Title: Recruitment and Retention in Intimate Partner Violence Research Series: Final Report Author: Mary Ann Dutton; Amy Holtzworth-Munroe; Ernest Jouriles; Renee McDonald; Satya Krishnan; Judith McFarlane; Cris Sullivan Unpublished: Subject: Violence against women 60 pages 154,000 bytes --------------------------- Figures, charts, forms, and tables are not included in this ASCII plain-text file. To view this document in its entirety, download the Adobe Acrobat graphic file available from this Web site or order a print copy from NCJRS at 800-851- 3420 (877-712-9279 For TTY users). --------------------------- Cover Sheet: The author(s) shown below used Federal funds provided by the U.S. Department of Justice and prepared the following final report: Document Title: Recruitment and Retention in Intimate Partner Violence Research, Final Report Author(s): Mary Ann Dutton; Amy Holtzworth-Munroe; Ernest Jouriles; Renee McDonald; Satya Krishnan; Judith McFarlane; Cris Sullivan NCJ 201943 Received: September 5, 2003 This report has not been published by the U.S. Department of Justice. To provide better customer service, NCJRS has made this federally funded grant final report available electronically in addition to traditional paper copies. This report was supported by the National Institute of Justice under Purchase Order #2002M_183. Opinions or points of view expressed are those of the author(s) and do not reflect the official position or policies of the U.S. Department of Justice. -------------------------- Recruitment and Retention in Intimate Partner Violence Research Mary Ann Dutton, Ph.D. Georgetown University Medical Center Amy Holtzworth-Munroe, Ph.D. Indiana University Ernest Jouriles, Ph.D. and Renee McDonald, Ph.D. University of Houston Satya Krishnan, Ph.D. New Mexico State University Judith McFarlane, R.N., Dr. PH, FAAN. Texas Women's University Cris Sullivan, Ph.D. Michigan State University This document is a research report submitted to the U.S. Department of Justice. It was supported by the National Institute of Justice under Purchase Order #2002M_183. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not reflect the official position or policies of the U.S. Department of Justice. --------------------------- TABLE OF CONTENTS INTRODUCTION MARY ANN DUTTON Review of Research Examining Recruitment and Retention --Methods of Recruitment --Recruitment of Special / Vulnerable Populations --Review of Studies Examining Retention Recruitment and Retention in Intimate Partner Violence Research --Attrition in Studies Involving Batterer Treatment --Retention in Longitudinal Studies Involving Intimate Partner Violence CASE STUDIES INVOLVING INTIMATE PARTNER VIOLENCE RESEARCH Issues in the Sampling, Recruitment and Retention of Women and Children Seeking Services at Women's Shelters ERNEST N. JOURILES AND RENEE MCDONALD Documenting Intimate Partner Violence in Border Communities: Research along the U.S. - Mexico Border SATYA KRISHNAN Intimate Partner Violence Victims in Urban Health Care and Justice Settings JUDITH MCFARLANE Safely and Successfully Retaining Battered Women in Longitudinal Research CRIS SULLIVAN Recruiting and Retaining Convenience Samples of Maritally Violence Men AMY HOLTZWORTH-MUNROE GUIDING PRINCIPLES FOR INTIMATE PARTNER VIOLENCE RESEARCH: RECRUITMENT AND RETENTION MARY ANN DUTTON REFERENCES --------------------------- Participant recruitment is an essential task in any empirical research study involving humans. Indeed, the effectiveness of recruitment has important implications for the validity of results, as well as for the management of ethical and human subjects concerns. Attending to the potential vulnerabilities of the population from which study participants are sampled is critical as a human subjects issue. Yet, including special populations in research and also protecting their vulnerabilities is a skillful balancing act (Meaux & Bell, 2001). Finally, recruitment procedures that fail to sample adequately the population of interest can potentially skew study results or reduce their generalizability. Yet, few studies have addressed the effectiveness of recruitment procedures (Froelicher & Lorig, 2002), especially for intimate partner violence research. There has been little research examining recruitment and retention issues in intimate partner violence research. Both recruitment and retention issues are relevant to research involving victims and offenders, women, men, and children. Successful recruitment is important in research, since it is necessary to obtain adequate sampling of the population under study. Further, recruitment methods vary in their cost. Unsuccessful recruitment strategies unnecessarily expend time and resources. Recruitment is an important issue for another reason. Inadequate methods may increase the risk of unintended disclosure of information by participants; of safety concerns, such as retaliation using threats, violence, or other acts of abuse against intimate partner violence victims by the offenders or other family members; and of feeling stigmatized or experiencing emotional discomfort. Retention in research involving intimate partner violence is a second major methodological concern. Poor follow through with program referral and high attrition rates compromise efforts to experimentally examine outcomes in batterer treatment studies (Gondolf, 2001). Low retention rates compromise the scientific quality of research studies. Maintaining adequate retention rates requires careful planning and resources. Retention issues are most relevant for research involving multiple data collection points, such as those employing follow-up data collection, longitudinal designs, and intervention studies. This report is organized into four major sections. First, the literature pertaining to recruitment and retention in research studies is reviewed. This review focuses primarily on methods of recruitment, recruitment within special or vulnerable populations, and retention. Second, recruitment and retention in intimate partner violence research is addressed with particular attention to attrition in batterer treatment programs and retention in longitudinal studies. Third, case studies highlighting recruitment and retention methods in five different intimate partner violence research programs are described by their respective investigators. Here we highlight five successful research programs. We acknowledge that successful recruitment and retention are difficult tasks. Our hope is that the approaches developed from many past lessons learned will assist others in their efforts to recruit participants successfully into research studies involving intimate partner violence and to retain them safely and effectively for follow-up participation. Finally, as a synthesis of the material that precedes it, a set of guidelines is offered for recruitment and retention in intimate partner violence research. This is intended to help guide the field by offering suggestions that can lead to successful recruitment and retention efforts in a way that maintains the safety of participants and researchers alike. REVIEW OF RESEARCH EXAMINING RECRUITMENT AND RETENTION The direction for intimate partner violence researchers can be informed by the findings of researchers in other areas. Accordingly, the literature concerning recruitment and retention generally is reviewed next. It provides a context within which to later consider these issues within INTIMATE PARTNER VIOLENCE research. Methods of Recruitment Recruitment methods are an important consideration in study design. Methods of recruitment vary, yet it is only recently that researchers have begun to compare empirically various methods. For example, in a study of sexual coercion involving college students (Senn & Desmarais, 2001), gender of both participant and recruiter influenced recruitment success. One study that explicitly compared recruitment methods examined recruitment and retention of healthy community-based women (age 40-48 years) from three ethnic groups (African Americans, non-Hispanic European Americans, Mexicans/Central Americans) into a subject pool for a longitudinal study (Gilliss et al., 2001). They found that face-to-face recruitment produced the highest yield of eligible participants (84% eligible) with the lowest attrition rate (7%), but required more resources than did other methods. Media advertisements produced the largest group of potential subjects, but also the highest rate of ineligible women and the highest attrition rates. Printed matter produced large numbers of potential subjects, but ineligibility was high (53%). Direct referral was reasonably efficient (57%) as a means of recruitment. These authors recommended multiple approaches for recruitment for maximal success. Another study (Nazemi, Larkin, Sullivan, & Katon, 2001) also recommended the simultaneous use of both waiting room screening and physician referral to recruit the greatest number of primary care patients in the least amount of time into a study of depression. In this study, males and older patients were more likely to refuse participation in the waiting room recruitment method. Telephone recruitment of 368 African American and White youth and all household family members in 58 neighborhoods was used in a study of risky health-related behavior (Duncan, Strycker, Duncan, He, & Stark, 2002). Of all calls, 45% resulted in contact; 1% qualified, of which 75% agreed to participate. Door-to-door recruitment was used as a supplement to recruitment. This study found the average cost to telephone recruitment was $169 per family. Another study examined cost of recruitment in a geographically diverse study on smoking (McIntosh, Ossip-Klein, Spada, & Burton, 2000). Multiple recruitment channels were used, including 1) multiple paid newspaper advertisements, 2) free media (TV and radio) advertisements, 3) referrals, 4) HMO newsletters, 5) targeted mailings, 6) face-to-face contact, and 7) passive recruitment. The most reliable, cost-controlled method was paid newspaper advertisements ($18-$19). Face-to-face contact was the most inefficient and costly ($140) per enrolled subject. A study of community-dwelling older adults experiencing generalized anxiety disorder found that media sources produced both the greatest number of inquiries and study participants, whereas healthcare provider referrals accounted for the least number (6%) of participants. Researchers suggested a need for greater collaboration between medical and mental health practitioners to improve recruitment efforts. Two additional studies examined the use of community-based resources for recruitment. One study recruited African Americans using church rosters (Carter-Edwards, Fisher, Vaughn, & Svetkey, 2002) for a health-related survey conducted in the church or in participants' homes. This low-cost effective recruitment tool involved frequent contact with pastors and church representatives, presentations, standard and tailored recruitment approaches and biannual progress reports. Rapport with church representatives and congregations was deemed critical, as was trust. The second study of these two studies (Dietze et al., 2002) utilized ambulance attendees (Melbourne, Australia) or researchers traveling with ambulance paramedics (Sydney, Australia) to recruit revived residents at an overdose scene. This recruitment method in Melbourne involved distribution of 281 cards over a 7-month period which resulted in an overall 23% (n = 66) contact rate and a 13.8% (n = 39) interview rate. In Sydney, 170 initial contacts resulted in 82% willing to answer questions at the scene (n = 139) and 28% completing an interview. Authors concluded that recruitment through contact with an ambulance service in a novel method with distinct advantages. Collaboration with Community Agencies As noted above in several studies, effective recruitment often requires collaboration with community agencies or organizations, especially when potential participants are clients involved in that agency or organization. Collaboration is a two-way effort and there must be a benefit to both researchers and community-agencies for participating in research collaboration. A recent article (Thomas, 2002) suggested that providing active feedback to agencies through a clinical report and through a research-agency liaison was a successful strategy for developing a collaborative atmosphere that resulted in improved recruitment, motivation and reduced attrition among therapists and clients as research participants. Further, a report of focus groups involving researchers and practitioners (National Violence Against Women Prevention Resource Center, 2001) provided guidelines for collaboration in intimate partner violence research. These include recommendations to: 1) discuss all aspects of the collaboration, 2) establish a shared vision and goals for the joint project, 3) be certain that goals are clearly stated and understood by key participants, 4) involve both researchers and practitioners/advocates in the planning of each phase, 5) ensure that all questions are answered adequately, 6) ensure that responsibilities are divided in ways that are reasonable, fair, and sensitive to time constraints, 7) provide material and other support, and 8) make it a goal to secure funds for all involved in the collaboration. Recruitment of Special / Vulnerable Populations Intimate partner violence research often involves special or vulnerable populations. These may include individuals who have been recently victimized by violence and abuse, children who have witnessed intimate partner violence or have been abused directly, or persons who are subject to criminal or civil sanctions resulting from an arrest for engaging in intimate partner violence. Research on recruitment of special populations has described several obstacles to recruitment. These have focused on older adults (Akkerman et al., 2001; Boles, Getchell, Feldman, McBride, & Hart, 2000; Hawranik & Pangman, 2002), Asian immigrants (Weiss & Weiss, 2002), Latinos with HIV/AIDS (McQuiston & Uribe, 2001), family members with Alzheimer's disease (Connell, Shaw, Holmes, & Foster, 2001), American Indians (Stoddart et al., 2000), African-Americans (Hatchett, Holmes, Duran, & Davis, 2000; McNeilly et al., 2000; Sinclair et al., 2000; Zhu et al., 2000), mothers of child abuse victims (Kinard, 2001), and Chinese-American caregivers of dementia patients (Hinton, Guo, Hillygus, & Levkoff, 2000). Numerous problems were encountered in these studies. They included difficulty locating potential participants; potential participants identifying the focus of the study (e.g., dementia) as "normal" rather than as a problem; participants' concern about confidentiality, class differences between researchers and participants; lack of phones for follow-up contact; hesitancy by participants in providing feedback to researchers; participants' unwillingness to travel to the research site; participants' reluctance to give up their choice (e.g., treatment alternatives) required by involvement in a treatment study; recruiter/participant gender match; participants' physical, social, psychological, and age- or health-related issues that made participation more difficult; language differences between researchers and participants; participants' skepticism about the research process sometimes related to prior research abuses; no direct benefit for participation; participants' perception of research as intrusive; participants' lack of time and resources required to participate; participants' stigma or difficulty accepting the diagnosis (e.g., Alzheimer's) for which the recruitment was targeted; participants' concern that involvement in the research process would create excessive worry for themselves; and agencies' overwhelming service demands and high employee turnover. Indeed, barriers exist with community agencies, gatekeepers, and individual participants alike (Levkoff, Levy, & Weitzman, 2000). Barriers have also been noted in recruitment of professionals working with target populations. Specifically, recruiting rates of mothers have been shown to be lower in families involving child abuse, especially when the child abuse was sexual, vs. physical abuse (Kinard, 2001). In a study recruiting general practitioners to screen for mental health issues (Harris, 2000), a key barrier was the small number of patients with the identified mental health issue in the general practitioners caseload. Recommended incentives were specialist mental health support from the research team and inclusion of practitioners on the research team. Methods for increasing recruitment success have been suggested in the literature. Researchers studying African American and White Alzheimer patients' family members suggested efforts to increase the perceived benefit of research participation by providing regular personal contact with research staff, information about health status changes in the research participant, and short- and long-term results of the research studies in which they are participants (Connell et al., 2001). Researchers studying elderly African Americans suggested that three factors were critical for successful recruitment: 1) degree of convenience for the research participants, 2) degree of perceived physiological or psychological invasiveness of research procedures, and 3) trust established between the researcher or research institution and the community at large. Another group of researchers involving older African Americans suggests that researchers must 1) be sensitive to population concerns, 2) build working relationships with community "gatekeepers", 3) establish relations with the elderly African American community, 4) be aware of the impact of staff's demographic characteristics on the population, 5) provide transportation, 6) address safety concerns, 7) provide project materials that are clear and easy to read, and 8) provide sign language and hearing-enhanced devices, as needed. A study of breast cancer screening among older African American women (Zhu et al., 2000) reported that a strategy which tailored recruitment to the cultural, perceptive, and cognitive characteristics of the population was effective. One model for recruiting minorities into health care research (Levkoff et al., 2000) suggested formation of a community advisory group, sensitivity to words used on forms, and a consent process that respected family hierarchy, mixing quantitative and qualitative methods. They indicate that the match between the goals of a minority and research communities is important in determining recruitment success and that recognizing and understanding the culture of each community is a prerequisite to a collaborative match. Commitment from the community is commonly viewed as essential (Stoddart et al., 2000). Review of Studies Examining Retention Retention in research involving intimate partner violence is a challenge, whether the research is testing the effectiveness of interventions or is based on a longitudinal or follow-up design. Retention in research studies has been shown to be more difficult with ethnic minorities, younger participants or the elderly, those with greater psychiatric problems, and the unemployed. Generally, more unstable status is associated with worse retention. A study of client matching in substance abuse treatment - involving outpatient, intensive outpatient, or residential - (Klein, di Menza, Arfken, & Schuster, 2002) found evidence that client-treatment setting interactions influenced retention. They warned that client characteristics may bias estimates of effectiveness generally, or for specific populations. Much of the research on retention involves drug treatment studies. A study of factors related to retention in the National Institute on Drug Abuse Collaborative Cocaine Treatment Study (Siqueland et al., 2002) found that younger, African-American, and unemployed patients were retained for fewer days. Gender differences were also found. Greater psychiatric severity kept men in treatment, but put women at risk for dropping out sooner. A study of HMO members recruited into a substance abuse treatment study (Green, Polen, Dickinson, Lynch, & Bennett, 2002) found that retention for women was enhanced by legal/agency referral and alcohol or opiate diagnoses. For men, fewer mental health diagnoses, higher education, being a victim of intimate partner violence, and prior 12-step attendance were related to more time spent in drug treatment. Two studies of coercive strategies used to compel offenders in the criminal justice system to attend a long-term residential treatment program (Young, 2002; Young & Belenko, 2002) found that perception of legal pressure was related to retention in treatment. Researchers suggested that the use of structured protocols for informing clients about the conditions and legal contingencies of treatment participation and of how their participation will be monitored, as well as developing the capacity to enforce threatened consequences, are effective approaches to retention in mandated treatment programs. A study of pregnant crack users (Fiocchi & Kingree, 2001) found that prior psychiatric hospitalization was associated with worse retention. Previously hospitalized women were more likely to depart the treatment facility prior to delivery. However, no other client characteristics were associated with retention. In contrast, another study of retention of incarcerated substance abusers in a therapeutic community (Nielsen & Scarpitti, 2002) failed to find predictors of retention from demographic or background factors, criminal history, prior substance abuse, prior treatment experience, motivation and readiness, psychological factors, or legal pressure to attend treatment. Gender differences in substance abuse treatment have been demonstrated. In a study of men and women in Detroit's publicly funded substance abuse treatment system, women, who had significantly more problems at assessment, were less likely to stay in treatment for 30 days or to complete treatment compared to men. Another study which examined the impact of mixed-gender vs. single-gender substance abuse treatment on retention in a treatment study (Bride, 2001) found that the single-gender settings did not significantly increase treatment retention and completion. The authors suggested that gender-specific treatment must do more than provide traditional treatment in a single-gender environment. Therapeutic alliance in intervention studies has been considered an important predictor of retention in treatment studies. In a study of cocaine-dependent outpatients (Barber et al., 2001), greater therapeutic alliance in treatment was associated with retention in some treatment conditions (supportive-expressive therapy, individual drug counseling), but not others (cognitive therapy), even though cognitive therapy had the greatest retention rate compared to the other types of intervention. These findings suggest the need to examine the relationship between alliance and retention in future studies. A study examining the influence of staff experience on retention in long-term residential drug treatment (De Leon, Hawke, Jainchill, & Melnick, 2000) found that intervention by senior staff significantly reduced the likelihood of dropout after 30 days. These results were especially evident for new patients, with the lowest level of motivation. A study of patients involved in a multidisciplinary team model of HIV care (Sherer et al., 2002) found that support services increased retention in HIV primary care. These authors suggest that health services tailored to the needs of the patients lead to better improved health outcomes. Conclusion Previous studies illustrate the wide array of recruitment strategies that have been previously employed. Notably, the "success" of recruitment can be measured in several different ways. Maximizing the number of people who are potential participants vs. those who are eligible or who consent to participate are separate recruitment considerations. Depending on the nature of the study and the scarcity or characteristics of the target population to be studied, the required emphasis may differ. These studies point to the importance of reporting recruitment success as part of a research report which, even without multiple recruitment methods from which to make comparisons, would contribute to the field by providing some guidance to other researchers. Examination of previous studies provides evidence that retention in research studies is influenced by many different factors. These studies suggest that variables related to greater social (e.g., ethnic minority status, elderly, educational status) or emotional (e.g., prior psychiatric hospitalization or greater severity of problems) vulnerability increase the difficulty in retaining participants in research studies. Importantly, however, the findings were not consistent in this regard (see Nielsen & Scarpitti, 2002). Nevertheless, these findings point to the importance of recognizing the types of vulnerability which may characterize the participant sample and tailoring retention strategies to address those potential concerns. --------------------------- RECRUITMENT AND RETENTION IN INTIMATE PARTNER VIOLENCE RESEARCH Recruitment and retention are relevant for intimate partner violence research in numerous ways, yet there has been little research focusing directly on either of these areas. However, two related issues can be discussed based on the literature to date. These include attrition in batterer treatment and retention in studies involving multiple assessments over time. Nevertheless, those studies who report findings describing factors specifically related to recruitment or retention are few. Attrition in Batterer Treatment Attrition in mandated batterer treatment programs is a serious concern for researchers examining treatment outcomes (Gondolf, 2001), for court systems that mandate their participation, and, perhaps most importantly, for abused partners who are often overly optimistic about the likelihood of their successful completion in batterer treatment programs (Heckert & Gondolf, 2002). Two types of attrition affect mandated batterer treatment programs. The first is failing to follow through with a judge's order to attend a batterer treatment program or attrition prior to entering the program. The second type of attrition refers to the failure to complete the program or premature dropout, once treatment has begun. Even among those who successfully enter batterer treatment programs, the rates of attrition, or failure to complete the required sessions, is alarmingly high. For example, in a study of 15 treatment groups for intimate partner violence involving 104 men, the attrition rate was 68.3% (completion rate of 31.7%) (Tutty, Bidgood, Rothery, & Bidgood, 2001). Most importantly, program dropout has been associated with worse outcomes, including both physical (Gondolf, 2000; Gondolf & Jones, 2001) and nonphysical (i.e., control, verbal, threats, combined) (Gondolf, Heckert, & Kimmel, 2002) abuse. Research examining completion rates in culturally-focused counseling for African American men arrested for intimate partner violence (Gondolf & Williams, 2001) found higher rates in both specialized (50%) and conventional (50%) counseling groups, compared to racially-mixed groups (37%) (Gondolf, 2003). For African-American men with high cultural identification, the rate was even higher in culturally-focused groups (70%), compared to the total African American group (50%), and racially-mixed groups (32%). These results show considerable promise for improving batterer program completion among African American men. Further, they suggest promise for developing culturally-relevant services more generally for both offenders and victims. Several studies have examined predictors of attrition in batterer treatment programs. Several factors (class status, instability) are thought to influence attendance by offenders (Daly, Power, & Gondolf, 2001). Indeed, Daly and colleagues found that the less educated, unemployed at intake, not court ordered, and those with a history of alcohol-related problems completed fewer sessions. However, another study of 91 male court mandated batterers (Buttell & Pike, 2002) found only few demographic or psychological variables to predict program dropout. Finally one study considered the deterrent effect of both certain and severe sanctions (Heckert & Gondolf, 2000). These investigators found that neither perceived certainty nor severity of sanctions was predictive of program dropout (or reassault). Gondolf and his colleagues (Gondolf & Jones, 2001; Jones & Gondolf, 2002) have used instrumental variable analysis to address the issue of the relationship between program dropout and batterer reassault, a method which takes into account possible confounding due to unmeasured traits of program completers vs. dropouts. Retention in Longitudinal Studies Involving Intimate Partner Violence Retention in longitudinal studies involving intimate partner violence has been a challenge, especially considering the circumstances under which they are often recruited. Current or recent victims of intimate partner violence are typically dealing with safety issues, coping with traumatic reactions to violence and abuse, and making decisions and difficult transitions in their lives. This means that they may be involved in transient living situations, in crisis, or at least distracted by dealing with the aftermath of violence and abuse in their lives. These factors add to the challenges of recruiting and retaining samples from economically oppressed or unstable populations. Yet, researchers have developed effective methods for obtaining high levels of retention, with at least some samples. In the case studies below, see an example involving intimate partner violence victims recruited from a shelter (Sullivan & Bybee, 1999; Sullivan, Rumptz, Campbell, Eby, & Davidson, 1996) and heterosexual couples involving a community sample of male perpetrators of intimate partner violence (Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2003). -------------------------- CASE STUDIES INVOLVING INTIMATE PARTNER VIOLENCE RESEARCH Below, five case studies involving intimate partner violence research are presented. They include an advocacy-based intervention study involving recently sheltered victims (Sullivan), a study involving minority intimate partner violence victims, especially Hispanic and Native American women (Krishnan), a study of women exposed to intimate partner violence recruited from urban health care and justice settings (McFarlane), a study of women and children seeking services at a domestic violence shelter (Jouriles & McDonald), and a study involving a community sample of men as perpetrators of intimate partner violence (Holtzworth-Munroe). -------------------------- Issues in the Sampling, Recruitment and Retention of Women and Children Seeking Services at Women's Shelters Ernest N. Jouriles and Renee McDonald Description of Studies The focus of our research is on children who accompany their mothers to women's shelters. Shelter residence is typically prompted by violence directed at the mother by an intimate partner; thus, the mothers are the most obvious victims of violence in shelter settings. However, a growing body of research suggests that the children residing at these shelters are victims as well. Specifically, many children brought to women's shelters experience significant mental health problems, and remain at risk for such problems following shelter departure (see Jouriles, Norwood, McDonald, & Peters, 2001; for a review). Because the shelter setting readily affords opportunities for service delivery, knowledge about the children brought to these settings can serve an important public health function. We have conducted a number of studies designed to help us better understand and ameliorate the problems of children brought to women's shelters. For example, we are among those investigators who initially attempted to document a relation between domestic violence and child problems, ruling out important alternative explanations (e.g., Jouriles, Murphy, & O'Leary, 1989). We have begun to test theory regarding the processes that link domestic violence to child problems (e.g., Grych, Fincham, Jouriles, McDonald, 2000; Jouriles, Spiller, Stephens, McDonald, & Swank, 2000) and, in this context, have examined links between domestic violence and child abuse (e.g., Jouriles & LeCompte, 1991; Jouriles & Norwood, 1995). We have examined the stability of child problems following shelter departure (Ware, Jouriles, Spiller, McDonald, Swank, & Norwood, 2001) and have also begun to evaluate interventions to reduce conduct problems among children brought to women's shelters (Jouriles, McDonald, Spiller et al., 2001). The families (women and children) who seek services at the women's shelters in and around Houston are culturally diverse. In fact, these shelters, as a group, house approximately equal proportions of families of African American, Hispanic, and European American descent. As one might expect, many of the families come to these shelters from urban neighborhoods. Yet, a sizable proportion are from rural settings as well. A common characteristic among most of these families is that they are extremely socioeconomically disadvantaged. Most of our studies have involved gathering information directly from the children and their mothers. Although many of our studies have been limited to cross-sectional assessments conducted during shelter residence, several have tracked families following their shelter departure. In most of our shelter studies, our participation rates for our research are over 85%. In our longitudinal research, retention rates exceed 90%. Critical Issues in Sampling, Recruitment, and Retention There are many issues pertaining to the sampling, recruitment, and retention of families seeking shelter because of domestic violence. Of course, the specific nature of these issues depends, in part, on the scope and aims of the research. Our focus here is on research with samples of families seeking services at a single shelter or, perhaps, a small sample of shelters. Some of the points we make are applicable to research with other samples. Some are perhaps less relevant or not cost-effective for research requiring broadly representative samples (e.g., a sample that is nationally representative of families seeking shelter). The issues that we cover are also applicable to research studies with varied aims, ranging from studies attempting to understand the processes linking domestic violence with child problems to more applied investigations, such as those evaluating interventions designed to assist families seeking shelter services. Prior to initiating research involving the clients of a shelter, we attempt to develop a sound working relationship with the leadership of the agency sponsoring the shelter - which in our experience has typically been a women's center - and with the staff at the shelter. We do this by providing volunteer services to the agency, helping where our capabilities fit the agency's needs. Such activities have been quite broad-ranging. For example, we have served as a source of information and referrals, helped to develop and evaluate new programs, provided staff training, helped with agency fund-raisers, and we regularly collect and donate personal items to the shelter or the agency thrift shop. In short, we learn and demonstrate recognition of the needs of the agency and its clientele and provide concrete evidence of a commitment to help the clients. At a couple of the shelters, we have also staffed and supervised an after-school program for the children in the shelter. Through such activities, a trusting and positive working relationship is established and maintained - we are viewed as partners in the efforts to help the women and children who are victims of domestic violence and are considered to be a resource to the agencies. Sampling. One advantage of sampling residents at any particular shelter is that there is a clearly defined sample frame (set of families that has a chance to be chosen). However, obtaining a "representative" sample of residents at a shelter can be quite challenging. An initial step necessitates consideration of the definition of the term representative. That is, does the research require a sample that is representative of a shelter on a given day? Month? Year? Such consideration is necessary because of the variable duration of most families' shelter stays. Specifically, most of the shelters we work with grant families an initial stay of up to 30 days, with some families obtaining extensions. Many families, however, remain in shelters for only a few days, and some depart with little or no warning. On any given day, a shelter may house mostly "long-term" families (those that are staying at the shelter for several weeks or more). Yet, a representative sample of shelter residents over the course of the year will most likely consist of a greater proportion of "short-term" families (those staying at the shelter for only a day or two). Thus, to obtain a sample that is representative of a shelter in a given month or year, the shelter census may need to be checked frequently (e.g., every day or two in "busy" shelters) over a period of time to identify and recruit new residents before they exit. This issue makes it very unlikely that a sample that represents "typical" shelter residents can be obtained if recruitment and/or data collection is limited to a very short time frame (e.g., one or two days). Another challenge is obtaining an adequate sample with respect to size. Because the number of families who can be housed at any given time is limited by the capacity of the shelter, the number of potential research participants is limited as well. This figure is reduced even further when investigators must impose specific eligibility criteria (e.g., families with a child within a specific age range) for research participation. One possible solution is the simultaneous recruitment of families from multiple shelters. However, given that women's shelters typically serve a particular community (e.g., city, county), they are often geographically somewhat distant from one another. Consequently, research that involves multiple shelters often incurs nontrivial costs related to additional staff required for recruiting from additional shelters and expenses for travel to, from, and between shelters (i.e., travel time, mileage and fuel costs). Recruitment. We have found a number of strategies helpful in the recruitment of shelter residents for research. In line with our efforts to develop sound working relationships with agency and shelter staff, we demonstrate a commitment to helping the individual families seeking services at the agency. Again, this can be accomplished through a variety of activities. For example, we help women access goods and services for which they are entitled or in need; we volunteer to supervise and play with the children for brief periods, so women can attend to other urgent matters or simply relax. Our efforts on behalf of individual agency clients - who form our pool of potential research participants - lends us credibility and increases their trust that we care about their well-being. It also keeps us well in tune with the real issues confronting our research population. The importance of these human relationship issues to successful recruitment efforts should not be underestimated. They can greatly influence the willingness of others (individuals as well as agencies) to participate in research endeavors, and in the communal living environment of shelters, information and reputation spread rapidly. Importantly, our own staff is responsible for all of our research recruitment; that is, we do not rely on shelter staff for assistance with our research. Shelter staff are often over-burdened with their own responsibilities and thus are not likely to be able to recruit families in a systematic manner. They may not be able to respond to questions about the research accurately and may see the "extra work" they are being asked to do as superfluous or burdensome. Although it requires more labor and time to do all of the recruitment activities with project staff, the returns in research participation rates and maintenance of good working relationships with agencies is more than worth the effort. In our efforts to recruit research participants, we show an appreciation for the immediate circumstances of the families. Many women in these shelters are emotionally distressed and exhausted from the circumstances that led them to seek shelter in the first place, and worried about what the future holds for themselves and their children. Moreover, during shelter residence, most women must necessarily focus on urgent, immediate concerns, such as physical injuries or illnesses; finding a job and a permanent, affordable place to live; and attending to legal issues such as protective orders, child support and divorce. Needless to say, their immediate circumstances are often not very conducive for recruitment to participate in research. Our project staff engage in rapport-building interactions with the mothers prior to talking with them about a specific research project. Often, this is simply visiting with mothers while they are at the shelter, expressing sympathy and understanding of the families' circumstances, and facilitating access to resources in the shelter and community. We typically do not attempt to recruit families for participation in research until they have had a chance to settle in at the shelter for a day or two. We do whatever we can to make it easy and comfortable for women and children to participate in our research. For example, project staff are available during various times of the day so that women's schedules can be accommodated. The recruitment efforts as well as the actual research are scheduled according to the families' availability; evening and weekend appointments are always available. In our efforts to explain research to families, we present the broader scope and objectives of our research program, which is "understanding the ways in which domestic violence affects children," or "developing effective treatments for children exposed to domestic violence," as well as the narrower objectives of specific research studies. Whenever possible, mothers are compensated financially for the time spent participating in the data collection efforts. We provide child-care for families while we explain the research and answer questions about participation, and we let mothers know that we provide child-care during the research interviews as well, so that child-care does not pose an obstacle to participation. In our research, we interview mothers and children privately and independently from one another. To minimize the possibility of embarrassment for women with low educational attainment or poor reading skills, we always offer mothers the option of completing questionnaires themselves or having the questions read aloud by the interviewers, indiciating that our familiarity with the questionnaires sometimes makes it "go faster" if we read the questions aloud. Questionnaires are always read aloud to the children. In addition, we try to make the assessment process as fun as possible for the participants, especially the children. We play lots of games with the children over the course of an assessment period, and we break for snacks or meal time if warranted. Interestingly, we have been told by many of the mothers that their children really enjoyed their participation, and we have observed some children eagerly ask their mothers to let them participate again because it was so much fun. The cultural and ethnic backgrounds of shelter residents are also important to consider in recruiting samples that are reflective of the shelters from which they are drawn. In particular, in shelters in which large numbers of Hispanic women seek services, bilingual research staff are necessary to be able to fully explain what participation in research will involve, as well as to administer instruments in Spanish, if the research design allows. When determining women's interest in and eligibility for our research projects, we have asked the women if they have particular preferences regarding the gender or ethnicity of research staff who would be working with them. Only a very few women stated a preference, and it was usually to work with a female rather than a male. Although matching staff with families on the basis of ethnicity is not always feasible, one thing that has been feasible, and is always desirable, is to employ staff that are reflective of the diversity in the shelters and the greater community of which we are a part. One potential challenge for investigators in the recruitment of shelter families pertains to the location of the research and how families are to get to that location. At the shelters where we work, over half of the women do not own or otherwise have access to an automobile, and a number of the shelters are located in areas without public transportation. Under these circumstances, getting the women to and from a location where an interview or assessment is to be conducted can be difficult. Even in areas where public transportation is available, travel to a research site requires women to coordinate bus schedules, sometimes bus transfers, and either obtaining child care for their children or taking their children on the bus with them. While this may sound simple enough for many, it should be remembered that women in shelters are often financially very impoverished, and it can be cost prohibitive for a woman to transport herself, and her children across town on a bus - even if she is to be reimbursed for these expenses after she participates in the research. Such logistical issues complicate and increase the cost of collecting data at sites other than the shelters themselves. In addition, it has been our experience that shelters vary quite a bit with respect to available space for research. That is, appropriate space for meeting privately with research participants at these shelters (e.g., for interviewing or conducting assessments) is sometimes an issue. Retention. There are many challenges facing investigators who are interested in tracking families following their shelter departure. Most families who seek refuge at a domestic violence shelter were living in poverty prior to shelter residence. Many - especially families in which the mother does not return to her violent partner - must move into even more impoverished conditions following shelter departure. Their financial disadvantage, coupled with the fact that many of these women are attempting to hide from very violent men, contribute to difficulties in tracking over time. Specifically, safety concerns, inability to pay rent, and evictions, among other reasons, result in frequent moves for many of these families following shelter residence. In addition, many go for extended periods of time without phone service, making it more difficult and costly to contact them on a regular basis. We maintain very regular contact with families we follow longitudinally. We developed tracking procedures that not only ensured that we could continue to follow families but also communicated to them that we cared about their condition. Again, whenever possible, mothers are compensated financially for the time spent participating in the data collection efforts. Each family is contacted monthly, either by phone or in person, and at each of these contacts, we inquire about their needs. These monthly contacts also often result in the provision of donated tangible goods and in helping the families make contact with community social service agencies when possible. Each family is also contacted during the holiday season, and through our own volunteer efforts and in coordination with those of the shelters, families are often provided with donated goods (e.g., household items such as sheets, pots and pans, school supplies, Thanksgiving dinners, birthday and Christmas presents). Our tracking protocol includes other procedures as well. For example, we secure the names, addresses and phone numbers of two individuals certain to know the whereabouts of our participants throughout the study. However, in addition to securing these names, we often have the mothers introduce us (by telephone) to these individuals at the beginning of a longitudinal data collection, so that they will know it is permissible for them to disclose to us the woman's whereabouts. We have found this to be extremely helpful with this sample of women, given that other individuals who these women are trying to avoid (ex-partners, bill collectors) are sometimes trying to locate them as well. Since many of our project participants do not have phones, we arrange for alternative methods for them to contact us and vice-versa. On some occasions we provide money or a prepaid calling card so they can contact us by pay phone; we arrange for them to obtain a free voice mailbox through a local agency so that we can let them know when we need to talk with them. We equip our project staff with pagers to facilitate immediate communication with project participants if they call. We also think it is important, whenever possible, to have the same staff member follow a family from the beginning of their participation to the end. Again, we always attempt to make participation in our research as easy as possible for participants. We are very flexible regarding when we can collect data, and many of our assessments are conducted during the early evenings and on weekends. We provide child-care during the assessments, and provide snacks or meals when needed, such as when lengthy assessments are scheduled through the dinner hour. As the design of the research allows, we offer to conduct assessments at a location that is most convenient to the families: their home, our lab, or another, mutually agreed upon place. We reimburse families for transportation costs (i.e., bus fare) if they elect to come to our lab. In projects in which data collection has taken place in the participants' homes, a number of considerations have emerged. We try to be sensitive to the differences in the material circumstances of the participants relative to those of our staff. Many of our research participants are self-conscious about their situation in our presence. For example, women have apologized for the condition of their apartments (no air-conditioning, stained carpets and walls), their meager furnishings (we have conducted interviews on living room floors because there were no chairs), or their general neighborhood surroundings (poor, inner city or rural areas). Our approach to such situations is based on the fact that we are invited guests in their homes: we always strive to be gracious guests and attempt to help the participant feel at ease. After a families' participation in a study has ended, we leave the door open for them to contact us if they feel we can be of help. In studies that involve extensive contact over time, we might have a small "farewell" party for the family, providing pizza and soft drinks, and small, donated gifts for the children, immediately after the last data collection period. Of the hundreds of families who have participated, a few have called us for assistance after their participation in our research has concluded, and some give us periodic updates on positive turns of events. When families have called for assistance, we do our best to help them find or obtain the goods or services they need. One slightly unexpected outcome of our longitudinal research was how attached some of our staff became with project families. That is, a number of our staff wished for continued contact with families - especially the children - after the project was terminated. In such cases, we asked staff to make sure that they clarified with participants that the project was over. We reminded our staff that their relationship with project families was professional. If participants initiated contact after the termination of the project (letter, phone call), we encouraged staff to respond briefly and appropriately. We encouraged them to offer and facilitate referrals as indicated, but we discouraged staff from more "personal" contacts, such as going to the families' home for dinner or for a child's birthday party. An important issue not yet addressed - the most important issue in many ways - is safety, that of the participants as well as the staff. Because our assessments have always included measures of domestic violence, we have been able to keep abreast of threats to the safety of the women and children by their partners/former partners, and to help the women respond to threats that have emerged. In addition, many of our studies have included assessment of aggression and violence toward children, necessitating attention to child safety and to legal requirements for reporting abuse and neglect. In almost all instances in which we have had to make such reports, we have been able to do so cooperatively with the mothers of the children involved, and have not had any participants withdraw from a study because of a report to children's protective services. Regarding staff safety, we have explicit safety rules regarding assessments done in families' homes. These rules include guidelines about monitoring the safety of neighborhoods, the potential for risk should a partner be at home during the assessment, and suggestions for reducing risk of property or violent crime not directly related to the research (such as theft of property from a car, or being robbed). An overarching instruction to staff regarding safety is never to do any research-related task that feels like it might be unsafe for themselves or another staff member. Conclusions Although there are many challenges to recruiting children and their mothers from women's shelters for participation in research and then tracking these families over time, we believe there are compelling reasons for learning more about these families. As indicated above, women's shelters can be conceptualized as a point of entry into the health-care and social services system. Given that a large proportion of the children who are brought to these shelters are experiencing significant mental health difficulties, and many appear to experience problems later in life, developing a more comprehensive understanding of the children brought to these shelters can enhance our ability to develop and offer effective services to this very needy, high-risk group. In addition, children who are brought to these shelters have typically experienced very frequent and severe domestic violence - violence at levels rarely experienced by children in most other settings (Jouriles, McDonald, Norwood, & Ezell, 2001). Thus, such children offer the rare opportunity to study how very frequent and severe violence might influence child development. In short, there are numerous potential benefits in learning more about the families seeking services at women's shelters. -------------------------- Documenting Intimate Partner Violence in Border Communities: Research along the U.S. - Mexico Border Satya Krishnan, Ph.D. The border between United States and Mexico is defined by a 2,000 mile-long stretch of land across the continent. This simplistic physical representation fails to capture the unique and complex blend of cultural, economic, political, and social factors that ultimately constitute the territory known as the "borderlands" (Ford, Barnes, Crabtree, Fairbanks, 1998). These borderlands are largely rural with a few cities interspersed. This geographic vastness creates a physical isolation that often limits interaction among and between service providers and community members. In addition, these borderlands are greatly influenced by poor economic conditions, fewer opportunities for educational attainment, and by the melding of multiple cultures including the dominant Hispanic and Anglo cultural traditions and customs as well as those of the Native American tribes. Finally, the permeability of the border between the two countries plays an important role in defining the nature of the communities that emerge and sustain in these borderlands. The permeability not only accounts for some of the expansion in the population that lives in these communities but allows individuals to move back and forth between the two countries on a regular and frequent basis (Ortega, 1992). It is important to understand the above-described backdrop to appreciate the challenges faced in the sampling, recruitment, and retention of participants in research studies particularly longitudinal studies that investigate issues of interpersonal violence. Description of Program of Research For the past seven years, my collaborators and I have conducted research studies to document and understand the violence experienced by women in their intimate heterosexual relationships in the borderlands. We have focused our attention and efforts primarily on women from minority communities. This focus on minority women, especially Hispanic (Mexican/Mexican American) and Native American women who are often oppressed by class, culture, ethnicity, economics, and race is imperative in states such as New Mexico which have a culturally diverse population living in them (Krishnan, Hilbert, & VanLeeuwen, 2001; Krishnan, Hilbert, VanLeeuwen, & Kolia, 1997) In our ongoing research work, we have recruited participants from domestic violence shelters, homeless shelters, and from hospital emergency departments located in the border communities of Southern New Mexico (Hilbert & Krishnan, 2000; Krishnan, Hilbert, & Pase, 2001). The participants were recruited from these locations, not so much for their representativeness but because of what we believed and has been suggested by Wuest and Merritt-Gray (2001) that these study participants are "expert sources" of the knowledge and experience we hoped to document and understand. We began with small descriptive studies that provided us with useful information and experience in recruiting, retaining, and working with women living in border communities and experiencing violence in their intimate relationships (Krishnan & Hilbert, 1998) About four years ago we began using larger comparative and longitudinal sampling techniques and research methodology. From the start, we have blended qualitative and quantitative data collection methods (Krishnan, J., McNeil, & Newman, in press). This combination of longitudinal study design and blended data collection method have been well-suited to the population of interest and very helpful in providing detailed information about "trajectories" (or long-term patterns of behaviors) as well as about transitions (that evolve over shorter periods of time) that occur in the violence and the lives of our study participants. In our view, these approaches are particularly relevant to the study of interpersonal violence in the borderlands and over a life course that Caspi and colleagues (1990) have described as "a sequence of culturally defined age-graded roles and social transitions that are enacted over time" (Pg.15). As in other longitudinal studies, sampling, recruitment and retention of study participants have been critical components of our efforts that have offered us avenues to be creative and at times innovative while posing some unique challenges to our work in the borderlands. These critical issues are discussed next. Critical Issues in Sampling, Recruitment, and Retention Sampling. The sampling techniques that we have utilized in border communities appear to have a variety of advantages in the recruitment and retention of participants. First, study participants recruited at the shelters have usually sought the services at these places and have been interested and willing to participate in research studies. Additionally, they seemed to be in a more protective environment that allows them the opportunity, time, and space to participate in such studies. Further, our research methodology has included both qualitative and quantitative data collection techniques that allow participants to describe (often in great detail) the violence they have and continue to experience in their past and current relationships, the context, circumstances, and conditions in which this violence has occurred as well as the cultural and social influences, factors, and other considerations that define and shape their experiences. In our recent longitudinal study, participants were compensated for their time that progressive increased over the 18-month period to encourage long-term participation. The financial incentives ranging between $15.00-$40.00 (data collection every 3 months for 18 months), in our view did not amount to coercion but rather a necessary compromise between ethical and practical considerations. It has been important to adhere to the essential elements of sampling, which in our view include informed consent, voluntary participation and ability to withdraw from the study at any time. In addition, it also meant hiring research assistants and interviewers who are well trained and who speak Spanish and English fluently and comfortably, incorporation of strategies that help develop rapport and effective professional relationships with study participants, and provision of initial and ongoing training to research staff. We also considered and incorporated the following key elements in conducting our research on interpersonal violence among the population described earlier. These included allowing for data collection at times and places selected by study participants, protecting the privacy of participants and maintaining the confidentiality of the data collected, developing a data collection protocol that acknowledged and minimized the risks associated with the interviews/surveys themselves, including appropriate introductions, organization of questions and probes in the interviews/surveys, and a conclusion that incorporated debriefing and referrals when needed. Finally, acknowledging, appreciating and understanding the enormous influence of long-standing cultural factors, social and familial norms, and gender roles and expectations in the prevalence of interpersonal violence in the region has been important to our ability to design and conduct our research on a long-term basis. Recruitment. Despite the above-mentioned elements that we have attempted to incorporate into our research in the borderlands, conducting research in these communities has often posed some specific challenges to the recruitment and retention of study participants in our longitudinal studies. The challenges we faced in terms of recruitment included the following: 1. Suspicions: One of our challenges was to find ways of overcoming the existing strong suspicions potential study participants had about formal helping systems and researchers. These suspicions were relayed to us by the staff at the shelters as well as directly by the study participants. As a consequence, participants indicated that they had deliberated a long-time before entering a shelter and/or participating in a research study. 2. Secrecy: We as researchers had to devise non-threatening ways of breaking through the secrecy that a number of participants had maintained for years about the violence in their lives. 'Feeling trapped' and fear were often mentioned as reasons for this secrecy. 3. Distrust: Establishing trust with each study participant was paramount to the future of our longitudinal study. This process became easier after our initial pilot study. Shelter residents and staff from the hospital and shelters saw us often and began to trust us after a few months. 4. Fears: We had to find relevant ways to address the fears of study participants about the consequences of participating in a study, especially over a 18-month period of time. These fears were particularly high among those participants who did not have legal residence status in the United States. 5. Confidentiality and Privacy: It was critical for us to be able to assure our study participants living in small towns and colonias along the US-Mexico border that the information they provided and their identity was protected and confidential and that their friends, families, and neighbors would not be informed of their participation in a research study. In many of the places we visited people knew one another well and socialized with each other through their church or family gatherings. Retention. Similarly, with respect to retention of study participants in a long-term study, we encountered a variety of challenges that longitudinal studies in general face as well as the following two regional challenges that were primarily due to the geography and location of the borderlands. These challenges were:1) contacting, tracking, and interviewing study participants across a very porous border since participants often traveled cross the border and at times could not be contacted or tracked for weeks and 2) being constantly aware and cognizant of the prevailing cultural, social, and familial norms and the differences in them across the border. In addition, we had to learn ways to frame the issues of immigration and residency status that did not threaten participation in the study. Our experience in working in US-Mexico border communities for the past 7 years may be helpful in providing a better understanding of some of the key considerations for effective sampling, recruitment, and retention of study participants from these communities. In our experience the key considerations include the following: 1. Cultural Competency: Having research assistants and staff who were well aware and understood the border culture. This included proficiency in both Spanish and English and an understanding of the various customs, traditions, family norms and values. This proficiency and understanding helped in the initial recruitment of participants for the study and subsequently in building rapport with them for retention. We learned this lesson through a process of trial and error during the first three months of the study that resulted in a more thorough screening of potential research assistants and developing a more suitable training module. As a result, not only did recruitment improve but attrition was minimized after the first three months of our longitudinal study. 2. Tracking System and Protocol: Developing a tracking system that was personalized and detailed. This was critical because study participants listed a variety of family members and friends who lived on both sides of the border in Mexico and in the US as sources that we could contact over the course of the study. In addition, we quickly learned that every time we interviewed/surveyed participants, we needed to spend some time updating all contact and tracking information. We also developed a tracking system that was used between data collection points and included detailed record-keeping, phone calls, letter/notes, messages and visits that were agreed upon with each study participant. Developing this tracking system and protocol was time and labor intensive but is essential for better retention rates in longitudinal studies. This was particularly relevant in our research where participants moved freely between two countries and among very diverse communities. 3. Addressing Participants' Concerns: Finding ways of acknowledging participants' fears, suspicions, and mistrust about researchers and about participating in a long-term study. We were able to address this issue through rigorous training of research assistants and by assigning a cohort of study participants that each research assistant worked with from the initial time point to the end. We also learned that these concerns could and needed to be addressed over time and could not be rushed. It required communicating at a pace that was dictated by each participant. We had to be careful that our eagerness and over-zealousness did not get in the way of establishing rapport and trust in the long-term. 4. Avenues and Time for Expression: Providing study participants the time to deviate and talk about tangential issues that were on their minds and were important to them. We recognized that this opportunity to bring out and talk about issues that were salient to participants was important especially among the populations we worked with. Consequently, we learned to be flexible with our time and scheduled our interviews/surveys with plenty of time for introductions, story telling, breaks, and debriefing. 5. Scheduling and Convenience: We conducted interviews and surveys at times and in places (including in Juarez, Mexico and small border towns in the Southern New Mexico) that were selected and were suitable to study participants. This often meant driving several hours to collect data in a timely manner. This was necessary because many of the participants had no mode of transportation and could not travel around on their own. The process was time and labor intensive and required careful planning and coordination particularly as we moved into the later time points. 6. Closure and Termination: It was critical that we integrate a well-thought process of closure and termination. This was most challenging to us because when we began the study we did not plan for the day when it would end. We learned that by planning for closure and termination we could reduce the anxiety that our research assistants and study participants experienced and better prepare them for life after a research study. Conclusion Sampling, recruitment, and retention in longitudinal studies pose challenges that need to be anticipated and addressed in a timely manner. In addition, the border communities we worked in introduced a variety of challenges that were unique. Building rapport, having a staff that was qualified, professional, and able to understand the cultural and social norms and values of the region, designing a methodology that suited our purpose and was responsive to the needs of the study participants, and building an efficient tracking system played an important role in our ability to conduct a longitudinal study. Developing a closure/termination protocol was one of the areas that we were not as prepared or as effective as we should have been. Addressing the anxiety of both study participants and research assistants towards the end of the study was important and needed to be built into the study protocol. It was not just about completing the final report for the funding agency or presenting our findings to the shelter staff, or generating manuscripts for peer-reviewed journals but it was about bringing closure to the study participants and to the research assistants who played key roles in the successful implementation of our study. -------------------------- Intimate Partner Violence Victims in Urban Health Care and Justice Settings Judith McFarlane, R.N., Dr. P.H., FAAN Description of Research Program Abused women are users of both health and justice services. This section focuses on longitudinal studies of adult women who have reported physical and/or sexual assault by their intimate partner in urban health care settings, such as emergency rooms or outpatient clinics, or justice agencies, such as civil justice for protection orders or criminal justice for assault charges. A recently completed randomized clinical trial of 150 ethnically diverse English and Spanish speaking assaulted women, recruited through an urban justice agency, retained 100% of the sample at 18 months (McFarlane, Malecha et al., 2002). Strategies used to achieve this 100% retention are described in this paper. Other studies that have employed successful recruitment and retention strategies are referenced elsewhere (McFarlane, Campbell, Sharps, & Watson, 2002; McFarlane, Campbell, & Watson, 2002; McFarlane, Malecha et al., 2002; McFarlane & Soeken, 1999; McFarlane, Soeken, & Wiist, 2000; Parker, McFarlane, Soeken, Silva, & Reel, 1999; Wiist & McFarlane, 1999). Critical Issues Sampling within a Service Agency: The Wait is the Challenge. Sampling of abused women initially contacted in health care and justice settings requires doing research in a service delivery milieu. Emergency rooms, outpatient clinics, district attorney offices, and law enforcement agencies are designed to serve the persons that seek their services. The common goal is to provide the person with the services as quickly and efficiently as possible. A process always exists whereby clients are routed through several people that complete different tasks. For example, in most health clinics, everyone must register, and then patients without an appointment are usually triaged according to the urgency of the health complaint. Following triage, patients are assessed, and depending on the health findings, wait for further assessment, diagnosis and treatment. Further assessment usually involves additional waiting periods. Persons with an appointment also wait. Wait to be seen by the care provider. Wait to be seen in the laboratory. Wait for the results of the laboratory tests. Wait for the prescription. Wait for the next appointment. Each wait may be for minutes or hours, depending on the volume of patients and number of staff. The situation is similar in justice agencies. Every client must complete a process that includes both written forms and personal interviews. As in the health system, clients wait for varying amounts of time. One justice agency had six "stations" that each client was required to visit. The progression through stations could not be changed and the wait between stations was unpredictable, as was the time at each station. It is not uncommon for women to spend four to six hours at a health or justice setting to receive one hour of service. Researchers see this waiting time as research time. Although most people would much rather be doing something, including participation in your research, rather than sitting and staring; the waiting time is unpredictable. No client wants to miss a station and thereby extend his or her wait. Clients are frequently told, "If you do not respond when your name is called, you will be placed at the end of the line". An observation of any health or justice waiting room notes clients asking other waiting clients to listen for their name while they make a phone call, go to the restroom, or go to the vending machine to sooth a hungry child. Similarly, no service provider wants to wait for a client to complete a research questionnaire and thereby be delayed in completing his or her required work and is at risk for a reprimand. The challenge becomes how does a researcher work within the wait? Strategies for Successful Sampling within Health & Justice Agencies. The first strategy for successful sampling is to establish a partnership with the service agency. For successful sampling of abused women from health and justice service agencies, a working partnership between the service agency and the research team is essential. However, do not expect the health or justice system to be excited and eager to collaborate with you on a research study. Both health and justice systems are service focused. Research usually is not part of the mission of the agency. Research is perceived as an "extra" activity that other people do that may or may not benefit the agency. Health and justice agencies will relate experiences about the time they participated in research only to be misunderstood, misquoted, or have a client handled inappropriately by the research team. Without adequate rapport, trust and continuous nurturing, health and justice agencies will not be interested or willing to offer permission for subject sampling at their agency. To begin the working partnership for successful sampling, the investigator(s) and research team must understand the mission, goals, tasks, and demands of the health or justice service agency. Meet with the service agency director and staff to learn what they do and the challenges they face. Ask to shadow agency staff as they assist clients and complete the required tasks of client service, including the four block walk in the rain by the clerk to file papers or the 30 minute phone wait by the health care provider to discuss a patient's insurance coverage. It is important to observe all staff, especially the person, usually a receptionist, who is responsible for routing people. This information offers the researcher the best idea of how a research protocol could be implemented. As a researcher, volunteer to assist the agency with needed services, such as teaching a continuing education program on domestic violence, assistance with program evaluation or grant writing. When the researcher volunteers skills to the agency to promote their work, the agency will be far more likely to accommodate the research protocol. Explain to all agency personnel that may come into contact with the potential research subjects what the research study is about and how the study will benefit their clients. Support staff (e.g., receptionists, secretaries, clerks), licensed staff (e.g., attorneys, registered nurses, physicians, social workers), and administrators must be equally informed and knowledgeable about the research study. Explain how the proposed research can promote agency goals and client service. For example, in a large urban public health clinic, an abuse assessment and intervention program was proposed as a no cost service to be offered to all adult female patients. The abuse assessment and intervention was to be provided by the research team and would require no time from the staff. Since staff had been frustrated at their inability to meet the needs of abused women, administrators and staff viewed the research as an asset. Agency staffs are experts on available resources. Ask for staff recommendations on when, where and how the potential subjects could be recruited and interviewed. Staff and administrators know about the seldom-used storage room with a phone that can be converted to an interview room or the staff member, with a private office, going on leave for three months. Staff and administrators also know about the wait times. At one clinic, patients usually waited for up to an hour to be seen by the care provider. This provided an idea time to recruit and interview research subjects. At a justice agency, there were three 30-minute wait periods that were used to complete a 60-minute research interview in three 20 minute sessions. A second strategy relates to investigator presence and involvement at the agency. Once a solid partnership is formed with staff, the partnership requires an active presence and consistent nurturing by the investigator(s). I always arrive at the agency to recruit subjects at least 30 minutes early to allow time to visit with each staff, answer questions about the study, and ask each staff person how they feel the study is going. Staff personnel create an atmosphere and attitude about research at the agency that can encourage or discourage subject recruitment and retention. Of course, nothing is ever formally said, but if one staff member feels ignored or disrespected, the whole study could fail. It is the staff that the potential research subjects trust and relate to. It is nurses and physicians in health care settings and attorneys, law enforcement officers and counselors in the justice systems that abused women are seeking services from, not researchers. The potential research subjects are not coming to an agency to be part of our research study but rather to receive a specific service. If the atmosphere of the service agency is supportive of the research study and agency staff is convinced of the study's merit, the potential subjects will sense this "attitude" and be far more likely to participate. Nurturing agency staff, a third priority, includes social actions by the investigators and research team such as eating lunch with the staff, bringing food treats and participating in planned social events, such as birthdays and holiday luncheons. When the research study moves away from the agency, such as during follow-up, continue to offer frequent progress reports to the agency staff. In a recently completed study that had 100% subject retention at the end of 18-months, one staff member from the justice service agency was given 4 hours each week by the agency to participate in subject recruitment. The more integrated agency staff is into the recruitment and retention efforts, the more successful the project. An additional strategy to recognize the staff's contribution to the research study is in the acknowledgement section of all reports from the study, as well as offering and sharing authorship in published manuscripts. Finally, if appropriate and permitted by the granting agency as well as the service agency, include rent money for the agency in the grant proposal. If a room(s), furniture, utilities, housekeeping and security services are being used; monetary reimbursement is important. An equitable amount can be calculated based on market value of the space and associated support services. A fourth strategy is patience; establishing collaborations take time. How long does it take to establish a solid partnership? Usually the process takes several months to years. I always build on past partnerships to forge new partnerships. For example, after working with law enforcement for several years on a research study and routinely interacting with the District Attorney's office, I initiated a study with the District Attorney's office researching an issue of mutual interest. Always ask the agency what research they need and then use this information to guide your program of research. Agencies are always practice and outcome focused. This offers a nice complement to research, which can offer the documentation for best practices and favorable outcomes. Recruitment within a Service Agency: The "Research Team" is Everything. Successful subject recruitment within a service agency begins with selection and preparation of the research team. A well-prepared research team begins with adequate funding to hire sufficiently credentialed and experienced staff at a competitive market salary. A team that is ethnically, racially, and gender proportionate matched to the research subjects is optimum. For example, if sample inclusion criteria were limited to an equal proportion of African-American and Caucasian Hispanic adult women, then the team would optimally be composed of an equal proportion of African-American and Caucasian Hispanic female members. With equal ethnic/racial and gender sensitivity, an interviewer of the same gender and ethnic/racial group is preferable. A clear description of individual responsibilities is essential. Leadership must be understood. If a project manager is hired then the manager must be allowed to manage. If two principal investigators exist, then the unique responsibilities of each must be explained to the research team and practiced by the investigators. The research team needs detailed training in intimate partner violence as well as the specifics of the research protocol. A safety protocol for staff is imperative. If subject recruitment occurs in an agency, the research team follows the safety plan of the agency. Since longitudinal studies frequently involve fieldwork to locate subjects, a fieldwork safety protocol is needed. The following fieldwork safety plan has been used for more than 15 years of longitudinal research during which time there have been no threats or breaches of personal safety for any research team member, community resident, or research subject. 1. Work in pairs as often as possible; 2. Carry a cell phone on your person; 3. Make someone aware of the destination(s) and expected time of return to the office; 4. Maintain constant vigilance of the surroundings; 5. If a situation feels uncomfortable or unsafe, for example there is yelling or intoxicated persons present, leave immediately; 6. Do fieldwork only during the daylight hours, especially on weekends (a time found to be very good for contacts); and 7. Wear sensible attire (garments with pockets, sturdy walking shoes, and minimal or no jewelry). Nurturing the Research Team. Maintaining an optimally functioning research team requires lots of nurturing. Give the team members regular opportunities for debriefing and group problem solving. Frequent staff meetings that are attended by everyone provide a consistent opportunity for team building. Encourage freedom of expression and open discussion of issues. Offer career opportunities for each team member such as flexibility of the schedule to encourage attendance at an educational conference or course work toward a degree program. Offer the research team a lunch out, remember birthdays, and take time to chat. These are essential actions to retaining the team. A research team with high turnover will lose more research subjects than the stable team. It is difficult for abused women to share their trauma. The women form a bond with the research team member. If the researcher leaves the project, the woman may also choose to leave the study rather than form a new relationship. Subject Recruitment: The Importance of the Research Invitation. Subject recruitment is best done in person during a one-on-one encounter. Both health and justice agencies provide a secure and confidential setting for the recruitment and retention of abused women. The point during the processing of abused women at the agency that potential research subjects can be approached and invited into the research study will need to be carefully considered and piloted. It is essential to approach women in a consistent and confidential manner. Agency staff may be the best persons to approach potential research subjects and inform them about the study. If interested, the potential research subject can be referred to the researcher for further information. Since the staff member may have established rapport with the woman and previously assisted her with a requested service, such as health care, the potential research subject may be more receptive to talking to the researcher. Subject Retention Depends on a Safe Contact List. Once the potential research subject is introduced to the researcher, the written informed consent process begins. Following signed consent, it is important to obtain a safe contact list of persons that the research subject gives permission for the researcher to contact in the event the woman cannot be contacted. I always obtain a list of at least six persons, always striving for close relatives (i.e., mother, grandmother, sister, adult children) neighbors, friends, work colleagues, and other acquaintances. For each person, the name, relationship, address (home and work), and contact telephone numbers (i.e., home, work, and cell) are listed. The safe contacts will only be contacted if the researcher cannot contact the research subject. When contacted the safe contact is told the research subject is in a Women's Health Study and has given her permission for the researcher to contact persons that may know of her current address and/or phone numbers. To assist the women with the safe contact list, a phone directory is available during the interview. During each subsequent interview, both the contact information for the research subject and each safe contact is reviewed and updated before the interview begins. The safe contact list has proven the best method for maintaining contact with a research subject. Retention & Incentives. An additional strategy for successful retention is the use of an incentive for each interview completed. There are no legal sanctions regarding incentives and the Federal Common Rule does not address incentives. Commonly used incentives include money, gift certificates (i.e., food, clothing, entertainment such as movie tickets) and consumer goods, such as infant or child car seats or educational toys. Money is usually preferred. Deciding how much money or the amount of the gift certificate is difficult. One can use a wage payment model and offer a cash incentive equal to the amount of time the subject is asked to devote to the research activity. Usually the more invasive the topic and the longer the time commitment, the more money offered. I have found incremental monetary incentives to help retention and to be equally effective for all socioeconomic and racial/ethnic groups. For example if the research consists of six interviews, each lasting about one hour, over a two year period, the research subject would be given $30 for the first completed interview, followed by $40 for the second, $50 for the third, and so on with a $50 bonus for completing all six interviews. The money is offered in cash if the interview is completed in person and mailed as a United States postal money order if the interview is completed over the phone or through the postal mail. The incentives are explicitly explained in the signed informed consent. Initial interviews are always ended with giving the research subject the research team members' card with postal, electronic, and facsimile addresses as well as 24-hour voice contacts. A toll free telephone number is also helpful. All contact information is listed on a wallet-size card titled the Women's Health Study with the name of each researcher and, if safe, their credentials, such as registered nurse or medical doctor. Women should always be asked if it is safe for them to take the card with them. Retention & Safety Protocols. Retention of research subjects requires regular and personal contact between data collection cycles via letters, postcards, or phone calls. When contacting women we following a safety protocol that includes the following: 1. During the initial interview a) the woman chooses a day, time, and phone number for follow-up telephone interviews; b) the researcher supplies each woman with a business card identifying that the study is a "Woman's Health Study." ; c) The woman is informed that the researcher will always identify herself on behalf of the Woman's Health Study whenever telephoning. 2. When conducting the follow-up telephone interviews, the researcher will use telephone lines connected with the university or health clinic due to the Caller ID system that is in place in most homes. When the researcher telephones the woman's home and states the call is being made on behalf of the Woman's Health Study, the Caller ID digital reading will verify the call is coming from the university or health clinic. I do not recommend calling from justice agencies as this may alert the perpetrator to the woman's use of justice services. 3. When conducting the follow-up interviews, once the woman is on the telephone, the researcher will identify herself and then state clearly: "Please answer 'yes' or 'no,' is it safe for you to speak right now?" * If the woman answers 'yes,' the researcher will proceed with the interview. The researcher will ask the woman to disconnect at any time during the call if she feels unsafe. * If the woman answers 'no,' the researcher will say 'Thank you, Goodbye' and disconnect. The researcher will attempt the call on another day during a different time period. 4. When conducting the follow-up interviews, if another person answers the telephone, the researcher will identify herself as calling on behalf of the Woman's Health Study. The researchers will offer no further information. 5. Money orders will be mailed only to addresses supplied by the women. Subject Retention: Strategies for Regaining Contact with "lost" Subjects. Sometimes the subject loses contact with both the research team and her safe contacts. People move and phones are disconnected. The internet has a number of free people locators Web sites such as Switchboard and Anywho that can be used successfully. Also, if the phone is disconnected and the address is current, a letter requesting that the woman meet with a research team member in a safe and convenient public place of the woman's choice can reestablish connection. Also asking the agency about current addresses and telephone contact information of the research subject has proven helpful. Terminating the Research Relationship: The Final Step of Retention. Terminating the research relationship is part of every longitudinal study. Strong bonds frequently develop between researcher and research subjects. The researcher may be the first person with which the abused woman has shared the details of the abuse. Preparing the woman for termination of the research begins at the first interview when the length of the research study is explained. At each subsequent interview, it is important to remind the woman of the number of remaining interviews. If the woman was recruited from a health or justice service agency, it is important to remind the woman that the agency exists to serve the woman and, although your relationship is ending, the service agency will be there if she chooses to use it. Finally, we always ask the women if they would like a summary of the study findings and, if so, a safe address to which we can mail the report. Conclusions The advantages of sampling, recruitment and retention of abused women from health care and justice settings is client safety and the opportunity to sample from an ethnically, racially, and socioeconomic diverse population of service-seeking women. The most common services sought by abused women are justice services, both civil justice services, such as protection orders, as well as criminal justice services, such as perpetrator arrest and assault charge filing. Similarly, most abused women seek health care, either for routine well-woman or pregnancy care, or for trauma and associated health problems related to the abuse. Both health and justice agencies provide a safe and confidential setting for subject sampling, recruitment and retention. Additionally, to test interventions for abused women, a representative sample of abused women is essential. Both justice and health agencies provide representative samples of abused women seeking services. This paper has described strategies for successful sampling, recruitment and retention of abused women from health and justice service agencies. To summarize, the process begins with the establishment of a working partnership with the service agency that includes an active presence and consistent nurturing of agency staff by the investigators. Successful recruitment of research subjects begins with careful selection and adequate preparation of the research team followed by attention to the research invitation and a clear explanation of the study. Subject retention is best maintained with a detailed and frequently updated safe contact list, safe telephone and fieldwork protocol, and incremental monetary incentives. -------------------------- Safely and Successfully Retaining Battered Women in Longitudinal Research Cris Sullivan, Ph.D. This case study describes the strategies used to safely retain women participating in a longitudinal study that also employed an experimental intervention (The Community Advocacy Project)[i]. As other case studies in this document have effectively described issues pertaining to sampling and recruitment, this example will focus on successfully and safely locating women with abusive partners over time. Description of the Research The current research study was created in collaboration with victim advocates and survivors of intimate partner violence. Every phase of the research -- including recruitment strategies, intervention design, measurement development, and retention -- was decided collaboratively in order to design the most culturally appropriate, meaningful intervention study possible. As a white, female researcher from a middle class background I believed it was important to work with others from a variety of other backgrounds and experiences to inform this study. All of the many advocates and survivors with whom I have collaborated over the years are equally responsible for the successes of this research. In this study 278 women were recruited after they had exited a domestic violence shelter program in a mid-sized urban city located in the Midwest. Immediately upon completion of the first interview, half the women were randomly assigned to the experimental condition and received the free services of trained advocates for 10 weeks. Women in the control group were not contacted again until their next interview, and simply received services-as-usual from the community[ii]. All women were interviewed six times for this project: preintervention, postintervention, and at 6-, 12-, 18-, and 24-month followups. Interviews were conducted in the community at women's convenience, often in their homes. Interviews lasted approximately 1.5 to 2 hr, and women were compensated for their participation ($10, $40, $60, $80, $90, and $100, respectively). Women were also contacted halfway between each of their interview time points (at 3, 9, 15, and 21 months) to update their contact information and to remind them of upcoming interviews. Critical Issues Locating any research participants over time is time-consuming, difficult, and expensive, but women with abusive partners and ex-partners can be especially difficult to find. While battering occurs across all socio-economic groups, survivors who are likely to come to the attention of domestic violence shelter programs, the police, or emergency rooms are often more likely to have low incomes. This means they are likely to move frequently, and their telephone service can be extremely erratic. For the current study an elaborate protocol was developed with a multicultural team of advocates and survivors to safely locate women over time. Lengthy discussions about the difficulties in safely and respectfully following battered women over time resulted in an elaborate and extensive protocol being put into place to maximize retention. While many of the individual components of this retention protocol have been used by other researchers, it appears to be the combination of these techniques that produced the extremely high retention rate in this study (94% or more retained at any time period). The first phase of the retention protocol involved setting the stage for future contacts. This phase involved four components: First, we gained the trust of potential research participants by stressing that we were conducting this research to improve the community response to domestic violence. The best way to engage women is to repeat at each time point why the research is being conducted and how it can be helpful to other women in their situations. Second, we gathered detailed information from women about alternate contacts in their lives who could help us locate them should we have difficulty finding them over time. For each alternate contact (e.g., family members, friends, coworkers, clergy) each participant signed a Release of Information Form giving the contact permission to tell us where the woman was living and how to reach her. Each participant receiving governmental assistance was also asked to (voluntarily) sign a Release of Information Form for the caseworker handling her case. We also noted best times to contact each participant in general, as well as inconvenient or unsafe times. The third component of Phase I involved giving participants a tangible reminder of their next interview. Each woman received a business card with our toll-free number, the date of her next interview, and the dollar amount she would receive. On the back of each card was a reminder that we could do interviews by telephone for women who moved out of the area. The fourth component involved compensating women for their time and expertise. Payment was increased at each interview to encourage women to contact the project if they moved and to thank them for their continued involvement. Phase II of the retention protocol included locating women for subsequent interviews by attempting to contact them directly, either by phone or in person. If these approaches provided unsuccessful we moved on to Phase III, using women's social networks to contact them. This phase involved calling, visiting and sending letters to each alternate contact. Retention Rate by Strategy Used. Telephoning and going out to participants' homes repeatedly to locate them for interviews resulted in an average retention rate of 47% across all time periods. Sending letters to participants secured an additional 1% to 9%. The Phase III strategies of phoning, visiting or sending letters to women's alternate contacts secured an additional 23% to 43% of the participants across the different time points. The combination of Phase II and Phase III techniques, then, resulted in an overall retention rate of 96% at 10 weeks, 94% at 6 and 12 months, 96% at 18 months, and 97% at 24 months. It is important to note that if the study had relied solely on retention strategies that could be implemented from the office (phone calls and letters), the vast majority of the sample would have been lost over time. Specifically, only 31% of the original sample would have been retained at 12-months, 25% at 18-months, and 22% at the 24-month follow-up. We compared the "easy to find" women with the "difficult to find" women at the 6-month follow-up point and found that the easier to locate (through office-based strategies) women were more likely to be white, were more highly educated, were less depressed, and had experienced less physical and psychological abuse compare to the women who required fieldwork to be located (Sullivan et al., 1996). Attending To Safety Concerns. This project used a number of strategies to maximize women's safety during their participation in the study. First, all staff who might have any contact with research participants were intensively trained to (1) understand the complexities of domestic violence, (2) guide a woman through basic safety planning if necessary, (3) actively listen and empathize with participants, (4) maintain confidentiality, and (5) recommend community resources as needed. It is not enough to simply "follow a script;" women's lives are multifaceted, and project staff must have the knowledge and skills to handle unanticipated situations. This type of training can often be obtained from local domestic violence service programs. Indeed, some service programs will only work with researchers who first successfully complete their extensive trainings (see (McGee, 1999). For those researchers who have any questions or qualms about protecting women's safety, this type of training can be extremely beneficial. We also attended to issues of safety by staying vague about the nature of the research study on our business cards, on our answering machine, and in the Release of Information Forms which noted only that the woman we were looking for was participating in a study at the local university. Finally, safety was a primary concern when contacting women by telephone and when conducting interviews in their homes. With all of the new and changing technology around telephones (e.g., caller ID, call blocking, call unblocking) an important consideration for any study is deciding from where to place calls. While some research teams have chosen to block their numbers, there are some drawbacks to this decision. First, some people will not answer their telephones if the number is blocked from their Caller ID because they assume the caller is a telemarketer. Second, some people can "unblock" supposedly blocked calls, and the research team should be prepared for this possibility. I have found two strategies to work well, depending on situation and resources. First, unblocked calls can be made from an office phone as long as everyone who answers that telephone (1) is aware of the study, (2) does not answer the phone with a revealing name like "Domestic Violence Project," and (3) has a prepared script for answering the question "Why did you just call my house?" There should also be an answering machine with a vague message attached to this number as well. In cases where it would be difficult to guarantee that all people answering the office phone would handle difficult situations adequately (e.g., too many people in and out of the office), calls can also be placed from cell phones. The same protocol would apply with handling call-backs, and voice mail would need to be vague as well. When making a cold call to a home, a protocol should be in place for handling the possibility of the assailant answering. Will the staff member ask for the woman anyway? Pretend they have a wrong number? Hang up? We have made the decision to ask for the woman directly (as long as each participant agrees to this), because if we pretend to have the wrong number we have fewer options for calling back. If the staff member is asked why they want to talk to the woman, a pat response should be prepared ahead of time. For instance, the staff member could say they are calling about a women's health study. Even if a woman answers the telephone, the first question should be "May I speak with '_______'" since, even if the person on the phone is a woman, she may not be the intended respondent. Once we have the woman we want to speak with on the telephone, we always assume the assailant is either standing right behind her or is on an extension. While this will not always (or even often) be the case, precautions should always presume worst-case scenarios. Again, studies can be presented in a vague fashion until there is some assurance the woman can speak freely. The interviewer might say the study pertains to personal relationships, which would lead to asking if the woman is alone and can speak openly. Interviewers should be trained to listen for indirect communications that might indicate a woman is uncomfortable or unsafe -- speech hesitations, uncomfortable silences, background noises -- and have standardized ways to discontinue the conversation if necessary. For this study most interviews were conducted in-person in women's homes, unless they did not feel it was safe. We stressed to women ahead of time that no other adult could be present during the interview, and if there was any chance that the assailant might interrupt we worked out an "explanation" for our presence (e.g., a study interviewing people about their neighborhoods, or someone selling cosmetics) and postponed finishing the interview. If women did not feel it was safe to do the interview in their homes we picked them up and took them to a private location such as a library, community center, or quiet coffee house. We also considered issues of race and class in determining where to conduct interviews. For example, in our area there are a number of low-income, predominantly Black neighborhoods in which some of our sample lived. A young white woman knocking on a door in that neighborhood is immediately perceived by many neighbors as being some sort of case worker or social worker, which can be stigmatizing to the family participating in the study. We therefore talked extensively with our interviewers about these types of issues, and took care to send interviewers who would not "stand out" into neighborhoods. This included discussing how to dress, how to hide the interviewing materials and tape recorders, and how to "blend in" to one's surroundings. While there is no one way to deal with all issues of race and class as they pertain to entering neighborhoods and conducting interviews, conversations are needed to discuss the complexities surrounding this. It also helps if the research team is not only multicultural but diverse in terms of class background, age, sexual orientation, and religious/spiritual beliefs so that a variety of experiences and views can inform all aspects of the study. It is also important, if researchers expect women to leave their homes for interviews, that they consider providing transportation (after considering issues of liability), or at the very least provide them with cab fare (as opposed to bus fare, which, while cheaper for the researcher, is less comfortable and less convenient for participants). Discussion should occur with women about that transportation: Where will women be picked up? Will it look odd to anyone in the woman's home or neighborhood to have an unfamiliar car or taxi arrive at her door? Is there any chance the assailant might follow the woman, and if so, how should that be handled? Given the individuality of each woman's situation, there is no one way to handle any of these issues. Safety issues need to be tailored to each woman's individual life circumstance. It is also important to remind women at each time point that they can discontinue their participation for any reason. In addition to this, there may be times when the research team suspects that even tracking a woman might be endangering her (e.g., by the tone of the man who consistently answers the telephone and will not let her speak), and in these cases the decision should be made to discontinue the woman's participation. Women's situations can change dramatically over even brief periods of time, and it is the responsibility of the research team to consider this and prepare for it. For example, at one time point a woman might insist that contacting her at any time of the day or night will be safe because she has no intention of ever seeing her assailant again. Yet by the next contact she could very well be living with him or he may have gained access to her home or answering machine. It is therefore critical to ask women at each time point how best to contact them in the future, but to also always take certain precautions when contacting women, regardless of their situations at prior interviews. Related to this, it is extremely important to remember that women might still be in danger even after their relationship ends. It is quite common for batterers to not only continue but to escalate their violence after the relationship ends (American Psychological Association, 1996; Browne & Bassuk, 1997; Fleury, Sullivan, & Bybee, 2000; Mahoney, 1991). When conducting research with women who have been abused, the presumption should always be that each woman could be in danger. Preparing for such a possibility will serve to minimize the risk, and will also send a message to the research participant that the study team is aware of and sensitive to her situation. The Role of Children in Protecting Women's Safety. Many assailants use their children as a means to continue threatening, stalking, or harming their intimate partners and ex-partners (McMahon & Pence, 1995; Shepard, 2991). It is not uncommon for them to interrogate the children about their mothers' activities, as a way of monitoring all aspects of the women's lives. This is often done in subtle ways so that the children do not realize they are being manipulated. Researchers need to be aware of this issue and consider the likelihood as well as the consequences of children informing assailants of their mothers' participation in a research study. This is especially likely if the research directly involves the children, but is a consideration even if children might be present during the interview or during any telephone or in-person contacts. In this study, for example, we never left messages with children unless the woman specifically indicated that this was acceptable. We also made sure we conducted interviews privately. Conclusions This study demonstrates that it is possible to locate low-income, abused and formerly abused women if a comprehensive system of techniques is built into the research design. It is important to gain the trust of participants from the beginning, to gather information about people in their lives who can help locate them, to provide them with a tangible reminder of their next interview, and to compensate them for their time and assistance. Further, it should be expected that trained interviewers will actively look for participants in the community, as opposed to relying only on office strategies. And perhaps most importantly, the retention protocol must attend to issues of safety in a variety of ways. This involves making retention an important and distinct component of the research project, and it is ideal if a research team is diverse across many dimensions in order to bring a variety of experiences and perspectives to the table. 1. This research was supported by National Institute of Mental Health Grant R01 MH44849, and grant 98-WT-VX-0013 from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. 2. For more information about the effectiveness of the intervention, please see Bybee & Sullivan, 2002; Campbell, Sullivan, & Davidson, 1995; Sullivan, 1991, 2000; Sullivan & Bybee, 1999. -------------------------- Recruiting and Retaining Convenience Samples of Maritally Violent Men Amy Holtzworth-Munroe, Ph.D. Description of Research Program Our research program studies male perpetrators of intimate partner violence, comparing violent and nonviolent husbands and identifying subtypes of male batterers. One option in doing such work is to recruit and study clinical samples of men beginning domestic violence treatment programs, who often have been arrested and court-referred to treatment. Such samples have the obvious advantage of providing meaningful data to clinicians and others in the criminal justice system. In more recent studies, however, we have chosen to recruit study samples from the community, for at least two reasons. First, while clinical samples usually identify severely violent men, we are interested in studying the full range of husband violence, from infrequent and "mild" physical aggression to very severe physical violence. We believe that by comparing men engaged in differing levels of violence, we can begin to understand what differentiates men who only occasionally engage in low levels of aggression from men who frequently use high levels of violence. Second, when studying severely violent men, we are interested in studying men who have not been "detected" by the criminal justice or therapy systems (i.e., before they receive official interventions). There are many men who engage in marital violence but never come to the attention of authorities or therapists; studying such men may increase the generalizability of our findings. Critical Issues in Sampling, Recruitment, and Retention Sampling Issues. When recruiting a community sample, it would be ideal to recruit a representative sample; for example, to conduct random digit dialing phone calls to identify maritally violent men and then invite them into the lab for further assessment. Yet, when attempting to recruit men who engage in severe partner violence, one faces the dilemma that only a small percentage of the population engages in severe husband violence. Indeed, while national surveys suggest that 12-14% of couples experience husband physical aggression in any year, the prevalence of severe husband violence may be as low as 1% of couples (Straus & Gelles, 1990). Thus, researchers in this area must confront the large expense involved in recruiting a representative sample with large enough numbers of severely violent men to provide adequate sample sizes. For example, one must consider the number of random phone calls that must be made to reach a couple, how many couples screened might be willing to come to the lab for further assessment, etc. Working with our university's survey institute, we calculated that to recruit a sample of up to 100 moderately to severely violent men, a total of almost 35,000 phone calls would need to be made and almost 13,000 couples would have to be interviewed. The survey institute's cost estimate for such a recruitment effort was over $1 million! Given such an expense, many past researchers recruiting community samples, including us, have relied on convenience samples (e.g., recruited with ads). Of course, such samples are not ideal. We never know how the subjects who volunteered to be in the study compare to the individuals who did not respond to our ads, limiting the generalizability of our findings. We can, however, at least try to recruit samples that are demographically representative of the area where the research is conducted, comparing the study sample's demographics to census data regarding the area of recruitment. Our recent studies have used convenience samples of men who are either married or cohabiting with an intimate female partner. To recruit such men, we have used a variety of methods. This paper is based on experience gained in our most recent study, a longitudinal study designed to identify subtypes of maritally violent men and follow them for three years (Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2000; Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, in press). In this study, we recruited a total of 164 couples, 102 of whom were experiencing husband violence. We followed the 102 husband violent couples over time, contacting them at 1.5 and 3 year follow-ups. Recruitment. One lesson we learned is that we could reduce our rate of "no shows" at assessment appointments (i.e., last minute cancellations or failure to come to an appointment) by scheduling both the man and his female partner to come to the lab together, rather than inviting the man to come to the lab alone. For example, when we schedule both partners to come together, approximately 1/3 of our assessment slots are "no shows", compared to a "no show" rate of 50 - 75% when we invite only men to come into the lab. We assume that having both spouses involved increases the chances of their remembering the appointment and making the necessary arrangements to attend (e.g., babysitting, transportation). Of course, another possible interpretation is that women are more responsible than men, but we haven't fairly tested this by examining our "no show" rate when just women are scheduled to come to the lab! Given these informal observations, along with our scientific interest in gathering data from the women, our recent studies have involved recruiting couples experiencing husband violence. Although the focus of our research is often on the male perpetrator, we invite both spouses to the lab and gather data from both. We should note, however, that such samples may not be representative of all violent couples, as our study samples include partners who are willing to come to the lab together. In contrast, some battered women may not want to participate in a laboratory study with their abusive partner (e.g., being too disgusted, disengaged, or fearful to do so). Similarly, perhaps the violent men who enter our study are unique; relative to other violent men, they might be more interested in maintaining their relationship or more likely to blame their wives, thinking that by coming together to the lab, they can demonstrate to others their wife's role in the relationship violence. We have used a variety of methods to recruit our convenience samples. When they call us to obtain more information about the study, one thing we have learned is to ask potential subjects is where they heard about the study. This allows us to track which recruitment efforts are working and which are not. By far, our most successful method is the placement of advertisements in the newspapers. We have used both classified and display ads. We have tried placing classified ads in various sections, with the most successful being "temporary employment" (as our subjects earn money for study participation). We systematically place display ads both in sections of the paper likely to attract men (e.g., sports) and to attract women (e.g., lifestyle). Almost 75% of our sample said they had learned about our study through newspaper ads. Two other methods each yielded another 10-12% of our sample. One involved posting flyers around town. We posted flyers in locations we thought our sample might frequent, including economically disadvantaged areas of town, laudromats, etc., given data demonstrating that husband violence (particularly severe husband violence) is more prevalent among couples with lower socioeconomic status (e.g., see review of research on SES and husband violence in (Holtzworth-Munroe, Smutzler, & Bates, 1997). We also were creative in trying to figure out where to recruit men for our sample and had great success with flyers in such places as striptease joints and liquor stores! The other method, which probably was not worth the cost, was a mailing to professionals whom we thought might be in contact with maritally violent men; we included a flyer to be posted in the professional's office. We mailed study information to therapists and lawyers (e.g., divorce and criminal lawyers). Finally, we also found that subjects would tell their friends about the study; we recruited almost 4% of our sample in this manner. Incentive to Participate. We have found that a few subjects participate in our studies "for the good of science" or to learn more about themselves and their relationship. These tend to be the happily married couples in our comparison samples and/or women. In contrast, the vast majority of subjects, particularly violent men, report that they participate in our studies for the financial compensation. In fact, they explicitly tell us that we need to pay them as much (or more), per hour, as they make at their jobs or it "isn't worth their while" to participate. As such jobs include construction or factory work, in our area of the country, we find that our recruitment efforts are most successful when we are able to pay study participants $15 - $20 per hour or more. Recently, ethical discussions have involved the question of whether high financial reimbursement for study participation is coercive. In other words, given the lower socioeconomic status of many of our study participants, are individuals who would normally be unwilling to participate in the study agreeing to do so only to receive high levels of subject payments? We do believe that such payment is necessary to recruit subjects. Yet, while payments must be high enough to attract potential subjects' attention, we have not set our reimbursements far above the subjects' possible hourly wages, so as not to coerce subjects into participating. In essence, we believe that we are fairly compensating subjects for their willingness to take time out of their very busy lives to participate in our studies. There is also the question of whether an abusive husband might force his wife to participate in the study to obtain subject payments. We take what steps we can to prevent this. Partners are interviewed separately on the phone and, in the lab, go over the informed consent procedures separately. Both are informed that we have many reasons for not accepting a couple into our study. Thus, should the wife choose not to participate, we never tell the husband that this is the reason the couple is not participating. Instead, we say that we are screening out the couple, reminding the husband of the many reasons we screen out couples (e.g., a subject cell is full). If a couple comes to the lab and then the wife chooses not to participate, we still pay the couple for the first assessment session, to compensate them for their time and effort involved in coming to the lab. In addition, men and women receive separate checks for their participation in the laboratory assessments, to try and give the woman financial independence for her participation. Finally, during any longitudinal assessments, each partner is individually reimbursed and their reimbursement for participation is not contingent upon the participation of their partner. Other Safety Procedures. Of course, in the lab, we take additional steps to minimize risks of study participation for the wives. For example, partners complete measures in different rooms and, when possible, in different suites of rooms, so that they can not overhear each other (e.g., answering interview questions) nor check on one another. As another example, husbands and wives do not complete exactly the same measures and, when possible, each completes measures in a randomly determined order. They are informed of this, to minimize the chances that a husband will demand to know how his partner completed a particular measure. When we ask couples to engage in videotaped discussions of marital problems, we separate the spouses to independently check whether each partner is comfortable with discussing the selected topics; if not, we work with the spouse to choose a topic they will be comfortable discussing. As a final example, if a study participant seems upset or reports being so, they are immediately given a break. We have found that in almost all such cases, a 10-15 minute break (for eating or smoking) is enough to calm the subject down to the point where they wish to continue the study. In addition, each laboratory session ends with careful debriefing and safety procedures. Each spouse independently completes a questionnaire measure regarding how they are feeling at that time (e.g., sadness, anxiety, anger, fear, concern that they will fight, concern for their safety). Using their answers as a basis for discussion, each spouse is then separately interviewed about his/her feelings and any concerns about disagreements that may occur as a result of their participation in the session. Any concerns are dealt with, first with attempts to control them (e.g., taking a time-out break) and, if these steps are not successful, with more explicit steps to protect the wife's safety (e.g., paying for a cab to take her to a safe place; calling the police to intervene). In our recent typology study, which included 102 husband violent couples, not one couple required any procedures beyond taking a brief break. In addition, during later assessment sessions and phone interviews, not one couple reported that our study procedures had directly led to any fights or violence at home. Retention. Our typology study was designed to examine how subtypes of maritally violent men change over time. Thus, during all initial contacts with potential subjects (e.g., phone screening, the initial assessment), participants were informed of the longitudinal nature of the study. At the first assessment, each spouse was asked to provide us with information that would help us contact them for later assessments. Specifically, each spouse was asked to provide his or her address, home and work phone numbers, and the names, addresses, and phone numbers of three people who "should always know how to find you (e.g., parent, close friend)". Following the initial assessment, every six months, each spouse was mailed a letter that included all of this information and asked them to update the information. If we didn't receive a reply, we attempted to contact the spouse and update their information by phone. If we couldn't do so, we would call the listed contacts to obtain such information. In retrospect, we wish we had made these contacts every three months, instead of every six months, as it was often difficult to track subjects after a six month period had elapsed. Also, we believe that such tracking work would have ideally been done by assigning one research assistant to track each couple, to stay in contact with them over time, forming a relationship with the couple and their contacts. Unfortunately, given limited resources, we relied on undergraduate research assistants to help us track subjects and there was frequent personnel turn-over among these assistants. At each follow-up point, each couple was contacted and invited to participate in a follow-up assessment. If the couple was still together, they were asked to come to the laboratory together. If they were separated, they were asked to come to the laboratory separately. If a participant had moved out of the area and/or was unwilling to come to the laboratory, questionnaire packets were mailed to the subject. In other words, we tried to make participation as easy as possible for subjects, to accommodate their current living situation. Also to decrease attrition, subjects were paid for their follow-up assessments, at a higher hourly rate than they had received at the initial assessment and a progressively higher rate at each subsequent follow-up. For example, the second follow-up assessment only took each subject approximately a 1/2 hour, but they were each paid approximately $20. These procedures allowed us to obtain at least some information (from one or both spouses) on 97 of the 102 husband violent couples at both the 1.5 and 3 year follow-up assessments (i.e., an attrition rate of only 5% over 3 years). We found it difficult to track some subjects, given the chaotic nature of these subjects' lives. Two of the men in our sample died. In addition, we had difficulty tracking men who had separated from their wives and were in the military, as the military was relatively unaccommodating in helping us to locate subjects. We also had difficulty with men who were in jail at the time of the follow-up assessments. While we could reach such men, and they were usually eager to participate (reporting that they had nothing better to do with their time and would like the money), prisoners are considered a special population, in need of additional human subjects protection. Thus, we would have to seek special permission to contact each man in prison. Another issue was that some of these men apparently engendered so much fear or hostility in others that their contact informants, including their own mothers, no longer had contact with them (e.g., the mothers of two men told us something to the effect of, "I don't know where that bastard is, and I don't care! I actually hope that I never see him again!")! Of course, this is also a problem for female partners who have been able to leave an abusive husband. Many such women have worked hard to lose track of their abuser and needed to be assured that we would not share information about them with their partner if we located him. A final issue that made it difficult to track subjects was the chaotic nature of their relationships, particularly among the most severely violent men. This created two problems. One was the issue of how to treat couples who permanently end their relationship and have no further contact with one another. Clearly, such relationships no longer afford the "opportunity" for continued male violence. Thus, it does not make sense to include such couples in data analyses predicting the course of violence over time. At our 1.5 year follow-up, a total of 8 (of 102 husband violent couples) were in this category, but this figure grew to 20 couples (1/5 of the sample) by the 3 year follow-up. Of course, researchers can continue to follow such individuals to observe whether violence emerges in new relationships, but we found that very few men in our sample formed stable new relationships over the course of 3 years. The second issue was that many of the relationships of couples in our study simply defied the typical research categorizations of couples as "together" versus "separated/divorced". For example, we had couples who told us, during the follow-up phone interviews, that they had separated during the follow-up period, but were currently living together and wanted to come to the lab together, only to have one spouse arrive at the lab alone, saying that they had filed for divorce and separated again. Thus, to examine potential group differences in relationship status, we divided couples into those whose relationships were "stable" at both follow-up assessments (i.e., there had been no separations during the follow-up periods and the couple was still living together) versus those who had experienced separations during the follow-up and/or were separated or divorced at the time of the follow-up assessment. Clearly, new methods of considering relationship change over time may need to be implemented when studying violent couples. Conclusion Researchers in this field need to grapple with the issue of whether convenience samples are scientifically acceptable; if not, then either substantial resources will need to be devoted to recruit representative community samples or researchers studying severe husband violence may have to limit research to "identified" samples of men in the criminal justice system and treatment programs. Funding agencies must provide researchers with adequate funds to motivate potential subjects to participate in studies, and longitudinal studies require increasing amounts of subject reimbursement over time, along with the financial resources to keep in contact with subjects (e.g., hiring research assistants to work for the full length of the study). Researchers conducting longitudinal studies of severely violent men will need to address the issue of the instability in these relationships and how to handle changing relationship status and the break-up of relationships (with the inherent end to the "opportunity" for further relationship violence) in study designs and data analytic plans. Such issues provide many opportunities for researchers to pilot various methods, to share information, and to learn from one another's experiences. -------------------------- GUIDING PRINCIPLES FOR INTIMATE PARTNER VIOLENCE RESEARCH: RECRUITMENT AND RETENTION The preceding case studies have illustrated successful recruitment and retention strategies in research involving intimate partner violence. The following is offered as a set of guiding principles for recruitment and retention in research involving intimate partner violence, whether the participants include victims, offenders, or children. Safety * Develop an individualized protocol to minimize safety concerns in recruitment and participation in intimate partner violence research. Safety concerns can include risk of intimate partner violence, risk of child victimization associated with participation in the research study, risk of harm to self or others. * Develop a protocol for assessing and reassessing safety issues on an ongoing basis and modify the strategies for minimizing safety concerns accordingly. * Request that participants introduce their contacts to the study personnel - by phone, in person, or by letter - to ensure they are aware of participants' involvement in the study. * Include clear guidelines about the safety of project staff when conducting research tasks (e.g., interviews in participants homes, traveling in unfamiliar neighborhoods, interviews when contact with an abuser becomes a potential concern) in order to minimize the risk to staff members, as well as to participants. Cultural Sensitivity / Special Groups * Develop cultural competency among research staff so they are aware of the cultural nuances of the selected target samples. * Tailor recruitment and retention strategies based on the cultural or other special groups involved, taking into account participants' historical involvement in research activities and particular risks associated with membership in certain groups (e.g., undocumented immigrants). . * Consider additional burdens or barriers to participation based on the vulnerabilities of the cultural or other special groups * Include researchers that are reflective of the diversity of the communities from which the participants are drawn. Collaboration with Community Agencies * Develop partnerships with community agencies involved in the research effort to collaborate in all aspects of the research process, as mutually desired, especially in helping to maintain the focus or mission of the community agency's efforts. At minimum, develop methods of meaningful and periodic inclusion of community collaborators in the research process. * Demonstrate a commitment to the welfare of the research participants beyond their research involvement. * Establish clear and negotiated expectations concerning the community agency's contribution of effort in the research endeavor, taking care not to increase the burden on an already overextended staff. Impact of Participation * Develop recruitment protocols that recognize participants' immediate needs and priorities. * Develop research protocols that respect the privacy of individual participants, especially parents and their children. * Develop a protocol for minimizing emotional distress and stigma associated with recruitment and participation in the study * Develop a protocol to assess research participants' reaction to research participation * Develop an ending or closure process that recognizes the extent of the participants' involvement and the potential emotional loss that the discontinuation of the study may involve. Provide resources for participants who may wish to maintain contact after the study ends. Support Research Participants * Pay participants for their time. * Develop a protocol for assessing critical needs of research participants in order to incorporate attention to them within the research process, as needed. * Develop a protocol for providing needed support to research participants. Support may be include regular contact with participants, crisis support services, emotional support and validation, information, provision of tangible goods as needed, and referral for services. * Where possible, build in same researcher/participant pairs to facilitate relationship continuity within a longitudinal research project. * Maximize flexibility of participant's involvement in terms of time, place, method of participation (i.e., self-report, interview), * Maintain the demeanor of an invited guest when conducting interviews in participants' homes. * Offer participants an opportunity to receive a description of study findings as a means of offering potential benefits of the research to study participants, as well as to others for whom the research may benefit at later times. Support Study Personnel * Develop a mechanism for assessing the secondary victimization and burnout of study personnel * Develop mechanisms for providing support to study personnel. Support may include, but not be limited to, emotional support and validation, feedback about their work, opportunity for expressing emotional reactions to research work, Report Recruitment and Retention Issues in Published Studies * Routinely report on information concerning recruitment and retention in published research studies (DiMattio, 2001; Kinard, 2001) -------------------------- References Akkerman, R. L., Stanley, M. A., Averill, P. M., Novy, D. M., Snyder, A. G., & Diefenbach, G. J. (2001). Recruiting older adults with generalized anxiety disorder. Journal of Mental Health & Aging, 7(4), 385-394. American Psychological Association. (1996). Violence and the family. Washington, D.C.: Author. Barber, J. P., Luborsky, L., Gallop, R., Crits-Christoph, P., Frank, A., Weiss, R. D., Thase, M. E., Connolly, M. B., Gladis, M., Foltz, C., & Siqueland, L. (2001). Therapeutic alliance as a predictor of outcome and retention in the National Institute on Drug Abuse Collaborative Cocaine Treatment Study. Journal of Consulting & Clinical Psychology, 69(1), 119-124. Boles, M., Getchell, W. S., Feldman, G., McBride, R., & Hart, R. G. (2000). Primary prevention studies and the healthy elderly: Evaluating barriers to recruitment. Journal of Community Health: The Publication for Health Promotion & Disease Prevention, 25(4), 279-292. Bride, B. E. (2001). Single-gender treatment of substance abuse: Effect on treatment retention and completion. Social Work Research, 25(4), 223-232. Browne, A., & Bassuk, S. S. (1997). Intimate violence in the lives of homeless and poor housed women: Prevalence and patterns in an ethnically diverse sample. American Journal of Orthopsychiatry, 67(2), 261-278. Buttell, F. P., & Pike, C. K. (2002). Investigating predictors of treatment attrition among court-ordered batterers. Journal of Social Service Research, 28(4), 53-68. Carter-Edwards, L., Fisher, J. T., Vaughn, B. J., & Svetkey, L. P. (2002). Church rosters: Is this a viable mechanism for effectively recruiting African Americans for a community-based survey? Ethnicity & Health, 7(1), 41-55. Connell, C. M., Shaw, B. A., Holmes, S. B., & Foster, N. L. (2001). Caregivers' attitudes toward their family members' participation in Alzheimer disease research: Implications for recruitment and retention. Alzheimer Disease & Associated Disorders, 15(3), 137-145. Daly, J. E., Power, T. G., & Gondolf, E. W. (2001). Predictors of batterer program attendance. Journal of Interpersonal Violence, 16(10), 971-991. De Leon, G., Hawke, J., Jainchill, N., & Melnick, G. (2000). Therapeutic communities. Enhancing retention in treatment using "senior professor" staff. Journal of Substance Abuse Treatment, 19(4), 375-382. Dietze, P., Fry, C., Sunjic, S., Bammer, G., Zador, D., Jolley, D., & Rumbold, G. (2002). Using ambulance attendances to recruit people who have experienced non-fatal heroin overdose. Drug & Alcohol Dependence, 67(1), 99-103. DiMattio, M. J. K. (2001). Recruitment and retention of community-dwelling, aging women in nursing studies. Nursing Research, 50(6), 369-373. Duncan, S. C., Strycker, L. A., Duncan, T. E., He, H., & Stark, M. J. (2002). Telephone recruitment of a random stratified African American and White family study sample. Journal of Ethnicity in Substance Abuse, 1(3), 57-73. Fiocchi, F. F., & Kingree, J. B. (2001). Treatment retention and birth outcomes of crack users enrolled in a substance abuse treatment program for pregnant women. Journal of Substance Abuse Treatment, 20(2), 137-142. Fleury, R. E., Sullivan, C. M., & Bybee, D. I. (2000). When ending the relationship does not end the violence: Women's experiences of violence by former partners. Violence Against Women, 6(12), 1363-1383. Froelicher, E. S., & Lorig, K. (2002). Who cares about recruitment anyway? Patient Education & Counseling, 48(2), 97. Gilliss, C. L., Lee, K. A., Gutierrez, Y., Taylor, D., Beyene, Y., Neuhaus, J., & Murrell, N. (2001). Recruitment and retention of healthy minority women into community-based longitudinal research. Journal of Women's Health & Gender-Based Medicine, 10(1), 77-85. Gondolf, E. W. (2000). A 30-month follow-up of court-referred batterers in four cities. International Journal of Offender Therapy & Comparative Criminology, 44(1), 111-128. Gondolf, E. W. (2001). Limitations of experimental evaluation of batterer programs. Trauma Violence & Abuse, 2(1), 79-88. Gondolf, E. W. (2003). Program completion in specialized batterer counseling for African American men: An experimental clinical trial. Paper presented at the International Research Conference on Family Violence, Portsmouth, NH. Gondolf, E. W., Heckert, D. A., & Kimmel, C. M. (2002). Nonphysical abuse among batterer program participants. Journal of Family Violence, 17(4), 293-314. Gondolf, E. W., & Jones, A. S. (2001). The program effect of batterer programs in three cities. Violence & Victims, 16(6), 693-704. Gondolf, E. W., & Williams, O. J. (2001). Culturally focused batterer counseling for African American men. Trauma Violence & Abuse, 2(4), 283-295. Green, C. A., Polen, M. R., Dickinson, D. M., Lynch, F. L., & Bennett, M. D. (2002). Gender differences in predictors of initiation, retention, and completion in an HMO-based substance abuse treatment program. Journal of Substance Abuse Treatment, 23(4), 285-295. Harris, M. (2000). General practice recruitment for schizophrenia prevention studies. Australian & New Zealand Journal of Psychiatry, 34(Suppl), S137-S139. Hatchett, B. F., Holmes, K., Duran, D. A., & Davis, C. (2000). African Americans and research participation: The recruitment process. Journal of Black Studies, 30(5), 664-675. Hawranik, P., & Pangman, P. (2002). Recruitment of community-dwelling older adults for nursing research: A challenging process. Canadian Journal of Nursing Research/Revue canadienne de recherche en sciences infirmieres, 33(4), 171-184. Heckert, D. A., & Gondolf, E. W. (2000). The effect of perceptions of sanctions on batterer program outcomes. Journal of Research in Crime & Delinquency, 37(4), 369-391. Heckert, D. A., & Gondolf, E. W. (2002). Predicting levels of abuse and reassault among batterer program participants. Washington, D.C.: National Institute of Justice, U. S. Department of Justice. Hilbert, J. C., & Krishnan, S. P. (2000). Addressing barriers to community care of battered women in rural environments: creating a policy of social inclusion. J Health Soc Policy, 12(1), 41-52. Hinton, L., Guo, Z., Hillygus, J., & Levkoff, S. (2000). Working with culture: A qualitative analysis of barriers to the recruitment of Chinese-American family caregivers for dementia research. Journal of Cross-Cultural Gerontology, 15(2), 119-137. Holtzworth-Munroe, A., Meehan, J. C., Herron, K., Rehman, U., & Stuart, G. L. (2000). Testing the Holtzworth-Munroe and Stuart (1994) batterer typology. Journal of Consulting & Clinical Psychology, 68(6), 1000-1019. Holtzworth-Munroe, A., Meehan, J. C., Herron, K., Rehman, U., & Stuart, G. L. (2003). Do subtypes of maritally violent men continue to differ over time?, Journal of Consulting & Clinical Psychology (Vol. 71, pp. 728-740). US: American Psychological Assn, US, http://www.apa.org. Holtzworth-Munroe, A., Meehan, J. C., Herron, K., Rehman, U., & Stuart, G. L. (in press). Do subtypes of maritally violent men continue to differ over time? Journal of Consulting and Clinical Psychology. Holtzworth-Munroe, A., Smutzler, N., & Bates, L. (1997). A brief review of the research on husband violence. Part III: Sociodemographic factors, relationship factors, and differing conseuqences of husbnad and wife violence. Aggression & Violent Behavior, 2(285-307). Jones, A. S., & Gondolf, E. W. (2002). Assessing the effect of batterer program completion on reassault: An instrumental variables analysis. Journal of Quantitative Criminology, 18(1), 71-98. Kinard, E. M. (2001). Recruiting participants for child abuse research: What does it take? Journal of Family Violence, 16(3), 219-236. Klein, C., di Menza, S., Arfken, C., & Schuster, C. R. (2002). Interaction effects of treatment setting and client characteristics on retention and completion. Journal of Psychoactive Drugs, 34(1), 39-50. Krishnan, S. P., & Hilbert, J. C. (1998). In search of sanctuary: addressing issues of domestic violence and homelessness at shelters. Womens Health Issues, 8(5), 310-316. Krishnan, S. P., Hilbert, J. C., & Pase, M. (2001). An examination of intimate partner violence in rural communities: Results from a hospital emergency department study from southwest United States. Family & Community Health, 24(1), 1-14. Krishnan, S. P., Hilbert, J. C., & VanLeeuwen, D. (2001). Domestic violence and help-seeking behaviors among rural women: Results from a shelter-based study. Family & Community Health, 24(1), 28-38. Krishnan, S. P., Hilbert, J. C., VanLeeuwen, D., & Kolia, R. (1997). Documenting domestic violence among ethnically diverse populations: Results from a preliminary study. Family & Community Health, 20(3), 32-48. Krishnan, S. P., J., H., McNeil, K., & Newman, I. (in press). From respite to transition: Women's use of domestic violence shelters in rural New Mexico. Journal of Family Violence. Levkoff, S. E., Levy, B. R., & Weitzman, P. F. (2000). The matching model of recruitment. Journal of Mental Health & Aging, 6(1), 29-38. Mahoney, M. R. (1991). Legal images of battered women: redefining the issue of separation. Michigan Law Review, 90, 1-94. McFarlane, J., Campbell, J. C., Sharps, P., & Watson, K. (2002). Abuse during pregnancy and femicide: urgent implications for women's health. Obstet Gynecol, 100(1), 27-36. McFarlane, J., Campbell, J. C., & Watson, K. (2002). Intimate partner stalking and femicide: Urgent implications for women's safety. Behavioral Sciences & the Law, 20(1-2), 51-68. McFarlane, J., Malecha, A., Gist, J., Watson, K., Batten, E., Hall, I., & Smith, S. (2002). An intervention to increase safety behaviors of abused women: results of a randomized clinical trial. Nurs Res, 51(6), 347-354. McFarlane, J., & Soeken, K. (1999). Weight change of infants, age birth to 12 months, born to abused women. Pediatr Nurs, 25(1), 19-23. McFarlane, J., Soeken, K., & Wiist, W. (2000). An evaluation of interventions to decrease intimate partner violence to pregnant women. Public Health Nursing, 17(6), 443-451. McGee, S. (1999). Commentary on domestic violence research. http://www.ssw.umich.edu/trapped/commentary.html. Retrieved September 9, 2002, from the World Wide Web: McIntosh, S., Ossip-Klein, D. H., Spada, J., & Burton, K. (2000). Recruitment strategies and success in a multi-county smoking cessation study. Nicotine & Tobacco Research, 2(3), 281-284. McMahon, M., & Pence, E. (1995). Doing more harm than good? Some cautions on visitation centers., Ending the cycle of violence: Community responses to children of battered women. (pp. 186-206). Thousand Oaks, CA, US: Sage Publications, Inc. McNeilly, M., Musick, M., Efland, J. R., Baughman, J. T., Toth, P. S., Saulter, T. D., Sumner, L., Sherwood, A., Weitzman, P. F., Levkoff, S. E., Williams, R. B., Jr., & Anderson, N. B. (2000). Minority populations and psychophysiologic research: Challenges in trust building and recruitment. Journal of Mental Health & Aging, 6(1), 91-102. McQuiston, C., & Uribe, L. (2001). Latino recruitment and retention strategies: Community-based HIV prevention. Journal of Immigrant Health, 3(2), 97-105. Meaux, J. B., & Bell, P. L. (2001). Balancing recruitment and protection: Children as research subjects. Issues in Comprehensive Pediatric Nursing, 24(4), 241-251. National Violence Against Women Prevention Resource Center. (2001). Fostering Collaborations to Prevent Violence Against Women: Integrating Findings from Practitioner and Research Focus Groups. Charleston, SC: Author. Nazemi, H., Larkin, A. A., Sullivan, M. D., & Katon, W. (2001). Methodological issues in the recruitment of primary care patients with depression. International Journal of Psychiatry in Medicine, 31(3), 277-288. Nielsen, A. L., & Scarpitti, F. R. (2002). Predicting retention in a therapeutic community for incarcerated substance abusers. Journal of Offender Rehabilitation, 34(3), 47-65. Parker, B., McFarlane, J., Soeken, K., Silva, C., & Reel, S. (1999). Testing an intervention to prevent further abuse to pregnant women. Research in Nursing & Health, 22(1), 59-66. Senn, C. Y., & Desmarais, S. (2001). Are our recruitment practices for sex studies working across gender? The effect of topic and gender of recruiter on participation rates of university men and women. Journal of Sex Research, 38(2), 111-117. Shepard, M. (2991). Child-visiting and domestic violence. Child Welfare, 71(4), 357-365. Sherer, R., Stieglitz, K., Narra, J., Jasek, J., Green, L., Moore, B., Shott, S., & Cohen, M. (2002). HIV multidisciplinary teams work: Support services improve access to and retention in HIV primary care. AIDS Care, 14(4,Suppl1), S31-S44. Sinclair, S., Hayes-Reams, P., Myers, H. F., Allen, W., Hawes-Dawson, J., & Kington, R. (2000). Recruiting African Americans for health studies: Lessons from the Drew-RAND center on health and aging. Journal of Mental Health & Aging, 6(1), 39-51. Siqueland, L., Crits-Christoph, P., Gallop, R., Barber, J. P., Griffin, M. L., Thase, M. E., Daley, D., Frank, A., Gastfriend, D. R., Blaine, J., Connolly, M. B., & Gladis, M. (2002). Retention in psychosocial treatment of cocaine dependence: Predictors and impact on outcome. American Journal on Addictions, 11(1), 24-40. Stoddart, M. L., Jarvis, B., Blake, B., Fabsitz, R. R., Howard, B. V., Lee, E. T., & Welty, T. K. (2000). Recruitment of American Indians in epidemiologic research: The Strong Heart Study. American Indian & Alaska Native Mental Health Research, 9(3), 20-37. Straus, M., & Gelles, R. (1990). Violence in American families: Risk factors and adaptations to violence in 8,145 families. New Brunswick: Transaction Publishers. Sullivan, C. M., & Bybee, D. I. (1999). Reducing violence using community-based advocacy for women with abusive partners. Journal of Consulting & Clinical Psychology, 67(1), 43-53. Sullivan, C. M., Rumptz, M. H., Campbell, R., Eby, K. K., & Davidson, W. S., II. (1996). Retaining participants in longitudinal community research: A comprehensive protocol. Journal of Applied Behavioral Science, 32(3), 262-276. Thomas, V. (2002). Conducting research with community agencies: Meeting recruitment and collaboration challenges. Journal of Marital & Family Therapy, 28(1), 9-14. Tutty, L. M., Bidgood, B. A., Rothery, M. A., & Bidgood, P. (2001). An evaluation of men's batterer treatment groups. Research on Social Work Practice, 11(6), 645-670. Weiss, J. W., & Weiss, D. J. (2002). Recruiting Asian-American adolescents for behavioral surveys. Journal of Child & Family Studies, 11(2), 143-149. Wiist, W. H., & McFarlane, J. (1999). The effectiveness of an abuse assessment protocol in public health prenatal clinics. American Journal of Public Health, 89(8), 1217-1221. Young, D. (2002). Impacts of perceived legal pressure on retention in drug treatment. Criminal Justice & Behavior, 29(1), 27-55. Young, D., & Belenko, S. (2002). Program retention and perceived coercion in three models of mandatory drug treatment. Journal of Drug Issues, 32(1), 297-328. Zhu, K., Hunter, S., Bernard, L. J., Payne-Wilks, K., Roland, C. L., & Levine, R. S. (2000). Recruiting elderly African-American women in cancer prevention and control studies: A multifaceted approach and its effectiveness. Journal of the National Medical Association, 92(4), 169-175.