Abstract
Declining social capital is one explanation for lower response to household surveys. By intentionally developing an awareness of social capital among its interviewers, the Por Ahí Dicen research study encouraged the use of interviewer social capital as a mechanism to achieve a response rate of 65.2 percent for baseline and post-intervention household studies of Puerto Rican mothers (n = 413). These surveys were conducted in a “hard-to-count” urban environment designated by the U.S. Census Bureau. The interviewer trainings highlighted three domains of social capital: reciprocity and generalized trust, group or social cohesion, and cultural affirmation. By stressing the importance of social capital as an engagement tool, interviewers more easily made research participation salient and successfully leveraged Puerto Rican mothers’ participation in the community study.
Introduction
The success of survey research depends heavily on engaging a representative sample of individuals in a target population. However, survey research has experienced an increase in nonresponse to household and other surveys in the United States over the past several decades (Atrostic, Bates, and Silberstein 2001; Brick and Williams 2013; Curtin, Presser, and Singer 2005; Groves, Cialdini, and Couper 1992). Understanding the factors contributing to the decision to participate in a research survey is important, as high nonresponse rates challenge the inferential value of statistical tests. Several factors have been identified as having a general impact on survey participation; they include social context, survey design, and characteristics and psychosocial elements influencing the interaction between respondents and interviewers (Groves, Cialdini, and Couper 1992). Researchers should expect to obtain lower/higher response rates based on the respondent type, suggesting that certain techniques or strategies to increase participation may be more or less effective depending on the particular population sampled (Anseel et al. 2010).
Obtaining a representative sample of Latinos and other “hard to reach populations” poses additional challenges in survey research: the tendency to consider Latino populations as a homogeneous group, participant hesitation due to immigration status or cultural misunderstandings, and issues relating to the research design itself such as recruitment methods that are not tailored to the unique cultural and linguistic needs of a Latino subpopulation (Martinez et al. 2012; Mendez-Luck et al. 2011; Villarruel et al. 2006). Lower levels of literacy and residential segregation have also been associated with reduced rates of participation among Latinos (Frack et al. 1997; O’Hegarty et al. 2010). In addition to these sociodemographic factors, a complex research design that includes cumbersome protocols and fails to provide accessible sites for participation also influence the survey response of these populations (Margitić et al. 1999).
Declining social capital is another explanation for decreasing survey response (Brick and Williams 2013). Social capital is defined as access to material and personal resources activated through participation in social networks and other institutions, and not accessible without such participation. Although participation in survey research appears on the surface an individual choice, the decision to participate is contingent on the connection participants make between the research and their needs or the needs of their community. Leverage-salience theory posits that a potential participant’s decision to engage in research is informed by a combination of his or her interest in the survey topic (salience) and attitudes toward other factors that are part of the survey request (leverage). This highlights the need for survey design methods to identify leverage points that are important in the recruitment process. The interviewer/recruiter’s goal is to make salient these opportunities of interest to the selected research participants to elicit research participation and compliance (Groves, Presser, and Dipko 2004; Groves, Singer, and Corning 2000). However, with declining social capital, fewer leverage points exist to make the research salient.
Although declining social capital is posited as an explanation for increasing nonresponse (Brick and Williams 2013; Putnam 2000), social capital offers a mechanism for increasing survey response among Latinos. Many social conditions in Latino neighborhoods suggest that individuals living there lack social capital, but upon deeper investigation, social structures exist to promote social capital (Small 2004). For example, specific cultural structures (e.g., shared ideology, norms, values, and practices) in Puerto Rican neighborhoods create a context that holds promise for research design incorporating social capital to highlight the research’s salience (Granberry and Torres 2017; Torres 1998). Latinos accumulate social capital through participation in neighborhood structures (Granberry 2014; Granberry and Marcelli 2007), and neighborhood relationships may be key for promoting research participation. By focusing on developing an awareness of social capital among research interview team members, the research design presented in this study trained interviewers to highlight more leverage points through accessing their social capital to make the research agenda more salient to a potential respondent, and thus increase research participation. This process involves unacknowledged background work that informs survey research methods and is glossed over in the literature (Lindsay 2005).
In this article, we explore how our probabilistic household surveys successfully recruited Puerto Rican mothers for pre- and post-intervention assessments with a combined 65.2 percent response rate (n = 413). We attribute our success to a methodological strategy to activate social capital among research interviewers and other staff, including undergraduate students, community residents, and investigators. We invested time in training that focused on developing research and community literacy and provided these individuals with material and personal resources related to research and community development (Granberry et al. 2016).
Increasing Puerto Rican survey participation, and in this case for health intervention research, is necessary to better understand social and health conditions related to health disparities among this at-risk minority population in the United States. The U.S. Census Bureau projects that Latinos will grow by 118 percent by 2060 and comprise 28 percent of the U.S. population (Colby and Ortman 2015). Successfully recruiting Puerto Ricans and other Latinos in research will be important to address health disparities and other socioeconomic issues. Latino recruitment in research studies has been difficult (Frack et al. 1997), but in this article, we describe the elements of a strategy that obtained high response rates. We proceed with an overview of (1) the study and its relevant methods, (2) the different types of social capital significant to this study, (3) the final dispositions from the sample frame, and (4) the relevance of using social capital to increase survey response.
Field Method and Processes
Our study, Por Ahí Dicen, evaluated the effectiveness of a Spanish media campaign on mother-child communication about sexuality. Our community partner, the Puerto Rican Cultural Center, provides educational and cultural programs to support Puerto Rican residents in Springfield, MA. This medium sized multiethnic city has not yet transitioned as other cities in the East Coast from an industrial to knowledge-based economy. The study was built off of a previously successful pilot study (Gubrium and Torres 2013). A Community Advisory Committee (CAC) was established that comprised 51 percent community residents and 49 percent professional public health educators, policy makers, and researchers. This committee participated in developing the survey instruments, research protocols, and survey implementation’s logistics.
Criteria for participation in the study included being a Spanish-speaking Puerto Rican mother or primary caregiver of at least one child between the ages of 10 and 19 living in one of the randomly selected census blocks of the sample frame. The goal of the first phase of fieldwork was to identify a household address list for possible selection in the survey. The multistage area probability sampling design randomly selected census blocks from census tracts with at least 7 percent of households meeting two criteria: (1) having a Spanish-speaking Puerto Rican mother and (2) at least one child age 10 to 19. From a possible 298 census blocks meeting this criteria, we ethnographically mapped 100 randomly selected blocks, identifying 4,828 possible housing units to comprise our sampling frame. The ethnographic mapping research activity located, listed, and identified households in the sample frame (Granberry et al. 2016; Groves and Couper 1998). We chose ethnographic mapping because it was an iterative process that allowed frequent interaction between research team members. Six teams consisting of one undergraduate student and one community member enumerated housing units in the selected census blocks. Once we recorded the enumeration data, we randomly selected 2,252 addresses from the sampling frame for a baseline survey and 2,164 addresses for a post-intervention study.
The second fieldwork phase was devoted to recruiting participants. Each address was visited up to four times to make contact with a household member to determine survey eligibility. Recruitment was centered on approximately five-minute conversations in which interviewers engaged with eligible sample elements; the details of this procedure will be discussed shortly. This process identified 263 eligible mothers in households and successfully recruited 203 mothers from July 2013 to January 2014 for the baseline survey, and identified 257 eligible mothers in households and successfully recruited 210 mothers from May to August 2015. Across both the baseline and post-intervention surveys, 65.2 percent of our attempts to recruit sample elements in these households were successful.
Recruitment and Training of Research Working Group
Recruitment strategies were based on recommendations from the literature as well as our prior experience with the study population (Marcelli 2014; Marcelli et al. 2009; Torres 1998). Our approach focused on having a bilingual, bicultural staff and members of the target population on the recruitment teams. The university and community-based organization partnership was built using both groups’ strengths. The research-team-building process centered around developing an awareness of social capital by promoting research and community literacy on two training occasions. The first session consisted of an 8-week program that introduced the research process and resulted in ethnographic mapping of all 100 census blocks that formed the basis of the sample frame. This training was conducted by two faculty members and used a Freirean approach (Freire 2000) of “learning by doing” with reflection on daily research activities. From the start, it was clear that team members came to the project with different perspectives. By linking Latino undergraduate students and community residents in the research setting, the research design created an atmosphere allowing for the transfer of material and personal resources through mutual trust and reciprocity. Undergraduate students implemented research methods by applying material from their coursework (research literacy) that community residents lacked. Community residents contributed experiential knowledge (community literacy) that the students lacked. The second training session, consisting of three training days with additional research assistants, occurred before the survey instrument implementation. Building on our previous work, this training’s purpose was twofold. It first addressed how the research was intended to help the community address problems related to sexual health. It then provided technical training on how to implement field research protocols. Throughout the training and accompanying reflection process, interviewers were exposed to how social interaction facilitated the transfer of material and personal resources.
By linking with our community partner, the research project was able to access existing networks in the community to demonstrate social capital’s utility to interviewers. One of our interviewers, who was a member of the local public safety taskforce, leveraged the participation of a community service officer from the local police department in the interviewer training. The officer made a presentation about safety in the community and was able to demonstrate how our community partner was linked into broader community initiatives in the neighborhoods that made up the sample frame. This presentation not only focused on public and personal safety but also linked the research project’s goal to broader networks engaged in community development in the surveyed city. Besides demonstrating how the efforts of the study would be connected to other community resources, it grounded the research study in the citywide community development agenda.
The second and third fieldwork phases consisted of implementing the study’s rapid ethnographic assessment. After completing three-day interviewer trainings, 15 students from a large urban university and 15 residents from the research area started the recruitment process. If a household contained an eligible Puerto Rican mother, the interviewer team introduced the survey and attempted to recruit the mother to participate. As a symbol of appreciation for their time, mothers received a US$25 gift card to a local supermarket on completion of the questionnaire. After consent was obtained, the interviews were conducted in Spanish and averaged 55 minutes. Interviews were audio recorded if the mother consented. The survey instrument not only consisted primarily of questions about communication regarding sexuality and sexual health but also included sections on (1) household members, (2) neighborhood and social capital, and (3) access and exposure to the media (Torres et al. 2016). Interviewers recorded the responses on paper questionnaires.
Participant Recruitment and Cooperation
The recruitment process began with a brief unannounced interaction at a potential participant’s home. Interviewers were provided a script that highlighted the motivation of the research and the need for participation by the selected community members. Interviewers were encouraged to tailor their approach to be responsive to the needs and interests of each individual (Groves, Singer, and Corning 2000; Morton-Williams 1993). After a brief introduction to the research project and a series of questions to determine eligibility for participation, there was a limited window to elicit research participation. Interviewers employed strategies that others had shared with them, and they became skilled at communicating the individual- and community-level benefits of participation.
Faculty Participation
A unique component of the research design called for engagement by faculty and investigators throughout all aspects of the research fieldwork. This participation allowed for daily interaction with interviewers to ensure successful implementation of the research protocol. In debriefing sessions led by the investigators, interviewers shared insights into their successful recruiting of survey participants. In addition, their participation created opportunities for the activation of interviewers’ material and personal resources.
The Role of Social Capital in Community Survey Research Recruitment
Social capital takes many forms, and it can be actualized and activated for different purposes by different networks in a community (McPherson, Smith-Lovin, and Cook 2001; Portes 1998; Putnam 2000; Sampson and Raudenbush 1997). In this article, we focus on three social capital domains that were activated by the study’s fieldwork strategy for recruiting and interviewing Puerto Rican mothers in an urban environment. We describe our experience using the domains of reciprocity and generalized trust (Glaeser, Laibson, and Sacerdote 2002; Putnam 2000), group or social cohesion (Sampson and Raudenbush 1997), and cultural identity affirmation (McPherson, Smith-Lovin, and Cook 2001; Small 2004).
The Domains of Social Capital
Through interviewer training and supervision, our project staff were exposed to three domains of social capital. (1) At the heart of Putnam’s (2000) initial research on social capital, he stresses that dense reciprocal networks develop through civic engagement and in turn can lead to the transfer of resources. When relationships are reciprocal, trust develops among partners. The combination of trust and exchange allows for the potential development and accumulation of more resources. Glaeser, Laibson, and Sacerdote (2002) make the distinction that this trust must be shared because, by itself, it does not guarantee reciprocity. Trust then is built over time through interactions in social networks via a common goal. (2) Group cohesiveness (also called social cohesion) arises when bonds link social network members to the larger group processes. Although group cohesion is a multifaceted process, it can be broken down into four main components: social relations, task relations, perceived unity, and emotional attachment. In this regard, Sampson and Raudenbush (1997) further nuanced social cohesion by identifying collective efficacy: a process of activating or converting social ties among neighborhood residents to achieve collective goals, such as public order or the control of crime. By developing collective efficacy, individuals link to social networks that build trusting relationships while activating mechanisms for social change. (3) Underlying both trusting relationships and group or social cohesion is a shared experience. Social networks develop among individuals with shared interests. Thus, shared group identity is important because the activation of social capital can be shaped by a group’s cultural identity. For example, Small’s (2004) research on another urban Puerto Rican neighborhood demonstrates that social capital exists in the unique context of people’s experiences in their environment. Often, these expressions of social capital are not visible to the outside eye. Thus, a group’s identity must be considered when measuring social capital.
Reciprocity and Generalized Trust
Trust between researchers and community is at the heart of successful research implementation (Schensul 1999). Models of community-based participatory research (CBPR) have identified two barriers to developing trust: past negative experiences and challenges to reconciling community and institutional priorities (Zambrana 1996). These barriers reflect difficulty in addressing power imbalances, related to different skill levels and experiences, between academic researchers (including both faculty and students) and community residents. In practice, as described by Schensul (1999), this means that researchers and community residents must commit to sharing their skills and experiences, so that they learn from and value each other’s contributions to a mutually beneficial research agenda. However, the transference of trust between interviewers and survey participants has been generally overlooked in the literature.
The research training emphasized the salience of collecting important information needed to act on an identified community problem (Lippard and Price 2011) and interviewers’ competence to implement the research protocols. Trust was enhanced by having group sessions to develop research literacy about ethics and human subject protection protocols, including obtaining certification by the Collaborative Institutional Training Initiative (CITI). In this process, interviewers saw the importance of their role in meeting the research protocols for collecting information to inform the research study’s media campaign. After a field test day in which the interviewers piloted the survey instrument, they experienced how having trust in their own ability to implement the research protocols (including the application of ethical principles) would be positively received by the community. In a subsequent debriefing session, the interviewers realized that their belief in the objective of the research was key to recruitment of participants because they were providing an opportunity to collect data that would be used to assist in the broader community development mission that was promoted by our community partner.
We posit that there is a triangular relationship between investigators, interviewers, and survey participants that begins when the goals of the investigators, the interviewers, and the community are aligned. In our case, these relationships were built from previous attempts to address the research problem. Our community partner had been actively advocating for the inclusion of sexual health in the school health curriculum. In addition, the organization served as a site for a family-oriented sex education project. Our present project seeking to design and test the effectivness of a linguistically and culturally responsive media campaign to encourage maternal communication about sexuality was a response to a salient community need. Likewise, investigators recruited undergraduate students to serve as interviewers through the Latino Leadership Opportunity Program that aimed at training a new generation of Latino professionals. This history served as the foundation for developing a trusting relationship between the investigators and the leadership of our community partner that served as a vehicle for recruiting interviewers from the community. This sense of trust was central to the creation of group cohesion and identity within the project.
Social (Group) Cohesion
A key characteristic of an effective CBPR partnership is group cohesion (Schulz et al. 2003). Feminist theory suggests that individuals who share particular social statuses often have meaningful experiences that generate shared knowledge about the social world (Harnois 2013). Most of our local interviewers were Puerto Rican women who resided in the general research area and were close in age to the participants, but most of our student interviewers were from other Latino countries and spoke Spanish. Curiosity about each other’s worldviews enriched the two groups’ sense of collective efforts and common goals to improve the community’s sexual health.
We tapped into an existing local community network built by our community partner and added a group of undergraduate students and researchers from outside of the community to help implement our study. Integrating the two groups required creating intentional opportunities for the exchange of individual resources such as knowledge and skills. This learning by doing approach served as a mechanism to create an environment supportive of developing intimate interpersonal relationships between the two groups. During the three-day training for interviewers, we paired a member of each group to practice the recruitment protocol. This focus on successfully adhering to fieldwork protocol for recruitment provided the opportunity to learn from each other’s personal resources such as in-depth community knowledge, persuasion skills, interest in meeting the community’s needs, and linkages to other local resources. Instead of seeing the practice exercises as a mechanism simply to learn how to implement the research protocols, group members modeled successful skills that might not have been available to them if only one group had been present. This bidirectional learning process continued though the fieldwork period during regular meetings that allowed for sharing successful strategies to recruit participants. The sense of unity was strengthened by the common experience of shared success in attaining the study’s recruitment goals. We further enhanced this process through celebratory meetings that allowed further interpersonal interactions, including the sharing of personal stories. Through this series of interpersonal exchanges, the group members increased their sense of commitment to the study’s recruitment goals.
Cultural Affirmation
As the only Latin American population that brings to the United States all the rights of naturalized citizens and prior exposure to American laws, customs, and language, Puerto Ricans’ self-identity tends to be influenced by the cultural dynamics and the level of organization in the receiving stateside community (Torres et al. 2008). For some Puerto Ricans arriving in this city, our community-based organization provided the first link to existing social networks and opportunities for community engagement, thus affirming their identity and sense of belonging in the new place. The value of this type of sociocultural brokerage by local community-based organizations has been documented in the literature (Cordero-Guzmán 2005; Rusch, Frazier, and Atkins 2015).
Our partnership highlighted the salience of the ethno-linguistic and national identity in the high levels of trust and social capital of our working group. Through a subcontract, our community partner assisted in the recruitment of members of the CAC, staffed its monthly meetings, and represented the project in community events. A group of five women associated with our community partner formed the local leadership team for our study and assumed the role of peer research assistants in the way described by Price and Hawkins (2002) and Greene et al. (2009). These community peer research assistants were trained to perform ethnographic mapping, recruitment, and interviewing tasks for the household enumeration, survey, and focus group phases, respectively. By virtue of their local expertise and relations, they also served as key informants throughout the project’s implementation process. Their shared experiences provided the cultural foundation for the training, content, and delivery of the project to potential participants and for the celebration of milestones. Perhaps the most memorable manifestation of cultural pride and joy occurred in the early phase of the project when a local group of five coordinated and led an informational community meeting to introduce the project to the larger community, to welcome additional volunteers for the CAC, and to generate a list of questions for the survey instrument. This community forum ended with a typical Puerto Rican lunch. While waiting in the lunch line, one of the community research assistants told the second author, “This is a Puerto Rican celebration . . . this is the first time that I am in a place where my language and my culture are affirmed.” At a time that researchers and professionals are seeking strategies to increase participation, this is a reminder that the cultural funds of knowledge and practice can be integrated into program activities.
Grounding this study in the experience of Puerto Ricans in their local community provided the milieu that shaped the entire research team’s identity and helped facilitate interactions where individuals with different identities shared how their cultural experiences differed from the Puerto Rican experience. This process was enhanced by different groups having the same desire to bring sexual health education to bear on addressing high rates of teen pregnancy and sexually transmitted infections, including HIV. The study’s overarching Puerto Rican identity created an environment that enabled team members to use their cultural resources in the recruitment of survey participants.
Response to the Community Study
From the sampling frame of 4,828 addresses identified during our ethnographic mapping fieldwork, 2,335 addresses were randomly selected for the baseline probabilistic household survey (Figure 1). From this list, 63 identified households were found not to exist. This was attributed to demolitions that occurred to two larger multiunit structures. In addition, 20 of the original addresses were found to be commercial and not residential, and 200 addresses were identified as not occupied. After four attempts to recruit for participation in the survey, 319 households did not comply with the research protocol because no one responded to the interviewer team’s contact at the door of the selected address. Another 13 households refused the screening protocol. Of the 1,720 remaining households, 1,451 did not meet the eligibility requirements of having a Spanish-speaking Puerto Rican mother with a child between the ages of 10 and 19 years, and six mothers were never able to be screened for eligibility. Of the remaining 263 eligible households with sample elements, 58 refused to participate and two households with mothers agreed to participate but never complied with the invitation to be interviewed. At the end of the interview process, 203 eligible households with Puerto Rican Spanish-speaking mothers had completed the survey questionnaire.
Final dispositions of sample elements for the Por Ahí Dicen study.
The remaining 2,235 households comprised the sample for the post-intervention survey. There were 71 addresses found not to exist and 94 addresses identified as unoccupied. After four attempts to recruit for participation in the survey, 362 households did not comply with the research protocol because no one responded to the interviewer team’s contact at the door of the selected address. Another 11 households refused the screening protocol. Of the 1,697 remaining households, 1,434 did not meet the eligibility requirements of having a Spanish-speaking Puerto Rican mother with a child between the ages of 10 and 19 years, and six mothers were never able to be screened for eligibility. Of the remaining 257 eligible households with sample elements, 47 refused to participate. At the end of the interview process, 210 eligible households with Puerto Rican Spanish-speaking mothers had completed the survey questionnaire.
We present the American Association for Public Opinion Research (AAPOR) Response Rate 3 (AAPOR 2016) given the expected high rate of ineligible units. This response rate includes a weighted proportion of the sampled households for which eligibility could not be determined given the inability to contact an adult in the household. The weighted proportion, e, is computed as the ratio of households known to be eligible in the numerator and households known to be eligible plus households known to be ineligible in the denominator. This proportion, 0.144, is used to weight the cases for which eligibility could not be determined (705), and is included in the denominator of the response rate computation with other cases known to be eligible whether completed or not. The final response rate for the study is 65.2 percent. This response rate presents a plausible middle ground between other potential response rate calculations, which make one of two problematic assumptions: that all cases for which eligibility cannot be determined are eligible, yielding a far too pessimistic estimate (AAPOR Response Rate 1), or are not eligible, likely too optimistic (AAPOR Response Rate 5).
The Salience of Social Capital in Recruitment for a Community Study
Developing an awareness of social capital among our study’s interviewers was an integral component of the research project’s internal design that was related to the study’s response rates (65.2 percent) of eligible Spanish-speaking Puerto Rican mothers in 2013 and 2015. Social capital enhanced interviewers’ competence to apply recruitment protocols. The less-than-five-minute window in which the recruitment occurred depended on the interviewer team’s competence to make the project salient to a selected Puerto Rican mother and then use the interpersonal connection to leverage her participation. By experiencing how participation in social structures had developed material and personal resources for themselves, interviewers were able to make the research topic more salient. They highlighted how a potential respondent’s participation would not only benefit the community by addressing a health disparity but also benefit herself by developing a better understanding of the significance of maternal-child communication about sexual health. This five-minute window for recruitment involved a negotiation for the proposed one-hour time needed to complete the full interview. By experiencing the added value of social capital, the interviewers developed better skills to identify recruitment leverage points.
Faculty participation in the training and fieldwork of the research project was instrumental for social capital development. Faculty have access to information and resources that can benefit both community members and undergraduate students. Faculty participation allowed for relationships to develop that were conducive to the transfer of personal and instrumental support for fostering social capital among interviewers. Nurtured through social relationships, the interviewers were motivated to promote the research topic and provide rationale for following research protocols whose logic was not always initially evident to the community. Beyond the day-to-day interpersonal interaction in the fieldwork, the faculty scheduled periodic group reflection sessions that focused on lessons learned from the recruitment process and developed suggestions to improve the study’s implementation. In addition to the group interactions, faculty members worked with interviewers to identify areas of personal and professional growth, and provided information and contacts to achieve this growth.
Leveraging social capital among interviewers increased the research project’s effectiveness during the data collection process. Our review of the project budget suggests that each of 423 completed interviews cost approximately US$450, representing cost for interviewers’ time and direct field supervision by the principal investigators. This cost estimate does not account for the previous investments in the foundation previously laid by the community-based organization and the Latino research institute. This previous partnership had generated information about the community, developed social networks committed to the salience of the topic, and established a trusting relationship that served as a foundation for the study (Lindsay 2005). Neither did this estimate include the contribution of the research team, whose investment of time in the fieldwork component augmented the social capital that helped produce the high response rate.
Conclusion
To recruit hard-to-reach populations for a health disparities study, it is necessary to invest in developing the social capital infrastructure that may or may not exist at the time in the community research setting. The methods for our study are reproducible, and an approach that maximizes the collaboration of social resources will likely bring an increased response rate. Our results suggest that an existing social capital foundation needs to be leveraged through appropriate research methods. These methods activate interviewer resources to enhance their ability to implement recruitment protocols by transmitting the salience of the project and leveraging participation in the study.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The research, Por Ahí Dicen, was funded by a 5-year NIH P60 award for the Center of Health Equity Intervention Research (CHEIR) from the National Institute of Minority Health and Health Disparities (NIMHD) #P60MD006912.