Sex and relationships on the street: How homeless men judge partner risk on Skid Row

INTRODUCTION

A recent critique of the HIV prevention literature decries the over-emphasis on women as passive victims of male risk behavior, arguing that men who have sex with women have not only been understudied, but also that their own HIV risk has been underplayed [1]. As a result of assuming a simplistic, power-driven model of men as powerful perpetrators and women as powerless victims, the authors argue that the literature on gender and HIV risk has largely ignored the structural, social, and cognitive factors that put men at risk and drive male risk behaviors, especially among high risk subgroups of men who have sex with women (MSW). Furthermore, understanding such men’s risk behavior is crucial for reducing rates of HIV among women, who are most likely to get HIV from sex with male partners [2].

Homeless men in the United States represent a sub-group of MSW who are at greater risk of HIV than other men. Among MSW, homeless men have higher rates of HIV/AIDS and exhibit more sexual risk behaviors than housed men [3, 4], including lower rates of condom use [5]. Estimates of HIV seroprevalence among homeless men vary widely by city [6], but are generally 10% or higher [3, 7]. A significant portion of this risk comes from heterosexual behavior. The CDC estimates that heterosexual sex is responsible for 31% of new HIV cases among men, versus 12% for injection drug use and 53% for sex with men [8]. Even among homeless injection drug users, sexual behavior is a strong predictor of new HIV infections over time, [3, 9].

Homeless men in the U.S. represent a large and growing population. Between 2.3 and 3.5 million people per year, mostly men, experience homelessness in the U.S. [10], and the economic recession of the last several years is steadily forcing more men out of housing and onto the street [11]. In order to reduce HIV infection rates among homeless men and their partners, including homeless women – who not surprisingly also exhibit elevated HIV risk due to unprotected heterosexual intercourse [12, 13] - it is important to understand the determinants of sexual risk behavior among homeless men.

A critical component of sexual decision-making involves judging the riskiness of sexual partners, a topic that has seen both qualitative and quantitative research among diverse populations. Watson and Bell [14] found that trust and the drive for intimacy underlie women’s decisions to forego safe sex, both among high SES middle aged women and younger women living in inner-city areas. Notably, sheltered homeless women living in Skid Row expressed similar tendencies to forego protection use due to feelings of intimacy and warmth, even for male partners they knew were not monogamous [15]. Meanwhile, Masaro et al. [16] found that visitors to a sexually transmitted infection (STI) clinic (both male and female) used trust, but also the social reputation and physical and socioeconomic characteristics of partners when judging whether to use condoms. In a sample visiting an STI clinic (predominantly male) in Southern California, Hoffman and Cohen [17] found that individuals used appearance and cleanliness, social reputation, and the location or venue of meeting to judge the riskiness of casual sexual partners.

Consistent with normative decision theory [18], these findings indicate that risky sexual decisions are motivated by the expectation of emotional and physical benefits, which makes attractive those heuristics that are easily available (e.g., a sense of trust, social reputation, appearance) for engaging in potentially unsafe sexual encounters. This may be especially pertinent among lower SES men, whose opportunities for sexual encounters are particularly restricted [19]. Homeless men represent the extreme low end of the SES continuum, creating unique considerations for the intersection of masculinity and social class [19, 20], and are at elevated risk for receiving and transmitting STIs [3, 4]. This begs the question of how homeless men judge partner risk, and whether their heuristics are similar to those found in other populations.

While previous research has used scales to examine general perceived STI risk among homeless men [21], or has focused on perceptions of STI transmission to homeless men’s partners [22], we know of no study that examines homeless men’s perceptions of sexual risk from a qualitative standpoint, nor of any research that examines the perceived riskiness of sexual partners among homeless men. We used the existing literature on partner risk perception, as well as drawing upon theories of gender and class and normative decision theory, to design a qualitative investigation of perceived sexual partner risk among homeless men living on Skid Row. In particular, Bourgois’s theory of “lumpen masculinity” [19] - which details how social and economic marginality accentuates male violent, risk-taking and abusive behavior via role frustration - prompted us to train interviewers to be attentive to how partner risk perception might be affected by a drive to realize traditional male roles and associated misogynistic views of women.

When developing research with understudied populations or regarding a phenomenon that is not well understood, qualitative interviewing can be particularly useful. Specifically, event-based interviews that obtain narrative descriptions of risk perception and decision-making [23] can aid the development of more structured questionnaires and surveys to investigate the determinants of health risk behaviors. This is a strategy our research group has followed with sexual decision-making among homeless women [15]. To address the gap in understanding homeless men’s perceptions of sexual risk, we conducted exploratory qualitative interviews with men using shelters and meal lines in downtown Los Angeles (Skid Row), focusing on risk perceptions and decision-making in sexual encounters and relationships with women. The goal of this study was to elicit narratives about recent sexual encounters and to analyze these narratives for common patterns and themes regarding sexual risk perception and decision-making.

METHODS

Recruitment

We used stratified random sampling to recruit participants from meal lines and shelters in the Skid Row/Downtown Los Angeles area. Approximately 4,000 homeless individuals are estimated to live in Skid Row, some 60% of them adult males [24, 25]. The Shelter Partnership directory of services for homeless people indicates that this area includes 15 service provision sites representing 25 total eligible shelter programs [26], ranging in bed capacity from 15 to 400. The Shelter Partnership directory also indicated that the Skid Row area contains eight meal programs that together serve 17 meals per day (each meal program serves between one and three meals per day). Across these eight programs, six breakfasts, five lunches, and six dinners are served daily. The number of meals served daily by these programs ranges from 40 for the smallest to 3300 for the largest program.

We used these figures to stratify shelter sites by size: small (less than 37 men served), medium (between 37 and 57 men served) and large (more than 57 men served). We then randomly selected one shelter from each stratum. Similarly we stratified the meal lines by the type of meal served: breakfast, lunch and dinner and randomly selected one breakfast, one lunch and one dinner. In sum, we selected a total of six sites: three shelters and three meal lines. We then allocated the target of 30 men equally across the six sites.

We aimed to obtain five completed interviews per selected site, which required multiple visits to each selected site. Men were randomly selected both from shelters and meal lines via a table of random numbers. For shelters, the table was applied to the bed list after known ineligibles had been excluded. For meal lines, men were randomly selected depending on their position in line. Eligible participants were men 18 years old or older, able to complete an interview in English, who had been sexually active with a woman in the past 6 months and had experienced some homelessness (“stayed at least one night in a place like a shelter, a public or abandoned building, or a voucher hotel, a vehicle, or outdoors” due to a lack of other lodging options) in the past 12 months.

All the selected sites agreed to participate. However, the small shelter selected had only six men residing in the facility (even though the site has 15 beds available for men). Five of the six men refused to be screened and one man was not eligible for the study due to not being sexually active with a woman in the previous six months. Because of this, we decided to recruit a larger number of respondents from the meal lines.

Sampling & Participants

We sampled 77 men to achieve 30 complete interviews. Two men were ineligible due to cognitive impairment. Of the remaining 75 men, 18 refused to be screened, yielding a screening refusal rate of 24%. Of the 57 men who were screened, two men were ineligible as they were unable to converse in English and 19 were ineligible because they were not sexually active with a woman in the previous six months, yielding an overall eligibility rate of 63%. Of the 36 men who screened eligible for the study, four refused to participate for a refusal rate of 11%. Two interviews were broken off midstream (one due to respondent’s psychological state and one due to a scheduling conflict). Thus, 30 men completed the interview in its entirety for a completion rate of 83%.

Of the 30 participants, 23 (77%) self-identified as Black, four self-identified as mixed race, one self-identified as Asian or Pacific-Islander, and four participants indicated that they were Hispanic. Men ranged in age from 22 to 54, with a median of 44 and mean of 43.7 ± 1.5 [SE]. Men who were ineligible or refused the interview did not differ in age (mean = 45.8 ± 3.3) from participants. The average time homeless across respondents’ lifespan was 4.7 years ± .7 [SE] (median = 4), with a minimum of two weeks and a maximum of 16 years.

Semi-structured Interviewing Guide

The content of the semi-structured interviewing guide (SSIG) was based on the existing literature on perceived partner risk [14, 16, 17], conversations with Skid Row service providers with extensive experience working with homeless men, and our team’s field research regarding interactions with male partners and sexual decision-making among women living on Skid Row [15]. It was designed to elicit the risk perceptions and sexual decision-making heuristics of homeless men by using narratives of sexual events, and focused on eliciting descriptions of two of the respondents’ recent sexual encounters. First, we asked men to describe their most recent sexual encounter with a woman that involved vaginal or anal sex. We then asked men to describe a sexual encounter that contrasted in some way to this event – either in condom use (e.g., condom use in first event, no condom use in second), substance use, or a different partner – to obtain information about heuristics used in a different decision-making context.

Qualitative interviews followed a matrix-based SSIG. This allows interviewers to cover a list of topics and sub-topics according to two intersecting dimensions. These dimensions are represented by an X-axis and a Y-axis, and intersections between X- and Y-axis topic headings produce suggested probes for use during interviewing. In this case, the X-axis represented different stages of the sexual event - the background/pre-existing relationship with partner, the time immediately before intimacy, intimacy, and post-intimacy. The Y-axis represented different aspects of the sexual event - the social context, their and their partners’ behaviors, their thoughts and decisions, their emotions and feelings, alcohol or drug use, and condom use. In addition, respondents were asked to compare the encounter with other sexual encounters. Intersections of topic headings along the X- and Y-axes produced specific interview probes. For example, the intersection of “Thoughts and Decisions” and “Pre-intimacy” yielded questions such as “How did you decide to have sex?” (see electronic supplemental material).

Interviewers were trained to let the participants guide the order of topics, using the SSIG only to make sure that they had covered the full time course of the sexual event as well as all of the topics of concern. Thus, participants were allowed to skip around the matrix as he narrated the sexual event, and interviewers were allowed to follow up on volunteered information in the order that it naturally arose in the conversation. This format allows qualitative interviews to follow a natural conversational flow, but also provides a mechanism by which interviewers can check off topics as they are covered so as not to miss crucial portions of the interview. Our research group has used this technique successfully in qualitative interviews regarding the sexual behavior of homeless women [15]. Interviews lasted approximately 60–90 minutes, and were audio recorded. All participants knew they were being audio recorded, consented to the interview after reading and listening to a description of the study and its risks and benefits, and were given $25 in cash for participation. Human subject protections and data safeguarding procedures were approved by the RAND and University of Southern California Human Subject Protection Committees.

Qualitative analysis

All interviews were fully transcribed to facilitate qualitative analysis. Interviewers checked each transcript for quality and completeness, correcting or filling in missing portions when possible.

A team of three coders was involved in the qualitative data coding. First, the lead and second authors engaged in a process of “open coding,” sampling only the de-briefing notes initially [27]. This purely inductive process was designed simply to let the “gestalt” of the interview content spontaneously produce codes for later discussion by the group and crystallization into formal codes.

Several rounds of group discussion and coding yielded the formal code “Risk,” pertaining to risk perception and partner selection. Following accepted protocol for rigorous qualitative coding [28], inclusion and exclusion criteria, typical and atypical exemplars, and “close but no” examples were generated and placed in a codebook for use by the coding team (see Table 1).

Table 1.

Risk codebook

Mnemonic	Risk
Short description	Sexual and relationship risk assessment
Detailed description	Ideas, beliefs, or strategies to assess risk from sexual activity, sex partners, or romantic/sexual relationships – includes statements of not caring about risk, as well as comments on others’ perceptions of risks.
Inclusion criteria	Must be ideas or narratives related to assessing and/or responding to the potentialnegative consequences of sex or sexual relationships. Can include any type of risk, including disease, physical danger, financial problems, loss of opportunity, damaged reputation, negative emotional effects, etc.
Exclusion criteria	Assessments of risk not as a result of sex or sexual/romantic relationships.
Typical exemplars	“Take precautions - if she’s a street walker, make sure her feet are clean.” “I try to stay away from serious relationships on the street; the only thing they bring is heartache, because the woman always ends up leaving you.”
Atypical exemplars	“If you date a drug addict, then you have to worry about all her debts and who might be coming after her.” “I had to break it off because homeless men shouldn’t be in relationships. We have to look out for ourselves.”
Close but no	“It’s dangerous to be a homeless man. You can’t trust anyone on the street.” “I wore a condom that time. She brought the condom, but I knew she expected me to put it on.”

Two of the coders then read all of the interview transcripts (separately, in parallel) and applied the Risk code. All of the text coded by each interviewer was then systematically compared line-by-line to examine inter-coder reliability. A total of 75% of the text coded by each coder in parallel was completely overlapping. Coders then met to arbitrate differences in coding and converge on a final, agreed-upon set of coded text. This resulted in excluding or including portions of non-overlapping text from each coder.

The results contained in this study all come from the Risk code, which was further sub-coded by the lead author. Sub-codes pertained to specific risky outcomes perceived by homeless men on Skid Row, as well as specific heuristics used by these men when choosing partners to minimize such risks. We then summarized sub-coded text into bullet points with quotations, reporting the frequency of each sub-code by number of respondents. Initial results were discussed with the entire research team, and sub-codes modified or added according to group input. A final research report was generated, including all themes and sub-themes, as well as examples of content from each.

To examine the characteristics of sexual events, descriptions of sexual events were coded using an Excel matrix. This matrix included a row for each of the two sexual events described by each respondent, as well as columns including brief qualitative descriptions (e.g., “Characteristic of relationship with partner”). Using these qualitative descriptions, columns indicating dichotomous variables were created (e.g., “Explicit trade for sexual encounter” or “Ex-partner (used to be monogamous but now not”).

RESULTS

Sexual events

As context for the discussions of sexual risk and partner selection, we provide details on the sexual events that homeless men described. Overall, our 30 respondents reported 58 sexual events (two men could only recall having had sex with a woman once in the last 12 months). Twenty-six of these events (less than 40%) involved condom use, and close to 20% (11 total events) involved the explicit trade of money, drugs, or other goods for sex.

Five men reported sexual events with monogamous partners (a total of 10 sex events overall). Four men reported having sex with a formerly monogamous “ex-” (with whom they were no longer monogamous). Fifteen men reported having sex with casual partners (totaling 20 sex events overall) – either long-term non-monogamous partners, short-term casual partners, or friends (partners with whom friendship was described as primary and sex secondary or occasional). Seven men (11 total sexual events) reported encounters with sex workers, in which there was an explicit trade of money, drugs, or other resources in return for sex.

Awareness of STI risk & STI testing

Not surprisingly, men were very aware of the STI risks involved in sexual encounters; 27 out of 30 men mentioned STIs as possible risks for sexual encounters. Many men (18 out of 30) also mentioned getting a woman pregnant as a negative potential outcome of sexual encounters that was to be avoided through the use of condoms or other methods. These concerns were often bundled together in the same (or nearby) statements; for example, “…with one-night stands it’s like - with all these little viruses and diseases and things going around - ‘Are you on the pill? You’re not HIV positive?’ You’ve got protection.”

Some men described how the one clear cue for a low STI risk partner was to obtain the results of a diagnostic test. For example, when describing why he did not use condoms with a particular partner, one respondent explained, “…she come up with the paperwork [test results].…maybe once a year…they look up in her, open her up and swab.”

A total of 18 men out of 30 mentioned being personally tested for HIV at some point within their memory. Considerably fewer knew the HIV status of their sexual partners’ test results. Four men mentioned being tested with their long-term sexual partner; two of these men were periodically tested with their partner, and the other two were tested with their partner at the beginning of the relationship. Three men described their female sexual partners showing them “papers” to prove that they had tested negative. Another three men simply assumed that their partners had been tested and did not have an STI, one because “the place she worked at, she had to go through a lot of physicals and stuff like that.”

Heuristics and strategies for partner selection and risk minimization

When describing partner selection to minimize risks, respondents indicated a variety of heuristicsthat they used to assess the risk of a potential partner – including the partner’s appearance, her reputation, and the location or region in which the sexual partner was first encountered.

Visual and behavioral cues

Homeless men often claimed to use cues about a woman’s appearance or behavior patterns to judge whether she might be infected with an STI (as well as sometimes whether she was just a generally a “risky” person to have sex with). The most common heuristic used by men to judge riskiness was observing the behavior of sex partners (or potential partners); overall, 17 men described using some method of behavioral observation to gauge partner risk. This generally pertained to risk of infection, but also other negative outcomes. Men considered six broad categories of behaviors considered to be indicators of a potentially risky partner: obvious drug use, prostitution, behavioral indicators of mental illness, signs of promiscuity, assorted behavioral clues regarding a woman’s “decency” or respectability, and several indicators of overtly dangerous or predatory behavior (e.g., being a gang member, selling drugs, flirting with men when benefit checks are distributed, etc.).

Obvious indicators of drug use or addiction and actively soliciting men for sex (especially in exchange for resources) were commonly mentioned as warning signs of a high risk partner. The following quote exemplifies these beliefs:

You’ll find women - you might find some that are drunk and just out of their mind and you want to have sex, which had happened to me and the three friends we was with. We just got away from it, walked away from it. ‘Oh, I love you, I want - come on, let’s go down the street.’ And she was drunk out of her mind. And then there’s those that are out there…selling their body for drugs. And then in either situation, that is guaranteed you should use protection and to exchange as less skin contact as possible in some situations, I would say.

Another man asserted that female drug use should cause suspicion about a woman’s assertion that she is STI-free; “Now, she was smoking [crack]; she said she was clean. But, you know, crack addicts are the biggest liars in the world.”

Additionally, men used mental health as an indicator of female partner risk. One man described how, often, homeless women’s “minds are messed up,” so “you’ve got to be careful.” Another respondent stated, “You might even have a woman that’s beautiful and she might be crazier than 5150 [California police code for involuntary psychiatric commitment].”

Men also described using various behavioral cues to attempt to assess a woman’s relative promiscuity (or selectiveness) with men. Surprisingly, homeless men used this heuristic strategy even when judging the riskiness of paid sex partners. In the following case, the respondent had sex with a woman, paid her, and did not use a condom. In his explanation of why he felt safe not using a condom, the key indicator of safety was that his partner was selective in the men she slept with and did not directly solicit money for sex. Rather, she made this request gradually and subtly:

All I’m trying to say is if you [the woman] choose[s] your [her] man, she’s not the street hooker type. She won’t screw everybody. She know you, and she’s pretty attractive…and she get like a rapport with you then she might say, you know, “If you want to do something, I got to get myself some money or I’ve got to pay the house note. My phone bill is due.” Something like that.

Similarly, men often looked for general indicators of “decency” in a female partner, explaining that “I got to know her and her habits” or “Me and her worked together and everything, and she seemed like a pretty decent woman”; both of these reasons were offered as justifying not using a condom in a casual relationship.

Men also used a variety of seemingly idiosyncratic behavioral indicators, from avoiding women who exhibited “switch ass walking” [pronounced hip movement], to avoiding women who seem to go after men opportunistically during the time of the month that benefit checks are disbursed. The following quote exemplifies the broad range of characteristics that men described using to assess a potential partner’s decency and decide whether condom use was necessary. Interestingly, the respondent juxtaposes his partner’s “wildness” with characteristics indicating relative safety.

Yeah, she’s plain, she’s got a good job and all that and she’s an exotic dancer. She’s wild, just like an exotic dancer, she’s wild, but friendly, cool, but she smoked PCP. And you know, she don’t hustle, she don’t sell her body, she don’t sell dope, she kept two jobs. And she’s just like the type of woman you think that would never mess around.

Twelve men claimed that they could use physical characteristics of the woman – simple cues based on looks, but also sometimes feel and smell – to tell whether a she had an STI or not. Men used three categories of cues: general appearance, specific cues, and vaginal cues. Examples of general appearance and specific visual cues (e.g., clean hair or clothes, well kept feet) are presented in Table 2. Two men also asserted that they could assess the disease risk of a partner through vaginal cues such as look, feel, and smell (which could reasonably apply to some STI’s but not HIV).

Table 2.

Visual cues used by men to judge partner risk

General appearance	Specific cues (indicators of risk in italics)
“attractive” “looks clean” “cute and chubby” “looks like the type of girl that you’d love to take home to meet your mama” “her presentation”	“My father told me if a girl’s hair ain’t combed, it means she’s not real clean.” “her clothes were clean” “dirty feet”/Has to “take care of her feet” [2 different men] “…a woman that covers up. I don’t need you to have no Daisy Dukes on or a miniskirt up here.” presence of visible sores

Reputation and location

As well as direct behavioral observations, men also described accessing shared knowledge from their social networks concerning a woman’s behavioral history and overall reputation. Eleven men described using such cues to judge partner risk. One respondent described the general importance of shared social knowledge regarding potential sex partners:

Because homeless people, you’re out in the street, you see more; you’re out in the street. And see, now, if I lived in an apartment complex or in the projects or something like that, there would be more of by mouth of what girl is doing what in an apartment complex or in the projects.

Such knowledge obtained from others was used to make decisions about safety and whether to use a condom or not. For example, one respondent indicated that, “I could feel safe around her because I know she doesn’t screw around. She’s considered a good girl. She doesn’t screw around, she’s not a whore.” Another respondent described how his partner’s “reputation wasn’t inconsistent with anything of what I asked. You know, she wasn’t a person that has been known for [sexual] activity in that area.”

Another common strategy for judging partner risk concerned the place in which the respondent originally encountered his female partner. Twelve men mentioned the location, place, or region in which they met their sexual or romantic partner as indicative of the kind of risk they would face from this partner.

Women living “Downtown” (the Skid Row area and surroundings) were seen as likely to be carrying disease or to put men at risk in other ways. Being from “Downtown” seemed to function as an indicator of destitution, homelessness, and general risk. After describing a casual sexual encounter that occurred outside of the Los Angeles area (without condom use), one respondent described how he would never do this in the local area; “Now picking up one of these things on the street, of course not, you’re not going to mess with them…These women mess with everything around here. You’d be foolish.” Interestingly, one man described being initially rejected by a woman because she assumed that he had been having sex with women from downtown Los Angeles. He reported her saying to him, “Ah, you’re running around here with all them tramps, all that nasty stuff Downtown. I ain’t giving you nothing [no sex].”

Several men talked about the type of establishment as a determining factor in partner risk (i.e., church, upscale bar, nightclub). Two men mentioned meeting women in church as safer, but both of them also doubted that this was truly a good heuristic; “…you’ve got some church women that do what they want to do, they still have another man on the side. You find that too.” Two men also described using the “classiness” of a location to determine risk. For example, one man compared the relative danger of a “nightclub” with the perceived safety of an “upscale bar.” “If you meet a woman in a nightclub [after hours dance club], most guys are going to use protection. If you meet a woman in an upscale bar, a lot of the guys wouldn’t.” Similarly, one man claimed that the most important diagnostic indicator of the risk presented by a woman was “…the environment. You know, if a guy meets a woman, for example, in a bar, he may or may not use protection.

” Respondents also used regional geography to assess partner risk. Three men perceived there to be HIV “safe zones” – in these cases, rural areas, one’s hometown, and the entire state of California. One man perceived less metropolitan areas to be safe; “…in that particular area, I don’t think it was a high risk for people with-they can contract that virus like in the metropolitan cities. It wasn’t a high risk area for those cases.” Another respondent perceived the entire state of California to be a relative “safe zone” for HIV; “I’m just saying statewide as far as our state goes, it’s [HIV] at an all time low, especially here in California.” And another was assured because his partner came from his home town (outside Los Angeles). Expounding on this, he said:

Because the rest of the women be like, I meet on the street, walk up and down here, or I meet at a dope house, or I meet at the liquor store, or I meet at a club or something. Some of them are used to doing this, but [name of partner] - Because when she was a cheerleader, I used to sit right at the front row, she was cute. But with her I didn’t use a rubber. I trusted her [emphasis added].

Trust and relationship status

Sixteen men mentioned trust and other characteristics of relationships, including the perceived status of relationships as “serious” – as indicating low partner risk. Trust was specifically mentioned by seven different men as a reason for foregoing condom use. In one respondent’s words, “I kind of trust [partner name], you know. Like you can feel it, you know? ….with her I didn’t use a rubber. I trusted her.” Men also described cessation in condom use as a natural progression as partners got to know each other better. As the relationship progresses, “condoms come off. Because then I trust her, and she’s not messing around, and I know I’m not messing around.” Indicating reasons why he felt comfortable not using a condom, one man involved in a casual relationship with a woman from work described:

See, I was able to get to be the closest one to get to her. And, I feel as though sometimes she would tell me, too; I don’t think she would lie to me and stuff. And, you know, we talked somewhat in confidence, you know. And that’s what she liked about me; we could talk.

After spending a considerable length of time in prison separated from his girlfriend, another man indicated that he did not use a condom with this girlfriend after getting out of prison because “she still gave me her trust and love that there was no one else.” In several cases, men mentioned that they would not use condoms in any relationship that they considered “serious.”

Respondents also described cases in which existing rapport or trust formed barriers for honest discussions about STI risk. One man was estranged from his wife, but would still sleep with her on occasion (he was sexually active outside this relationship, including casual encounters). He described how he did not discuss STI/HIV status nor use condoms with his estranged wife, because the conversation would be too uncomfortable: “Yeah, she’ll try to put it off on me and like I really, really super get upset. So I kind of like stay away from that [conversation], but I realize in my heart that this isn’t the thing to do.” Another described how conversations about HIV status were off limits in a “serious” relationship: “…with all the faithfulness and the total commitment to each other, that would be the furthest thing from each other’s mind to think that they wandered off….that means you’re not trusting me, if you’re going to ask me some shit like that.”

In contrast, one respondent described how his female partner attempted to use the trust-based decision rule, but he rejected this logic. He reported a conversation in which she said “Oh, I’ve known you for years, it [condom use] don’t matter.” The respondent thought this was a bad heuristic to use; “I’m like, ‘No, you know, I’m married and I know my wife is cheating and I don’t know what she gave me, you don’t know what I’m doing, and I know you’re not really sexually active.’” In this case, the respondent insisted on using a condom, out of concern for her safety. Still another respondent rejected the heuristic of trust in general; “…trust builds and you’re going to take it off and have sex, which is still actually a poor judgment.”

Folk beliefs

Five respondents described idiosyncratic beliefs or practices to assess partner STI risk that were inconsistent with common medical knowledge. One man described how he offset the risk of an unprotected casual encounter by “washing up.” Another man expressed the belief that there was a “cure for AIDS,” based on the fact that Magic Johnson was still alive. Still another respondent believed that he had immunity to STIs, given that he had not been diagnosed with anything sexually transmitted for a long time; “I ain’t caught nothing since 2003….So I’m feeling pretty confident, I ain’t caught nothing. Maybe I’m immune.”

One respondent indicated that he believed “pulling out” and having an orgasm in his partner’s mouth was an effective HIV prevention technique. Another respondent indicated what he saw as a failsafe way for “home testing” of a woman’s infectious status; “Take your finger and take a little earwax out of your ear and stick it in the woman’s vagina. If she hollers or she burns and has an unpleasant reaction to that, you don’t want to go up in her without a condom.”

Other perceived risks of romantic and sexual relationships

When describing sexual and romantic encounters and relationships, men described trying to minimize multiple personal risks that could result from engaging in such encounters and relationships while living on the street. Listed from highest (mentioned by the most participants) to lowest frequency, these concerns were: contracting STIs, making a woman pregnant, emotional harm, remaining homeless or addicted to drugs, losing resources, getting into trouble with law enforcement, and the risk of physical harm.

Besides STIs and unwanted pregnancies, the next most common risk mentioned by participants was the possibility of emotional harm or damage to themselves; 15 men (50%) related narratives of emotional harm they had incurred in sexual or romantic relationships or of potential emotional harm and specific attempts to avoid this harm. Men were particularly concerned about infidelity and abandonment, and in some cases linked such risks specifically to their being homeless. For example, “Now, they might even break up with you if you be homeless. It’s just that stigma about being homeless, unless that woman…really, really have love for you.” Six men expressed not being able to trust women, or that women were inherently manipulative. For example, “They will find your weak spot. Once they find your weak spot, they dominate….And once they defeat you, they leave you.”

Thirteen men mentioned that relationships with women carried the risk of prolonging negative aspects of their current situation – either by accentuating or causing relapses in substance use, or through the mutual reinforcement of other behaviors associated with homelessness. For example, when asked about relationships on the street, one participant stated, “…when you start getting into a relationship, and you start laying in boxes and all the other stuff with a woman…you created it [the situation] yourself….Yeah, so you got to be careful who you’re getting with. Similarly, “So if you can start a relationship with a woman in here [homeless shelter], yeah, that’s cool but I really wouldn’t suggest it because if you’re in a drug program and one of you relapses, her feelings or his feelings are caring for you, but he may end up relapsing too, you know what I mean?”

Nine men mentioned the loss of money or other resources as a risk inherent in sexual or romantic relationships. Two men described being robbed or tricked by sex workers, while a third expressed his fears of being robbed by a casual sex partner that he met at a bar. The other six men focused on the general loss of resources that can be incurred when one invests in a relationship; in one case, a participant described how certain women in Skid Row would intentionally establish brief relationships with men at the time that government benefit checks were distributed, in order to extract these resources from homeless men.

Five men out of 30 were concerned that their relationships with women could get (or had already gotten them) into trouble with law enforcement. Two men claimed that female behavior tended to incite them to violence, which in turn caused legal troubles for them. An additional two men were simply concerned that police would catch them on the street during sexual activity with female partners. Finally, one man blamed missing his court date on being distracted by an ongoing relationship with a woman.

Men also mentioned physical risks to their safety as a potential risk of engaging in sexual or romantic relationships with women; four men discussed such physical risk. In two cases, these potential risks were described as crossing a jealous lover of their partner. In two other cases, participants mentioned that a woman might be physically dangerous herself. One participant mentioned being “cut in the face” by a woman who suspected him of cheating, and another participant was generally wary of partnering with violent women: “…they have female gang members too; a lot of women that I see around here that will sell drugs and carry weapons, and stuff like that; like the woman who shot her mother in the head.”

DISCUSSION

Homeless men are a growing population at high risk for HIV, and yet the existing knowledge base regarding homeless men’s perceptions of STI risk is sparse. Based on our review of the literature, this is the first study to examine homeless men’s perception of risks posed by specific female partners and the heuristics and strategies that homeless men use when making decisions about sexual activity with women. In comparison with characterizations of partner risk perception in other populations, homeless men showed some striking similarities but also unique patterns suggesting possibilities for intervention.

Homeless men perceived a wide variety of risks from sexual encounters and relationships, including STI risk but also emotional harm, legal consequences, and even mortal danger. Perhaps the most unexpected pattern with respect to perceived risks of relationships was the emphasis that homeless men placed on emotional vulnerability. Data analysis revealed that many homeless men on Skid Row felt emotionally scarred and vulnerable from past relationships with women, often causing them to shy away from long-term partnerships. Men often tied narratives of loss and fear of abandonment explicitly to being homeless, expressing the belief (credibly) that most women would have no interest in a relationship with a homeless man.

To mitigate emotional, physical, and other risks, homeless men described using multiple heuristics for partner selection, focusing on a diverse set of characteristics that men believed were associated with higher- or lower-risk female partners. These included a range of strategies, from those that were highly likely to be successful (e.g., viewing a woman’s current STI test results) to those with little or no connection to actual risk (e.g., perceiving a woman who “switches her hips” as dangerous or a woman who is “chubby and cute” as safe), to those based on outright misinformation (e.g., telling a woman’s STI status by the feel of her vagina).

Such “folk beliefs” concerning medically inaccurate beliefs about ascertaining a woman’s STI status were especially striking. Interestingly, one described strategy, the “earwax test” was documented by researchers working with high risk populations in Southern California more than ten years before this study [17]. This begs the question of whether some heuristics and strategies for judging partner risk become local cultural knowledge that is transmitted not because of its effectiveness in offsetting risk, but because they require few resources and allow men to feel safe and assured about their partners in high risk environments. Similarly, one of the more dangerous beliefs espoused by men on Skid Row was that women from certain regions were safe or “clean,” a belief that researchers also noticed in this area more than a decade prior to this study [17].

It might be easy at first pass to assume homeless men’s risk perception and partner selection heuristics are idiosyncratic or extreme. However, a wide variety of other populations also appear to use cues about a partner’s appearance, such as physical attractiveness and a “clean” look, to judge partner riskiness [16]. More strikingly, both homeless men and high SES women in their 40s and 50s [14], as well college students, adolescents, and others [16, 29], report using intuitive feelings of trust and personal perceptions of relationships seriousness when judging whether or not to use a condom. Not surprisingly, homeless women on Skid Row describe also similar processes of risk-taking linked with lingering feelings of warmth or attachment from ex-husbands and ex-boyfriends [15].

Such emotional indicators of when to cease condom use might have some limited utility in certain circumstances. However, respondents described several circumstances in which such processes would increase HIV risk, such as feelings of trust based on very limited (or even first time) encounters, as well as situations in which lingering feelings of trust and intimacy inhibited frank discussions about sexual activity outside the relationship. This latter situation could be especially risky in ongoing relationships with ex-wives and ex-girlfriends, in which partners continue to avoid condom use despite the fact that such relationships are no longer monogamous. The reliance on trust for risk judgment – as well as the fact that homeless men often reported feeling trust in short relationships or one-time encounters – underscores the way in which homeless men and women’s emotional vulnerability can also place them at physical risk.

Among homeless men, heuristics for partner selection are likely tied to high risk sexual behavior both due to misinformation and through some degree of post-hoc rationalization. Homeless men and their partners face a range of structural constraints on establishing stable relationships, not the least of which is the lack of privacy or any shared dwelling space. As a result, opportunistic or short-term encounters are often the only option available to homeless men. While many may be misinformed of the risks of unprotected sexual activity or having multiple or concurrent partners, even those who are well informed may resort to flexible heuristics to judge partner safety when the drive for intimacy or sexual contact is high. As a result, intervention efforts should be attuned both to the provision of accurate information about HIV risk, and also to providing possibilities and resources that make stable relationships real possibilities for homeless men. Additionally, as only 18 out of 30 men mentioned being tested for HIV (but 27 out of 30 mentioned it as a concern), efforts that encourage and provide access to HIV testing should also be supported.

LIMITATIONS

Despite the contribution this study makes to better understanding perceptions of risk of heterosexual sex among men facing severe structural obstacles, there are several limitations to our findings. A sizable minority of the homeless male population desists from sexual activity with women altogether [19]. The men we encountered on Skid Row seem to be no different; 35% of those who we sampled screened out of the interview because they had not had sex with a woman in the past 6 months. Further research should focus on homeless men’s reasons for desisting from sex with women. Both our focus on men who had been recently sexually active, as well as those who are living on Skid Row, may limit the generalizability of our results. Additionally, our sampling frame excluded men who do not visit shelters and meal lines, but previous research suggests that such men are a small minority (less than 15%) in Skid Row [30].

Furthermore, despite our carefully designed sampling strategy which was intended to maximize diversity in our sample, the relatively small sample size (n=30) cautions against generalizing too broadly or strongly about the patterns observed in this study. Another limitation is that the semi-structured nature of qualitative interviewing means that questions were not asked in precisely the same way of all men. Systematic, structured surveys with a larger sample are better designed for strong population-level inferences.

Finally, it is not possible to determine whether the heuristics men described were actively used during partner selection or functioned more as post-hoc rationales used to either second guess or feel better about engaging in unprotected sex. With this caveat in mind, it is still useful to examine the statements made by men “after the fact,” as research on risk perception shows that post-hoc rationalizations can become predictors of future engagement in risk behavior [31].

CONCLUSIONS

Many of the heuristics concerning perceived partner risk were idiosyncratic or medically inaccurate (for example, the belief that one could assess the HIV status of a potential partner by examining the condition of her feet). Given that so many men expressed these sorts of beliefs without specific prompting, and that this study went to great lengths to assure random sampling, we can be reasonably confident in making the assumption that a sizable proportion of the homeless men in downtown Los Angeles hold such medically inaccurate “folk beliefs.” These would seem rife for correction via evidence-based interventions designed to elicit and softly counter or insert doubts around such belief systems, such as motivational interviewing [32, 33]. Additionally, given that both homeless men and homeless women [15] described engaging in risk behaviors driven by feelings of trust and intimacy or lingering attachment to ex-partners, future interventions could be designed to point out the potential danger of such emotional dynamics in sexual decision-making.

Stable, long-term monogamous relationships present one common path to STI risk reduction. Thus, one can interpret at least some of the risk behaviors engaged in by homeless men as linked to the social and emotional turmoil imposed by homeless men’s intense marginalization, stigmatization and relative absence of resources [34], which together make them often unmarketable as long-term partners. Given the multiple barriers to successful, stable relationships expressed by homeless men, it is worth considering that intervention efforts might do well to increase homeless men’s emotional and material capacity to engage in stable relationships with women. Indeed, existing research shows that marriage among homeless couples is linked with lower HIV risk behaviors [35]. Supporting relationships among homeless men would involve some departure from current housing strategies, which generally provide space for single men, single women, or women with children. A growing body of research has shown that lack of stable housing is significantly associated with HIV risk [13, 36, 37]. Our study lends support to this body of research and extends it by indicating that housing assistance must be provided in a manner that supports relationships.