Psychology of Violence

Advancing Domestic Violence Program Evaluation: Development and Validation of the Measure of Victim Empowerment Related to Safety (MOVERS) Lisa A. Goodman, Lauren Bennett Cattaneo, Kristie Thomas, Julie Woulfe, Siu Kwan Chong, and Katya Fels Smyth Online First Publication, November 10, 2014. http://dx.doi.org/10.1037/a0038318

CITATION Goodman, L. A., Bennett Cattaneo, L., Thomas, K., Woulfe, J., Chong, S. K., & Fels Smyth, K. (2014, November 10). Advancing Domestic Violence Program Evaluation: Development and Validation of the Measure of Victim Empowerment Related to Safety (MOVERS). Psychology of Violence. Advance online publication. http://dx.doi.org/10.1037/a0038318 Psychology of Violence © 2014 American Psychological Association 2014, Vol. 5, No. 1, 000 2152-0828/14/$12.00 http://dx.doi.org/10.1037/a0038318

Advancing Domestic Violence Program Evaluation: Development and Validation of the Measure of Victim Empowerment Related to Safety (MOVERS)

Lisa A. Goodman Lauren Bennett Cattaneo Boston College George Mason University

Kristie Thomas Julie Woulfe and Siu Kwan Chong School of Social Work, Simmons College Boston College

Katya Fels Smyth Full Frame Initiative, Greenfield, Massachusetts

Objective: As budgets tighten and demand grows, domestic violence (DV) programs are facing enormous pressure to demonstrate the impact of their work. A critical challenge to doing so is the absence of outcome measures that reflect DV programs’ missions and survivors’ goals for themselves. Academic- community partnerships are critical to developing such measures. The 2 aims of this study were to (a) develop and validate a measure—the Measure of Victim Empowerment Related to Safety (MOVERS) – that taps a key goal shared by DV program staff and program participants, and (b) draw on and model the benefits of community-based participatory research. Method: We evaluated the factor structure, reliability, and validity of MOVERS through a 2-stage process in which we developed a university- community partnership with 17 DV programs across the Northeast and administered a survey to 230 help-seeking survivors. Results: A scree plot and parallel analysis supported a 3-factor solution, with subscales assessing the extent to which a survivor (a) has developed a set of safety-related goals and a belief in her ability to accomplish them, (b) perceives that she has the support she needs to move toward safety, and (c) senses that her actions toward safety will not cause new problems in other domains. Each subscale demonstrated good internal reliability and construct validity. Conclusion: MOVERS provides a tool for assessing a key dimension of survivors’ experience and enables the evaluation of domestic violence program practices in ways that are consistent with core program and survivor goals.

Keywords: community-based participatory research, domestic violence, empowerment, evaluation, intimate partner violence, outcome measure, safety

Intimate partner violence (IPV)—that is, physical abuse, sexual term shelter, advocacy, counseling, and legal and economic sup- assault, psychological abuse, stalking, and coercion—causes dev- port (Davies & Lyon, 2013; Goodman & Epstein, 2008). astating physical and psychological damage to millions of people In recent years, these programs have come under enormous in the United States each year (Black et al., 2011). In response, pressure to assess and demonstrate the impact of their work, often residential and community-based organizations have sprung up in without the guidance or resources necessary to do so (Sullivan, communities across the country to provide temporary and longer- 2011; Sullivan & Alexy, 2001). Perhaps the most fundamental This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Casa Myrna, DOVE, Family and Community Resources, Health Imper- Lisa A. Goodman, Department of Counseling and Developmental Psy- atives, Jeanne Geiger Crisis Center, Journey to Safety, New Hope, chology, Lynch School of Education, Boston College; Lauren Bennett Passageway, The Second Step, The Women’s Center of Rhode Island, Cattaneo, Department of Psychology, George Mason University; Kristie Transition House, REACH, Renewal House, Respond, Sojourner Thomas, School of Social Work, Simmons College; Julie Woulfe and Siu House, and Women Against Abuse, and the YWCA of Central Massa- Kwan Chong, Department of Counseling and Developmental Psychology, chusetts. We are also grateful to the many mental health counseling MA Lynch School of Education, Boston College; Katya Fels Smyth, Full Frame students and social work students who supported this study through Initiative, Greenfield, Massachusetts. survey administration and data entry, and to Craig DiGiovanni, who This research was supported in part by The Lynch School of Educa- helped with statistical analyses. tion at Boston College, The Simmons School of Social Work, and the Correspondence concerning this article should be addressed to Lisa A. Full Frame Initiative. We are deeply indebted to the staff and partici- Goodman, Department of Counseling and Developmental Psychology, pants of the domestic violence programs that constitute the Domestic Lynch School of Education, Boston College, Chestnut Hill, MA 02467. Violence Program Evaluation and Research Collaborative, including E-mail: [email protected]

1 2 GOODMAN ET AL.

obstacle is the absence of outcome measures that reflect programs’ empowerment and social support and satisfaction with services missions and survivors’ goals for themselves (Goodman & Ep- (Corrigan, Faber, Rashid, & Leary, 1999) and depression (Rogers stein, 2008; Kulkarni, Bell, & Rhodes, 2012; Song, 2012). Indeed, et al, 1997). Next, we turn to the theoretical foundations of the evaluators, practitioners, and scholars have long noted the need for measure. outcome measures that are relevant to both DV programs and their participants (Sullivan, 2011), and rigorously developed, tested, and refined (e.g., Bennett, Riger, Schewe, Howard, & Wasco, 2004). The Conceptual and Practice Foundation of MOVERS Although few scholars have attempted to identify key outcomes that extend across various types of DV programs (see Sullivan, Empowerment Scholarship and the Empowerment 2011 for an especially thorough discussion), they tend to reflect Process Model specific program activities (e.g., “client obtained a restraining order”) rather than survivors’ progress toward goals they have set Although rarely measured or even consistently defined, the for themselves. Moreover, they tend not to be theoretically based notion of empowerment—encompassing themes of personal or rigorously tested (Cattaneo & Goodman, in press). This study’s choice; finding voice; a focus on strength versus deficit; and primary aim was to address this gap. Through a university- transcending oppression—remains an animating principle of most community partnership, we developed and validated a measure- DV programs, as it has from the earliest days of the domestic ment tool, called the Measure of Victim Empowerment Related to violence movement (Cattaneo & Goodman, in press; Kasturiran- Safety (MOVERS; see the Appendix for English and Spanish gan, 2008; Song, 2012; Macy, Giattina, Sangster, Crosby, & versions) that taps a construct central to DV programs’ missions Montijo, 2009). The pivotal role of empowerment in DV program and many survivors’ goals for themselves: Empowerment within ideology and practice is supported by research showing its asso- the domain of safety. In this literature review, we describe the ciation with other key aspects of survivor wellbeing, such that scholarly and practice-based rationale for the selection and con- those who experience a greater degree of empowerment (variously ceptualization of this construct. defined) in the short-term appear more likely to be satisfied with A secondary aim of this study was to draw on and model the services and to reach positive longer-term outcomes such as men- benefits to both DV researchers and DV programs of community- tal health and safety (e.g., Cattaneo & Goodman, 2010; Buzawa & based participatory research. Referred to variously as community- Hotaling, 2003; Johnson, Zlotnick, & Perez, 2011; Zweig & Burt, engaged research, participatory action research, and empowerment 2007). Thus, the construct of empowerment has the potential to be evaluation, partnership approaches to research and evaluation have useful as a measure of the proximal or immediate impact of DV gained momentum across academic disciplines (Israel, Eng, services and as an indicator of distal or longer-term outcomes that Schulz, & Parker, 2005). Although such approaches involve com- any organization working with IPV survivors would target. munity partners in many aspects of the research process, collabo- Although the evidence for the centrality of empowerment to DV ration usually does not extend to the process of developing mea- program missions and survivor wellbeing is compelling, there is sures to assess intervention outcomes (Hausman et al., 2013). little consensus in the literature both inside and outside of IPV on Instead, researchers tend to draw on academically derived and how to define or measure it. At the same time, several key themes validated outcome measures. When community stakeholders are cut across scholarly discussions of the construct: Empowerment left out of the process of defining the outcomes that would be most scholars often describe empowerment as a shift in an individual’s relevant to them, however, the resulting information is also less psychological sense of control as well as a person’s actual power relevant and important program effects may be missed (Hausman to affect the social world; an iterative process rather than an et al., 2013). By contrast, when community stakeholders actively outcome at a single point in time; a process that can be hindered or engage in the process, not only do they develop a sense of facilitated by others, though not gifted by one person to another; ownership, but they also increase the validity and utility of eval- and driven by priorities that are personally meaningful rather than uation efforts for all involved (Fetterman, Kaftarian, & Wanders- imposed (see, e.g., Gutiérrez, 1990; Rappaport, 1987; and Zim- man, 1996). merman, 1995 for critical discussions of the idea of empowerment; To develop the measure at the heart of this study, we established and see Cattaneo & Chapman, 2010). For example, in foundational an ongoing collaboration called the Domestic Violence Program work on empowering practices with women of color, Gutierrez This document is copyrighted by the American Psychological Association or one of its allied publishers. Evaluation and Research Collaborative (DVPERC), composed of touches on each of these themes, noting that empowerment repre- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. researchers, advocates, supervisors, and executive directors from sents on an ongoing process of growth and change that combines 17 DV organizations representing residential (shelter and transi- “a sense of personal control with the ability to affect the behavior tional living) and community programs across northeastern United of others” (p. 150). Empowering practices therefore entail strate- States. The first part of the Method section describes the gies that address both the structural and the psychological imped- DVPERC’s process of developing the initial draft of the measure. iments to gaining power. As should be evident from this descrip- The second part describes the procedures we used to evaluate its tion, the very richness of the empowerment construct that makes it psychometrics, including an exploratory factor analysis and eval- so appealing also makes it difficult to define and measure with uation of the measure’s reliability and validity. In addition to precision. administering background questions and the MOVERS itself, we Consistent with the consensus on key aspects of empowerment, administered a set of standardized measures of depression, self- the Empowerment Process Model (EPM), a conceptual model of efficacy, social support, and satisfaction with services to test empowerment that integrates prior research and theory (see Cat- MOVERS’ convergent validity. These variables were selected taneo & Chapman, 2010 for original model and Cattaneo & Good- based on prior research demonstrating the relationship between man, in press for model revised for DV context), defines empow- MEASURE OF VICTIM EMPOWERMENT RELATED TO SAFETY 3

erment as a meaningful shift in the experience of power attained people (e.g., abusers’ friends and family members) and to the through interaction in the social world. This shift occurs itera- social service and justice system actors that so often revictimize as tively through a process in which a survivor (a) sets personally much as they help. Thus, DVPERC members agreed that a mea- meaningful goals oriented toward increasing power; (b) takes sure of empowerment that is tailored to survivors’ experience must action to accomplish those goals, drawing on both internal (e.g., focus on empowerment as it relates to safety, defined broadly as self-efficacy, knowledge and skills) and external (e.g., formal and freedom from physical or emotional abuse from another person. informal support) resources; and (c) makes meaningful progress Principle #2: The measure should be sufficiently flexible to allow without creating new sources of disempowerment in the process. It for variability in survivors’ safety-related goals. Consistent with is possible for others (e.g., service providers, friends and family) to the overall goal of empowerment, DVPERC members were com- facilitate or hinder any part of the process, from the refinement of mitted to a survivor-defined approach to their work (Davies & goals to the building or degrading of resources, to the curtailing of Lyon, 2013; Kulkarni et al., 2012). That is, to the extent possible, new problems (sources of disempowerment) that might arise as a they saw their role as facilitating survivors’ efforts to become safer person works toward her goals. in whatever way worked best for them, rather than following a To develop a measure of the empowerment of DV survivors rigid protocol laid out by the program. Within a relationship with that would be theoretically grounded as well as ecologically a current or former intimate partner, the dynamics that interfere valid, we proceeded along two highly integrated tracks, one with safety can vary widely, and interact in infinite ways with focusing on scholarship, highlighting the EPM, and the other other aspects of a survivor’s situation; the route to safety varies reflecting the on-the-ground-expertise brought by the members just as widely. of academic-community partnership we established and the Indeed, within the domain of safety, survivors describe a broad voices of survivors with whom they talked. These two tracks range of goals (Davies & Lyon, 2013; Goodman, Dutton, Vankos, quickly became indistinguishable as we moved from scholarly & Weinfurt, 2005). For example, a survivor may want to find a discussions of the empowerment literature to conversations new place to live; gain financial independence to increase her about what we were learning through focus groups at each range of options within or outside the relationship; or increase member organization and then back again (a process described her social support system so that she will have people to help her under Methods, below). Ultimately, these discussions became pursue safety. Within the context of broad goals such as these, the basis for four key principles that integrate scholarship and there is further variety. For example, the survivor seeking financial practice observations, described next. These principles guided independence might wish to find a job, seek public benefits to help the development of the measure. with expenses such as childcare, or decrease expenses by moving in with a family member (Cattaneo & Goodman, in press). More- Principles Resulting From Integration of Scholarship over, these goals may be short-term (e.g., learning about available and Program Expertise services and resources) or longer-term (e.g., gaining job skills). In sum, a measure that takes this complex reality into account must Principle #1: The measure should be domain specific, with a allow for survivors to define their own route to safety. Thus, focus on safety. Research on empowerment suggests that it takes DVPERC members agreed that the measure needed to be stan- on different meanings across specific populations and settings dardized so as to produce a quantitative assessment of survivors’ (Zimmerman & Warschausky, 1998), and that even within specific safety-related empowerment, yet sufficiently flexible to accommo- populations and settings, most people are empowered in some date survivors’ individualized safety-related goals. areas but not in others. Scholars have noted that DV survivors can Principle #3: The measure should bridge the psychological be enormously empowered with respect to parenting or profes- (internal) and the social (external). As research makes clear, sional accomplishment, for example, but disempowered with re- empowerment is neither solely a psychological phenomenon nor spect to physical safety at home (Goodman & Epstein, 2008). solely a social or contextual one, but involves interaction between DVPERC members found this to be the case for many of their individuals and their social contexts (Christens, 2012; Gutiérrez, program participants, and wanted to assess survivors’ empower- 1990; Neal & Neal, 2011; Riger, 1993). In the case of DV, a ment within a specific domain. survivor’s belief (internal) in the potential helpfulness of formal As for which domain to target, DVPERC members were unan- supports informs her likelihood of using those supports; and the This document is copyrighted by the American Psychological Association or one of its allied publishers. imous on this question: They named increased safety from abuse supports’ responses (external), whether positive or negative, in- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. as the most salient (though by no means the only) program-related form her future choices. The measure, therefore, needed to reflect goal for DV survivors accessing their services. Though this may this dynamic interplay and include internal psychological aspects seem obvious, in fact it is not: First, DVPERC members made (e.g., goal identification) and external contextual aspects (e.g., clear that freedom from physical abuse is only one dimension of access to community resources). safety and that survivors in their programs struggled to find free- Principle #4: The measure should assess the trade-offs that dom from emotional abuse as much as from physical abuse. This seeking safety creates as a component of empowerment. Finally, observation is consistent with research showing that emotional both scholarship and DVPERC discussions highlighted a complex- abuse may be particularly damaging to survivors’ psychological ity experienced by programs and survivors: The very process of wellbeing (Coker et al., 2002; Dutton et al., 2006). Second, increasing safety from an abusive partner can create other prob- DVPERC members echoed a growing chorus of voices (e.g., lems of equal or even greater magnitude (Goodman et al., 2009). Goodman, Smyth, Borges, & Singer, 2009) suggesting that for Does the survivor take the job that gives her financial indepen- some survivors, the quest for safety extends beyond the context of dence but leave her children alone with the abusive partner? Does their relationships with their partners to relationships with other she seek a shelter that means moving away from the immigrant 4 GOODMAN ET AL.

community that has been her sole support since she moved to this feedback. During in-person meetings, we discussed the literature country? In these situations, responses that might externally appear and results of our conversations with staff and survivors. “empowered” in that they seem to represent movement toward a Very quickly, these conversations resulted in our selection of the specific goal could, in reality, cause greater levels of instability construct of empowerment as our outcome of interest. This con- and crisis and ultimately even compromise safety further (Hamby, struct rose to the top for several reasons: First, all DVPERC 2014). DVPERC members agreed that empowerment should in- member programs felt that it was a core part of their mission and volve a reduction in these kinds of difficult trade-offs so that the had been from the beginning. Second, it was less contingent on survivor can move out of crisis orientation and toward longer-term outside conditions than other key priorities such as safety alone goals. Put differently, it is not enough to measure empowerment in (dependent on abuser actions) and residential stability (dependent terms of movement toward a specific goal without also assessing on community’s level of affordable housing stock). Once we had the degree to which that very movement creates new and equally established empowerment as the outcome of interest, we turned to difficult challenges. The measure therefore needed to capture the identification of key principles for defining the construct, as participant trade-offs related to safety. described above. As already noted, these principles, in turn, shaped Building on these principles, we generated items for the mea- our decision to focus still further on the idea of empowerment sure, and piloted the resulting draft. This process is described in within the domain of safety. The EPM both influenced these detail next. principles and was influenced by them, ultimately resulting in a formal revision of the model. Finally, during this process we Method determined that we wanted the measure to be available in English and Spanish (the two most common languages spoken at member The method is presented in two sections: The first describes the programs), and that it had to be relatively brief to be usable. DVPERC and its process for developing a draft measure. The The remainder of the year was devoted to generating items for second delineates the psychometric evaluation of the draft mea- each piece of the revised EPM, taking draft items back to program sure, including procedures for recruiting participants, administer- staff meetings for further discussion and refinement along the lines ing the larger survey in which our target measure was embedded, of our principles, and then drafting new items. Overall, this iter- and conducting an exploratory factor analysis. ative process went through at least four rounds before we finally settled on a set of 36 items that became the penultimate draft of Stage I: The DVPERC Process of Developing MOVERS. The last step was to pilot test the draft with the first 20 the MOVERS participants recruited into the study (15 English speakers and five Spanish speakers). We asked these participants to complete the An explicit secondary aim of this project was to develop a survey and give us feedback about confusing items. On the basis strong partnership between academics and DV program staff and of these miniconversations, we changed the wording of a number participants for the purpose of developing useful tools for program of the items to render them clearer to participants. Once we had evaluation. This process proceeded through several phases, de- arrived at the final draft measure (as elaborated below under scribed next. measures), we were ready to begin the psychometric component of DVPERC establishment and composition. The collabora- the study. Every single DVPERC member program signed on for tion started when the directors of three local DV organizations this next stage. asked the first author to consult with them as they worked to develop an outcome measure they could use to respond to internal Stage II: Psychometric Evaluation of the MOVERS and external demands for evaluation. We met biweekly for two months to refine our focus, during which time other organizations Participants. We collected data from a convenience sample heard about the project and asked to join. Eventually, we decided of survivors currently accessing DV services at 17 urban and to open the project to others and it quickly expanded to include suburban organizations in three states in the Northeastern United researchers from several institutions, doctoral students, and repre- States. All programs provided safety planning, counseling, and sentatives of DV programs across the northeast. Most of these information and referrals in person and over the phone to survivors organizations joined in the first four months of the project, though living in the community. Thirteen programs also provided emer- This document is copyrighted by the American Psychological Association or one of its allied publishers. a few did not join until 10 months in. By the end of 12 months, 17 gency shelter, with stay lengths of several weeks to approximately This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. programs had joined and we had adopted our formal title. six months. Shelter sizes ranged from small (i.e., five families) to DVPERC process. Over the course of a 12-month period, medium (i.e., approximately 15 families) to large (i.e., approxi- DVPERC members met monthly to define and generate items for mately 50 families), though most were medium in size. Five of the an evaluation tool. Those researchers who could not be at the programs that provided emergency shelter also provided transi- meetings themselves provided consultation to the group through tional living programs (TLP), with stay lengths of up to two years. the first author, and the three programs that were too far away to An additional two programs provided TLP only. All programs send representatives participated mainly in the second stage of the were “mainstream” in that they did not focus services on one project, the survey administration. Between meetings, three of the particular population. researchers (the first, third, and fourth authors) collected and Eligible participants were (a) aged 18 or over and (b) disseminated relevant research and conducted focus groups with English- or Spanish-speaking. Participant retention in the study staff at seven of the member programs and with survivors at one of was high; only one participant decided to withdraw partway the programs. Meanwhile DV program representatives talked more through because she felt overwhelmed, having arrived at the informally with staff and survivors to elicit further guidance and shelter the day before. From the original sample of 309, we MEASURE OF VICTIM EMPOWERMENT RELATED TO SAFETY 5

eliminated data from the four men who participated in the (88.3%) had children; of those, 45.8% reported that the abuser study, and from four participants who responded to less than was the father of at least one child. 50% of the entire survey. After preliminary analysis (as pre- The sample was evenly divided between residential program sented in the results section), we excluded 71 participants who (48.7%) and community-based program (51.3%) participants. endorsed “not applicable” and/or blank responses among the Among community participants, 35.6% reported current contact retained MOVERS items (see Figure 1 for a description of the with the program about once a week; the next most frequent progress of sample selection). We did not impute these data responses were a few times a month (15.3%) and every few because we considered “not applicable” a substantive response. months (10.2%). Among residential participants, 28.6% of partic- This left a final sample size of 230. Compared with the final ipants had been at the shelter or transitional living program for less sample, the 71 excluded participants included an overrepresen- than one month, 44.7% for one to six months, and 18.8% for more tation of Hispanic/Latina-identified women, as well as partici- than six months. pants who reported being born outside of the U.S. who com- Procedures. Staff at each program posted flyers, described pleted the survey in Spanish, or reported reading and speaking the study at client meetings, and scheduled administration times. English “not well” or “okay.” There is an obvious need to Depending on the number of expected participants, anywhere from replicate these analyses with diverse samples (see Limitations one to four members of the research team administered the surveys below). Nevertheless, this sample size was sufficiently large for to participants in groups of up to 15 people. While survey admin- analyses given that it provided statistical power of .80 to detect istration was done at programs for survivors’ convenience, no correlations of .18 in magnitude at the .05 alpha level. program staff was ever present during survey administration. The The final sample of 230 women had a mean age of 36.2 survey administration team included the first, third, and fourth (SD ϭ 10.9) with a self-identified racial and ethnic composition authors, along with 11 trained graduate students pursuing degrees of 37.8% White, 28.3% Black/African American, 20.0% His- in mental health counseling or social work. Research team mem- panic/Latina, and 11.7% other. Most of the sample (71.7%) was bers described the study and procedures, administered the in- born in the United States. Participants’ socioeconomic status formed consent form orally, and answered questions while partic- was mixed, with about half (52.2%) reporting they attended ipants filled out the survey. For administrations with Spanish- some college, graduated from college, or pursued an advanced speaking participants, a bilingual member of the research team was degree; but 54.8% were unemployed, and almost three fourths present and all instructions and materials were provided in Span- (71.3%) reported receiving some type of need-based govern- ish. To prevent anyone with literacy issues from feeling uncom- ment assistance. A minority (17.0%) reported a physical dis- fortable, we offered to read the survey to any participant who “did ability, but half (49.6%) reported mental health concerns, not feel like” doing the survey on her own, and we stressed that no 72.8% of which interfered with daily functioning. The majority survey information would be shared with program staff. Snacks (87.4%) identified as heterosexual; 20.4% reported being in a and childcare were provided, and all participants received a $20 relationship, 27.7% of whom reported the relationship was with TARGET gift card. Survey administrations lasted between 60 and the partner whose abuse led them to seek DV services. Most 90 minutes. Enrollment Preliminary analysis

Completed survey (n = 309)

Excluded (n = 8) • Identified as male (n = 4) • Responded < 50% of survey (n = 4)

Used for evaluating the

This document is copyrighted by the American Psychological Association or one of its allied publishers. quality of the 36 MOVERS

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. items for EFA (n = 301)

Excluded (n = 71) • Endorsed “not applicable” option for one or more items among the retained 23 MOVERS items (n = 47) • Left one or more items unanswered without endorsing any “not applicable” option for the retained 23 n

Analysis MOVERS items ( = 24)

Used for Conducting EFA (n = 230)

Figure 1. The progress of sample selection. 6 GOODMAN ET AL.

Measures bilingual doctoral students suggested refinements to this transla- tion to enhance its clarity and cultural sensitivity across Spanish Background questions. As reported in participant character- dialects. A second translator then translated the Spanish version istics, we asked about demographics and use of DV services within back into English as a final check for consistency of meaning. We the past year. used this same process for all measures used, except those already Measure of Victims’ Safety Related to Empowerment translated and published in Spanish. When we compared the par- (MOVERS). MOVERS started as a draft with 36 items and, ticipants who completed the survey in Spanish (or in English with following the factor analysis, ended as a 13-item scale composed assisted translations) who were excluded (n ϭ 22) with their of three subscales: Internal Tools (six items), Expectations of counterparts who were not excluded (n ϭ 30), we noted no Support (four items), and Trade-Offs (three items). Participants significant difference in any demographics or scale scores. In responded using a 5-point Likert scale (from never true to always addition, results based on participants who completed the survey in true). The measure introduction defined safety as freedom from English were very similar to those based on the full sample that physical or emotional abuse from another person and encouraged included the 30 Spanish-speaking participants. Therefore, unless respondents to think broadly, if applicable (i.e., When you are stated otherwise, we reported results based on the full sample. responding to these questions, it is fine to think about your family’s safety along with your own if that is what you usually do). Information on reliability is presented below. Results Depressive symptoms. The 20-item Center for Epidemiolog- We used exploratory factor analysis (EFA) to identify latent ical Studies Depression Scale (CES-D; Radloff, 1977) was used to constructs (factors) underlying MOVERS items and to reduce the measure severity of depressive symptoms. Participants responded number of items in the measure (DeVellis, 2012; Worthington & to items (e.g., I could not get going) on a 4-point Likert scale Whittaker, 2006). Although we had a theoretical model of empow- ranging from rarely to all of the time in the past week. This scale erment guiding our work, the model was too new and untested to has been used extensively with community samples of low-income assume that the items would map onto its dimensions precisely. women and has strong construct and concurrent validity when EFA is appropriate in such cases, where there is no assumption that compared with clinical diagnostic criteria, as well as to other items conform to existing theory. All of the following analyses self-report scales of depression (e.g., Belle, 1982; Goodman, were conducted using SPSS 20. 1991). Cronbach’s alpha was .93. Self-efficacy. The 10-item General Self-Efficacy Scale (GSE; Schwarzer & Jerusalem, 1995) measured level of perceived self- Preliminary Analysis efficacy. Participants responded to items (e.g., I can usually handle Before conducting EFAs, we evaluated the quality of the whatever comes my way) on a 4-point Likert scale ranging from not MOVERS items by examining their corrected item-total correla- at all true to exactly true. The GSE has demonstrated high construct tions and the frequency of “not applicable” responses (DeVellis, validity across diverse countries and samples (Luszczynska, 2012; Worthington & Whittaker, 2006). Unless there was an Gutierrez-Dona, & Schwarzer, 2005). Cronbach’s alpha was .91. overriding conceptual reason to keep them, we eliminated items Social support. The12-item Social Support Network Scale that were correlated below .20 with the total scale score to ensure (SSN) from the Chicago Women’s Health Risk Study (Block et al., factorability; and items with a “not applicable” response rate of 2000) measured informal social support. The scale, developed for over 4% to confirm applicability and legibility. Through this adult women challenged by abuse or poverty, assesses three as- process, we removed most of our negatively valenced items (10 pects of informal social support: acceptance and support (five out of 13), in addition to three others, leaving 23 items for further items), emergency help (four items), and access to resources analysis (see Table 1). Item-total correlations were between .22 related to basic needs (three items). We chose to use the first two and .76 for the retained items. of these subscales only since the third did not fit the situations of participants living in a shelter who had temporary access to re- Exploratory Factor Analyses sources they might not normally have. Participants responded with either agree or disagree. The SSN has been shown to have high Estimating factorability. The Kaiser–Meyer–Olkin index This document is copyrighted by the American Psychological Association or one of its allied publishers. reliability and construct validity with abused women (Block et al., (.93; on a possible range of 0–1) indicated excellent sampling This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 2000). Cronbach’s alpha for the first two subscales combined was .84. adequacy and Bartlett’s test of sphericity was significant (p Ͻ Client satisfaction with services. The Client Satisfaction .001), suggesting that the correlation matrix was appropriate for Questionaire-8 (CSQ-8; Attkisson & Greenfield, 2004) measured EFA. level of satisfaction with the program from which participants Determining number of factors. We determined the number were recruited. Participants responded to eight items (e.g., How of factors to extract from a scree plot and parallel analysis would you rate the quality of services you received?) on a 4-point (DeVellis, 2012; Worthington & Whittaker, 2006). Our scree plot scale depending on the nature of the question (e.g., poor to supported a three-factor solution. Next, using the SPSS syntax excellent vs. quite dissatisfied to quite satisfied). The CSQ-8 has developed by O’Connor (2000), we conducted a parallel analysis been demonstrated high internal consistency and predictive valid- with a generation of 100 random permutations from our data set. ity in a variety of samples, including inpatients and outpatients in Given the absence of consensus in the field as to which technique mental health settings. Cronbach’s alpha was .95. is best, we conducted the analysis using both principal-axis fac- Spanish translation. A bilingual translator who was also a toring (PAF) and principal components analysis (PCA) DV advocate conducted the initial translation of MOVERS. Four (O’Connor, 2000). Our analyses based on PAF and PCA suggested MEASURE OF VICTIM EMPOWERMENT RELATED TO SAFETY 7

Table 1 Factor Loadings of the Pattern Matrix and Extracted Communalities for MOVERS

Structure coefficientsa Structure coefficientsb Item F1 F2 F3h2a F1 F2 F3 h2b

Clear sense of goals .86 .00 Ϫ.07 .67 .83 .05 .00 .66 Making progress .85 .07 Ϫ.06 .63 .77 .07 .05 .61 Know my next steps .84 Ϫ.01 Ϫ.01 .69 .85 .00 Ϫ.02 .71 (Strategy) seems to be working .82 Ϫ.03 Ϫ.07 .61 .68 Ϫ.06 .05 .55 Can try something else when something doesn’t work .69 Ϫ.02 .06 .54 .64 Ϫ.03 .10 .52 Feel safe .68 Ϫ.11 Ϫ.07 .46 .59 Ϫ.10 .01 .41 Confident in decisions I make .68 .03 .15 .61 .69 Ϫ.01 .10 .58 Can do what it takes to keep safe .67 Ϫ.05 Ϫ.02 .44 .67 Ϫ.07 Ϫ.06 .43 Can cope with challenges .61 Ϫ.06 .07 .45 .57 Ϫ.12 .07 .46 Can picture my life in 6 months .57 .04 Ϫ.03 .30 .64 .06 Ϫ.09 .32 Have skills I need to keep safe .56 .09 .34 .66 .50 .08 .38 .62 Know what to do in response to threats .55 .02 .21 .49 .53 Ϫ.02 .18 .46 Know about my legal rights .50 Ϫ.03 .19 .43 .35 Ϫ.01 .38 .45 Know ways to keep safe .48 .13 .37 .57 .43 .08 .41 .56 Have a good sense of my needs .46 Ϫ.11 .10 .32 .65 Ϫ.07 Ϫ.04 .43 New problems for me .01 .84 Ϫ.10 .75 .00 .80 Ϫ.08 .69 New problems for people I care about Ϫ.08 .82 .08 .68 Ϫ.09 .80 .10 .65 Have to give up too much .00 .48 Ϫ.04 .24 .01 .50 Ϫ.06 .27 (Know) kinds of support from programs and services Ϫ.09 Ϫ.05 .93 .80 Ϫ.03 Ϫ.03 .87 .74 Feel comfortable asking for help Ϫ.03 Ϫ.08 .66 .44 .01 Ϫ.08 .60 .40 (Know) kinds of support from people in community .09 Ϫ.04 .66 .54 .10 Ϫ.05 .62 .51 Know about community resources available .25 .03 .60 .62 .13 .02 .71 .63 Programs and services able to provide support .17 Ϫ.03 .52 .43 Ϫ.05 Ϫ.02 .74 .50 Note. F1 ϭ Internal Tools; F2 ϭ Trade-Offs; F3 ϭ Expectations of Support. Items and structure coefficients in italic indicate items uniquely loaded onto a specific factor. Items and structure coefficients in bold indicate items retained in the final version of MOVERS. a Statistics were calculated based on the full sample (n ϭ 230). b Statistics were calculated based on the sample of participants who completed the survey in English (n ϭ 200).

extracting six and three factors, respectively. With this range, we minimizing the length of the final measure. This resulted in a total used PAF to separately extract the three-, four-, five-, and six- of 13 MOVERS items—six for Internal Tools, three for Trade- factor solutions. For each factor solution, we applied an oblique Offs, and four for Expectations of Support (see Table 1, bold rotation using the direct oblimin method to aid factor interpretation items). A final EFA confirmed these changes had not altered our due to our assumption that the factors would be correlated. We factor structure (Worthington & Whittaker, 2006). The final three- then considered whether the rotated factor solutions fit our con- factor solution accounted for 56.9% of the shared variance in the ceptual priorities and interpretability, demonstrated satisfactory 13 items (eigenvalues for unrotated Factors 1–3 were 5.18, 1.41, internal consistency, and consisted of factors with at least three and 0.80, respectively). Communalities were in an adequate range items (Worthington & Whittaker, 2006). Among all factor solu- of .24–.83. The absolute values of the two strongest factor load- tions, the three-factor solution was best in meeting the above ings across each item differed by at least .32, which indicated a criteria. simple factor structure. Specifically, the six items for Internal tools Developing MOVERS subscales. The development of sub- loaded most strongly on Factor 1 with their coefficients ranging scales included a statistical and a collaborative element. Statisti- from .61 to .87. I know what my next steps are on the path to keep cally, we interpreted the factors from the three-factor solution as safe was its top-loading item. The three items for Trade-Offs follows: (a) Internal Tools (a general assessment of a survivor’s loaded most strongly on Factor 2 with their coefficients ranging This document is copyrighted by the American Psychological Association or one of its allied publishers. internal resources relevant to safety; 13 items); (b) Trade-Offs (the from .48 to .86. Working to keep safe creates (or will create) new This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. extent to which the survivor feels that her efforts to achieve safety problems for people I care about was its top-loading item. Lastly, trigger new difficulties, three items); and (c) Expectations of the four items for Expectations of Support loaded most strongly on Support (the extent to which the survivor has knowledge about and Factor 3 with their coefficients ranging from .49 to .92. I have a access to useful support from formal supports and informal net- good idea about what kinds of support for safety I can get from works, five items). We retained items based on two criteria: First, programs, services, and/or government agencies was its top- to ensure a strong relation of each item to the construct it assessed, loading item. we retained items if their highest structure coefficient was at least .40 in absolute magnitude. Next, to ensure each subscale was Descriptive Statistics, Reliability, and distinct, we retained the item if the difference between the absolute Validity of MOVERS values of the two strongest structure coefficients was at least .25. These criteria led to retention of 21 items (italicized in Table 1). We computed subscale scores by averaging item scores corre- In collaboration with our partners through a consensus process, sponding to each subscale, and generated a total score by averag- we next eliminated several redundant items, with the goal of ing item scores with the Trade-Offs items reverse-scored. Table 2 8 GOODMAN ET AL.

Table 2 The MOVERS’ Structure and Content Subscale Correlations, Means, Standard Deviations, and Internal Consistency The structure of MOVERS that emerged through the EFA fits well with the theoretical model of empowerment (the EPM) that helped to 1 2 3 Mean SD ␣ guide item generation, described earlier. The EPM suggests that a survivor experiences a meaningful shift in the experience of power ءء ءء 1. Internal tools — Ϫ.35 .62 3.94 0.79 .88 related to safety from an abusive partner through a process of setting 74. 1.03 2.10 ءءϪ.33 — ءءTrade-offs Ϫ.26 .2 3. Expectations of personally meaningful goals oriented toward increasing safety, and ءء ءء support .62 Ϫ.29 — 3.78 0.96 .83 taking action to accomplish those goals—drawing on both internal Mean 3.95 2.10 3.80 SD 0.83 1.06 0.99 and external resources—making meaningful progress without creat- ␣ .88 .74 .83 ing new sources of disempowerment in the process (Cattaneo & Goodman, in press). The three factors that emerged from the EFA Note. Statistics under the diagonal are based on the entire sample (n ϭ 230) whereas those above the diagonal are based on the sample of which encompass all of the components of the original model, grouping only participants who completed the survey in English were included (n ϭ them together in a way that is parsimonious and clear. 200). Specifically, Internal Tools, the first factor of MOVERS, mea- .p Ͻ .01 ءء sures the extent to which a survivor has developed a set of safety-related goals and a belief in her ability to accomplish them, encompassing the goals, self-efficacy, knowledge, and skill com- lists subscale correlations, means, standard deviations, and Cron- ponents of the model. Higher scores on this subscale represent the bach’s alpha estimates. The correlations among subscales were in survivor’s perception that she knows what she wants to do to the expected directions, with Trade-Offs negatively correlated with increase safety and that she has the internal resources to accom- Internal Tools and Expectations of Support. Further supporting the plish those goals. Trade-Offs, the second factor of MOVERS, scale’s construct validity, its total score was significantly related to measures the survivor’s sense that action toward the goal of safety the other study measures as hypothesized: MOVERS was nega- will create new problems to contend with. Higher scores reflect a tively associated with depressive symptom scores on the CES-D greater sense of risk in moving toward safety (and need to be r ϭϪ ( .42), and positively associated with General Self-Efficacy reverse scored in order to obtain a total MOVERS score). This (r ϭ .50), the two subscale version of the Social Support Network factor encompasses the EPM idea that the process of empower- Scale (r ϭ .35), and the Client Satisfaction Questionnaire (r ϭ ment entails the survivor’s evaluation of his or her progress .35). Correlations with subscales were also consistent with theory: toward goals in light of whether that progress has brought about Among the three MOVERS subscales, Internal Tools was most new problems (e.g., new sources of disempowerment) that may be strongly associated with the General Self-Efficacy Scale (r ϭ .55); Expectations of Support was most highly correlated with the just as daunting as the threats to safety. Expectations of Support, Social Support Network Scale; and the Client Satisfaction Ques- the third factor, measures the survivor’s sense that others can and tionnaire (r ϭ .38 and .41). will assist her as needed, incorporating the formal and informal support component of the model. Higher scores reflect the survi- Discussion vor’s perception that the support she needs exist, and that she can access it. These factors fit with the principles described earlier, and This study attempted to address a major challenge of DV program are thus consistent with both the scholarly and on-the-ground evaluation through the development of a theory and practice-informed expertise that undergirded the project. measure of safety-related empowerment. Members of the DVPERC, MOVERS demonstrated satisfactory psychometric properties. The a university–community partnership including the authors and repre- communalities of the final 13 items were adequate (i.e., above .20). sentatives of 17 DV programs across the Northeast collaborated in an The factor structure was distinctive, as reflected by a lack of equally iterative year-long process to create a measure that could be used high loadings across factors. The reliability of the subscale scores across program models and timeframes to assess program partici- were satisfactory (i.e., Cronbach’s alphas above .70). Finally, the pants’ empowerment as it relates to their safety from domestic vio- This document is copyrighted by the American Psychological Association or one of its allied publishers. construct validity of the MOVERS, both as a full measure and as a set lence. An exploratory factor analysis based on 230 survivors seeking This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. of subscales, was excellent: As expected, MOVERS was related to support from DV programs and validation of the measure based on its global measures of psychological wellbeing (i.e., depression, general relationship with other hypothesized variables supported the mea- self-efficacy) and perceived quality of support systems (i.e., sure’s reliability and validity. MOVERS (available from first author upon request) assesses empowerment within a single domain (safety), informal social support, satisfaction with services), though cor- is sufficiently flexible to accommodate participants’ individualized relations were moderate. Also consistent with expectations, safety-related goals, bridges the psychological and social dimensions Internal Tools was most strongly associated with general self- of empowerment, and incorporates measurement of the trade-offs efficacy and Expectations of Support was most highly corre- inherent in seeking safety. lated with informal social support and satisfaction with ser- This section reviews the factor structure of the measure, the value vices. of the university-community partnership that created it, and the lim- In sum, MOVERS appears to be a theoretically grounded, ecolog- itations of the study. We then discuss some of the research questions ically valid, and psychometrically sound measure that assesses a to which the measure lends itself, as well as the measure’s potential distinct and key component of survivors’ experience in DV programs uses in practice contexts. and potentially beyond. MEASURE OF VICTIM EMPOWERMENT RELATED TO SAFETY 9

University–Community Partnerships for Outcome change over time. However, DVPERC members are committed to Measure Development using MOVERS to assess program participants at multiple time points, providing an opportunity for the partnership to fill this gap in This project represents a unique contribution to the literature on the future. community-based participatory evaluation. Despite compelling evi- dence of the value of research and evaluation efforts that engage stakeholders, the critical task of developing outcome measurement Implications for Research and Evaluation tools has often been left out of the collaborative process (Hausman et al., 2013). This project countered that pattern by integrating practi- Assuming that a CFA confirms the factor structure of the MOV- tioner and survivor input throughout the entire measure development ERS, it’s development and validation opens the way for several new process. Researchers contributed literature on empowerment and a lines of evaluation and research. First, despite the central role of conceptual model that community stakeholders could work with, as empowerment as a guiding concept in the DV movement, there has well as the expertise required to develop a psychometrically valid been relatively little empirical research on how to conceptualize or measure; and community members brought their deep knowledge of measure it. This project provides a measure that can be used to the target population, the mission and practices of their programs, and evaluate program practices in ways that are consistent with core the pragmatic requirements for a measure that could be useful in the program and survivor goals. It also enables the evaluation of practices context of sometimes chaotic and generally underresourced settings. in terms of their effectiveness for specific survivor subgroups over This collaboration contributed dramatically to the quality of the final time, and with respect to the specific subscales of MOVERS. measure and ensured its conceptual integrity, relevance, face validity, In terms of research, the development of MOVERS allows for and utility. It provides an example of how academic–community investigation of a range of new questions, including what individual, partnerships can produce something powerful and useful that neither interpersonal, programmatic, and larger contextual factors contribute academics nor practitioners could produce alone. to safety-related empowerment, whether certain subgroups of survi- vors have different trajectories in their path toward empowerment, and most importantly, whether safety-related empowerment indeed Limitations predicts longer-term safety and wellbeing. Most of these questions require longitudinal follow-up, a challenge for researchers who focus Several sampling issues limit the generalizability of this study’s on marginalized and often residentially unstable groups. Here too, findings. First, although the sample was ethnically diverse and drawn university–community partnerships can enhance researchers’ capac- from programs across the Northeast, thereby increasing ecological ity to engage survivors safely and over time. validity and generalizability, we were not able to recruit from other regions of the country where safety-related empowerment might be felt differently. Second, the small number of Spanish-speaking survi- Implications for Practice vors in the sample made it impossible to conduct a separate EFA with this subsample, and we recognize the urgent need to replicate the MOVERS enables the concrete evaluation of programs’ efforts to study with non-English speaking groups. In particular, given that facilitate safety-related empowerment—a central tenet of their prac- Hispanic/Latina participants who were born outside of the United tice. For DVPERC programs and many others, facilitating a survi- States, who completed the survey in Spanish, or who reported speak- vor’s empowerment means partnering with her to envision a set of ing English “not well” or only “okay” were overrepresented among goals and strengthen the internal resources necessary to move toward participants dropped from the final sample, generalization to this them, learn about and connect with potential external sources of group requires additional study. Third, the sample was entirely female assistance, and access some of the resources necessary to prevent the and almost entirely heterosexual. The extent to which the measure emergence of impossible double-binds and trade-offs. Not only does would be useful for men or gay, lesbian, bisexual, transgender, and/or MOVERS allow for research and evaluation related to these activities, queer survivors is unclear. Despite these limitations, it is heartening in it also opens up the possibility of individual assessment of survivors’ terms of generalizability to note that demographics of this sample experience and progress. Administering MOVERS periodically could (including race/ethnicity, immigration status, and educational back- increase staff’s capacity to conduct ongoing assessment of their own ground) are almost identical to those of a national sample obtained in delivery on the three aspects of program support described above. This document is copyrighted by the American Psychological Association or one of its allied publishers. a study of community-based domestic violence programs (Lyon, Indeed, noting that a laser focus on survivors’ highest priorities This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Bradshaw, & Menard, 2011). This suggests that the measure may well sometimes resulted in other important issues falling off the proverbial be generalizable to survivors seeking help from mainstream DV table, several DVPERC member programs have begun to use the programs. Still, further investigation is needed to learn whether the measure as an ongoing check on how survivors are fairing within each measure is applicable to those survivors who do not access formal of the three dimensions of empowerment—internal resources, exter- support. nal supports, and a diminishment of difficult trade-offs. The Trade- In terms of sample size, although 230 participants was a large Offs subscale in particular may be valuable to programs, reminding enough sample to conduct an EFA, it was not large enough for a them to assess the existence and severity of difficulties that are random split to conduct a confirmatory factor analysis (CFA) for external to but intertwined with the central issue of safety and pointing cross-validation. For this reason, the findings presented here represent the way toward various types of system-level collaborations. Indeed, preliminary analyses only. A CFA will be required to learn whether documentation of the nature and extent of program participant the MOVERS factor structure stays stable in a new sample (Van tradeoffs may enable programs to fight successfully for greater Prooijen & Van Der Kloot, 2001). Finally, the cross-sectional nature system-wide collaboration and increased funding for new types of of the sample prevented evaluation of the measure’s ability to capture support. 10 GOODMAN ET AL.

Conclusion Fetterman, D. M., Kaftarian, S. J., & Wandersman, A. (1996). Empower- ment evaluation: Knowledge and tools for self-assessment and account- In this article, we describe a process of collaboration that was ability. London: Sage. arduous at times but that we feel was well worth the investment. It Goodman, L. A. (1991). The relationship between social support and is our hope that the measure it generated will be useful to programs family homelessness: A comparison study of homeless and housed and researchers as we move toward our common, overriding goal mothers. Journal of Community Psychology, 19, 321–332. http://dx.doi of survivor safety. .org/10.1002/1520-6629(199110)19:4Ͻ321::AID-JCOP2290190404Ͼ3 .0.CO;2-8 Goodman, L., Dutton, M. A., Vankos, N., & Weinfurt, K. (2005). Wom- References en’s resources and use of strategies as risk and protective factors for Attkisson, C. C., & Greenfield, T. K. (2004). The UCSF Client Satisfaction reabuse over time. Violence Against Women, 11, 311–336. http://dx.doi Scales: I. The Client Satisfaction Questionnaire-8. In M. Maruish (Ed.), .org/10.1177/1077801204273297 The use of psychological testing for treatment planning and outcome Goodman, L. A., & Epstein, D. (2008). Listening to battered women: A assessment (3rd ed., pp. 799–811). Mahwah, NJ: Erlbaum. survivor centered approach to advocacy, mental health, and justice. Belle, D. (1982). Lives in stress: Women and depression. Beverly Hills, Washington, DC: American Psychological Association. CA: Sage. Goodman, L. A., Smyth, K. F., Borges, A. M., & Singer, R. (2009). When Bennett, L., Riger, S., Schewe, P., Howard, A., & Wasco, S. (2004). Effec- crises collide: How intimate partner violence and poverty intersect to tiveness of hotline, advocacy, counseling, and shelter services for victims of shape women’s mental health and coping. Trauma, Violence and Abuse: domestic violence: A statewide evaluation. Journal of Interpersonal Vio- Special Issue on the Mental Health Implications of Violence Against lence, 19, 815–829. http://dx.doi.org/10.1177/0886260504265687 Women, 10, 306–329. Black, M. C., Basile, K. C., Breading, M. J., Smith, S. G., Walters, M. L., Gutiérrez, L. M. (1990). Working with women of color: An empowerment Merrick, M. T.,...Stevens, M. R. (2011). The National Intimate perspective. Social Work, 35, 149–153. Partner and Sexual Violence Survey (NISVS): 2010 summary report. Hamby, S. (2014). Stronger Than You Know. Oxford University Press. Atlanta, GA: National Center for Injury Prevention and Control, Centers Hausman, A. J., Baker, C. N., Komaroff, E., Thomas, N., Guerra, T., Hohl, for Disease Control and Prevention. B. C., & Leff, S. S. (2013). Developing measures of community-relevant Block, C., Devitt, C. O., Fonda, D., Engle, B., Fugate, M., Martin, C.,... outcomes for violence prevention programs: A community-based par- Walker, G. R. (2000). The Chicago Women’s Health Risk Study: Risk of ticipatory research approach to measurement. American Journal of Com- serious injury or death in intimate violence: A collaborative research munity Psychology, 52, 249–262. http://dx.doi.org/10.1007/s10464-013- project. Washington, DC: Department of Justice, National Institute of 9590-6 Justice. http://dx.doi.org/10.1037/e300372003-001 Israel, B. A., Eng, E., Schulz, A. J., & Parker, E. A. (2005). Methods in Buzawa, E., & Hotaling, G. (2003). Domestic violence assaults in three community-based participatory research for health. San Francisco, CA: Massachusetts communities: Final report. Washington, DC: National Institute of Justice. Jossey-Bass. Cattaneo, L. B., & Chapman, A. R. (2010). The process of empowerment: Johnson, D. M., Zlotnick, C., & Perez, S. (2011). Cognitive behavioral A model for use in research and practice. American Psychologist, 65(7), treatment of PTSD in residents of battered women’s shelters: Results of 646–659. http://dx.doi.org/10.1037/a0018854 a randomized clinical trial. Journal of Consulting and Clinical Psychol- Cattaneo, L. B., & Goodman, L. A. (in press). What is empowerment ogy, 79, 542–551. http://dx.doi.org/10.1037/a0023822 Anyway? A model for domestic violence practice, research and evalu- Kasturirangan, A. (2008). Empowerment and programs designed to address ation. Psychology of Violence. domestic violence. Violence Against Women, 14, 1465–1475. http://dx Cattaneo, L. B., & Goodman, L. A. (2010). Through the lens of therapeutic .doi.org/10.1177/1077801208325188 jurisprudence: The relationship between empowerment in the court Kulkarni, S. J., Bell, H., & Rhodes, D. M. (2012). Back to basics: system and well-being for intimate partner violence victims. Journal of Essential qualities of services for survivors of intimate partner vio- Interpersonal Violence, 25, 481–502. http://dx.doi.org/10.1177/ lence. Violence Against Women, 18, 85–101. http://dx.doi.org/ 0886260509334282 10.1177/1077801212437137 Christens, B. D. (2012). Toward relational empowerment. American Jour- Luszczynska, A., Gutiérrez-Doña, B., & Schwarzer, R. (2005). General nal of Community Psychology, 50, 114–128. http://dx.doi.org/10.1007/ self-efficacy in various domains of human functioning: Evidence from s10464-011-9483-5 five countries. International Journal of Psychology, 40, 80–89. http:// Coker, A. L., Davis, K. E., Arias, I., Desai, S., Sanderson, M., Brandt, dx.doi.org/10.1080/00207590444000041 This document is copyrighted by the American Psychological Association or one of its allied publishers. H. M., & Smith, P. H. (2002). Physical and mental health effects of Lyon, E., Bradshaw, J., & Menard, A. (2011). Meeting survivors’ needs

This article is intended solely for the personal use of the individual userintimate and is not to be disseminated broadly. partner violence for men and women. American Journal of through non-residential domestic violence services & supports: Results Preventive Medicine, 23, 260–268. http://dx.doi.org/10.1016/S0749- of a multi-state state. Final Report. Washington, DC: National Institute 3797(02)00514-7 of Justice. Corrigan, P. W., Faber, D., Rashid, F., & Leary, M. (1999). The construct Macy, R. J., Giattina, M., Sangster, T. H., Crosby, C., & Montijo, N. J. validity of empowerment among consumers of mental health services. (2009). Domestic violence and sexual assault services: Inside the black Schizophrenia Research, 38, 77–84. http://dx.doi.org/10.1016/S0920- 9964(98)00180-7 box. Aggression and Violent Behavior, 14, 359–373. http://dx.doi.org/ Davies, J., & Lyon, E. (2013). Domestic violence advocacy: Complex 10.1016/j.avb.2009.06.002 lives/difficult choices. Thousand Oaks, CA: Sage. Neal, J. W., & Neal, Z. P. (2011). Power as a structural phenomenon. DeVellis, R. F. (2012). Scale Development: Theory and applications (3rd American Journal of Community Psychology, 48, 157–167. http://dx.doi ed.). Chapel Hill, NC: Sage. .org/10.1007/s10464-010-9356-3 Dutton, M. A., Green, B. L., Kaltman, S. I., Roesch, D. M., Zeffiro, T. A., O’Connor, B. P. (2000). SPSS and SAS programs for determining the & Krause, E. D. (2006). Intimate partner violence, PTSD, and adverse number of components using parallel analysis and velicer’s MAP test. health outcomes. Journal of Interpersonal Violence, 21, 955–968. http:// Behavior Research Methods, Instruments, & Computers, 32, 396–402. dx.doi.org/10.1177/0886260506289178 http://dx.doi.org/10.3758/BF03200807 MEASURE OF VICTIM EMPOWERMENT RELATED TO SAFETY 11

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for Sullivan, C., & Alexy, C. (2001). Evaluating the outcomes of domestic research in the general population. Applied Psychological Measurement, violence service programs: Some practical considerations and strate- 1, 385–401. http://dx.doi.org/10.1177/014662167700100306 gies. Harrisburg, PA: VAWnet, a project of the National Resource Rappaport, J. (1987). Terms of empowerment/exemplars of prevention: Center on Domestic Violence/Pennsylvania Coalition Against Domestic Toward a theory for community psychology. American Journal of Com- Violence. Retrieved February 8, 2014 from http://www.vawnet.org munity Psychology, 15, 121–148. http://dx.doi.org/10.1007/BF00919275 Van Prooijen, J. W., & Van Der Kloot, W. A. (2001). Confirmatory Riger, S. (1993). What’s wrong with empowerment. American Journal of analysis of exploratively obtained factor structures. Educational and Community Psychology, 21, 279–292. http://dx.doi.org/10.1007/ Psychological Measurement, 61, 777–792. http://dx.doi.org/10.1177/ BF00941504 00131640121971518 Rogers, E. S., Chamberlin, J., Ellison, M. L., & Crean, T. (1997). A Worthington, R. L., & Whittaker, T. A. (2006). Scale development re- consumer-constructed scale to measure empowerment among users of search: A content analysis and recommendations for best practices. The mental health services. Psychiatric Services, 48, 1042–1047. Counseling Psychologist, 34, 806–838. http://dx.doi.org/10.1177/ Schwarzer, R., & Jerusalem, M. (1995). Generalized Self-Efficacy Scale. 0011000006288127 In J. Weinman, S. Wright, & M. Johnston (Eds.), Measures in health Zimmerman, M. A. (1995). Psychological empowerment: Issues and illus- psychology: A user’s portfolio. Causal and control beliefs (pp. 35–37). trations. American Journal of Community Psychology, 23, 581–599. Windsor, UK: Nfer-Nelson. http://dx.doi.org/10.1007/BF02506983 Song, L. Y. (2012). Service utilization, perceived changes of self, and life Zimmerman, M. A., & Warschausky, S. (1998). Empowerment theory for satisfaction among women who experienced intimate partner abuse: The rehabilitation research: Conceptual and methodological issues. Rehabil- mediation effect of empowerment. Journal of Interpersonal Violence, itation Psychology, 43, 3–16. http://dx.doi.org/10.1037/0090-5550.43 27, 1112–1136. http://dx.doi.org/10.1177/0886260511424495 .1.3 Sullivan, C. M. (2011). Evaluating domestic violence support service Zweig, J. M., & Burt, M. R. (2007). Predicting women’s perceptions of programs: Waste of time, necessary evil, or opportunity for growth? domestic violence and sexual assault agency helpfulness: What matters Aggression and Violent Behavior, 16, 354–360. http://dx.doi.org/ to program clients? Violence Against Women, 13, 1149–1178. http://dx 10.1016/j.avb.2011.04.008 .doi.org/10.1177/1077801207307799

Appendix Measure of Victim Empowerment in Relation to Safety

You may be facing a variety of different challenges to safety. When we use the word safety in the next set of statements, we mean safety from physical or emotional abuse by another person. Please circle the number that best describes how you think about your and your family’s safety right now. When you are responding to these statement, it is fine to think about your family’s safety along with your own if that is what you usually do.

Your Safety

Never Sometimes Half the Mostly Always Statement true true time true true true

1. I can cope with whatever challenges come at me as I work to keep safe. 1 2 3 4 5 2. I have to give up too much to keep safe. 1 2 3 4 5 3. I know what to do in response to threats to my safety. 1 2 3 4 5 4. I have a good idea about what kinds of support for safety that I can get from people in my community (friends, family, neighbors, people in my faith community, etc.). 1 2 3 4 5 5. I know what my next steps are on the path to keeping safe. 1 2 3 4 5 6. Working to keep safe creates (or will create) new problems for me. 1 2 3 4 5 7. When something doesn’t work to keep safe, I can try something else. 1 2 3 4 5 This document is copyrighted by the American Psychological Association or one of its allied publishers. 8. I feel comfortable asking for help to keep safe. 1 2 3 4 5 This article is intended solely for the personal use of the9. individual user and isWhen not to be disseminated broadly. I think about keeping safe, I have a clear sense of my goals for the next few years. 1 2 3 4 5 10. Working to keep safe creates (or will create) new problems for people I care about. 1 2 3 4 5 11. I feel confident in the decisions I make to keep safe 1 2 3 4 5 12. I have a good idea about what kinds of support for safety I can get from community programs and services. 1 2 3 4 5 13. Community programs and services provide support I need to keep safe. 1 2 3 4 5 Note. Measure of Victim Empowerment in Relation to Safety (MOVERS). © Lisa Goodman, 2014. No part of this measure may be reproduced or distributed without the prior written permission of the first author.

(Appendix continues) 12 GOODMAN ET AL.

Su Seguridad

No es Aveces es La mitad del tiempo Casi siempre Siempre verdad verdad es verdad es verdad es verdad

1. Puedo enfrentarme a cualquier reto para mantener mi seguridad. 1 2 3 4 5 2. Tengo que renunciar demasiadas cosas para mantener mi seguridad. 1 2 3 4 5 3. Yo se= como responder a amenazas a mi seguridad. 1 2 3 4 5 4. Yo se= que tipos de apoyo con respeto a seguridad puedo obtener en mi comunidad (amigos, familia, vecinos, gente de mi espiritual) 1 2 3 4 5 5. Yo se= cuales son los siguientes pasos para mantenerme seguro/a. 1 2 3 4 5 6. Mis intentos para mantener mi seguridad crean o van a crear nuevos problemas para mí. 1 2 3 4 5 7. Cuando algo no está funcionando para mantener mi seguridad, yo puedo intentar algo diferente. 1 2 3 4 5 8. Me siento co=modo/a pidiendo ayuda para mantener mi seguridad. 1 2 3 4 5 9. Cuando pienso en mi seguridad, tengo claras mis metas para el futuro. 1 2 3 4 5 10. Mis intentos para mantener mi seguridad crean o van a crear nuevos problemas para la gente que yo quiero. 1 2 3 4 5 11. Me siento seguro de las decisiones que hago para mantener mi seguridad. 1 2 3 4 5 12. Tengo una buena idea de qué tipo de apoyo puedo conseguir de las programas comunitarios para mantener mi seguridad. 1 2 3 4 5 13. Hay programas en mi comunidad y servicios sociales que pueden proveer el apoyo y los recursos que yo necesito para mantener mi seguridad. 1 2 3 4 5 Note. Measure of Victim Empowerment in Relation to Safety (MOVERS). © Lisa Goodman, 2014. No parte de esta medida puede ser reproducida o distribuida sin obtener permiso escrito de la primera autora.

Received February 18, 2014 Revision received August 7, 2014 Accepted August 8, 2014 Ⅲ This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.