=communication; Plan=planning and decision-making; UAI=unprotected anal intercourse.

^an=56

^*p<.05

^**p<.01

^***p<.001

Communal Coping Strategies Scale

Correlations among the three subscales and validation measures are presented in Table 8. The subscales were all significantly related to additional relationship characteristics. The Joint Effort scale was negatively correlated with having had unprotected anal intercourse with their main partner; however, the other two subscales were not. Joint Effort and Planning and Decision-making were negatively correlated with having had unprotected anal intercourse with a casual partner and with having kept their agreement suggesting strong criterion-related validity, and positively related to perceived personal gay-related stigma. All three subscales were positively related to HIV-related social support, and none were related to education level.

Table 8.

Correlations among Communal Coping Subscales and Validation Measures

Validation Measure	Communal coping-Joint	Communal coping-Comm	Communal coping-Plan
Relationship Satisfaction	.23***	.28***	.29***
Relationship Happiness	.20***	.25***	.24***
Communal Confidence	.07	.07	.06
Conflict Style	−.04	.00	−.04
Communication Style	.20***	.22***	.24***
Agreement	.16***	.22***	.21***
Broke agreement	−.11**	−.12**	−.14**
Attitudes Agreements	.16***	.19***	.22***
HIV-Social Support (−)	.23***	.32***	.26***
HIV-Social Support (+)a	.45**	.43**	.40**
Personal stigma	.13**	.06	.12**
Local stigma	−.02	.04	.06
UAI-main partner	−.17***	−.07	−.00
UAI-casual	−.18***	−.07	−.10*
HIV status	−.09*	−.08*	−.09*
Education level	.04	−.01	−.03

Note: Values in boldface type denote >.40. Communal coping= Communal Coping to Reduce HIV Threat scale; Joint=joint effort; Comm.=communication; Plan=planning and decision-making; UAI=unprotected anal intercourse.

^an=56

^*p<.05

^**p<.01

^***p<.001

Discussion

Measurement is necessary for understanding health behaviors through research and programs. A fundamental challenge involves measuring intangibles and theoretical constructs. Good measurement is critical to the advancement of HIV prevention as it determines what the target of our efforts will be and whether our efforts are truly having an impact. With the advent of dyadic-level HIV prevention interventions that target MSM couples, measurement of key dyadic characteristics specific to MSM is essential. Thus, this article describes the development of several scales that are theoretically driven to measure dyadic characteristics of MSM in relationships.

We designed scales to measure two “predisposing factors” drawn from interdependence theory—perceptions of the severity of HIV infection and preferences for general lifestyle and sexual health outcomes and three scales related to the communal coping process. All scales were developed from a systematic process, included an expert review and then assessed psychometric properties with a sample of MSM recruited from the Internet. Although our sample was relatively homogenous, our methodology nonetheless illustrates how data can be collected from hard to reach population groups with minimal effort.

Our results suggest that all of the scales are reliable with the target population. The factor analysis of Perceived Severity of HIV Scale indicates three factors constitute the scale. Although there are other scales in the published literature that measure this construct, the items for this scale cross three relevant domains and were informed based on qualitative responses from MSM, who had been in relationships with a main partner for at least 3 months. Another advantage of this scale is that it consists of only 13 items, which makes it desirable from an administrative perspective.

Although the factor analysis demonstrates that Perceived Severity of HIV Scale has construct validity, the scale correlates with only a few other variables. Moreover, the scale was not related to any of the communal coping constructs; however, based on the integrative model in Figure 1, perceived severity of the health threat “predisposes” people to view the health threat as significant in terms of their relationship (relationship-centered) rather then themselves (self-centered). This relationship-centered focus is deemed transformation of motivation and in turn influences communal coping processes. Thus, perceived severity of the health threat is related indirectly to communal coping processes according to this integrative model and thus, significant direct associations should not be expected.

Nonetheless, we did find some significant associations with the Perceived Severity of HIV Scale. The scale is associated with higher levels of perceived local and personal gay-related stigma—two constructs that could act as predisposing factors. These results provide evidence for the influence of the social context on MSM couples and their perceptions of a health threat such as HIV. It stands to reason that MSM who perceive more stigma related to being gay would also perceive that HIV is more severe than men who do not have such perceptions. In terms of relationship markers, which are considered predisposing factors in the model, the results were mixed: the scale is positively related to a constructive conflict style, but negatively related to a positive communication pattern. In addition, the scale is not associated with any of the sexual risk behaviors, which according to the model suggests that predisposing factors such as perceived severity may be related directly to behavior. Together, these results suggest more research may be needed to further validate this scale.

The factor analysis results for the Preferences for Outcomes Scale suggest construct validity. We forced a two-factor solution to determine whether there were indeed two distinct underlying dimensions as theorized. The results indicate that the two-factor solution is consistent with our understanding of the nature of this construct. In addition, more evidence for construct validity was found through the significant correlations between the two subscales and other positive relationship markers.

According to interdependence theory, a couple’s preferences for outcomes indicate the degree of conflict within the relationship and can serve as either a motivator of or a barrier to behavior change (Lewis et al., 2006). Thus, for MSM who are in a relationship, whether both are HIV negative or serodiscordant, understanding their degree of correspondence regarding reducing the threat of HIV infection provides a better understanding of their motivations to engage in mutually protective behaviors. As researchers in HIV prevention, we need to first understand this important dynamic in MSM couples so that effective dyad-level HIV prevention interventions can be designed that consider this characteristic and the role it plays in motivating behavior change. In fact, this is especially relevant given our results. We found that MSM, who perceive higher levels of agreement with their partner regarding sexual health outcomes, report higher outcome and couple efficacy, engage in communal coping strategies more frequently, break their agreements less often, have unprotected anal intercourse with a casual partner less often and are less likely to be HIV positive than MSM who perceive lower levels of agreement. These findings suggest that working with MSM couples to get them on the “same page” and to develop their abilities to work together and reduce conflicts so as to avoid HIV infection might be an effective intervention strategy.

All three of our communal coping measures demonstrated construct validity. Performing a factor analysis was not appropriate for these three scales due to the fact that all of the respective subscale items are the same; only the stems differ. All three subscales (Joint Effort, Communication, and Planning and Decision-making) show adequate reliability. Moreover, all three demonstrate construct validity based on consistent and significant correlations with each other and with a host of relational markers and social variables including risk behavior. All subscales were negatively related to having unprotected anal intercourse with a casual partner. Only one scale, the Communal Coping Strategies Joint Effort subscale, was related negatively to having unprotected anal intercourse with a main partner. This is an important finding as other research has shown that sexual risk behavior with a main versus a casual partner is a significant contributor to HIV infection (Sullivan et al., 2009). Thus, engaging MSM couples to utilize strategies where they work together to reduce their HIV risk (e.g., CVCT) may be an effective approach versus working with MSM individually. This is plausible considering that these three aspects of communal coping are related directly to risk reducing behaviors in addition to HIV-negative status. Thus, a challenge of future research with MSM couples is to identify how to implement communal coping within the context of HIV prevention.

The results of this study may have implications for those who provide direct services to MSM (e.g., psychologists, therapists, health care providers) and for public health practitioners who design preventive interventions. We acknowledge that sexual health-enhancing behavior is determined by many factors; however, the results suggest that one particularly salient factor involves the joint effort of the couple. Focusing on relationship dynamics such as those described in this integrative model when providing health services or intervening with MSM may be an effective approach to behavior change. Decades of literature surrounding effective therapeutic approaches for heterosexual couples to improve their relationships suggest that MSM couples could also benefit from similar approaches. Thus, an approach such as CVCT for MSM couples may be feasible and effective.

Although the dynamics for MSM couples may be similar to heterosexual couples (Holmberg & Blair, 2009; Kurdek, 2004; 2005; 2007), there are some differences. For example, we found that several of these key dyadic constructs correlate with perceptions of gay-related stigma, suggesting the influence of social factors on MSM romantic relationships. More research is needed with MSM couples to identify other factors as possible mediators of these relationships and which can help explain the mechanisms through which social factors such as gay-related stigma affect these dyadic constructs.

As with any study, this study is limited in several ways. First, we used an online convenience sample of MSM, who were mostly White and highly educated. These results may not generalize to Black MSM or other subgroups of MSM. Future research with additional samples of MSM is needed to determine if these measures are generalizable to other racial/ethnically diverse populations of MSM or whether the psychometric properties are stable across different subpopulations and different samples.

Another limitation is that we did not recruit both members of the MSM dyad. We acknowledge the interdependence that exists between dyad members especially when assessing dyadic factors. Mutual dyad influences in the present study would include “partner effects”, in that the characteristics of one person affect the outcomes of the other. Given the unit of analysis was the individual versus the dyad, we cannot say whether partner effects would have modified the results. Thus, these results should be viewed as one part of the psychometric process. Future research should assess these scales using the dyad as the unit of analysis.

In addition, given we did not provide a cash incentive and we used passive recruitment, the sample may be biased toward MSM who are high in altruism or differ in some other unmeasured factor. Also, as with most sexual risk behavior research, the data were self-reported potentially resulting in social desirability bias; however, completing the survey online and anonymously may have attenuated this type of bias (Catania, Gibson, Chitwood, & Coates, 1990). Last, although we attempted to be systematic and empirical in our development of these scales, and we elicited expert feedback for face and content validity, there may be other aspects of these constructs that are relevant but not measured. Despite these limitations, the data reported here suggest that these psychometrically sound and theoretically derived scales may be useful for research and practice with MSM couples.