Reactions of Heterosexual African-American Men to Women’s Condom Negotiation Strategies

Abstract

This study describes responses of 172 single heterosexual African American men, ages 18–35, to condom negotiation attempts. Strategies used included reward, coercive, legitimate, expert, referent, and informational strategies, based on Raven’s (1992) influence model. The purpose was: 1) to identify strategies influencing participant acquiescence to request, and 2) to identify predictors of participant compliance/refusal to comply with negotiation attempts. Participants viewed six videotape segments showing an actress, portrayed in silhouette, speaking to the viewer as a ‘steady partner’. After each segment, participants completed measures of: request compliance, positive and negative affect, and attributions concerning the model and themselves. No significant differences were found in men’s ratings across all vignettes. However, differences in response existed across subgroups of individuals, suggesting that while the strategy used had little impact on participant response, the act of suggesting condom use produced responses that differed across participant subgroups. Subgroups differed on levels of AIDS risk knowledge, STD history, and experience with sexual coercion. Also, the “least-willing-to-use” subgroup was highest in anger/rejection and least likely to make attributions of caring for partner. Effective negotiation of condom use with a male sexual partner may not be determined as much by specific strategy used as by partner characteristics.

Interventions to assist persons in reducing risk for HIV infection have drawn upon a number of behavioral science theories, such as the Theory of Planned Behavior (Ajzen, 1991); social cognitive and self-efficacy theory (Bandura, 1988); the Information-Motivation-Behavior model (Fisher & Fisher, 1992); and the AIDS Risk Reduction Model (Catania, Kegeles, & Coates, 1990). Although these theories approach the issue of HIV risk behavior changes from different perspectives, they all posit that behavioral skills–including the ability to communicate condom use requests with sexual partners--are important determinants of success in risk reduction behavior change.

State-of-the-art skills building interventions aimed at promoting HIV risk reduction behavior change have often emphasized the acquisition of sexual assertiveness, communication, and negotiation skills to be used with sexual partners. Methods often used to teach these skills include the modeling, role-playing, and practice of negotiation tactics, and emphasis of communication with sexual partners as an important precursor of condom use. Negotiation skills training has been incorporated into HIV risk reduction intervention programs for many at-risk populations including adolescents (DiNoia & Schinke, 2007; Fisher, Fisher, Bryan & Misovich, 2002; Hovell et al., 2001), inner-city men (Calsyn et al., 2009; Kalichman, Cherry, & Browne-Sperling, 1999) and women (Carey et al., 2000; DiClemente & Wingood, 1995), gay men (Bowen et al., 2008; Kelly et al., 1997), and adults with serious mental illness (Otto-Salaj, Kelly, Hoffmann, Stevenson, & Kalichman, 2001; Sikkema, Meade, Doughty-Berry, Zimmerman, Kloos, & Snow, 2007; Susser et al., 1998). In several meta-analyses of HIV risk reduction interventions, negotiation skills training has been cited as a common component of many HIV interventions that have been deemed efficacious in changing HIV risk behaviors (e.g, meta-analyses by Albarracin, Gillette, Earl, Glasman, Durantini & Ho, 2005; Crepaz et al., 2006; Darbes, Crepaz, Lyles, Kennedy & Rutherford, 2008; Herbst, Kay, Passin, Lyles, Crepaz, & Marin, 2007). Further, a synthesis of the results of HIV intervention meta-analyses suggests that increased condom use in particular may account for the greatest effect sizes in HIV prevention intervention results (Noar, 2008). In support of this finding, results of a meta-analysis examining the impact of sexual communication on condom use (Noar, Carlyle & Cole, 2006) showed that communications about condom use had the largest effect size of all the variables they examined, indicating that discussion of condom use in some fashion may play a key role in facilitating use.

There is still much to be learned about how different approaches to negotiating safer sex are perceived and responded to by sexual partners, and what factors influence reaction to negotiation attempts. Traditional models of persuasion suggest that success or failure to achieve compliance may indeed vary with type of message used, characteristics of the negotiator or their relationship, or characteristics of the sexual partner for whom the message is targeted (Hovland, Janis, & Kelley, 1953).

There are several reasons why we need to further investigate reactions to condom negotiation strategies. First, women in power-imbalanced relationships may perceive that they have a lot to lose from unsuccessful negotiation efforts (e.g., Beadnell, Baker, Morrison, & Knox, 2000; Wingood, Hunter-Gamble, & DiClemente, 1993; Wingood & DiClemente, 1997). This makes sense: If a negotiation attempt is effective, it could prevent risk, and much could be gained. If it is ineffective, however, it could produce negative partner outcomes: a partner may refuse to comply with the request, may become angry or defensive, and the individual attempting the negotiation may abandon further efforts. This issue is especially pertinent to heterosexual women’s negotiation skill attempts. Whether or not condoms are used is strongly influenced by the male partner in a male/female relationship; safer sex efforts may require negotiation skills if a male partner is initially reluctant to use a condom during sexual activity. In addition, gender role stereotypes and gender scripts have traditionally discouraged women’s initiation of sexual discussion, so initiating sexual negotiation may be difficult for many women.

Many studies have investigated gender differences in strategy use (Aida & Falbo, 1991;Bird, Harvey, Beckman, & Johnson, 2001; Bui, Raven, & Schwarzwald, 1994; Noar, Morokoff, & Harlow, 2004); however, findings have been inconsistent. Several studies have found differences in the comfort level felt by men and women when using different forms of sexual negotiation. For example, Carter and colleagues(1999) found in their examination of roles in negotiation of condom use that women play a more active role and men a more reactive role in negotiation; similar results were found by Noar, Morokoff, and Harlow (2002). Further, Carter, McNair, Corbin, and Williams (1999) found that when men’s partners were more active in the process of deciding whether or not to use condoms, the relationship between intentions to use condoms and past condom use increased: it seemed as if men’s intentions to use condoms were based on the assumption that a female partner would initiate the decision to use a condom. Some studies examining gender (e.g., Allen, Emmers-Sommer & Crowell, 2002; Noar, Morokoff & Harlow, 2002) suggest that women may be generally more communicative about safer sex than men. Other studies(Aida & Falbo, 1991; Bird, Harvey, Beckman, & Johnson, 2001; Buss, Gomes, Higgins, & Lauterbach, 1987; Noar, Morokoff, & Harlow, 2004) have not found gender differences in strategy use in intimate relationships. For example, in a study of 90 heterosexual couples, Bird and colleagues (2001) found that study participants reported that the strategy used most often was to engage in a conversation about disease prevention, pregnancy prevention, or condom use; they next reported using assertion of needs or wants directly, followed by threatening to withhold sex. However, fewer studies have been performed examining a partner’s perceptions of specific condom use negotiation strategies. Johnson, Gant, Hinkle, Gilbert, Willis, and Hoopwood (1992) found that in their sample of young African-American men and women, men were much more likely to react with anger when asked how they would react if a sexual partner proposed using a condom. Our research team (Otto-Salaj, Reed, Brondino, Gore-Felton, Kelly, & Stevenson, 2008) found similar results in our qualitative study of sexual negotiation in African-American adults.

There are little data showing use of specific strategies as predictors of successful negotiation. Margillo and Imahori (1998) examined use of negotiation strategies using semi-structured interviews with 21 African-American women between the ages of 15 and 30 who were economically impoverished and who used condoms inconsistently. Successful use of six types of “compliance-gaining” strategies was assessed, based on a typology developed by Howard, Blumstein, and Schwartz (1986). The typology includes autocracy, bargaining, bullying, disengagement, manipulation, and supplication. Autocracy includes asserting authority, claiming greater knowledge, making self-centered statements, or flatly insisting; bargaining involves use of reasoning or compromise; bullying uses threats, insults, ultimatums, or violence; disengagement relies on indirect methods such as sulking, making the partner feel guilty, or removing oneself from the situation; manipulation utilizes flattery, seduction, hints, or any practice which is secretive; and supplication includes pleading or acting ill or helpless. Results showed women reported successfully using bargaining/reasoning most often with male sexual partners, followed by bullying and autocracy. However, the study did not assess perceptions of the strategies by male partners, or unsuccessful use of strategy types by women participants.

Noar, Morokoff, and Harlow have conducted several studies (2002 (2004) examining use of condom influence strategies. Their 2002 study focused on college students, and identified six particular influence strategies (withholding sex, direct request, nonverbal seduction, relationship compartmentalizing, provision of risk information, and deception) used by the students in their sample; further, they found that use of these strategies were related to intentions for condom use, and that women endorsed use of different strategies emphasizing verbal approaches (withholding sex, direct request, risk information, and relationship conceptualizing) compared to men. In their 2004 study, they explored use of these strategies in a diverse community sample of 113 heterosexually-active men and women, and found that participants endorsed the aforementioned six strategies, plus a strategy based on pregnancy prevention. However, they did not find gender differences in strategy use. They also found that willingness to use condoms, sexual assertiveness, and being sexually communicative was linked to greater endorsement of strategy use. Their results suggest that becoming more ready to use condoms on a regular basis is related to greater likelihood of using strategies to influence a partner to use condoms.

Other studies (e.g., Dunn & Cowan, 1993; Lam & Barnhart, 2006; Lam, Mak, Lindsay, & Russell, 2004) have indicated there may be potential influence of ethnicity or culture on negotiation attempts. This may be important, as HIV risk reduction messages are increasingly targeted to populations identified as being at greater risk for infection. In the epidemiological literature on HIV seroprevalence, the rate of AIDS diagnoses for black adults and adolescents in 2005 was 10 times the rate for whites and nearly 3 times the rate for Hispanics. Further, the rate of AIDS diagnoses for black women was nearly 23 times the rate for white women (CDC, 2007a; 2007b). Although African-American women comprised an estimated 13% of the U.S. population in 2005, they also comprised a disproportionate 66% of HIV/AIDS cases(CDC, 2007b). For these reasons, African-American men and women are often targeted for inclusion in HIV risk reduction interventions, including those focusing on negotiation skills training (e.g., Carey et al., 2000; DiClemente & Wingood, 1995; DiClemente et al., 2004; Jemmott, Jemmott & Fong, 1992; Kalichman, Cherry & Browne-Sperling, 1999; Sikkema et al., 2005). In particular, many interventions appear to target African-American women (DiClemente & Wingood, 1995; DiClemente et al., 2004; Sikkema et al., 2000; Wechsberg, Lam, Zule, & Bobashev, 2004).

Cultural differences in sexual and safer sex behavior have been reported in some studies(e.g. Catania, Coates & Golden, 1994), but not others(e.g. Seal & Palmer-Seal, 1996). Numerous studies have examined the influence of cultural norms and values on sexual decision-making; of these, several(Belk, Snell, Garcia-Falconi, Hernandez-Sanchez, Hargrove, & Holtzman 1988; Dunn & Cowan, 1993; Lam & Barnhart, 2006; Lam, Mak, Lindsay, & Russell, 2004; Steil & Hillman, 1993) have examined cultural context as a possible determining factor in the use of influence strategies. Lam and colleagues (Lam, Mak, Lindsay, & Russell, 2004) found that Asian American students used verbal indirect strategies (e.g., deception, flattery, dropping hints) significantly more than White American students. A study of African-American, Latina, and Caucasian-American women (Steil & Hillman, 1993) found that ethnicity was associated with decision-making, in that African American women were the least willing or able to communicate with partners about contraceptive decision-making. Wingood and DiClemente (1998) suggested that a reason for this may be the strong association among African American women of condom negotiation attempts with the belief that asking a partner to use a condom implies infidelity.

Wingood, Hunter-Gamble, & DiClemente(1993) conducted a series of focus groups with 18 unmarried African-American women between 18 and 25, in order to gain a greater understanding of the process by which women initiate communication about safer sex and negotiate condom use with regular and casual partners. Wingood and her colleagues found that while many women reported the ability to initiate discussions concerning safer sex with male sexual partners, few reported that they were able to negotiate condom use. Factors such as trust, conflict avoidance, attitudes toward using condoms, and sexual self-efficacy were identified as barriers to condom negotiation. In particular, most of the women in this study did not perceive themselves as having the power to make their partner wear a condom. Further, women who did feel comfortable exercising their power possessed either well-developed sexual negotiation skills or greater sexual self-efficacy. Finally, several women expressed fear of violence or described personal experience with violent reactions upon attempting to negotiate condom use with their partner. The researchers concluded that future research be targeted to determine the nature of sexual negotiations among African-American men and women and predictors of successful sexual negotiation among these populations, and to understand the role of power dynamics between African-American men and women in negotiating condom use.

In keeping with the need to investigate power dynamics in the context of negotiation, our investigational team (Otto-Salaj, Reed, Brondino, Gore-Felton, Kelly & Stevenson, 2008) completed a qualitative study of perceptions and responses of 51 African-American men and women to six different types of negotiation strategies. These strategies included reward, coercive, legitimate, expert, referent, and informational strategies, and were based on Raven’s (1992) Power/Interaction Model of Interpersonal Influence. The following statements were used to represent each of the six types of negotiation strategy, to be used with a hypothetical ‘steady’ sexual partner:

• Coercion: If a steady sexual partner said to you that if a condom wasn’t used, s/he wasn’t sure that s/he wanted to have sex with you or continue the relationship:

• Reward: If a steady sexual partner said to you that s/he thought condoms made it easier to “let go”, that s/he thought you both could have sex for a longer time when using one, and that s/he knew some things that would make you feel really good while a condom was being used:

• Legitimate: If a steady sexual partner brought up condom use as the “responsible thing to do”:

• Referent: If a steady sexual partner brought up condom use by saying that s/he respects you a lot, that condom use would be the best thing for you both, and how great you would be if you went along with it:

• Expert: If a steady sexual partner said that s/he’s learned a lot about using condoms, and that the more s/he learns about them, the more s/he wants to use them, so it’s a good idea that you use one:

• Informational: If a steady sexual partner brought up using condoms by saying that s/he saw a TV show that talked about HIV, and how effective condoms are in keeping people safe from HIV, and so what do you think about using a condom:

Results demonstrated gender differences in response to different strategy types: female participants responded best to referent, legitimate and reward strategies, and worst to the use of informational tactics. On the other hand, male participants responded best to reward strategies, and worst to coercion to use condoms. Further, responses given by a subset of both women and men indicated that use of negotiation tactics involving coercion to use condoms may result in negative and angry or violent reactions by those sexual partners. This was interesting, given that when female participants were asked to generate strategies they would use to negotiate with a sexual partner, most participants replied that they would use strategies that would be classified as coercive, and could not generate strategies based on other types of power bases.

Thus, we set out to examine the types of negotiation styles women might use to maximize the likelihood of safer outcomes while minimizing the likelihood of negative social outcomes, as well as the characteristics of those partners who may respond well to different types of condom use negotiation attempts. This study describes the reactions of a sample of heterosexual African-American men to several condom use negotiation attempts by a hypothetical female partner, involving six different types of approaches purportedly used by a steady sexual partner. Our purpose was: 1) to assess the differences in reactions to each type of negotiation strategy and attempt to identify strategies more likely or less likely to result in participant acquiescence to the request, and 2) to identify predictors of participant compliance or refusal to comply with condom use negotiation attempts. More information may help us inform more specifically communication components of HIV risk reduction interventions.

Method

Setting and Participants

A sample of 172 men was included in the present set of analyses. Participants met the following inclusion criteria: 1) they were male; 2) of African-American ethnicity; 3) were heterosexually-active in the three months prior to interview, 4) were between the ages of 18 and 35, 5) were unmarried, and 6) agreed to give written informed consent for project participation. All participants were recruited from an inner-city community-based service center providing outreach, social service, and recreational programs for African-American men and women.

Procedures and Measures

In the community center, recruitment for the study was performed by project staff using face-to-face meetings, brochures, and posters. Interested persons were briefly screened to determine whether or not they met criteria for study entry. Informed consent was obtained from men who met screening criteria, and participants were then interviewed in a private room at the center. Assessments were conducted by an ethnically-matched research team experienced in the administration of assessment measures. Assessors also had previous experience in sexuality education and HIV prevention. Moreover, all assessors received extensive training in the study protocol and assessment measures before conducting the fieldwork.

In individual sessions, participants completed measures assessing:

Demographic data and health characteristics

Data were obtained on participant age, education level, relationship status, employment status, and income. In addition, participants were asked how often they had had an STD, and whether or not they ever had an HIV antibody test (and, if so, its result).

AIDS risk behavior knowledge

In previous research, we developed, pilot tested, and normed objective tests of knowledge concerning risk practices and risk reduction steps (Kelly et al., 1990; Kelly, St. Lawrence & Brasfield, 1991). Recently, we have modified the measure’s vernacular and phrasing and have added items that better assess HIV risk knowledge in low-literacy populations. This 17-item version has sound psychometric properties (Otto-Salaj, Heckman, Stevenson, & Kelly, 1998), and is scored by summing the number of items correct.

Risk History Survey (RHS)

The RHS assesses respondent sexual practices and substance use over the preceding 3 months. The RHS has proven to correlate with other measures pertinent to HIV risk-taking such as risk knowledge (Kelly et al., 1990), and to exhibit acceptable reliability characteristics (Kauth, St. Lawrence, & Kelly, 1991). The RHS elicits information on sexual activities and substance use occurring over the past three months including number of male and female sexual partners, frequency of unprotected vaginal and anal intercourse, and frequency of condom use. The format and language used in this instrument was found to be reliable and valid when used in a previous study (Otto-Salaj, Heckman, Stevenson, & Kelly, 1998).

Condom attitudes

The Condom Attitude Scale is a ten-item scale assessing participant’s attitudes towards condom use. Each item consists of a statement (sample items: “Condoms ruin the mood,” “Condoms interrupt foreplay”) and a 6-point Likert scale to indicate level of agreement (1=”Strongly Disagree” to 6=”Strongly Agree”). This scale produces scores ranging from 10 to 60, with higher scores indicative of more positive attitudes toward condom use. The scale has demonstrated adequate internal consistency and reliability in a previous study (alpha=.70; Otto-Salaj, Heckman, Stevenson, & Kelly, 1998).

Videotape segments

In order to examine the relative effectiveness and social consequences of different theory-based strategies that might be employed by women to negotiate condom use with male partners, videotape segments representing each of six types of negotiation strategies— coercion, reward, referent power, legitimate power, expert power, and informational power—were shown to participants.

Each videotape segment showed a female actor speaking directly into the camera as if speaking directly to the viewer of the tape; participants were instructed to view the tapes as if the woman was a ‘steady’ sexual partner. The actor was represented in silhouette, to obscure personal features or characteristics which might confound participants’ assessments of negotiation attempts. Edgar, Noar and Murphy (2007) suggest that although many HIV risk reduction interventions may employ negotiation strategy use, these strategies are often not sufficiently linked to theory or empirical evidence of strategy use. Thus, each segment showed the model using a different negotiation strategy; strategies portrayed were based on Raven’s (1992) social influence tactic conceptualizations, but informed and tailored for content, vernacular, and expressive style. Raven proposes six bases of power from which people derive strategies attempting to influence the behavior of others. These bases of power include: (a) reward; (b) coercion; (c) legitimate; (d) expert; (e) referent; and (f) information. According to Raven, reward and coercive power can refer to tangible rewards and real physical threats, but can also include personal approval or rejection. Legitimate power is derived from the structural relationship between the influencing agent and the target; the agent may implicitly or explicitly communicate that s/he has a “right” to ask the target to engage in some behavior, and that the target has an obligation to comply. Raven explicates several dimensions of legitimate power, including reciprocity (“I did that for you, so you should feel obliged to do this for me”), equity (“I have worked hard and suffered, so I have a right to ask you to do something to make up for it”) and responsibility or dependence (“the obligation to help others who cannot help themselves or others who are dependent on us”). We chose to focus on the dimension of responsibility/dependence for the purpose of this study. Expert power is acting on the assumption that the powerholder is “correct”, while referent power refers to engaging in a behavior because of a sense of connection with the influencing agent. Raven suggests that these forms of power may also have a “boomerang” effect in which the target does the opposite of whatever the influencing agent wishes, perhaps because s/he perceives the agent as acting in his or her best interest, or because people often dislike an influencing agent. Finally, informational power is based on logical argument that the influencing agent can present, either directly or indirectly, to the target in order to implement change.

In constructing the text of the video portrayals, a focus group was held with 6 African-American men and women, during which videotape text was edited and rewritten to reflect cultural tailoring of language, accurate portrayal of each power base, and social validity of presentation. In addition, we attempted to control for length of negotiation message/attempt, to make them similar across attempts. Video segments were presented in counterbalanced order across participants, as a control for possible influences of order and primacy or recency effects on participant ratings of the segments. Table 1 illustrates how the influence tactics were reflected in the text used in the videotape segments.

Table 1.

Videotape Strategy Types, Premises, and Associated Text

Social Power Base	Influence Tactic Utilizing Base	Condom Negotiation Script Reflecting This Tactic
Coercion	Threatening punishment (disapproval, relationship change) for not adhering to the request	“You know, I really want to have sex with you, and I know you want to have sex with me. But I’ve got to tell you, I’ve made a rule for myself, and that is to use condoms when having sex. I can’t respect someone who doesn’t use them, so we’re going to use them or we’re not having sex. Period.”
Reward	Promising or insinuating reward (personal approval, sexual satisfaction) for adhering to the request	“I know you want to have sex, and I want to do it with you, too. We can really get each other off using condoms – just the thought of getting together with you and using them really turns me on. I can make this very hot, and we can try some things that make us feel really good.”
Legitimate	Invoking principles of reciprocity, responsibility, or dependence	“I care about you. You really seem to care about me, and I think you want to have sex with me. I know you want to use a condom, because you’re the kind of guy that looks out for a woman. And that’s what people who care about one another do, it’s the responsible thing to do.”
Referent	Stressing target of influence as a “role model,” someone the powerholder admires	“We seem to have a lot in common, and I respect and admire you. We both want the same thing--I want to sleep with you, and you want to sleep with me. Talking with you about condoms is no problem, because a man like you is always prepared —that’s one of the great things about you. I know you want to do it safely.”
Expert	Invoking powerholder as the “expert,” the one who knows best what to do, without associated knowledge (informational power, see below)	“I want to have sex with you, and think you want to have sex with me. I really like learning new things, and think I’m pretty good at a lot of things. I know all about using condoms and how great they are for staying healthy. The more I learn about using them, the more I know it’s a good idea to use one.”
Information	Emphasizing specific factual information supporting adherence to the request	“I think you want to sleep with me, and I really want to sleep with you, but I’m concerned about HIV. I saw a TV show the other day on HIV, you know--the virus that causes AIDS? More of us are being infected with it every day, and although there’s drugs to treat it, there’s no cure. Condoms are a good way to keep us safe, and we won’t have to worry about pregnancy or HIV.”

After viewing each video segment, each participant completed measures describing his reaction to the video segment he just observed, using six-point Likert scales to indicate level of agreement (where 1 = Strongly Disagree, and 6 = Strongly Agree) with statements reflecting the following domains:

Compliance with the condom use request

Six items were used to measure the participant’s likelihood to use condoms as influenced by the condom use request (e.g., “If she said this to me, I would use a condom if we had sex, even if I didn’t really want to”; “I would refuse to use a condom if she said this to me”);

Positive and negative affect

Six items measured the participant’s emotional reaction and liking for the model as influenced by her request (e.g., “If she said this to me I would be happy”; “If she said this to me, I would like her less”);

Attributions concerning the woman model

The five items on this dimension measured the participant’s perceptions of the model’s motivation for engaging in the request (e.g., “If she said this to me, I would think she only cares about herself”); and

Attributions concerning himself

Five items measured the motivations of the participant regarding his reaction to the request (e.g., “If she said this to me I would think I need to find another partner”).

Thus, we were able to assess each participant’s reported likelihood of condom use request adherence, positive and negative affect, and attributions concerning himself and the hypothetical woman partner in response to each of the six portrayals. These items are shown in Table 2.

Table 2.

Videotape Rating Items

Content	Items
Condom Use Compliance:	(1) If she said this to me, I’d use a condom if we had sex even if I really didn’t want to.
	(2) There would be nothing a woman could say to get me to use a condom.
	(3) I would refuse to use a condom if she said this to me.
	(4) Sex with a condom would be okay if she suggested it like this.
	(5) There is nothing this woman could say to get me to use a condom.
	(6) If my real-life sexual partner said this to me, I’d use a condom.
Positive and Negative Affect	(7) If she said this to me, I would be happy.
	(8) If she said this to me, I would feel angry.
	(9) If she said this to me, I would feel anxious.
Liking for the Model:	(10) If she said this to me, I would like her more.
	(11) If she said this to me, I would like her less.
	(12) If she said this to me, I might not be able to have sex at all.
Attributions for the Model:	(13) If she said this to me, I would think that she cares about me.
	(14) If she said this to me, I would think she only cares about herself.
	(15) If she said this to me, it’s because she doesn’t trust me.
	(16) If she said this to me, she must have been having sex with someone else, and that’s why she brought it up.
	(17) If she said this to me, it would show that she is thinking of both of us.
Attributions about Himself (the participant):	(18) If she said this to me, I would think I need to find another partner.
	(19) If she said this to me, I would go along with it because I care about her.
	(20) I make the decisions about how we have sex, so I wouldn’t really care how she feels about condoms - I’ll decide.
	(21) Having sex is more important than whether or not we use condoms.
	(22) If she said this to me, I wouldn’t trust her anymore.

Data Analysis

Ratings on each item across vignettes were compared using a repeated measures general linear model to determine whether participant’s ratings of each item varied by vignette. Cluster analysis was applied to the ratings from each vignette. Specifically, SPSS version 16 was used to run average linkage hierarchical agglomerative cluster analyses using the squared Euclidean distance method of measuring similarity. The 22 items following a vignette were entered into the cluster analysis and dendograms were examined to determine which items grouped together to form clusters. This was done for each of the 6 vignettes. Cluster analysis results from each vignette were compared to determine whether items clustered similarly for each vignette. The items and clusters which were consistent across vignettes were selected and average composite scores were created for each item cluster within each vignette. These cluster analysis derived composite scores were aggregated by averaging them across vignettes. This produced four composite scores representing the following constructs: Anger/rejection of the partner; Condom use refusal; Attributions of the partner as selfish, cheating, and/or uncaring of the respondent; and Attributions of the partner as caring and agreement to use condoms. Since interest in the current study is on condom use, the sample was sorted in terms of their responses to the new index representing the degree of condom refusal averaged across the vignettes. Subgroups were identified who would definitely use condoms, were less likely to do so, and who were least likely to do so. Exploratory analyses were conducted to determine how these groups differed from one another in their responses to the vignettes on the remaining composite scores and other variables in the data set.

Results

Mean age of participants was 31.5 years (SD=3.3). Most participants (74%) were unemployed. The mean education level of the sample was 12.1 years (SD=1.89). Thirty-five percent of the participants were not currently in a relationship, 24% were in a relationship that was not long-term, 21% were in a long-term relationship but not living with their partner, and 18% were living with their partner.

Aggregated Men’s Ratings

No significant differences were found in men’s ratings across the six vignettes. Specifically, a repeated measures general linear model was conducted for each item. None of the results reached significance, which indicates that there were no significant differences in men’s ratings across the six vignettes on any of the 22 items. Greenhouse-Geisser F-ratios ranged from 0.170 to 2.204, with p-values ranging from 0.057 to 0.971.

Cluster Analysis

The 22 items from each vignette were then placed into a cluster analysis to determine whether the items clustered in any way. This was done for each of the 6 vignettes. The dendogram from each cluster analysis was examined to look for items which clustered together and provided an interpretable solution (Norusis, 2009). Results were then compared across vignettes to look for similarity. It was found that each of the 6 cluster analyses revealed 4 clusters, which supported the clusters that were chosen by showing that the clusters were stable across vignettes (Clatworthy, Buick, Hankins, Weinman, & Horne, 2005). One cluster, termed anger/rejection of the partner included items 8, 11, 18, and 22. The condom use refusal cluster included items 2, 3, and 5. The cluster indicating attributions of the partner as selfish, cheating, and untrusting of the respondent included items 14, 15, and 16. Finally, the cluster which indicated agreement on condom use and attributions of the partner as caring included items 7, 10, 13, 17, and 19. Items on each cluster were averaged for each vignette to create 4 composite scores for each vignette. As it was already determined that there were no differences across vignettes, these scores were then averaged across vignettes to create 4 composite scores for each participant, representing their scores on each of the constructs listed above.

Since our interest is in condom use, the sample was then sorted in terms of their responses to the condom use refusal construct. Subgroups were identified who would either definitely use condoms, were less likely to do so, and who were least likely to do so. The individuals who were thought to definitely use condoms (n=41; 23.7%) responded to each item on this subscale with a 1, which indicated that they strongly disagreed that they would refuse to use condoms, giving an average on this construct of 1. Individuals who were thought to be less likely to use condoms (n=111; 64.2%) had an average on this construct between 1 and 3, which indicated that, on average, they disagreed that they would refuse to use condoms. Finally, individuals who were thought to be least likely to use condoms (n=20; 11.6%) had an average on this construct which was greater than 3. This indicated that they, on average, agreed that they would refuse to use condoms. Exploratory analyses were then conducted to determine how these groups differed from one another in their responses to the vignettes on the remaining composite scores and other variables in the data set.

Chi-square Analyses

Subgroups varied on whether or not they were forced into having sex when they didn’t want to in the past 3 months [χ²(2)=6.763, p<.04]. Specifically, participants who stated that they had been forced into having sex when they didn’t want to were more likely to be in the “less likely” and “least likely to use a condom” subgroups. On the other hand, participants who stated that they had not been forced into having sex when they didn’t want to were more likely to be in the “definitely would use a condom” subgroup than in the other two subgroups.

Subgroups also varied on whether or not they ever had a STD [χ²(2)=7.313, p<.03]. In this case, participants who stated that they had ever had a STD were more likely to be in the definitely would use a condom or in the less likely to use a condom subgroups than in the least likely to use a condom subgroup. On the other hand, participants who stated that they had never had a STD were more likely to be in the least likely to use a condom subgroup than in the other two subgroups.

ANOVAs

The subgroups were found to differ significantly from one another on measures of AIDS knowledge [F(2,169)=3.283, p<.05]. Specifically, the subgroup that would definitely use condoms had the highest AIDS knowledge mean (M=12.585), while the subgroup that was less likely to use condoms had a lower mean (M=11.333), and the subgroup that was least likely to use condoms had the lowest AIDS knowledge mean (M=11.100). Post hoc pairwise Bonferroni comparisons revealed that the only significant difference was between the group that would definitely use condoms and the group that was less likely to use condoms.

The subgroups were also found to be significantly different from one another on the anger/rejection of the partner construct which was created from the cluster analysis [F(2,167)=27.171, p<.001]. Specifically, the subgroup that would definitely use condoms had the lowest anger/rejection of the partner mean (M=1.556), while the subgroup that was less likely to use a condom had a higher anger/rejection of the partner mean (M=2.288), and the subgroup that was least likely to use a condom had the highest anger/rejection of the partner mean (M=3.042). All post hoc Bonferonni comparisons were significant, which indicated that each of the 3 groups were significantly different from one another. This indicated that likelihood of condom use was related to anger/rejection of the partner such that being more likely to use condoms was related to lower levels of anger/rejection of the partner.

In addition, the subgroups were found to be significantly different from one another on the agreement to use condoms and attributions of the partner as caring construct [F(2,167)=14.465, p<.001]. Specifically, the subgroup that would definitely use condoms had the highest mean (M=5.072), while the subgroup that was less likely to use condoms had a lower mean (M=4.375), and the subgroup that was least likely to use condoms had the lowest mean (M=4.040). Post hoc pairwise Bonferroni comparisons revealed a significant difference between the group that would definitely use condoms and the group that was less likely to use condoms as well as between the group that would definitely use condoms and the group that was least likely to use condoms. This indicated that the likelihood of condom use was related to agreement to use condoms and attributions of the partner as caring. That is, participants who were most likely to use condoms were also most likely to agree to use condoms and to view their partner as caring.

Putting this all together, a 3×3 mixed model ANOVA (condom use subgroup by composite score) was performed, which demonstrated consistencies across subgroups in responses on the outcome measures. The subgroup that was most willing to use condoms was lowest in anger/rejection of the partner and in attributions of the partner as selfish, and highest on agreement to use condoms and attributions of the partner as caring. The subgroup that was least willing to use condoms was highest in anger/rejection of the partner and in attributions of the partner as selfish, and lowest on agreement to use condoms and attributions of the partner as caring. The subgroup that was less likely to use condoms scored in between the other two subgroups on all three constructs. That is, there was a significant interaction between subgroup and composite score [F(2,162)=27.697, p<.001].

Discussion

Our results suggest that first, in comparing the reactions of heterosexually-active African American men to a number of condom use negotiation strategies used by a hypothetical female partner, there was no one strategy that was responded to significantly more positively or negatively. In other words, there was no “magic bullet’ – no strategy worked to facilitate positive responses and acquiescence to use condoms by project participants. However, we also found that none of the strategies elicited universally negative responses –in the form of negative affect or attributions regarding the hypothetical female partner. This differs somewhat from the findings of our previous qualitative study of perceptions of condom use negotiation strategies (Otto-Salaj, Reed, Brondino, Gore-Felton, Kelly, & Stevenson, 2008). In that study, we found that men generally responded best to reward strategies, and worst to coercion strategies.

Participants were divided into three subgroups based on the results of one of the cluster analysis clusters (condom use refusal). These subgroups appear to be related to responsiveness to negotiation attempts as a whole and willingness to use condoms. Further, our results indicate that these three subgroups significantly differed from one another on constructs that one would think would predict differences in willingness to use condoms, such as anger/rejection of the negotiator and attributions of the negotiator as caring. The group that indicated greatest willingness to use condoms also indicated positive attributions and least anger/rejection toward the negotiator, and were more likely to have been diagnosed with an STD in the past; further, this group significantly differed from the group indicating they were least likely to use condoms on these measures. Also, participants who indicated they were unwilling to use condoms were also more likely to have been forced into sex in the past. These results support the findings in our earlier study, and suggest that responses to negotiation strategies are rather complex. In our qualitative study, we also found that responses differed across subsets of women and men. Responses given by a subset of women and-- to a greater extent, men-- indicated that use of negotiation tactics involving coercion to use condoms may result in negative and even angry or violent reactions. Especially among some men, the potential for violent response to a “no condom, no sex” strategy was clear. Further, negative affect did not always correspond with refusal to use condoms, but may have other consequences such as increasing the likelihood of intrapersonal violence between sexual partners. Thus, in order to avoid potentially angry or violent reactions in response to negotiation attempts, we recommend that this particular strategy type be used with caution, and with attention paid to the characteristics of the partner.

Our current findings suggest that while the manner in which the female approached her request to use condoms had little impact on the participant’s response, the very act of suggesting condom use produced a response that differed across participants. Further, this act produced an extremely negative response among some men, no matter what strategy was used. As women are often targeted for training in condom use negotiation as a part of HIV risk reduction strategies, it is important for trainers to be aware that successful negotiation may be influenced as much by partner characteristics as by the method of training and strategies emphasized. In certain relationships, the use of negotiation may encourage negative relationship outcomes, with a potential for anger and violence from a subset of male partners in response to negotiation attempts. Little exploration of personality predictors of condom use has been conducted (Hoyle, Fejfar & Miller, 2000; Zimmerman et al., 2007). Hoyle, Fejfar and Miller (2000) performed a meta-analysis of the literature on personality traits and sexual risk taking, and identified sensation-seeking and impulsivity to be most indicative of sexual risk taking. This may help to explain our finding that men who were least likely to use condoms were also most likely to have had a previous history of an STD; however, as there is little integration of personality characteristics and condom use, we know of no studies examining personality predictors of response to negotiation attempts. Most studies examining partner characteristics and condom negotiation have focused on variables such as perception of partner’s willingness to use condoms. For example, a study of factors thought to promote safer sex communication in a sample of college students (DiIorio, Dudley, Lehr & Soet, 2000) found that perception of partner attitudes toward a safer sex discussion was related both to communication self-efficacy, communication outcome expectancies and safer sex communication; however, the relationship between partner attitudes and communication was weak, suggesting a complex relationship and potential mediation by other factors. Thus, research specifically exploring partner characteristics that influence perception of strategies and willingness to use condoms would be a welcome addition to the literature, and would facilitate the tailoring of HIV interventions to individuals having characteristics that predict unwillingness to use condoms.

As effective negotiation of condom use with a male sexual partner may be determined more by characteristics of that partner than by specific strategy used, teaching an array of negotiation strategies in HIV intervention skills training may allow intervention participants to be able to more accurately target use of a specific strategy for a particular partner. Also, increasing the repertoire of strategies among HIV intervention participants may increase the likelihood that 1) they actually engage in condom negotiation attempts; and 2) they are able to choose and apply a strategy that is more effective in persuading a partner to use a condom. The results also imply a need to more adequately explore gender differences not by negotiation strategy use or perception of strategies, but in the efficacy of HIV intervention components other than negotiation skills training. In a meta-analysis of HIV prevention interventions and components found to be efficacious with different subgroups, Albarracin, Gillette, Earl, Glasman, Durantini, and Ho (2005) found that men and not women were the ones who benefitted most from behavioral skills training, including training in condom use and communication skills; this suggests that efficacy of HIV intervention components may differ by gender of the target of the intervention. Further analysis of the efficacy of specific components of interventions – and how efficacy may differ by gender, age, ethnicity, and personality characteristics of those to whom the interventions are targeted – is suggested. Finally, it may also be equally effective for HIV risk interventions to attend to issues in both men and women regarding the empowerment of women, and creation of norms for: respect, empathy and caring for partners; rejection of anger, hostility, aggressiveness, and physical and emotional violence; and condom use and sexual safety.

Study Limitations

There are several limitations to our findings. First, we attempted to simulate a negotiation attempt as might occur in a relationship with a steady (that is, ongoing and relatively stable) sexual partner. There are numerous challenges in this context: it is difficult to simulate the dynamics in an existing relationship. We chose to characterize the relationship as ‘steady’ for several reasons. First, those are relationships in which it is often difficult to increase condom use. Studies suggest that condom use in most relationships is less consistent or diminishes over time (e.g. Manlove, Ikramullah, & Terry-Humen, 2008). Previous HIV intervention trials have shown outcomes reflective of HIV risk behavior change with new sexual partners, but no change with existing or long-term partners (e.g. Otto-Salaj et al., 2000). Also, being in a ‘steady’ relationship may also result in dynamics creating greater sexual risk for partners, because of the balance--or imbalance-- of power, and commitment to relationship resulting in potentially greater losses for a female negotiator if partners are non-monogamous and negotiation attempts go awry. Similarly, many public health campaigns endorsing condom negotiation do not propose using specific strategies with specific partner types (i.e. steady or casual) – they simply suggest using a strategy, without taking into account the length of relationship (e.g., “No condom, no way; “Just say no”); our recent study shows use of some strategies may potentially produce issues for established relationships (Otto-Salaj, Reed, Brondino, Gore-Felton, Kelly, & Stevenson, 2008). Thus, we wished to continue our work examining the context of negotiating condom use in a steady relationship.

However, several studies have found that context of relationship may be an important factor influencing use of negotiation strategies. In a study examining assertive communication with a sexual partner, Quina and colleagues (Quina, Harlow, Morokoff, Burkholder, & Deiter, 2000) found that women having a current sexual partner generally responded more negatively on most study measures, suggesting that just having a real-life partner may influence responses on measures assessing communication strategies. In a similar vein, DeVisser and Smith (2001) conducted a study of 103 heterosexual men and women who completed a condom use diary over a period of six months. They found that for relationships with both regular and casual partners, the dynamic of condom use was influenced not as much by the attitudes or beliefs of the partner, but more by the “characteristics of the interaction between the partners (e.g., prior agreements, use of other contraception). (p. 423).” DeVisser and Smith go on to suggest that condom use among steady partners may be heavily influenced by established patterns of behavior, whereas condom use among casual partners may be more influenced by the context of that specific encounter. Considering these studies, we need to interpret our results with caution; future research examining and comparing multiple contexts of relationship in the assessment of strategy perception and use is recommended.

In addition, in order to allow us to characterize the relationship with the hypothetical female partner as ‘steady’, we felt we needed to obscure most of the model’s physical characteristics so that perceptions of the physical attractiveness of the hypothetical partner would not influence our results. This controlled for some factors, although study participants were exposed to the model’s voice when listening to the vignettes, and we relied on the ability of the participants to transcend the context of laboratory research in the conduct of the study. Thus, the extent to which we were successful in facilitating our participants to realistically consider the hypothetical partner as ‘real’ is unclear. Also, it is likely that dynamics in condom negotiation with newer partners may vastly differ from the situations we were trying to assess.

Finally, the study was also designed with the intent of examining use of negotiation strategies based on Raven’s power bases, while at the same time making them ‘user-friendly,’ implementing language that people might use in negotiating use of condoms. To address this, we conducted the focus group and worked with African American men and women to construct and revise the wording for each strategy, based on the idea that these strategies would be used with a steady partner. Although the members of the group worked together to develop wording, and agreed at the end on the wording for each strategy, it is admittedly arguable how believable each strategy would be in a real life negotiation attempt. In particular, some strategies are more believable than others, because we had some sources of comparison in the form of messages similar to those advocated in public health campaigns. Also, we attempted to control for length of negotiation messages and make them similar across attempts. The example negotiation strategies used in our attempts may not accurately reflect the length of time spent in negotiation in real-life situations; our examples may be longer or ‘more wordy,’ or may be shorter than the phrasing people really might use.

Thus, when including condom use negotiation components in HIV risk reduction programs, using a “one size fits all” approach in advocating use of negotiation strategies – no matter what the strategy – may not be as efficacious in promoting behavior change as using a repertoire of several different negotiation tactics, and taking into account partner characteristics and relationship context in choice of strategies to use. Providing a “menu of options” – may be key in facilitating successful adoption and use of strategies appropriate for specific relationships. Further, additional research is recommended on the specific characteristics of partners that predict strategy efficacy, and the effects of relationship context of strategy use and efficacy of specific strategies.