Over the prior decade, structural change efforts have become an important component of community-based HIV prevention initiatives, as well as initiatives designed to address other health problems (Frieden 2010; Marmot 2005). Structural changes alter the social, economic, political, and environmental conditions that give rise to the risk of exposure to a disease such as HIV among vulnerable individuals and populations (Blankenship et al. 2000, 2006; Frieden 2010; Gupta et al. 2008; Sumartojo 2000). The vulnerability of those who are at risk for HIV may be reduced by changing features of the environment in which they live. As noted by Gupta et al. (2008), determining the root causes of HIV risk and the meaningful pursuit of structural change is an inherently local enterprise, as context informs the relevance of the approach and targets of change.

Gupta et al. (2008) note that structural factors are far-reaching, diffuse, and perceived as beyond the conventional limits of public health, which is often a barrier to their pursuit. Structural change may be particularly difficult to attain when the planned changes encounter political resistance or invoke conflicting values about what serves the larger social good (Blankenship et al. 2000, 2006). Entrenched political ideologies and socio-cultural values constrain the realm of possible structural changes and the malleability of the larger environment to change (Sabatier and Jenkins-Smith 1999). Many structural change interventions, particularly those addressing macro-systemic drivers of risk such as social and economic inequality, “require fundamental societal transformation” that is “often controversial” (Frieden 2010, p. 594). These changes are often of the second-order (Watzlawick et al. 1974) and therefore may lack support from the public and those who control resources and decision-making processes. Second-order changes often threaten vested political and economic interests because of their emphasis on addressing root causes and changing the status quo; politically or socially contentious second-order changes often meet with implementation barriers (Bartunek and Moch 1987). Like many systems change initiatives, structural change approaches “must deal with issues of conflicting values,” making their implementation a major challenge even if there is evidence of their effectiveness (McLeroy et al. 1988, p. 353). As Frieden (2010) suggests, it is not often a lack of money that makes social and policy change difficult or impossible, but a lack of “political will.”

Within the context of HIV infection, stigma is both a root structural cause and a source of resistance to changing its course (Herek 1999; Holtgrave et al. 2007; Mahajan et al. 2008; Parker and Aggleton 2003). Stigma generally refers to attributes of individuals that taint or discredit them socially (Goffman 1963). Social stigma may result from physical attributes, character traits or behaviors that are perceived as disordered or contra-normative, or from membership in a devalued social group. According to Link and Phelan (2001), stigma occurs when the differences among persons or groups are labeled and salient, rather than overlooked, and negative associations are attached to these differences. These negative associations produce a stereotyped view of the persons who possess these salient differences and lead to a set of social processes in which those people are held apart or separate, suffer loss of social status, and become subject to discrimination.

Structural changes may be more difficult to attain if they concern populations that are either highly socially stigmatized or engage in stigmatized behaviors. Needle exchange provides a salient example of an evidence-supported structural change initiative that has not been fully implemented because it provokes divisive political and community debate regarding whether or not access to clean syringes is in the public good (Rockwell et al. 1999; Tempalski et al. 2007). To some people, promoting needle exchange is akin to promoting injection drug use and normalizing a contra-normative behavior. Similarly, providing free access to condoms in high schools is akin to promoting sexual activity among youth. Parker and Aggleton (2003) argue that stigmas such as those associated with injection drug use and sexuality strengthen the inequalities that feed the HIV epidemic. For instance, levels of religiosity impacts the level of investment in HIV prevention targeting gay and bisexual men (Rosser and Horvath 2008). As religiosity in a community increases, targeted prevention funding for gay and bisexual men in that community decreases. As these examples suggest, stigmatizing community conditions may impede structural change, and ultimately HIV prevention efforts.

The potential negative impact of stigma on the process and success of pursuing structural change initiatives merits further investigation, since stigmatized groups and those engaged in stigmatized behaviors are probable beneficiaries of these initiatives. Parker and Aggleton (2003) underscore the need for research to clarify strategies that lead to success in responding to stigma and on community mobilization efforts to attempt to change structures that produce and reinforce stigma. The current study sought to examine the impact of stigma on the ability of community coalitions to achieve HIV prevention structural change objectives on behalf of high-risk adolescents. Coalitions are a commonplace approach to the pursuit of health-related structural change (Mizrahi and Rosenthal 2001; Roussos and Fawcett 2000; Watson-Thompson et al. 2008), though are often used in communities as planning and coordinating bodies rather than as grassroots mobilization efforts and agents of social change (Butterfoss 2006; Roussos and Fawcett 2000; Valente et al. 2007). Coalitions are temporary or enduring collaborations among diverse individuals, organizations, and constituents who agree to work jointly toward a common goal (Butterfoss 2006). As a vehicle for social activism, coalitions are asserted to be a promising means to challenge the existing social order precisely because they create alliances across diverse actors, pool voluntary effort, and via their collective effort, legitimate the outcomes which they pursue (Well et al. 2004). Most of the empirical evidence on coalitions’ ability to function over time and accomplish objectives has focused on features related to their internal functioning, development, and composition (see, for example, Cheadle et al. 2005; Collie-Akers, et al. 2007; Crowley et al. 2000; Florin et al. 2000). The literature on coalitions has paid scant attention to external contextual factors that may bear on the ability of coalitions to achieve objectives, such as local opposition and stigma, despite theoretical literature in areas such as social movement scholarship (e.g., Piven and Cloward 1979) implicating political and community openness as critical to the success of social change agents, including coalitions.

The Current Study

The current study uses data from the Connect-to-Protect (C2P) initiative of the Adolescent Trials Network (ATN) to explore the role of stigma on coalitions’ ability to achieve structural change objectives. C2P is an ongoing multi-site trial funded by the National Institutes of Health in which local coalitions were created and mobilized in 2006 to determine and pursue structural change objectives that would impact local adolescents’ risk of exposure to HIV (for more detail, see Straub et al. 2007; Ziff et al. 2006). Coalitions that remain part of the C2P initiative are currently entering their sixth year of operation. Most youth targeted by C2P are persons of color and all reside in low-income neighborhoods with high rates of sexually transmitted infections among adolescents. The C2P coalitions focus on adolescent women, gay and bisexual male youth, or injection drug using youth. These differing adolescent populations of focus provide a natural opportunity to explore the impact of seeking to benefit groups that vary in their degree of social stigmatization on structural change efforts. A solid body of evidence supports that gay and bisexual teens experience higher levels of distress, bullying and harassment relative to their heterosexual counterparts, experiences which are attributable to stigma associated with their same-sex attractions and sexual minority identities (Almeida et al. 2009; Balsam et al. 2005; Berlan et al. 2010; D’Augelli et al. 2002; Rosario et al. 2002; Russell et al. 2001; Taywaditep 2001; Williams et al. 2003). Emerging evidence suggests that injection drug using youth also experience severe stigma related to their behavior (Flom et al. 2001; Henkel et al. 2008). For the purpose of the present study, 13 sites’ coalitions were examined. The study was guided by two questions: Does working to achieve structural change on behalf of differentially stigmatized groups affect success? How does stigma affect the process of accomplishing structural change outcomes?

All C2P coalitions follow a national protocol for community mobilization that includes conducting an environmental scan, analyzing root causes of risk for adolescents in a defined geographic area, developing a logic model depicting local causes of risk, and formulating strategic plans and structural change objectives that are linked to the locally identified root causes of risk (Community Tool Box 2003; Fawcett et al. 2000a, b; Francisco et al. 2000). The coalitions’ action plans are dynamic and revised in an ongoing fashion to reflect the objectives that have been achieved, emergent objectives, new opportunities, and challenges encountered. Coalitions are encouraged to bring new actors into their coalitions as the need to develop new relationships becomes clear. Thus, the size of each coalition has varied over time and across sites, as has the number of objectives any one coalition has pursued. Over the lifespan of these coalitions, a total of 1,448 local actors have been engaged in their efforts; 214 objectives have been completed. Table 1 displays examples of the kinds of objectives that the coalitions have established. During the time period for the current study, the coalitions were staffed by two fulltime employees of the local adolescent medical trial units that compose the ATN. Coalitions were supervised by a national coordinating center staffed by five people who provided ongoing technical assistance, monitored coalition progress, and also kept independent documentation of coalition functioning and health.

Table 1 Examples of structural change objectives
Full size table

In addition to following a common protocol to coalition formation and the implementation of activities, coalitions use a common standardized monitoring system for reporting on the formulation of objectives, action steps toward the completion of objectives, obstacles encountered, strategic decisions, and coalition achievements and failures. For example, for every objective a coalition elects to pursue, an action planning worksheet is completed which details what strategies the coalition intends to employ to achieve the objective, what resources it may need to garner, and what new relationships it may need to develop in order to complete the objective. Each of the elements of this planning sheet provides a basis for subsequent reporting on the objective. Coalitions routinely complete logs documenting their progress toward completing objectives, any changes made to objectives, and the rationale for changes, including decisions to abandon an objective. Coalition members are also interviewed every 6 months on their perception of their coalition’s functioning. All of the data used in the current study are from the various coalition reports and monitoring documents created by the national coordinating center personnel assigned to oversee each coalition.

Methods

The current study focuses on the operations of the 13 coalitions from their inception in 2006 through the end of 2008. Thus, this study concerns the initial phase of these coalitions’ development. As noted, data for the current analyses come from the routinely filed reports (e.g., action planning reports, quarterly progress reports, meeting minutes, coalition member interviews, activity logs, strategic plans) that are provided routinely to the study’s coordinating center and its centralized data warehouse to document the objectives established by each coalition, the rationale for objectives, and progress on completing these objectives.

Records that were in narrative form were entered into a database in NVivo 8 (QSR International 2008), a qualitative data analysis software package. We developed a priori codes to reflect constructs such as whether the objectives set by coalitions reflected universal, selected, or indicated prevention goals (Gordon 1983); whether objectives were reported by the coalitions as having been attained, abandoned, or still active; whether objectives met SMART criteria (e.g., specific, measurable, achievable, realistic, time-bound objectives) (Drucker 1954), and to code the strategies that were employed to achieve objectives. Other codes were developed inductively. For example, the documents reflected a variety of reasons for abandoning an objective, such as discovering it was not feasible or that some other group in the community had already accomplished it. Table 2 provides a summary of key constructs used in the current paper. Additionally, some of the data were quantitative, such as the length of time, in months, that it took to a coalition to accomplish or abandon an objective from its first appearance in the coalition’s action plans.

Table 2 Description of coding categories for objectives
Full size table

All texts were coded by trained research assistants. A second research assistant, working independently, coded a randomly selected sub-sample of the documents. Inter-rater agreement rates were calculated using Cohen’s Kappa. For all double-coded documents, kappa coefficients of .90 or higher were obtained, indicating high rates of agreement among coders. Discrepancies in the application of codes were discussed by coders until consensus could be achieved on the appropriate code. In some cases, disagreements resulted in revisions to codes.

The analysis approach we employed followed the principles of analytic induction (Robinson 1951), as modified by Erickson (1986). Analytic induction has been advocated as a means by which to synthesize data of multiple types and from multiple sources for analysis (Smith 1997). It is conventionally used to establish the conditions that distinguish between categories of cases (Bloor 1978). In the analytic inductive process, inductive reasoning allows for modification of concepts and relationships throughout the analytical process. Tentative hypotheses or propositions are generated based on an initial reading of each case in the data and then are rigorously tested in the data case by case, with the goal of arriving at an accurate representation of the empirical reality that distinguishes the phenomenon. Negative cases—cases that do not fit the emerging propositions—are specifically sought in analytic induction. These cases force the analyst to take one of three courses of action: revise the tentative propositions so that they can be maximally generalized to apply to all of cases of the phenomenon under investigation; redefine the phenomenon; or, eliminate the deviant case if it is in fact not an instance of the phenomenon (Robinson 1951).

To follow the analytic inductive approach, we treated each coalition as the unit of analysis and the data about and from that coalition were analyzed as such. We began analyses by examining a single case closely to generate a set of case-relevant propositions and then proceeded to complete within case analyses of each subsequent case, keeping in mind the propositions that emerged from prior within case analyses. The within case analyses provided an in-depth understanding of each individual coalition’s experiences of implementation and provided a means to develop and refine basic propositions about how the target population probably impacted the coalitions’ experiences in achieving objectives. Cross-case and comparative analysis provided a basis for testing the evidentiary warrant for each proposition and lead to their further refinement. That is, in analytic induction one first determines what is probably true based on a thorough understanding of the data and then establishes whether and how strong the evidence is to confirm initial propositions through constant cross-case comparative analyses.

To test the propositions generated through our within case analyses and examine how well these propositions distinguished categories of cases, we categorized coalitions by their rate of achievement of objectives so that we could compare the strength of the evidence for our propositions by high and low achieving groups. Because coalitions had the latitude to establish as many structural change objectives per any unit of time as befit local priorities and capacity, simple counts of achieved objectives did not provide an appropriate indicator of success. We therefore calculated the percentage of objectives achieved over the total number proposed during the time period under investigation for each coalition. Since it was early in the life of these coalitions, a success cutoff based on an absolute performance standard was premature to employ, given how long structural change may take to achieve. We therefore placed coalitions into two groups based on the rate at which they had completed objectives, creating a group that could be considered relatively successful (above average performance) and one that could be considered comparatively less successful (below average performance).

We further classified coalitions into groups based on the target populations that were the intended beneficiaries of the coalitions’ structural change objectives. This yielded one group (n = 6) focused on less socially stigmatized adolescents (heterosexual females) and one group (n = 7) focused on more highly socially stigmatized adolescents (gay and bisexual males and injection drug users). We considered these groups as highly socially stigmatized because of their close association with HIV infection, pervasive heterosexism, and prejudice and discrimination against young gays and bisexuals and youth who engage in illicit drug use (Almeida et al. 2009; Balsam et al. 2005; Berlan et al. 2010; D’Augelli et al. 2002; Flom et al. 2001; Henkel et al. 2008; Herek 1999; Herek and Capitanio 1999; Rosario et al. 2002; Russell et al. 2001; Taywaditep 2001; Williams et al. 2003). The coalitions that focused on young women we considered less socially stigmatized because of hetero-normative social biases that normalize heterosexual social and sexual relationships and that are less stigmatizing of those who acquired HIV through heterosexual sex (Herek and Capitanio 1999). These classifications were also consistent with the pattern that emerged in our within case examinations, which indicated stigma was a prominent concern of the coalitions that worked with gay and bisexual young men. Although racial and ethnic minority background and low income are other means by which these coalitions’ target populations could be stigmatized, low-income persons of color are predominant among the adolescents on whom these coalitions are focused, so racial/ethnic and economic stigma could not be readily used to classify coalitions into high and low stigma groups. We iteratively tested each proposition regarding how a focus on a highly stigmatized target population might impact success. Each of the propositions was systematically tested across categories of cases, with particular focus on negative cases and on the strength of the data in support of or in contradiction to a proposition. Propositions were routinely revised, and occasionally eliminated, until these provided a reasonable fit to the evidence. When quantitative comparisons were made among categories of cases, we used Chi-square tests and Analysis of Variance tests with Welch’s correction to account for variance heterogeneity, as appropriate.

Throughout the analysis process, the research team met biweekly to discuss and reformulate propositions in light of the case-level evidence. A critical function of these meetings was to identify new propositions that could extend our analyses and to identify ways in which propositions could be verified from other sources of data. We also used these meetings to discuss instances that did not fit with our emerging understanding of the data, so that we could adequately explain negative cases. Following the completion of these stages of analysis, preliminary results were presented to members of the larger study team, which included representatives of several of the C2P coalitions and C2P’s coordinating body, to assess the degree to which they perceived that we had accurately, authentically, and fairly represented the data. We also desired to represent any disagreements they had with our analyses and interpretations in subsequent reporting. These occasions provided us with an opportunity to add new propositions we had not considered and expand the vantage points from which we understood the impact of target population on achievement of C2P objectives.

Results

We first present descriptive findings on rates of achievement. We then present findings on the ways in which the target populations impacted on coalitions’ achievement of their objectives. Findings are presented in analytic sequence, following the inductive process by which they were assessed and refined in the cross-case and categorical comparisons.

At the end of 2008, coalitions had accomplished between 30 and 65% of the objectives they had established during this initial period of their operations. Seven coalitions performed above average and were classified as high success coalitions based on their rate of completion of objectives (≥48% of objectives accomplished) and six coalitions were classified as low success coalitions based on their below average rate of completion of objectives (<48% of objectives accomplished). Though, on the average, rates of success increased over the study period, as would be expected, the rate at which coalitions completed their objectives was unrelated to the number of months that coalitions had been in operation (29–41 months). Coalitions had attempted an average of 23 objectives (range = 14–44), of which they completed an average of 10 (range = 7–17) and abandoned an average of 8 (range = 1–22). Coalitions had an average of 4.77 objectives that they were still attempting to complete (range = 1–12). During this period, the rate of completion of objectives was unrelated to completing objectives of optimal quality and which reflected the larger aim of creating structural change (Reed et al. in press; Miller et al. 2011; Willard et al. in press).

Do Success Rates Differ Between Coalitions Focused on Highly Versus Less Stigmatized Adolescent Populations?

We proposed that high stigma coalitions would be less successful than low stigma coalitions. However, high and low stigma coalitions were roughly equally distributed across the below and above average success groups. Consequently, we developed a second proposition: those coalitions that focused on structural change to reduce risk among highly stigmatized youth would have abandoned more objectives than would coalitions focused on those who were less stigmatized. Our initial within case analyses suggested that when coalitions set out to engage partners in making local changes, coalitions working with the comparatively more stigmatized populations faced steeper challenges in gaining needed cooperation and investment in their mission from local community stakeholders than did the coalitions whose populations were comparatively less stigmatized.

Abandonment of Objectives

In general, coalitions targeting highly stigmatized youth abandoned more objectives than did coalitions targeting less stigmatized youth (see Table 3). We found that, of the five coalitions that had low rates of discontinued objectives, nearly all were low stigma coalitions. In addition, of the eight coalitions that had high rates of discontinued objectives, nearly all were high stigma coalitions. In all, high stigma coalitions discontinued 41% of objectives whereas low stigma coalitions discontinued 26% of objectives. Moreover, the rate of abandonment of objectives was unrelated to being a high or low success coalition among the low stigma coalitions. However, among high stigma coalitions, the low success coalitions abandoned considerably more objectives than did the high success coalitions.

Table 3 Comparison of coalition stigma status by (a) success group on characteristics of objectives, (b) with symbolic stigma coalition reclassified
Full size table

Negative case analysis suggested that two coalitions were exceptions to the general pattern that high stigma coalitions abandoned considerably more objectives than low stigma coalitions; both exceptions were low achieving coalitions. In one instance, a low achieving low stigma coalition had abandoned more objectives than any other coalition in the study. Close examination of this coalitions’ data indicated that its members perceived that it operated within an extremely conservative city characterized by entrenched and pervasive community stigma surrounding HIV; most objectives encountered external resistance. In some sense, the coalition is not an anomaly so much as its experiences were generally more like those reported by the high stigma than low stigma coalitions. This coalition reported that it routinely experienced symbolic AIDS stigma as it is described by Herek (1999). Although we did not reclassify this coalition into the high stigma group, for all subsequent data analyses, we examined whether the reclassification of this coalition as high-stigma impacted on findings (see Table 3) in order to develop propositions that could account for it.

By contrast, the other coalition, a high stigma low achieving coalition, had the lowest rate of abandonment of objectives of all the coalitions and among the lowest rates of objective achievement. This coalition set few objectives overall and focused almost exclusively on achieving local legislative and organizational policy changes that were expected to take years to complete; unlike every other coalition, a majority of the objectives set in the initial study period were still being pursued. This coalition was also the sole coalition focused on IDUs. Close examination of all data from this coalition suggested it was an anomaly in multiple respects. This coalition also had high rates of missing narrative data, so provided us with insufficient evidence on its experiences to allow us to revise propositions to explain its experiences confidently. In analytic induction, some cases might prompt the investigator to redefine the phenomenon in ways that allow for case to be excluded (Robinson 1951). Inadequate evidence in a case may also prompt its exclusion because emerging hypotheses simply cannot be tested in the case. In light of the fact the coalition had not completed or abandoned most of its objectives and had little high-quality narrative data to use to understand the coalition, we excluded the IDU coalition from subsequent analyses and focused our understanding on the impact of stigma associated with trying to reduce exposure to HIV among gay and bisexual males versus young women through pursuit of structural changes. All analyses reported from this point forward exclude the IDU coalition.

To further explore differences in rates of completion, we proposed that the coalitions targeting stigmatized youth would take longer to complete objectives when compared with the low stigma coalitions. The data supported this proposition. The coalitions working on behalf of highly stigmatized youth took on average 9 months to complete an objective (SD = 8.63) whereas the low stigma coalitions took an average of 7 months (SD = 6.64) to complete an objective.Footnote 1

Quality of Objectives

Having confirmed that high stigma coalitions and low stigma coalitions can succeed equally, but that high stigma coalitions abandon more objectives and take longer to complete objectives than low stigma coalitions, we sought to rule out the possibility that these findings were due to differences in the quality of the objectives that coalitions set. We proposed that low success coalitions would set poorer quality objectives than high success coalitions. We expected no difference between high and low stigma coalitions in the quality of the objectives they set. As our first measure of quality, we examined the extent to which objectives met SMART criteria. In general, objectives that met SMART criteria were more likely to be achieved than were objectives that failed to meet SMART criteria (Reed et al. in press).

The proposition that low success coalitions set poorer quality objectives when compared to high success coalitions was confirmed. However, the finding was entirely attributable to the quality of objectives set by the high stigma low success coalitions. On the average, only 39% of objectives set by the high stigma low success coalitions were SMART, compared to 73% among the high stigma high success group, 64% among the low stigma low success group, and 60% among the low stigma high success group. Paradoxically, the high stigma low success coalitions were the least likely of any group to abandon their high quality objectives.

We also examined whether the need to revise how objectives were stated and the number of steps involved in attaining an objective impacted on the length of time it took to complete objectives. Improving on how objectives were stated and having objectives that required multiple steps to complete explained about half of the variation in completion time in the low stigma coalitions. In contrast, 33% of the variance in completion time at highly stigmatized coalitions was attributable to making adjustments to objectives and to the number of steps required to complete objectives. Thus, the need to revise how objectives were stated and the complexity of the objectives better explained time to completion for low stigma than high stigma coalitions.

Nature of Prevention Objectives

Our next measure of quality concerned the nature of the objectives that were crafted by the coalitions. According to Gordon (1983), prevention efforts are targeted at those who are not yet affected by a disease or suffering symptoms of it. Universal prevention efforts are those that are good for the health of everyone in a community, such as mandated seat belt use to prevent injury and fatalities in car accidents or fluoridating water to prevent dental caries. Universal prevention efforts are typically structural in nature. Achieving changes that will universally affect all members of a community are more challenging than achieving changes that target only those who form a subgroup of a population or those who have been diagnosed with specific, well-described risk factors (Frieden 2010; Gordon 1983). Selective interventions are those that target persons whose risk is above average, such as efforts to redesign residences to prevent falls among the elderly or focused condom distribution efforts in settings that attract gay and bisexual men. Selective interventions are typically programmatic rather than structural; in these examples, only particular residences in particular communities may be modified and only particular settings on particular occasions may receive condoms. Indicated interventions are those targeting individuals who possess a specific individual risk factor, such as providing HIV risk-reduction counseling to someone who has been diagnosed with syphilis or disclosed a pattern of unprotected sex with multiple concurrent sexual partners. Indicated interventions are typically clinical.

Gordon’s framework provides a heuristic for considering the degree to which coalitions’ objectives are structural in nature. We proposed that high stigma high success coalitions would achieve more of their universal prevention objectives than the high stigma low success coalitions, reasoning that those coalitions that were achieving at a high rate would demonstrate greater ability to succeed with the universal social change objectives that were perceived as an urgent priority to reduce risk among young gay and bisexual men. High stigma low success coalitions shared this sense of urgency, but we proposed they would be least likely to succeed in attaining these kinds of objectives.

Overall, selected and indicated objectives were far more commonly set and achieved than were universal objectives. Moreover, both high and low success coalitions achieved the universal objectives they set at roughly the same low rate (18 and 17%). Contrary to our proposition, the high stigma high success coalitions attained a small proportion of their universal objectives (17%) compared to high stigma low success coalitions (38%). The opposite pattern emerged among low stigma coalitions: less successful coalitions attained very few universal objectives (5%) compared with successful coalitions (18%).Footnote 2

We examined the history of the universal objectives the coalitions had attained to understand why our proposition was not supported. We observed that the high stigma coalitions had to modify 25% of their universal objectives so that they were perceived as less controversial. All of the modifications to reduce controversy occurred at the low success coalitions in the high stigma group. By contrast, only one universal objective set by a low stigma coalition was modified to quell controversy. This low stigma coalition was in the high success group. We also found that high stigma coalitions abandoned universal objectives at double the rate of low stigma coalitions. Thus, the data generally support the idea that achieving changes that affect all youth in the community was easier to accomplish when the reasons to pursue those changes was to benefit young women than was achieving universal changes when the reason to pursue those changes was to benefit young gay and bisexual men.

How Does Stigma Affect the Process of Achieving Structural Change Outcomes?

To better understand differences in what kinds of objectives are abandoned and completed and what role stigma may play in success, we examined the processes associated with completing and abandoning objectives. We proposed that high stigma coalitions would cite more factors external to the coalition as reasons to abandon an objective than would low stigma coalitions. We also expected that the specific reasons for abandoning an objective would differ qualitatively between the two groups.

Both propositions were supported by the data. The high stigma coalitions discontinued more objectives and did so more often due to external resistance and low commitment to the particular changes they pursued when compared to the low stigma coalitions. At high stigma coalitions, discontinued objectives were often no longer deemed feasible: the coalition members encountered resistance, such as from parents, community members, and law enforcement officials, or had difficulty obtaining commitment from sectors that were necessary to complete the objective. Moreover, the high stigma low success coalitions were more likely to discontinue their efforts in the face of contextual barriers than were the high stigma high success coalitions. By contrast, among low stigma coalitions the most common contextual factor that caused coalitions to abandon an objective was discovering it had already been accomplished by some other community entity.

Nature of Challenges to Completing Objectives

We further investigated the influence of contextual barriers on coalition success by coding all documents for words related to various forms of stigma and prejudice. Text searches were performed in more than 2,500 of the coalitions’ reports and documents for key words such as “stigma,” “heterosexist,” “homophobia,” “homophobic,” “racism,” “racist,” “sexism,” “sexist,” “bias,” “conservative,” “backlash,” “opposition,” “offensive,” “prejudice,” and “discrimination.” Passages of text that identified these and similar terms as a root cause of HIV incidence on which the coalition might focus its efforts were excluded from further qualitative analysis. Only text in which stigma was described as posing a challenge to carrying out the coalitions’ work was further examined. We proposed that stigma would be more commonly described as an impediment to the coalitions’ work in the high stigma coalitions when compared to the low stigma coalitions.

Of all mentions of stigma as an obstacle to completing objectives, 70% referred to barriers experienced at high stigma coalitions and 30% referred to barriers at low stigma coalitions. Among the stigma references at low stigma coalitions, these were most often barriers due to the cultural association of HIV with high stigma populations, such as community perceptions that HIV is a gay disease and therefore does not warrant their attention or concern, and general discomfort with HIV/AIDS.Footnote 3 More detailed examination of the data indicated that there were unique ways in which experiences of stigma hindered high-stigma coalitions’ ability to succeed. These data also provided insight regarding what differentiates those coalitions who achieve higher rates of success despite these challenges from those that do not.

Difficulties Engaging People in the Coalitions’ Work

In general, we observed that successful coalitions engaged more youth in their coalitions; in the data, the number of youth who were involved in coalition operations was positively correlated with rates of objective completion (Reed et al. 2011). Document analysis suggested that involving youth was more difficult for the high-stigma coalitions, in part because youth were hesitant to self-identify as a member of the coalitions’ target population. Youth who were involved in the high stigma coalitions “have to come out in a certain way” as gay or bisexual. Additionally, the high-stigma coalitions reported fewer providers in their coalitions with expertise in adolescents than did the low-stigma coalitions. For instance, the documents from two coalitions located in what are typically considered progressive, gay-friendly cities, describe how though there are gay and bisexual-friendly and HIV service organizations within their communities, these places are “not very much for young people.” Another document noted that although “there’s a lot of collaborative work that happens around HIV and HIV prevention,” it is not “youth specific.”

Coalition members’ inexperience with and prejudices regarding the target population required extra education and training from the high stigma coalitions’ organizers. These coalitions’ documents indicated that an extraordinary amount of time was invested in training coalition members on cultural and developmental competency and related matters. Doing so took up time and resources and was perceived as a necessary investment for the coalitions to succeed, even if it temporarily distracted their focus from their structural change objectives. Low stigma coalitions seldom documented this level of effort toward internal training in target population competence, though when they did it was often in regards to working with sexual minority adolescents even though these youth were not the target population of concern. As noted by a member of a low stigma high success coalition, “we’re going to have to deal with some of this (stigma directed towards gay and bisexual men)—even to help women.”

High stigma high success coalitions were more successful than the high stigma low success coalitions in overcoming the challenges associated with engaging youth and youth-competent providers. The most successful among these coalitions found ways to incorporate youth into the coalitions’ actions and were deliberate and strategic in their efforts to do so. Engaging youth proved important to coalition members because it helped them to feel “like we’re working with/alongside our target population as opposed to making decisions for them.” Members also “learned the value of having those youth involved and having them lead.”

Community Resistance to Coalition Objectives

A second challenge evident in the data concerned community resistance. Parental resistance was often associated with objectives focusing on the educational sector. High stigma coalitions attempting to work in schools encountered “criticism from both like church groups as well as from a number of parent groups saying…it was inappropriate or irresponsible to be handing out condoms (on school grounds), you know, in the public spaces because you’re encouraging people to have sex.” High stigma coalitions received negative press in the local papers regarding their school-based efforts. High stigma coalitions reported “back-lash from parents” because of their proposals to school boards. In contrast, low stigma coalitions did not mention that parents posed barriers to accomplishing objectives, even those targeted at schools, with the sole exception of the symbolic stigma coalition. This coalition encountered vocal community resistance about their efforts to intercede in schools.

The high stigma low success coalitions that initially worked primarily in their school systems discontinued 15 of their school-focused objectives, almost as many as were completed in this sector across all high stigma coalitions. In examining the objectives that were completed, these were mostly universal prevention objectives. Those objectives that failed within the school system often focused specifically on sexual diversity or gay and bisexual male students. As a coalition member said, “administrative homophobia will prohibit more focused efforts targeted at young MSMs.” “Administrative homophobia” may have lead high stigma low success coalitions to turn to universal interventions that schools found politically tolerable rather than pursue the objectives they had initially crafted to benefit young gay and bisexual men who were in school.

High and low stigma coalitions mentioned church-related resistance to HIV work; however, when church and faith-based barriers were encountered, low stigma coalitions were more likely to complete objectives and garner help from people within faith-based organizations than the high stigma coalitions. For example, despite the “stigma associated with HIV in this community,” one low stigma coalition was able to identify four churches willing to place HIV statistics in their bulletins, distribute HIV awareness ribbons to congregants, and provide HIV education and testing on church property. Another low stigma coalition was offered space from a church to hold meetings; another mentioned “seeing churches being progressive” in their community and “changing.” In contrast, high stigma coalitions did not report observing this “slow progression” within the religious community. One high-stigma coalition member discussed the difficulty engaging the church community:

So for instance, I’ll give you an example: if we as an agency have approached certain church groups about doing outreach and or education about IDU issues, they’ve been very receptive about that. But if we said that we want to come into the same church community where we’ve already been and have already talked about HIV and people who use drugs, and said we wanted to talk about HIV and persons that engage in MSM behavior, that’s a problem. And so, you know then we are not as willing to let you come into our community to talk about that issue.

The high stigma coalitions attained view objectives within the religious sector and, as a result, quickly avoided it as a focus of their work.

The high stigma coalitions’ documents described a tension between raising HIV awareness and further stigmatizing the members of their target populations. The members of these coalitions worried that their efforts to raise awareness and provide local data on HIV would lead to “further stigmatization of an already stigmatized population.” Nearly half of the high stigma coalitions were concerned “that the data would be taken out of context.” Coalition members were firm that there “are already enough barriers associated with sexual expression in this population (gay and bisexual men),” so “care must be taken in the public release of these data findings (HIV prevalence statistics).”

Coalition Dynamics

The documents from the high stigma coalitions provided evidence that members of these coalitions were keenly aware of the need to address “macro issues that support cultural change,” but also showed they were overwhelmed by the task of doing so. These coalitions often engaged in discussions about stigma at the subcommittee, coalition, and national project level and agreed that “homophobia would be high up there for us (as an issue to tackle in the community).” Issues related to homophobia and stigma “always come up in community meetings,” but addressing these macro system concerns is “a big task.” One coalition member divulged that the coalition has “some difficulty trying to put large structural change ideas (i.e., eliminating homophobia) into workable structural change objectives.” Because “stigma” is such a “major challenge,” coalition members believed “we can’t fix these big challenges.”

Though all of the high stigma coalitions were keenly aware of the scope of the challenges they faced, the high stigma low success coalitions were fraught with dissension among their members about how best to meet these challenges. The disagreements among or between coalition members at high stigma low success coalitions made it difficult to form objectives that satisfied everyone. By contrast, low stigma coalitions’ documents suggested they experienced more ease in targeting their efforts and less discord on how to proceed.

Success in the Face of Resistance

We further examined how the high and low success coalitions in the high stigma group responded to challenges and whether they did so in systematically different ways. The successful high stigma coalitions capitalized on contextual barriers and challenges. These coalitions expected barriers, used barriers as opportunities to create social change, and found ways to work around them. As one coalition member at a high stigma high success coalition noted, “we must realize that sometimes backlash makes things move forward.”

The successful high stigma coalitions sought to overcome community resistance through strategies such as working on their objectives in concert with other coalitions, engaging parents and guardians in their change efforts, and designing objectives that focused on sources of resistance, such as parents who had the potential to become allies. These coalitions sought out expert institutions in working with parents or youth to help them neutralize opposition and win opponents to their cause. Documents from the high stigma low success coalitions provided little evidence that these coalitions engaged sources of resistance in the coalitions, leveraged allies who could span the boundary between those who opposed them, or set objectives regarding reducing stigmatizing beliefs among their opposition. Additionally, as noted above, disagreements among the high stigma low success coalition members about how to proceed in the face of opposition often lead to paralysis and a lack of targeted response to specific challenges. Thus, whereas the high stigma high success coalitions created objectives to address the very barriers they encountered in completing other objectives, the high stigma low success coalitions did not turn barriers into opportunities and were seldom able to agree on how best to respond to barriers and challenges.

Discussion

We examined whether the ability of community coalitions to achieve structural changes aimed at reducing adolescent exposure to HIV is impacted by stigma and, if so, how. In so doing, we extend the literature on coalitions, as few empirical investigations have documented how the community context in which they operate impacts on their short-term success (Allen et al. 2008). We similarly extend the literature on stigma by identifying strategies successful coalitions employed to accomplish social change in the face of stigma.

We found that working on behalf of a stigmatized population of adolescents was not necessarily an impediment to success. Coalitions working on behalf of high-stigma populations could succeed in meeting their objectives, albeit more slowly and with having to give up more objectives when compared to coalitions working on behalf of low-stigma populations. Consistent with the observations of Kadushin and colleagues, the successful coalitions were highly focused and crafted objectives that were realistic, precise guides to action (Kadushin et al. 2005). The high stigma low success coalitions often failed to set precise, actionable objectives and, when they did, clung to them perhaps unwisely. The successful high stigma coalitions did not hesitate to change course, if necessary, to accomplish their goals. Changing course did not necessarily involve compromising on or altering objectives to increase their appeal to the opposition. Perhaps as a consequence, the successful high stigma coalitions were less likely than the high stigma low success coalitions to attain universal prevention objectives. The successful high stigma coalitions were best able to attain objectives that were programmatic and clinical over the 2-year timeframe we studied.

We observed that high stigma coalitions encountered unique challenges and faced greater community opposition than low stigma coalitions, but that the high stigma coalitions can succeed in the face of such challenges by using them as transformative opportunities and by forming focused and strategic rather than community-wide alliances. Chavis (2001) points out that unlike other entities that may be engaged in pursuit of social change, coalitions bring together people who represent diverse interests and levels of power. Navigating divergence is part of the normal routines of coalition operation. The differences we observed among the more and less successful high-stigma coalitions underscore the degree to which navigating such divergence successfully may assist a coalition to move forward with its social change agenda. The high stigma low success coalitions often allowed pluralism and member self-interest to have paralytic effects on setting attainable objectives.

Chavis (2001) argues that the inevitable conflict in coalitions is inherently paradoxical. He suggests that because well-composed coalitions represent the conflicts in the community, coalitions can transform their communities by successfully addressing the conflicts that manifest within coalitions themselves. Chavis refers to this as conflict transformation. Conflict and divergence can therefore aid a coalition, if it can capitalize on conflict and divergence to achieve socially just objectives. The high stigma high success coalitions in these data illustrate the power of Chavis’ paradox in that it is these coalitions that engaged the conflict they encountered directly as a transformative occasion, rather than allow such conflict to be internally divisive, a source of demoralization, or derailing of the coalition’s larger purpose. These coalitions actively sought to co-opt their opposition rather than be co-opted by it. Consistent with experiences of other youth-focused HIV coalitions (Ogusky and Tenner 2010) and writing on coalition best-practices (Wolff 2001), expanding the pool of allied collaborators, engaging youth directly and intensively in their work, and neutralizing or turning opponents into allies were the strategies these coalitions used to transform conflict. In contrast to the findings of Cheadle et al. study of issue-focused coalitions (2005), high stigma high success coalitions that made concerted efforts to engage key citizens, youth in particular, and invested in training all members for the work they would jointly pursue succeeded.

A primary strength of the current research resides in our ability to conduct a multiple case analysis using rigorous qualitative data analytic techniques and to do so with data collected longitudinally. Analytic inductive procedures offer qualitative researchers a clearly delineated logical process for establishing the evidentiary warrant for drawing conclusions that apply to all cases of a phenomenon and a reasoned basis for excluding cases. Additionally, because these coalitions all focus on adolescents and HIV, are allocated resources similarly, operate under common procedures, and have centralized quality assurance monitoring and technical assistance supports, we can have confidence that comparing these coalitions is reasonable. These elements, in combination, strengthen our ability to understand the means by which various forms of HIV-related stigma impact on the day-to-day work of coalitions and discover strategies that successful coalitions employ to achieve the outcomes they desire.

Despite these important strengths, there are limitations to our work that merit consideration. First, the commonalities among these coalitions that make it reasonable to compare them also serve to limit the circumstances to which the findings of this research may be appropriately applied. These coalitions, while they form the universe of coalitions in this particular initiative, are unique in ways that limit our ability to generalize our observations to the universe of coalitions. For instance, the focal parent organizations for these coalitions are local hospitals and clinics. Consequently, our findings reflect the experiences of coalitions organized by public health institutions rather than the experiences of coalitions founded by grassroots citizens groups or other institutional entities. Second, despite the common procedures and data collection protocols, there are variations across coalitions in the quality and thoroughness of some of the records they have produced, as well as variation in what was collected and how it was collected over time. In some cases, data that might have provided clarification on stigma-related impacts and processes were simply not collected with adequate depth and consistency to arrive at more than tentative conclusions. In one instance, data were of insufficient quality to support a complete analysis of that case. Finally, all of the data reflect the impressions of the coalition staff, supervisors and, to a lesser extent, coalition members. We do not have data that speak to community members’ and youths’ perceptions of the coalitions’ work. Importantly, we have no self-reported data from youth on their experiences of stigma related to their gender, sexual orientation, or injection drug use status against which to validate adult perceptions.

Conclusion

Structural change is a worthy and important component of comprehensive HIV prevention initiatives. However, structural change may be easier to achieve when it serves the interests of groups that are not heavily stigmatized. Stigma may impede structural change efforts, requiring structural and cultural change efforts designed to target stigma directly as a precursor step. Through this analysis, we identified coalition strategies to achieve success in pursuing such change in the face of stigma. These strategies include engaging invested supporters of the target population and the population itself, turning opponents into allies, and meeting obstacles head on so that these become opportunities for, rather than roadblocks to, social change. A transformative and uncompromising approach to stigma-driven resistance may be a necessary part of coalitions’ recipe for success.