EXAMINING ORGANIZATIONAL SUPPORTS WITHIN DOMESTIC VIOLENCE
PROGRAMS THAT SUPPORT OR HINDER RESPONSES TO REPRODUCTIVE
COERCION
By
Sara McGirr

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Psychology—Doctor of Philosophy
2017

ABSTRACT
EXAMINING ORGANIZATIONAL SUPPORTS WITHIN DOMESTIC VIOLENCE
PROGRAMS THAT SUPPORT OR HINDER RESPONSES TO REPRODUCTIVE
COERCION
By
Sara McGirr
Reproductive coercion (RC) is a newly identified but commonly experienced form of
domestic violence (DV) with serious potential consequences for women’s health and well-being.
Despite the high prevalence of RC against DV survivors, initial reports suggest that few DV
advocates regularly engage in RC-related practices with their clients. In order to better
understand the factors that may be impeding advocates’ RC-responsiveness, the study examined
data collected via a brief online survey of more than 300 domestic violence advocates across the
U.S. and its territories. Results identified critical barriers and facilitators to RC-responsive
practice in DV organizations on the intrapersonal and organizational ecological levels. While
intrapersonal factors (levels of comfort discussing sexuality and comfort discussing reproductive
health) influenced advocates’ frequency of universal and targeted RC practice, the level of RCresponsive supports provided by advocates’ organizations was much more impactful. In addition
to supporting the assertion that intervention on multiple ecological levels has the greatest
potential for successful change in professionals’ behavior, the study’s results also provide initial
insight into a minimum level of organizational supports that may be necessary to promote more
frequent RC-responsive practice in DV organizations. This guidance may prove useful for
agencies aiming to improve their RC-responsiveness; by focusing on those factors, DV programs
and their staff will ultimately be better prepared and better able to support survivors of RC in
regaining their reproductive control.

ACKNOWLEDGEMENTS

To my family, thank you for always supporting me in following my dreams, even when they take
me far away, for teaching me how to be a professional-ish adult, and for loving me no matter
what.
To my partner of three years, thank you for your steady support throughout the many
celebrations and disappointments that have accompanied this project, and for always making me
laugh.
To my advisor, Dr. Cris Sullivan, thank you for your mentorship, your training on how to trust
my instincts, and that keen editing eye.
To my committee members, Drs. Heather McCauley, Deb Bybee, and Amy Drahota, thank you
for your technical assistance and brilliant suggestions throughout my project.
To the Eco graduate students who helped to develop and implement the larger project from
which this data was drawn, Cortney Vandegrift, Heather Bomsta, and Dr. Katie Gregory, thank
you for the time and effort you contributed and your generosity in sharing this data.
To all of the domestic violence victim services practitioners who helped to develop the survey
and the advocates who participated, thank you for taking on this task on top of all of the other
incredibly meaningful and challenging work that you do daily.
To my team at Michigan Public Health Institute, including Jessica Hamel, Dr. Julia Heany, and
Dr. Chiharu Kato, thank you for believing in my abilities when it was hard for me to do so, and
for providing me with countless opportunities for professional development.

iii

To my student friends, including but not limited to Jenny, Nadeeka, three Katies, and Katherine,
thank you for the countless days and nights spent in coffee shops, bars, and online chats together
as we co-worked; without this supportive structure I would likely have made no progress.
Finally, to all those who I have befriended during my six years in Lansing, thank you for
bringing fun and balance to my life; for inspiring me to be creative and to play; and for providing
safe spaces for reflection and processing when things got tough. You made this college town a
home.

iv

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................................... vii
LIST OF FIGURES .................................................................................................................... viii
INTRODUCTION .......................................................................................................................... 1
LITERATURE REVIEW ............................................................................................................... 3
Reproductive Coercion ................................................................................................................ 3
Interventions to Address Reproductive Coercion ....................................................................... 7
RC-Responsive Practice in Domestic Violence Service Settings ............................................. 10
Identifying Barriers and Facilitators to RC-Responsive Practice in DV Settings .....................12
Barriers and Facilitators to RSH-Responsive Practice in Other Settings ..................................17
Individuals’ comfort discussing sex and sexuality ........................................................... 18
Individuals’ comfort discussing reproductive health topics ............................................. 23
Organizational factors that promote or hinder RSH responsive practice ......................... 27
Current Study ............................................................................................................................ 33
METHOD ..................................................................................................................................... 35
Survey Construction .................................................................................................................. 35
Survey Recruitment and Administration ................................................................................... 35
Measures .................................................................................................................................... 37
Analytic Techniques .................................................................................................................. 40
The need for Rasch measurement ..................................................................................... 40
Rasch analysis of measures ............................................................................................... 46
Comfort discussing sex and sexuality with clients ............................................... 46
Comfort discussing reproductive health with clients ............................................ 51
Universal reproductive coercion practices ............................................................ 53
Targeted reproductive coercion practices ............................................................. 58
RC-responsive organizational supports ................................................................ 61
Descriptive and parametric analyses ................................................................................. 64
RESULTS ..................................................................................................................................... 65
Sample Characteristics .............................................................................................................. 65
Descriptive Statistics ................................................................................................................. 66
Advocates’ comfort discussing sexuality and reproductive health ................................... 66
Organizational supports .................................................................................................... 67
Engagement in RC-responsive practices .......................................................................... 67
Demographic and Organizational Covariates of Study Scales .................................................. 78
Hypothesis 1 .............................................................................................................................. 79
Hypothesis 2 .............................................................................................................................. 80
Hypothesis 3 .............................................................................................................................. 81
Hypothesis 3a ............................................................................................................................ 81
v

Latent class probabilities as predictors of practice ........................................................... 89
Assigned class as a predictor of practice .......................................................................... 93
Hypothesis 4 .............................................................................................................................. 94
DISCUSSION ............................................................................................................................... 99
Identifying Key Levers for Change ......................................................................................... 102
Study Limitations .................................................................................................................... 106
Future Directions ..................................................................................................................... 108
Conclusion ............................................................................................................................... 109
REFERENCES ........................................................................................................................... 111

vi

LIST OF TABLES

Table 1. Descriptives, Fit Indices and Separation Indices for Rasch Analysis .............................48
Table 2. Frequencies & Descriptives: Comfort Discussing Sexuality with Clients ......................69
Table 3. Frequencies & Descriptives: Comfort Discussing Reproductive Health with Clients ....71
Table 4. Frequencies & Descriptives: RC-Responsive Organizational Supports Items ................72
Table 5. Frequencies & Descriptives: Universal Reproductive Coercion (RC) Practice Items ....74
Table 6. Frequencies & Descriptives: Targeted Reproductive Coercion (RC) Practice Items ......76
Table 7. Bivariate Correlations Between Advocate Characteristics and Study Scale Person
Measure ..........................................................................................................................................78
Table 8. Inter-item Correlations for RC-Responsive Organizational Supports .............................82
Table 9. RC-Responsive Organizational Supports Latent Class Analysis Fit Indices ..................84
Table 10. Characteristics of Each Latent Class of RC-Responsive Organizational Supports
(Final Four Class Solution) ............................................................................................................88
Table 11. Means and Standard Errors for Universal RC Practices and Targeted RC Practice Person
Measures by Latent Class Probabilities .........................................................................................91
Table 12. Bivariate Correlations Between Study Scale Person Measures .....................................95

vii

LIST OF FIGURES

Figure 1. Simplified Study Model for Hypothesis Four ...............................................................34
Figure 2. “Example survey rating scale. For the Q#5 scale, the “jump” between each of the
ratings is equal. For the second (Q#8) and third (Q#10) scales, the “jump” from each rating to the
next rating is not equal. Furthermore, the way the rating scale functions across the items is not
identical. All that a researcher can assert is that the rating scale is ordinal (SA > A > D > SD) for
each item” (Boone, 2016, p. 2; SA=Strongly Agree; A=Agree; D=Disagree; SD=Strongly
Disagree). .......................................................................................................................................43
Figure 3. Wright Map for Comfort Discussing Sexuality with Clients ........................................50
Figure 4. Wright map for Comfort Discussing Reproductive Health with Clients. ......................52
Figure 5. Wright map for Frequency of Universal RC Practice. ...................................................58
Figure 6. Wright map for Frequency of Targeted Reproductive Coercion Practice. .....................59
Figure 7. Wright map for Reproductive Coercion-Responsive Organizational Supports. ............62
Figure 8. Estimated Probabilities that Each RC-Responsive Organizational Support is Present for
Advocates in Each Latent Class .....................................................................................................87
Figure 9. Standardized Regression Path Coefficients for Person Measures as Predictors of
Universal Reproductive Coercion Practice ....................................................................................96
Figure 10. Standardized Regression Path Coefficients for Person Measures as Predictors of
Targeted Reproductive Coercion Practice .....................................................................................98

viii

INTRODUCTION
The connection between reproductive and sexual health (RSH) and intimate partner
violence (IPV) is well recognized, with links found between IPV and both riskier sexual
behavior and negative RSH outcomes (Bergmann & Stockman, 2015). An important but underresearched form of IPV that may explain this link is reproductive coercion (RC), defined as
“male partners’ attempts to promote pregnancy in their female partners through verbal pressure
and threats to become pregnant, direct interference with contraception, and threats and coercion
related to pregnancy continuation or termination” (Miller & Silverman, 2010, p.511). Recent
studies have found that between 8%-19% of women surveyed (including college students and
patients in OBGYN and family planning clinics) had experienced this type of abuse (Clark,
Allen, Goyal, Raker, & Gottlieb, 2014; Miller, Decker, McCauley, Tancredi, Levenson,
Waldman, Schoenwal, & Silverman, 2010; Sutherland, Fantasia, and Fontenot, 2015), and that
35%-74% of survivors of other types of IPV had also experienced RC (Miller et al., 2010; Thiel
de Bocanegra, Rostovtseva, Khera, & Godhwani, 2010).
While programs that support survivors of IPV offer a myriad of services, anecdotal
evidence suggests that providers are much less likely to offer support or services related to
reproductive and sexual health topics such as RC (V. Duplessis, personal communication,
January 21, 2015; L. Hofheimer, personal communication, January 26, 2015). Given the
prevalence of RC among survivors of IPV and its potential consequences, including unplanned
pregnancy or rapid repeat pregnancy (Miller et al., 2014), this gap in support should be remedied
to promote women’s reproductive health.
Targeted trainings on RC and related interventions are occurring, but at this point there is
little to no research on the extent to which domestic violence (DV) service staff employ such

1

techniques, or are even aware of the presence of this type of abuse. Anecdotal evidence from DV
advocate trainers suggests that few providers had ever heard of reproductive coercion before
attending an RC training, let alone incorporated related practices into their work with survivors
(L. Hofheimer, personal communication, January 26, 2015). DV staff cite a number of barriers to
doing so, ranging from individual-level factors, such as general discomfort discussing sexuality
or reproductive health topics, to organizational-level factors, such as availability of resources and
support from leadership.
Given the importance of addressing RC with survivors accessing DV services, the field is
eager to develop and implement approaches that successfully promote RC-responsive practice.
While a small number of interventions have been developed, in the time- and funding-strapped
world of DV service provision (NNEDV, 2015) administrators and staff alike may struggle to
implement all or even most of the suggested shifts. In light of these conditions, recommendations
are needed to provide guidance to agencies on which, if any, of the policy and practice changes
might provide the most leverage in the pursuit of more frequent RC-responsive practice. It is
with this goal in mind that the current study was conducted. Specifically, the study explores the
effects of critical individual- and organizational-level factors on the frequency of staff
engagement in RC-responsive practices. By exploring how these elements contribute to practice,
the study provides information that may aid DV programs across the world in promoting
effective responses to survivors experiencing RC.

2

LITERATURE REVIEW
Reproductive Coercion
Male perpetrated reproductive coercion (RC) is a type of IPV whereby an abusive male
partner attempts to control his female partner’s reproductive choices (Miller et al., 2010; Moore,
Frohwirth, & Miller, 2010). RC behaviors generally fall into three broad categories: pregnancy
coercion, birth control sabotage, and pregnancy outcome coercion (Chamberlain & Levenson,
2012; Grace & Anderson, 2016; Miller et al., 2010; Moore, Frohwirth, & Miller, 2010).
Pregnancy coercion involves male partners’ attempts to pressure their female partners to become
pregnant via verbal pressure or threats to the relationship or to the survivors safety. Birth control
sabotage involves male partners’ attempts to compromise women’s contraceptive(s) of choice,
such as by disposing of birth control pills or pulling out an IUD, or interfering with another
agreed upon method, such as breaking or deliberating removing a male condom. Finally,
pregnancy outcome coercion involves male partners attempts to force their female partners to
continue an unwanted pregnancy or to terminate a pregnancy against her will. RC may occur
alongside or independently of other types of IPV (Moore, Frohwirth, & Miller, 2010). In studies
with OBGYN patients and college students, researchers found that between 32-57% of women
who reported RC had also experienced IPV (Clark et al., 2014; Sutherland, Fantasia & Fontenot,
2015); however a synthesis of existing research suggests it is unclear which type of abuse usually
occurs first or if they transpire concurrently (Grace & Anderson, 2016).
Abusers’ specific motivations for engaging in RC behaviors have yet to be studied at
length (Grace & Anderson, 2016), however, scholars theorize that RC is a form of control over
women by their male partners (Moore et al., 2010). Indeed, this type of reproductive control
crosses the domains of gendered relations as initially identified by Connell (1987, as cited in

3

Moore et al., 2010): labor, as forced pregnancy and child rearing reinforces women’s domestic
duties; power, as RC seeks to assert authority over women’s sexuality and their biological
propensity to become pregnant; and cathexis, as abusive partners demand the investment of
women’s sexual, emotional, and intimate energy in their romantic and motherhood roles. The
dynamics of reproductive control in these gendered relations may manifest in specific
motivations for male perpetrated RC. For example, such motivations might include relationship
insecurity (i.e. when abusers seek to guarantee continued connection via pregnancy and shared
parenting); masculine gender ideologies around fertility (i.e. the belief that men who have many
children are thought to be more powerful); and beliefs about sexual ownership in relationships
(i.e. the belief that men have ownership of women and their biological ability to conceive; Miller
et al., 2007; Moore et al., 2010). Additionally, a qualitative study found evidence that social and
structural issues may contribute to RC victimization among African American women
(Nikolajski et al, 2015). African American study participants suggested that factors such as
incarceration, lack of social support, and structural barriers to stable housing and employment
may motivate men to secure an ongoing relationship with female partners via pregnancy.
RC is a common form of IPV against women of reproductive age (Clark et al., 2014). It
has been found to impact between 8% - 19% of women in general population, clinical and
college student samples (Black et al., 2011; Clark et al., 2014; Miller et al., 2010; Sutherland et
al., 2015), although disclosure rates may differ based on screening method (Thaller & Messing,
2016). Among women who have also experienced physical, sexual or psychological IPV, these
estimates may be much higher. In a sample of teens with a history of IPV drawn from multiple
community programs, one-quarter (26%) of the 53 participants interviewed reported that their
abusive male partners were actively trying to get them pregnant (Miller et al., 2007). In another

4

sample of women from family planning clinics (n=1278), participants who experienced physical
or sexual violence from an intimate partner were more than twice as likely to report experiencing
reproductive coercion as women who had not (35% v. 15% of respondents; Miller et al., 2010).
These figures suggest that many women seeking services from DV programs may also have
experienced some form of RC. A survey of over 3,000 survivors who called the National
Domestic Violence Hotline (2011) found that 25% of participants had experienced RC.
Interviews with heterosexual women using DV shelter services (n=54) echo this assessment; a
large proportion of participants had experienced birth control sabotage, resistance to condom use,
and forced sex by their partners (Thiel de Bocanegra et al., 2010).
RC victimization has a negative impact on the reproductive and sexual health of
survivors (Miller et al., 2010; Sutherland et al., 2015). Studies have demonstrated relationships
between experiences of RC and increased odds of having had one or more unintended
pregnancies (Miller et al., 2010; Miller et al., 2014; Sutherland et al., 2015), diagnosis of a
sexually transmitted infection (Jones et al., 2016) and history of abortion (Sutherland et al.,
2015). Perhaps unsurprisingly, RC has also been shown to be associated with increased use of
reproductive and sexual health services (Kazmerski et al., 2015). In a large sample of women
ages 16-29 from five family planning clinic (N=1262), experiences of recent RC without other
forms of IPV and RC in combination with other forms of IPV were found to be associated with
increased odds of seeking multiple pregnancy tests, using emergency contraception multiple
times, and seeking STI testing (Kazmerski et al., 2015). These results suggest that frequent
requests for pregnancy and STI testing and emergency contraception may indicate that clients
may have recently experienced RC.

5

While RC victimization has been reported by women from all backgrounds, several
studies have found that marginalized groups such as racial or sexual minorities may be more
likely to have experienced RC than their counterparts in dominant groups. Two studies of young
sexual minority women found that participants who reported having had sex with other women
were more likely to have experienced RC (McCauley et al., 2014; McCauley et al., 2015).
Likewise, several studies have found higher rates of RC victimization among non-Hispanic
African American/Black, multiracial, or Latina women compared to white women (Borrero et
al., 2015; Clark et al., 2014; Holliday et al., 2017; Miller et al., 2010; Miller et al., 2014;
Nikolajski et al., 2015; Sutherland et al., 2015; Upadhyay, Dworkin, Weitz, & Foster, 2014),
however conflicting results have been demonstrated elsewhere using a less robust RC measure
(Phillips, Bennett, Hacker & Gold, 2016). This disproportionately high prevalence of RC among
women of color may at least partially explain similar racial differences in unintended
pregnancies (Holliday et al., 2017). Qualitative studies have found that African American
participants had experienced more extreme, overt forms of RC compared to their white
counterparts, and that more African American women reported that their current or a past
pregnancy resulted directly from birth control sabotage and/or pregnancy pressure by a male
partner (Borrero et al., 2015; Nikolajski et al., 2015). Given these disparities, RC represents an
important social justice and reproductive justice issue (Chrisler, 2014).
While little research has occurred with abusers to date, several studies have revealed
patterns in survivors’ responses to RC. One such finding is that women may fear negotiating
condom use or resisting unwanted sex with their male partners (Raj, Silverman & Amaro, 2004;
Wingood & DiClemente, 2000). Qualitative research suggests that this fear may be related to
male partners’ threats of violence or other forms of retaliation, such as sexual and physical

6

violence, accusations of female STD infection, infidelity or threats to end the relationship (ElBassel, Gilbert, Wu, Go, & Hill, 2005; Miller et al., 2007; Mittal, Senn, & Carey, 2013; Seth,
Raiford, Robinson, Wingood & DiClemente, 2010). In such situations, survivors must weigh the
potential safety loss of increased violence in their relationship against the potential gain in
reproductive control by requesting condom use. Given these dynamics, condom use is not a
viable solution for most women in a violent relationship (Foster, Nunez, Spencer, Wolf, &
Robertson-James, 2015).
Despite this diminished power to negotiate condom use or to resist sex, studies have
revealed a variety of strategies that women have been able to use to resist their male partners’
attempts at RC. Such strategies include using more easily concealable forms of contraception,
such as Depo-Provera injections (Sutherland et al., 2015), hiding contraceptive or emergency
contraceptive use (Miller et al., 2007; Nikolajski et al., 2015; Thiel de Bocanegra et al., 2010),
lying about being pregnant (Miller et al., 2007), having abortions against their partners’ wishes
(Moore et al., 2010), lying to a partner about nonexistent fines for an intrauterine device (IUD)
insertion appointment (Thiel de Bocanegra et al., 2010), checking condom placement during sex
(Teitelman, Tennille, Bohinski, Jemmott, & Jemmott, 2011), promising a partner who pressured
for pregnancy termination that he would not have to pay child support (Thiel de Bocanegra et al.,
2010), and secretly leaving the abortion clinic after a pressuring partner dropped her off (Thiel de
Bocanegra et al., 2010).
Interventions to Address Reproductive Coercion
Given the prevalence of RC against women and its potential impacts on survivors’ health
and well-being, interventions aimed at identifying and addressing RC are imperative. One
intervention that has garnered initial empirical support is the Safety Card for Reproductive

7

Health. This wallet-sized safety card developed by the National Health Resource Center on
Domestic Violence (a project of Futures Without Violence), in partnership with the American
College of Obstetricians and Gynecologists, contains questions and information related to IPV
and RC. By taking 60 seconds to review the card with their patients, providers may be able to
help women see the link between IPV, RC, and their reproductive health (Chamberlain &
Levenson, 2012). The card also provides information on harm reduction strategies survivors can
use, tips for incorporating RC considerations into safety planning, and contact information for
national resources. Importantly, the card is useful regardless of whether patients choose to
disclose their RC experiences to their provider, as they are encouraged to take it with them to
consider in private or share with others.
An initial study of this intervention in family planning clinics found that women
introduced to the card who reported recent IPV at baseline had a 71% reduction in the odds of
pregnancy coercion at follow-up compared to participants who reported recent physical or sexual
IPV at baseline who solely received the standard of care (Miller et al, 2011). However, this
reduction in odds of pregnancy coercion was not found in reports of women who did not report
other forms of recent IPV at baseline. Additionally, women in the intervention clinics were more
likely to report ending a relationship at follow-up because it was unhealthy or unsafe regardless
of IPV status. However, when this approach was scaled to more clinics and women (n=4,009;
Miller et al., 2016), the intervention failed to demonstrated reductions in RC, other types of
partner abuse, or unintended pregnancies. The cards did improve two outcomes that may
contribute to women’s safety: awareness of partner violence resources and self-efficacy to
engage in harm reduction behaviors. Likewise, there was some evidence in this sample that
women exposed to the intervention who were experiencing multiple forms of abuse

8

demonstrated a reduction in RC. The authors of this study suggest that the limited effects of the
intervention might be due to inadequate implementation, and recommend that additional cliniclevel changes might be needed to facilitate delivery. Such elements might include the
development of protocols for intervention delivery, technology to facilitate this delivery, and the
incorporation of IPV counseling as a billable preventative service at clinics (Miller et al., 2016).
These results indicate that the Safety Card for Reproductive Health may be a promising
approach to screening for and educating patients on RC, but that certain organizational supports
may need to be in place to promote its effective implementation. The findings of an evaluation of
provider comfort and facility with this intervention in family planning clinic and home visiting
program settings provides important suggestions therein (Burton & Carlyle, 2015). Through
focus groups and interviews with 47 providers from multiple sites across one state, the study
revealed that providers valued the tools and felt that the intervention was important, but that they
struggled in knowing when and how to implement the necessary universal screening and
intervention protocols. Providers reported discomfort starting the conversation, and were unsure
about their own skills to respond should their client disclose RC. Additionally, clinicians in
particular worried that they were sacrificing limited time in which they could offer other
services, and home visitors worried that their clients were otherwise overwhelmed or not ready
to be honest about their relationships. Participants expressed a desire for additional training and
validation regarding their screening and intervention routines, including yearly “refresher”
learning opportunities and on-site “champions” of the intervention who could provide support
and accountability to ensure implementation was occurring. Tweaks such as these may help staff
feel more comfortable and help with prioritization of services in limited time frames in these

9

established settings; however, the extent to which this intervention may or may not be
appropriate for use in other venues, such as DV service programs, remains to be seen.
RC-Responsive Practice in Domestic Violence Service Settings
DV service and support organizations may be uniquely positioned to intervene in
situations of RC, as this type of abuse often co-occurs with other types of IPV, such as physical,
sexual, or psychological abuse, for which survivors may already be seeking services (Clark et al.,
2014; Miller et al., 2010; Sutherland et al., 2015). Just as DV program staff often seek to be
knowledgeable on the dynamics of physical, sexual, and psychological violence, professionals
working with survivors could benefit from education on reproductive coercion and related
interventions in order to meet the full range of survivors’ needs. While they need not be experts
in sexuality, they need to demonstrate a willingness to comfortably raise RSH-related topics, and
to respond in a client-centered, professional, and knowledgeable manner when their clients
present sexuality-related issues (McKay, 2015).
While DV services differ from program to program, they generally incorporate activities
such as providing information about adult and child survivors’ rights, options and experiences;
safety planning; skill building; and increasing access to community resources and social support
(Sullivan, 2016). Though the models for doing so are only in their initial stages of development,
RC-responsive practices can be woven into each of these activities. Researchers have called for
reproductive health issues, such as RC, to be universally included in the intake and
advocacy/counseling processes for all survivors who seek services from DV programs (Thiel de
Bocanegra et al, 2010). This should include elements such as education about RC and
concealable forms of birth control, information about how to access needed reproductive health
resources, and assessments of women’s barriers to consistent and effective use of contraceptive

10

methods (Duplessis & Levenson, 2014). Trainers in the field suggest that advocates should take a
“don’t ask, just tell” approach when engaging in RC-responsive practices, such that DV staff
share basic RC information with all women regardless of disclosure of RC experiences (L.
Hoffheimer, personal communication, January 26, 2015). The reassurance that advocates “talk
about this with everyone” can help to normalize what may feel like taboo conversations, and may
allow survivors who do not disclose RC to choose to use this information at a later time or share
it with others in need (Duplessis & Levenson, 2014). Researchers and advocates have likewise
called for programs to universally offer over-the-counter products such as emergency
contraception and pregnancy tests on site to support survivors in resisting RC or responding to its
consequences (Thiel de Bocanegra et al., 2010).
In the event that RC is identified, advocates should be prepared to engage in targeted
practice by providing information and support for dealing with potential health issues and
incorporating reproductive concerns into safety planning. This could include discussions of
options for emergency contraception and/or more easily concealable tamper-resistant birth
control (e.g. IUD, Depo-provera injections), training on safer condom negotiation skills, and the
development of strategies for staying safe while accessing reproductive health care such as
pregnancy testing, pregnancy termination or prenatal care. While none of these approaches is yet
linked to improved survivor outcomes in the DV service setting, they provide a compelling
starting point for the field to consider.
Trainings on RC and the implementation of RC-responsive practices are occurring in DV
programs around the country; however, at this point there is little to no research on the extent to
which staff employ such techniques, or are even aware of the presence of this type of abuse.
Anecdotal evidence from trainers suggests that few providers had ever heard of reproductive

11

coercion before attending an RC training, let alone incorporated related practices into their work
with survivors (L. Hofheimer, personal communication, January 26, 2015). Interestingly, even
once providers were introduced to the concept and its significance to survivors’ health and wellbeing, trainers still report experiencing resistance among some staff to discussing RC or even
broaching the topic of sexuality. Advocates cite a number of barriers to doing so, ranging from
time constraints and lack of knowledge or skills, to organizational policies and general
discomfort discussing reproductive or sexual health. The empirical identification of such barriers
and facilitators among DV staff could aid the field in developing more effective approaches to
promoting RC-responsive practice among advocates across the country.
Identifying Barriers and Facilitators to RC-Responsive Practice in DV Settings
A nonscientific survey conducted with advocates in the Pacific Northwest, conducted by
the Washington State Coalition Against Domestic Violence, has provided initial insight into
factors that may present the largest barriers to RC-responsive practice among advocates (L.
Hofheimer, personal communication, April 15, 2015). Among the top endorsed barriers were (1)
a general discomfort discussing sexuality with clients at all, and (2) a need for more training.
While these results may not be representative of all advocates everywhere, they do suggest that
further examination of the barriers based in individual advocates’ attitudes and comfort, as well
as in related organizational supports, may reveal important points of intervention for promoting
RC-responsive practices in DV service settings.
A useful theoretical framework for highlighting barriers and facilitators to RC-responsive
practice is the ecological model. Generally speaking, ecological models identify various “levels”
of contextual influences on human behavior. Several structures have been developed to map out
these levels. McLeroy et al.’s (1988) seminal model is commonly used and identifies the

12

intrapersonal, interpersonal, organizational, community, and policy levels. Furthermore, the
ecological perspective suggests that there are “links” between these levels, meaning that these
levels influence each other (Golden & Earp, 2012; Meadows & Wright, 2008). In the context of
organizations, this approach purports that the behavior of employees is influenced not only by
individuals themselves, but also by their relationships with coworkers and clients, agency
policies and practices, and the broader community and policy landscape, and finally by the
interactions among these levels.
Examining social issues through an ecological lens helps stakeholders partition the
environment into levels that can be used to focus attention on different types of behavioral
influences (e.g. barriers and facilitators; McLeroy, et al., 1988). When an organization (such as a
DV service agency) is looking to encourage a new behavior among staff (such as RC-responsive
practice), consideration of the barriers and facilitators to the adoption of this behavior at multiple
levels can lead to a more nuanced understanding of influencing factors. Several implementation
studies (e.g. Glisson, Schoenwald, Hemmelgarn, Green, Dukes, Armstrong, & Chapman, 2010;
Panzano, Seffrin, Chaney-Jones, Rot, Crane-Ross, Massatti, & Carstens, 2004) and reviews of
implementation frameworks (e.g. Fixsen, Naoom, Blasé, Friedman, & Wallace, 2005; Meyers,
Durlak, & Wandersman, 2012) have likewise supported this assertion. Indeed, such frameworks
often pose multi-level models of the implementation process that can guide organizations
attempting to institute change (e.g. Aarons, Hulrburt, & Horwitz, 2011). While a deep dive into
implementation literature is beyond the scope of this review, it is commonly acknowledged that
successful change to individual behavior in organizations often requires improvements at
multiple levels (Fixsen, Blasé, Metz & Van Dyke, 2015).

13

The most commonly examined level of the ecological model is the intrapersonal or
individual level (McLeroy, et al., 1988). This person-centered level of analysis highlights how
factors in staff’s individual characteristics, such as knowledge, attitudes, behaviors, and skills,
can act as barriers or facilitators to desired practice. Current training and coaching approaches
presently used to promote RC-responsive practice generally touch on a variety of individuallevel factors, which we may conclude are the barriers and facilitators to practice that the field
believes to be most pressing. These curricula generally include education on RC-related
knowledge, examination of how one’s personal beliefs impact one’s advocacy practice, and role
play scenarios where advocates can get comfortable with RC practice in a non-threatening space
(Duplessis & Levenson, 2014). This suggests that the field may currently consider the main
intrapersonal barriers and facilitators to RC practice to be knowledge, beliefs, and comfort
discussing RC.
While addressing these intrapersonal-level interventions may directly impact individual
comfort engaging in RC-responsive practice (by shifting knowledge, personal beliefs, fear of
negative response, etc.), a number of evaluations have failed to demonstrate that intervention on
this level alone actually changes practices on its own (e.g. Bouffard & Little, 2004). These
findings suggest that the identification of and shifts made to intrapersonal barriers and facilitators
may be necessary but insufficient if the aim is to change professionals’ behavior (Seidman &
Tseng, 2011). Without supportive organizational structures and leadership, the desired practices
are unlikely to be successfully initiated and maintained (Fixsen, et al., 2005).
The organizational level of the ecological model (McLeroy et al., 1988) is comprised of
social institutions with organizational characteristics such as formal and informal policies and
practices that operate within a defined domain. Identification of barriers and facilitators on the

14

organizational ecological level can lead to interventions that promote positive individual and
setting-level outcomes, with the settings themselves mediating improved outcomes for the
targeted population (Gregory, Henry, Schoeny, & METRO, 2007; Tseng & Seidman, 2007).
Service organizations (including DV programs) are complicated entities, characterized by
“multiple and often conflicting goals, unclear and uncertain technologies for realizing those
goals, and fluid participation and inconsistent attentiveness of principal actors” (Rosenheck,
2001, p. 1608). In this ambiguous, fluid environment (Fixsen et al., 2005), one method for
identifying barriers and facilitators (i.e. targets for change) on the organizational level is the
“Systems Framework for Understanding Social Settings” (Tseng & Seidman, 2007).
This theoretical framework focuses on three aspects of social settings: social processes,
resources, and organization of resources. Social processes include patterns of transactions
between two or more groups of people, including organizational norms, interactional patterns
and practices, and participation in activities. Resources include the traditional idea of economic
means, but also include human, physical and temporal resources available to an organization.
Finally, the organization of resources refers to how available resources are arranged, such as
how appointments are timed and the physical layout of office space. Tseng & Seidman (2007)
suggest that the quality and quantity of resources and their allocation within the setting lead to
patterns in social processes that ultimately influence outcomes in the organization. As such,
barriers and facilitators are generally related to the presence and/or allocation of local resources
that promote or hinder desired norms, relationships or participation patterns (Tseng & Seidman,
2007).
Current approaches presently used to promote RC-responsive practice also touch on a
variety of organizational-level factors as targets for changing the social process that occurs

15

between an advocate and a survivor seeking services (e.g. Duplessis & Levenson, 2014). Again,
the inclusion of these elements as targets for change suggests that these factors are the barriers
and facilitators to practice that the field believes to be most pressing on this level. In order to
shift the advocate-survivor social process, current interventions include alterations to resources
and their allocation within the organization. Examples of current organizational change strategies
include providing new physical resources (e.g. instituting new intake questions or scripts for
screening, purchasing RC informational materials to share with clients) or human resources (e.g.
hiring staff with experience in reproductive and sexual health fields, formalizing warm referral
processes to needed reproductive health services), or shifting the way that existing resources are
allocated (e.g. prioritizing the purchase of emergency contraception and pregnancy tests,
devoting a certain amount of staff time to maintaining an updated reproductive health resource
list, rearranging space to allow for private conversations). Finally, changes to other social
processes could be leveraged to promote this transformation; for example, shifts could be made
to the interactions of staff by regularly incorporating RC topics into supervision sessions, team
case review, and all staff meeting agendas.
While guidance from national and state DV leaders highlights the factors that these
organizations likely believe to be key barriers and facilitators to RC-responsive practice on the
interpersonal and organizational ecological levels, to date very little research has been conducted
in this area. The DV literature currently cannot offer empirical evidence that these factors are
indeed present in DV service organizations. Likewise, there is no evidence as of yet regarding
the extent to which these factors influence practice among DV advocates. While such research is
lacking in the DV service literature, studies do exist in the fields of social work and nursing that
may guide empirical examinations.

16

Barriers and Facilitators to RSH-Responsive Practice in Other Settings
The social work and nursing literatures provide an empirical evidence base that is
relevant to the identification of intrapersonal and organizational barriers and facilitators to RCresponsive practices in DV service settings. While the social work literature may be more closely
related to the practices of advocates, there are only a limited number of empirical studies of
practicing social workers’ barriers and facilitators to discussing RSH with their clients. The
nursing literature, however, presents a larger relevant body of research related to such RSH
discussions (Bulow, 2012). While nurses’ responsibilities are not equivalent to those of DV
service providers, it is logical to compare these groups for a variety of reasons. First, nurses and
DV service providers have similar relationships to those whom they are serving, as they are both
professionals who work closely with largely adult service seekers to offer specialized support or
assistance for a discrete amount of time. Additionally, the two professions share a values-driven
approach to their work. The nursing field’s commitment to patient-centeredness and holistic care
(Higgins, Barker, & Begley, 2008) could easily be compared to the DV service field’s core tenets
of survivor-driven and empowerment-focused services (Cattaneo & Goodman, 2015; Goodman
& Epstein, 2008) Finally, as a profession, the characteristics of nursing (close and ongoing
relationships with clients, values-driven approach, level of expertise and professionalism, etc.)
are more similar to DV service provision than the characteristics of other more specialized
healthcare work, such as that of physicians. Moreover, the literature on the lesser-trained
paraprofessional healthcare providers’ experiences of discussing sexuality with service seekers is
much less robust than that in the nursing field. As such, the DV field can draw from the
contributions and the lessons learned in other fields (in this case, nursing) in pursuit of promoting
common causes therein, namely the RSH and well-being of service-seekers.

17

Although the fields share a number of commonalities, nursing and the DV service fields
do have differences that are worth considering. Perhaps most notably, nurses generally work in a
health context that is driven by monetary gain (e.g., billable insurance hours). DV service
organizations are generally not as monetarily-driven, which could result in differential impacts
on staff’s likelihood to engage in RSH-responsive practices when compared with nurses. That
being said, both settings often experience a shortage of time to quickly serve what is often a very
large caseload, as well as pressures from external funding sources (e.g. insurance companies or
grant funders) to prioritize particular services (Nakopoulou, Papaharitou, & Hatzichristou, 2009;
National Network to End Domestic Violence, 2015; Teng, Hsiao, & Chou, 2010). Though the
sources of these time constraints and service prioritization considerations are different in largely
for-profit versus non-profit settings, the resulting impacts on providers and on service seekers’
experiences are likely comparable. These shared factors include: 1) comfort discussing sex and
sexuality, 2) comfort discussing reproduction-specific topics, and 3) RC-responsive
organizational supports.
Individuals’ comfort discussing sex and sexuality. One’s comfort level discussing sex
and sexuality is likely an important barrier to or facilitator of advocates engaging in RCresponsive care. A nonscientific survey conducted with a small number of DV advocates
revealed that a general discomfort discussing sexuality with clients was among the top barriers to
discussing reproductive coercion (L. Hofheimer, personal communication, April 15, 2015).
Multiple studies with both social workers and nurses also found that a lack of comfort or
confidence discussing sex was a top barrier to RSH-responsive practice (Bal & Sahiner, 2015;
Bulow, 2012; Dattilo & Brewer, 2005; King, Miree, Wilson, & Clayton, 2008; Kotronoulas et
al., 2009; Lavie-Ajayi, 2016; Mahieu, Van Elssen & Gastmans, 2011; Reynolds & Magnan,

18

2005; Strawgate-Kanefsky, 2000; Trotter, Brogatzki, Duggan, Foster & Levie, 2006; Vassiliadou
et al., 2008; Yildiz & Dereli, 2012). In response to these findings, scholars in these fields have
called repeatedly for more training focused on improving providers’ confidence and comfort
discussing sex in hopes of encouraging more frequent RSH practice (e.g. Higgins et al., 2008;
Magnan, Reynolds, & Galvin, 2006). In order for interventions to similarly aid advocates in
overcoming discomfort engaging in RC-responsive practices, they must target the factors that are
at the root of this feeling. Research with social workers and nurses suggests that the elements
contributing to discomfort likely include a lack of knowledge, embarrassment, personal beliefs
about sexuality, and fear of a negative client response.
The most often-cited barrier to practitioners feeling comfortable discussing sexuality
among both social workers and nurses was a lack of knowledge. This includes both knowledge
about the particular RSH issues that their service seekers may be experiencing, as well as the
basic understanding that their clients or patients might have sexuality-related concerns as part of
their condition (Algier & Kav, 2008; Ayaz, 2013; Bulow, 2012; Dyer & das Nair, 2013; Dyer,
Aubeeluck, Yates, & Nair, 2015; Kotronoulas et al., 2009; Mahieu et al., 2011; Nakopoulou et
al., 2009; Quinn, 2003: Strawgate-Kanefsky, 2000; Reebals, Brown, & Buckner, 2006: Stead,
Brown, Fallowfield, & Selby, 2003). Another area where nurses may lack knowledge is
regarding sufficient information about proper nursing interventions around sexuality concerns. In
three studies with nurses across Europe, practitioners expressed concerns that they might need
specialist knowledge in order to engage in conversations about sexuality, and were hesitant to
broach the subject given that they might not be well qualified to do so. Specifically, they were
worried that they might not have the most up-to-date knowledge about sexuality-related

19

developments in medicine or the correct sexuality-related language to use with patients (Gott,
Galena, Hinchliff, & Elford, 2004; Jaarsma et al., 2010; Nakopoulou et al., 2009).
Greater knowledge about how sexuality-related topics relate specifically to their clients’
care may result in advocates more frequently initiating RC-related discussions. One study
involved qualitative interviews with Swedish nurses several months after they attended a nursing
practice-focused human sexuality training (Saunamäki, & Engström, 2014). These interviews
revealed that nurses who had followed through in talking with their patients about sexuality after
receiving RSH training had done so partially because of their newly acquired knowledge. Nurses
reported feeling responsible for speaking with patients about sexuality because they had new
information about its significance to their patients’ well-being. The authors asserted that this
knowledge seemed to inspire nurses to set aside personal beliefs in favor of helping patients
address sexual concerns that may previously have been ignored. This claim has been supported
in the social work literature, where studies have found that RSH knowledge was significantly
related to frequency of RSH-related practice (Bulow, 2012; Strawgate-Kanefsy, 2000).
While many providers may cite a lack of knowledge about sexuality and related
interventions as primary obstacles to RSH care, such knowledge may not actually impact
practice. A survey of Swedish nurses found that over 90% of participants reported that they
understood how patients’ diseases and treatment might affect their sexuality; however, 60% of
participants did not feel confident in their ability to address patients’ concerns, and 80% of
respondents did not actually take the time to discuss sexual concerns when working with patients
(Saunamäki, Andersson, & Engström, 2010). This study highlights the lack of congruity between
service providers’ self reports of RC knowledge, and their comfort and confidence in carrying
out interventions or their likelihood of actually addressing sexual concerns. These results suggest

20

that knowledge of RC may be an important but not independently sufficient predictor of RCresponsive practice.
Another source of discomfort may be the belief that discussing sexuality-related topics is
embarrassing or taboo. Two literature reviews of sexuality and the nursing process (Ayaz, 2013;
Mahieu et al., 2011) highlighted feelings of embarrassment as one of the major barriers to RSHresponsive practice, as did two studies of social workers (Lavie-Ajayi, 2016; Trotter et al., 2006).
This embarrassment may be rooted in some providers’ belief that sexuality is a taboo subject
(Dyer & das Nair, 2013; Nakopoulou et al., 2009; Quinn, Happell, & Browne, 2011; Saunamäki
& Engström, 2014; Stead et al., 2003) or is too private or personal to be an appropriate topic for
conversation (Trotter et al., 2006) and as such should not be discussed. In general, talking about
sexuality is not seen as appropriate for public discussion in western societies (Ussher et al.,
2013). This private/public dichotomy may deter service providers from broaching what is seen as
a socially unacceptable topic—sexuality—in professional conversation (Lavie-Ajayi, 2016).
Beliefs and values around sexuality are another individual-level factor that impact an
individual’s comfort engaging in RC-responsive practices. Beliefs about patients’ sexuality are
often dominated by personal rather than professional values about the nature of individuals’
sexuality and their sexual rights (Ruane & Hayter, 2008). Individuals who report more sexpositive personal attitudes (meaning they were generally more erotophilic) may report greater
confidence, comfort, skills and knowledge in addressing sexuality-related issues. For example,
both social workers and nurses with more negative attitudes toward human sexuality were less
inclined to practice RSH-responsive care (Algier & Kav, 2008; Bulow, 2012; King et al., 2008;
Reynolds & Magnan, 2005; Sohocki, 2010; Strawgate-Kanefsky, 2000). Religious beliefs that
promoted more conservative views about sexuality were also considered a barrier to discussing

21

sexuality (Bal & Sahiner, 2015; Kotronoulas et al., 2009; Lavin & Hyde, 2006). Self-reports
from another group of nurses indicated that 40% of those surveyed reported that their personal
values sometimes affected their conversation on sexual problems with patients (Akinci, 2011).
Encouragingly, at least one study has demonstrated that the barrier of individual beliefs
about sexuality can be overcome. Interviews with psychiatric nurses in Quebec following a
discussion-based intervention on sexuality-related care found that nurses have the capacity to
reconcile their personal beliefs on sexuality with RSH-responsive care via their commitment to
quality patient care (Wright & Pugnaire-Gros, 2010). By focusing on a core principle of nursing,
the desire to provide holistic patient care, participants in this participatory action research project
and intervention were empowered to provide RSH care within their existing belief structures,
rather than being forced to provide RSH care despite their beliefs. If advocates were likewise
empowered, they could also potentially find the middle ground wherein they would be able to
stay true to their values while also providing effective RC care.
Finally, advocates may also feel uncomfortable discussing sexuality because they believe
that survivors could consider such conversations to be unexpected, intrusive and/or a violation of
their privacy. Surveys of nurses and nursing students in the U.S. found that a majority of
respondents did not believe that patients expect nurses to ask about sexual concerns (Reynolds &
Magnan, 2005; Magnan & Norris, 2008; Magnan et al., 2005). Interviews with nurses in the U.K.
revealed that providers had fears that asking about a topic as private as sexuality may offend
their patients, and that it could jeopardize the provider-patient relationship (Algier & Kav, 2008;
Gott et al., 2004; Jaarsma et al., 2010). Indeed, if providers believe that their relationships with
clients are based on trust, it is quite possible that they may avoid discussion of topics that could
breed mistrust by eliciting discomfort or anxiety (Magnan et al., 2005). Given the widely shared

22

taboo in western cultures around sexuality (Iantaffi, 2015) it is very possible that advocates may
similarly fear survivors’ response to questions regarding RC. That being said, general advocacy
practice commonly involves asking about very painful, private matters, so asking about sex may
not elicit a qualitatively different response.
Individuals’ comfort discussing reproductive health topics. In addition to one’s
comfort level talking about sex and sexuality, ease of discussing reproductive health can also
impact one’s willingness to engage in RC-responsive practice. Reproductive health care topics
such as contraception (including birth control methods, emergency contraception and forced
contraception), pregnancy termination, and adoption are frequently relevant in social work
practice settings (Alzate, 2009; Constantine, Jerman, & Constantine, 2009; Whitaker &
Arrington, 2008) and are particularly important when intervening in cases of RC. Indeed, social
workers may serve as one of the only points of contact for clients regarding reproductive health
questions (Alzate, 2009; Ely, Flaherty, Akers, & Noland, 2012; Whitaker & Arrington, 2008).
Despite the significance of addressing these topics, advocates may be uncomfortable discussing
reproductive choices with clients. Barriers such as a lack of knowledge about abortion or
emergency contraceptives, personal beliefs about restricting reproductive health options, and fear
of social stigma may contribute to this reluctance.
A lack of knowledge about the science, behavioral impacts, and methods for accessing
abortion and emergency contraception (EC) may negatively impact an individual’s comfort
engaging in RC-responsive practices. Studies with both social workers and nurses have found
that a lack of knowledge in this area may be a cause of a lack of discussion related to abortion or
EC with clients (Akers, Gold, Borrero, Santucci & Schwarz, 2010; Bell & Rubin, 2013; Miller et
al., 2011; Reed, Vaughn, & Pomerantz, 2012). Incorrect information about the science behind

23

these reproductive health tools may present one barrier to their use. A survey of licensed social
workers in Pennsylvania (N=197) found major gaps in participants’ understanding of how EC
works (Bell & Rubin, 2013). While items related to birth control and condom use were answered
correctly by more than 85% of the respondents, items related to EC had the fewest right answers,
with only 21-28% of participants answering correctly. These results demonstrated a gap in social
workers’ understanding that EC is not the same as the abortion pill, and that EC does not
interfere with a pregnancy that has already occurred. This gap in knowledge was also found
among social work students (N=116; Flaherty et al., 2012). Additionally, a qualitative study with
healthcare providers found that participants felt it was difficult to keep up with what was
perceived to be the rapidly changing pharmacology and technology used in EC (Akers, Gold,
Borrero, Santucci, & Schwarz, 2010).
Misconceptions about the social and behavioral impacts of access to abortion and EC
may also stand in the way of discussions on these topics. For example, two different studies
found that healthcare providers held several false beliefs about emergency contraception that
could deter them from recommending its use, such as assuming that access to EC would increase
risky sexual behaviors (Miller et al., 2011; Reed, Vaughn, & Pomerantz, 2012). Furthermore,
some healthcare providers erroneously believe that expanding access to safe and affordable
abortion necessarily leads to more abortions, a falsehood that has been disproven repeatedly (e.g.
Jones & Jerman, 2014). Providers who wish to promote individual responsibility and overall
RSH could be hesitant to bring up EC or abortion if they wrongly believed these options might
contribute to undesirable consequences.
Finally, a lack of knowledge about the rules and regulations around accessing EC or
abortion may contribute to advocates’ reluctance to incorporate RC-responsive care into their

24

practice. For example, more than three-quarters of social workers in one survey did not know
that a prescription is not needed to access EC (Bell & Rubin, 2013), and 41% of social work
students in another study did not even know if abortion was legal in their state (Ely, Flaherty,
Akers & Noland, 2012). Such gaps in understanding may contribute to providers’ discomfort
engaging in RC-responsive practices due to the fear of appearing incompetent or not being able
to respond adequately to clients’ needs. Interestingly, perception of knowledge may be more
powerful than actual knowledge in predicting likelihood of discussing reproductive health topics
with clients. Social workers who rated themselves as more knowledgeable were more likely to
discuss family planning with clients, regardless of their actual performance on an abortion
knowledge test (Bell & Rubin, 2013).
Advocates may hold personal beliefs that make them more or less likely to discuss
reproductive health options with survivors. Despite the common use of contraceptives, EC, and
abortion, the topic remains highly controversial in the U.S. (Guttmacher, 2016). Studies have
found a wide range of opinions about who should have access to birth control, EC, and abortion
and under what circumstances they should be allowed access in samples of social work students
(Begun et al., 2016; Ely, et al., 2012; Falherty et al., 2012) and healthcare providers (Akers et al.,
2010; Miller et al., 2011), as well as among the general public (Hess & Rueb, 2005). For
example, in one survey of social work students, almost one half of the participants said they
would not make a referral for an abortion. In another study, healthcare providers described being
hesitant to suggest EC to patients because of a belief that they should be punished for inadequate
contraception use, particularly in adolescents (Miller et al., 2011). Beliefs that champion the
restriction of reproductive health options may be influenced by individual providers’ upbringing,
religion, political affiliation, social context, or other factors (Alzate, 2009; Begun et al., 2016;

25

Ely, et al., 2012). The practice of “conscientious objection” to providing or referring clients to
reproductive healthcare based on personal beliefs has become a widespread barrier to care for
many women. Navigating the space between these closely held personal perspectives and one’s
professional responsibilities to the health, well being, and self-determination of clients can be
challenging for providers (Begun et al., 2016).
The sometimes-hostile atmosphere around reproductive health care in the U.S. may also
contribute to the stigmatization of advocates who do discuss reproductive health-related options
with their clients. Healthcare providers who perform abortions in the U.S. are often stigmatized
in the cultural discourse, law, politics, churches, and communities (Harris, Debbink, Martin &
Hassinger, 2011). A qualitative study with abortion providers found that these healthcare
professionals experience the manifestations of stigma in a variety of ways, including experiences
of violence or threats of violence, marginalization in the healthcare field, missed funding
opportunities, issues cultivating and maintaining friendships, and exclusion from participation in
local communities and institutions (Harris et al., 2011). This can result in a wariness in
interpersonal relations (Joffe, 2010), social isolation, a loss of self esteem, and may even result in
an exit from the field (Harris et al., 2011).
The fear of experiencing this stigma and resulting fallout may present a barrier to RCresponsive practice among advocates. Providers may choose to opt out of aiding access to birth
control, EC, or pregnancy termination out of fear of discrimination or negative legal or social
consequences (Faúndesa, Duarte, & Osis, 2013). A few participants in a qualitative study of
healthcare providers expressed suspicion that the main reason that providers were unwilling to
perform abortions was actually the fear of social stigma, rather than the often-given excuse of
“religious objection” (Sedgh et al., 2012). Given the severity of the consequences described

26

above, it is not hard to imagine that advocates may also choose to avoid engaging in practices
related to reproductive health rather than risk exposure. If an advocate was believed to be
promoting a “pro-choice” agenda by providing EC on site or offering referrals for pregnancy
termination, that advocate could potentially experience backlash in their personal and
professional life along the lines of that described by abortion providers. While the risk therein
likely varies greatly based on one’s organization and community, the potential could be enough
to make advocates think twice about engaging in RC-responsive care.
Organizational factors that promote or hinder RSH responsive practice. While
individual comfort discussing sexuality and reproductive health are likely significant
determinants of advocates’ likelihood to engage in RC-responsive care, there are likely settinglevel factors that also impact such outcomes. Formal and informal organizational practices,
policies and resources (aka “organizational supports”) may either hinder or facilitate effective
RC-related conversations between providers and clients. Anecdotal reports suggest that a lack of
clarity around RC-relevant agency policies at DV organizations, such as whether staff may offer
emergency contraception or the degree to which they may discuss abortion, may similarly stand
in the way of broaching RC topics with survivors (V. Duplessis, personal communication,
January 21, 2015). Conversely, incorporating clear, concrete RC-responsive elements into
advocates’ practice materials, such as requiring universal education about RC during intake
processes, may be a powerful facilitator (L. Hofheimer, personal communication, January 26,
2015). While very little research exists to date on organizational-level barriers and facilitators to
advocates engaging in RC-responsive practice, the factors identified by the social work and
nursing fields outlined below gesture toward potential contributors therein. These factors

27

represent potential points of intervention for DV organizations looking to shift their supports to
promote better RC care.
First, providers may be more likely to engage in RC-responsive practices if they have
access to physical spaces that feel appropriate for the task at hand. For example, one such critical
physical element is a private place to talk to patients. Providers reported feeling hesitant to bring
up sexuality if others would be able to hear the content of their conversations (Dyer & das Nair,
2013; King et al., 2008; Nakopoulou et al., 2009; Saunamäki & Engström, 2014; White, 2002).
They wanted privacy to discuss such sensitive topics to prevent the patient from becoming
embarrassed, however such space was often hard to come by among the shared rooms and wards
of clinical spaces where patients are often quite physically close to one another (Kotronoulas et
al., 2009). Similarly, nurses identified ambient noise in the environment, such as telephones
ringing across the ward, as another barrier (Kotronoulas et al., 2009). Having a quiet space can
help nurses stay focused and convey respect when having conversations on sensitive issues such
as sexuality. Similarly, working in DV service settings with a lack of private space may make
sensitive discussions, such as those related to sexuality, occur less often due to concerns that
confidential information may be overheard.
Another resource that may facilitate discussions of sexuality is educational materials for
teaching specific topics (Kotronoulas et al., 2009; Nagel & Neal, 2008; White, 2002). Materials
such as leaflets and posters can provide an informative jumping off point for RSH-responsive
discussions. However, one study found that only 30% of nurses interviewed were aware of
written materials on the subject available in their field that they could share with patients (Stead
et al., 2003). This suggests that having these resources only has the potential to be helpful to
clients if providers are also made aware of their presence. Likewise, a lack of posters or leaflets

28

that are relevant to the particular population being served and their potential RSH-related issues
can also present a barrier to such conversations (Gott et al., 2004). Such materials developed for
use among survivors have been reported to make RC conversations easier for DV service
providers (V. Duplessis, personal communication, January 21, 2015).
Having the appropriate resources to care for victims of RC and the power to provide them
to clients may contribute to a sense among providers that they can respond adequately should a
client disclose RC victimization. This could include, for example, resource lists with information
on specialists who can provide more in-depth treatment of reproductive health concerns, or
permission to access and distribute pregnancy tests and emergency contraception on site. In a
study with nurses, providers said their lack of power to provide medications or refer patients to
sexuality experts presented a barrier to RSH conversations (Gott et al., 2004). Likewise, a lack
of referral procedures or existing organizational connections with RSH specialists may also
interfere with providers’ likelihood to engage patients in sexuality-related conversations (Tsai,
2004). If advocates fear that they won’t be able to respond adequately to the complex
reproductive health needs of a patient who discloses RC by connecting them with needed
resources, providers may avoid asking about RSH concerns.
The presence of screening questions or scripts related to RC in regularly used documents
may encourage RC-responsive practice. The integration of relevant questions into intake and
discharge forms and RC-related services into resource sheets can build in opportunities for such
conversations to simply become part of standard procedures. Healthcare providers cited the lack
of reminders on exam forms or prompts within their electronic medical record systems as one
reason for neglecting to discuss EC with patients (Akers et al., 2010). Nurses in another study
echoed this sentiment, expressing that the inclusion of sexuality as a search word in electronic

29

medical records, as part of discharge talks, and as a topic in guidelines for treating different
diseases indicated to nurses that issues of sexuality are to be addressed with service seekers as
part of regular practice (Saunamäki & Engström, 2014).
An institution’s capacity to provide training opportunities around RC-responsive
practices may impact advocates’ likelihood of engaging in sexuality-related conversations.
Whether such trainings are provided to new staff at orientation or as part of a one-time or
ongoing professional development opportunity for staff, these trainings may be essential for
individuals to gain the necessary knowledge, confidence and comfort needed to engage in
effective RC practice. A number of studies have demonstrated that nurses identified their lack of
training or education as one of the most critical barriers to engaging in sexuality-related topics
with patients (Algier & Kav, 2008; Dyer & das Nair, 2013; Gott et al., 2004; Hautamäki,
Miettinen, Kellokumpu-Lehtinen, Aalto, & Lehto, 2007; Lavin & Hyde, 2006). However,
institutions may find it difficult to provide training to staff due to resource constraints and a lack
of interest from administration and/or staff (Gott et al., 2004). Studies on the impact of training
revealed that such activities may improve nurses’ relevant knowledge, attitudes, skills, and
comfort, and ultimately improve their likelihood to engage in RSH-relevant communication
(Akinci, 2011; Kim, Kang & Kim, 2011; Kotronoulas et al., 2009; Krebs, 2007; Magnan et al.,
2005; Quinn & Happell, 2012; Sack, Drabant, & Perrin, 2002; Saunamäki & Engström, 2014;
Sohocki, 2010; Sung & Lin, 2013; Weerakoon, Sitharthan, & Skowronski, 2008). As such, a lack
of organizational capacity to offer such training opportunities may present a significant barrier to
RC-responsive practice. Social workers who had received RSH training were likewise found to
have significantly greater RSH knowledge and more positive attitudes toward sexuality than
those without training (Strawgate-Kanefsky, 2000). While such trainings may benefit DV

30

providers, their organizations may similarly struggle to regularly provide these opportunities due
to decreased funding and staffing cuts (NNEDV, 2014).
Support in the form of formal coaching or mentoring by a more experienced peer or
supervisor can help professionals put knowledge into practice (Seidman & Tseng, 2011). In a
study with nurses, some providers reported that they avoided talking about sexual issues with
their patients because they had no role model to follow in developing these skills (Stead et al.,
2003). In another study, some nurses stated that their supervisors acted as barriers to discussing
sexuality with their patients, as these managers generally did not like them taking the initiative to
engage in such practices that might upset existing procedures (Nakopoulou et al., 2009). These
findings echo other scholars who suggest that by providing empathy, emotional support, and
practical know-how on implementing skills, supervisors or other leaders conversely may help
advocates to implement these practices more frequently (Seidman & Tseng, 2011).
Formal or informal rules about the prioritization of RC-related concerns relative to other
service needs may impact the frequency of RC-responsive practice. In some cases, providers or
their clients may view other practices as more immediately beneficial to their health and wellbeing, and choose to hold off on RC-related conversations. Alternatively, providers may be
required to prioritize particular facets of clinical practice or screening over RSH by virtue of
organizational regulations (Gott et al., 2004), which may be guided by administrative policies
linked to financial reimbursements from public or private insurance. This may result in limited
time and resources for engaging in non-prioritized areas, such as RC.
The lack of time is often cited as a key barrier to RC-responsive care among already busy
providers. Time constraints were commonly endorsed as one of the most significant barriers to
discussing sexuality in studies among nurses (Haboubi & Lincoln, 2003; King et al., 2008;

31

Reynolds & Magnan, 2005). Similarly, when asked to reflect on their reasons for not addressing
sexuality, most nurses interviewed in a qualitative study in Sweden felt that there was no time to
talk about this issue in their current stressful working environments (Saunamäki & Engström,
2014). They believed that sexual concerns should not be addressed in a hurried manner, and as
such, having ample time was essential to engage clients on this topic. Nurses in a mixed methods
study and a qualitative study also noted the lack of time for patient counseling as a contributor to
diminished conversations about EC (Akers et al., 2010; Miller et al., 2011). In other studies,
nurses described sexual health as a “can of worms,” meaning the topic was too complex to be
addressed adequately in the limited time they had with patients (Gott et al., 2004; Stead et al.,
2003). The knowledge that they may not be able to sufficiently address the time-consuming
issues that could be brought up by patients led nurses to avoid even beginning such
conversations. This time pressure was heightened when severe physical health problems were
present, as these concerns were seen to take precedence over sexual health when time with
patients was limited (Nakopoulou et al., 2009). Finally, providers also felt that they needed to
build a rapport with patients first before discussing sensitive issues, but that their tight clinical
timelines often did not allow for this (Gott et al., 2004). While the lack of time is a result of
differing pressures in the non-profit sector compared to the monetarily driven healthcare sector,
the impact of time constraints on DV staff’s availability is likely comparable, suggesting this
may be a factor impacting frequency of RC-responsive practice.
Finally, formalizing an organization’s overall policies and practices regarding RC
discussions may facilitate RC-responsive practice. The inclusion of formal policies supporting
RC-related practices in advocate guidelines may signal to providers that such issues are to be
discussed with clients as part of their job responsibilities. Nurses in one study indicated that

32

having written workplace guidelines for discussing sexuality was helpful, as these policies
lessened their hesitation to bring up the topic (Saunamäki & Engström, 2014). More specifically,
another study identified a lack of clarity around which healthcare team members were
responsible for initiating sexuality-related conversations as a significant barrier to RSH practice
among nurses (Nagel & Neal, 2008). These findings suggest that explicitly stating the
organization’s expectations about screening, education, and treatment of RC may help clarify
roles and improve the frequency of RC-responsive practice. Just as doctors at Catholic hospitals
are prohibited from providing contraception and abortion by the policies spelled out in the
Ethical and Religious Directives for Catholic Health Care Services (United States Conference of
Catholic Bishops, 2009), so too should DV organizations that wish to promote comprehensive
RSH services establish explicit guidelines for their advocates to follow to carry forward this
mission.
Current Study
The current study aimed to provide insight into the relative impacts of individual comfort
and organizational supports on the frequency of RC-responsive practice among DV advocates.
This research sought to help DV organizations make empirically informed decisions about which
intervention(s) to undertake in order to address the most impactful barriers to RC-responsive
practice. The study examined data from a multi-state online survey of domestic violence direct
service providers. Participants included paid staff and volunteers from DV victim service
focused agencies who carried out direct service tasks such as crisis line coverage, case
management, therapy, intake processing, children’s advocacy, or medical, legal, housing or
general adult advocacy. These participants, hereafter referred to generally as “advocates,” were
surveyed about their universal and targeted RC practices. They were asked about their comfort

33

level discussing sexuality and reproductive health, as well as the presence of RC-responsive
organizational supports. The study first describes the advocates’ practice and perceptions, and
then tests the following hypotheses:
1. Higher levels of comfort discussing sexuality will be associated with more frequent
universal and targeted RC practice.
2. Higher levels of comfort discussing reproductive health will be associated with more
frequent universal and targeted RC practice.
3. The number of RC-responsive organizational supports will be positively associated with
more frequent universal and targeted RC practice.
3a. Exploratory Research Question: Are certain combinations of RC-responsive
organizational supports more influential than other combinations?
4. The combination of comfort discussing sexuality, comfort discussing reproductive health,
and organizational supports will significantly predict both types of RC practice.

Comfort
Discussing
Sexuality

Frequency
of RC
Practice

Comfort
Discussing
Repro
Health

RCResponsive
Org.
Supports

Figure 1. Simplified Study Model for Hypothesis Four.

34

METHOD
Survey Construction
In order to develop an online survey related to RC practices, we first consulted with three
organizations around the country with expertise in this area: a national training and technical
assistance leader, a state domestic violence coalition, and a local DV service program. These
organizations have led the field in helping local programs incorporate reproductive and sexual
health-responsive practices into domestic violence organizations by developing innovative
approaches therein and/or providing related training and technical assistance. Representatives of
these organizations provided ongoing insight into advocates’ attitudes and practices in this arena
based on their experiences in the field, informed the researchers where additional information
was needed to inform future intervention efforts, and gave feedback on drafts of the survey.
Combining their input with the relevant literature from public health and nursing, we developed
a brief survey, which was then piloted by the staff of two domestic violence programs for clarity
and functionality. The survey was estimated to take 15-20 minutes to complete, and included the
option for participants to skip questions or end the survey early if they should wish to do so.
Survey Recruitment and Administration
Human subjects approval was obtained through Michigan State University’s Institutional
Review Board before the start of data collection. The finalized survey was distributed nationally
in August 2015 to all state domestic violence coalitions. At the request of the National Resource
Center on Domestic Violence and the investigators via a national coalition listserv email, state
coalitions were encouraged to use various modes of online contact to invite advocates in their
states to participate. State coalition staff who opted to help recruit participants then sent out
information about the survey via email or online bulletins to local domestic violence focused

35

programs who were members of their coalition. These entities generally included domestic
violence service programs, dual domestic violence and sexual assault focused programs, and
general crime victim services organizations. The sample recruitment emails provided to state
coalitions included language inviting direct service staff and volunteers of domestic violence
programs to participate in a brief anonymous online survey about their work with survivors. The
recruitment email intentionally left out the fact that the survey pertained to reproductive
coercion, in order to minimize self-selection bias based on either strong interest or disinterest in
the topic.
A reminder email from the National Resource Center on Domestic Violence and the
investigators was circulated via the same national coalition listserv two weeks after the launch of
the survey to encourage coalitions to distribute the information. Finally, in order to boost
participation the investigators extended the time frame for responses by 10 days and connected
via email with contacts at coalitions in states where few or no surveys had been completed three
weeks after the survey launch. These targeted invitations served to remind coalition members of
the opportunity to assist with recruitment, and to answer any questions they may have regarding
the survey. The survey was open for a total of five weeks, and was closed in September, 2015.
Program staff were eligible to begin the online survey if they lived in the United States or
a US territory, were at least 18 years old, were proficient in reading English, and were currently
employed by or volunteering for a program that primarily provides direct services to survivors of
IPV in a direct service role. All participants were consented on the initial screen of the survey,
and warned about the potentially sensitive nature of the questions. Participation in the brief
online survey was anonymous unless participants chose to share their email address for a chance
to win a $25 gift card in a drawing, in which case participants’ identities were kept confidential.

36

The survey was started 2377 times. 1879 of these potentially duplicated individuals consented to
participate and were randomized into one of four surveys related to reproductive and sexual
health practices in domestic violence service settings. A total of 471 participants were
randomized to complete the survey related to reproductive coercion- organizational supports. In
the initial demographics portion of the survey, participants were asked questions regarding their
work. As the target population for the survey was individuals who work in direct service with
survivors more than 50% of the time (regardless of number of hours worked or volunteered per
week), and have at least 50% of their clients fit the description of “women abused by men,” 97
individuals who did not fit this criteria were screened out following this section and were not
included in the analysis. Finally, three respondents were ultimately excluded from the analysis
because they failed to complete at least one of the substantive scales of interest, resulting in a
final sample of 371 advocates.
Measures
To date, no validated scales exist that capture information about provider practice related
to addressing reproductive coercion. In order to answer the study’s research questions, five
measures were created or adapted in collaboration with the consultant organizations,
incorporating relevant literature and studies of service delivery provider capacity. To ensure
participant comprehension of the concept of reproductive coercion, a note about its definition
was included at the top of each relevant survey page:
“Note: For this survey, we are defining Reproductive Coercion as behaviors an
abusive partner uses to maintain power and control around reproductive issues
(such as "attempting to impregnate a partner against her wishes (pregnancy
pressure), controlling outcomes of a pregnancy (pregnancy coercion), coercing a

37

partner to have unprotected sex (sexual coercion), or interfering with her attempts
to use birth control (birth control sabotage)").”
First, an existing scale was modified to capture participants’ comfort discussing sex and
sexuality. Comfort Discussing Sexuality with Clients is an 11-item scale using two subscales
from the Sexual Health Care Scale-Attitude (Kim, Kang & Kim, 2011): (1) discomfort in
providing sexual health care (7 items), and (2) feeling uncertain about patient’s acceptance (4
items). These items were modified to represent survivors of IPV rather than patients and were
answered using a three-point scale (1 = agree, 2 = uncertain and 3 = disagree). A mean scale
score of reverse coded responses was calculated for each participant that represented their
average level of comfort discussing sexuality with clients (M = 2.48, SD = 0.48, α = 0.90).
The 6-item measure, Comfort Discussing Reproductive Health with Clients, assessed
participants’ comfort level talking with clients about topics related to reproductive health (e.g.,
birth control; pregnancy termination; adoption; emergency contraception). The scale used a 4point response format, ranging from 1 = very uncomfortable, to 4 = very comfortable. A mean
scale score was calculated for each participant that represented their average level of comfort
discussing reproductive health topics with clients (M = 1.69, SD = 0.66, α = 0.87).
The Universal RC Practice scale included 10 items that asked the frequency with which
advocates engaged in particular skills relevant to reproductive coercion with all clients (e.g.,
providing information to all clients about emergency contraception; screening for RC).
Advocates were asked how frequently they engaged in each practice in the last year using a sixpoint response format (1 = never to 6 = always). A mean scale score was calculated for each
participant that represents their average frequency of engagement in RC-related practices with
their general client population (M = 2.16, SD = 0.92, α = 0.89).

38

The Targeted RC Practice scale included 6 items that asked the frequency with which
advocates engage in particular RC-related skills with survivors that they knew or believed to
have experienced reproductive coercion (e.g., helping acquire a pregnancy test; incorporating RC
considerations into safety planning). Advocates were asked how frequently they engaged in each
practice in the last year using a six-point response format (1 = never to 6 = always). Participants
were only asked these questions if they identified that they had worked with one or more clients
in the past year that they believed or knew to have experienced reproductive coercion (n=180). A
mean scale score was calculated for each participant that represents their average frequency of
engagement in RC-related practices with survivors of RC (M = 2.41, SD = 1.10, α = 0.88).
The RC-Responsive Organizational Supports index included 15 items that asked about
the presence or absence of a list of supports that are thought to promote RSH and RC-responsive
care (e.g. having a RSH resource list, displaying RSH materials, including intake questions about
unwanted pregnancy or birth control, offering EC or pregnancy tests on site; see Table 5 for
complete list of items). Advocates answered using a five-point response format (1=Yes, 2=I
think so, 3= I don’t think so, 4= No, 5= Have no idea). Responses were recoded into a
dichotomous Yes or No format (1&2 = Yes (1); 3&4= No (0); 5=Missing). An index score was
calculated for each participant that is equal to the sum of the “yes” answers, representing the
total number of supports in place at the advocate’s organization (M = 4.31, SD = 2.94).
Advocates were also asked for socio-demographic information, their role in their
organization, and the length of time they had been working with IPV survivors, among other
details regarding their service population and professional position.

39

Analytic Techniques
The need for Rasch measurement. Before data analysis began, the survey data were
reviewed using Rasch Measurement techniques to 1) understand how the measures were
operating, and 2) correct for common psychometric issues that can result in incorrect statistical
conclusions. This technique was chosen because the survey data were comprised of multiple
scales created for this study, some ordinal and some nominal, and each using different response
options. An ideal approach to analyzing data such as this is the Rasch Analysis technique. Below
is a brief description of Rasch Analysis, why this process was chosen, and how it addresses four
psychometric issues particularly relevant to this study (ordinal data, unequal difficulties among
items, data noise, and missing data).
Rasch measurement, in part, can be thought of as a psychometric technique by which
measurement scales (e.g., surveys, tests) can be built and used in studies. As a researcher, one
has to do many things, such as design an instrument, prove an instrument is reliable and valid,
revise an instrument, and compute scale scores that are used for parametric statistical tests. Rasch
measurement allows one to do all these things as part of one process. A full description of the
logic and math behind Rasch Measurement is beyond the scope of the current study; however,
detailed explanations are provided elsewhere (Boone, 2016). Briefly, the process involves a
number of activities, which often include testing to see if the data meet Rasch model
expectations; information on the quality of individual items, including individual item fit; testing
the assumption of unidimensionality; checking to see if the scale works in the same way across
groups (invariance as determined by Differential Item Functioning); and examining the
reliability and targeting of the scale to the sample. All of these activities provide a strategy by
which sample-independent item measures and item-independent respondent measures can be

40

computed.
While this approach may sound complex, the benefits of undertaking such steps rather
than a traditional psychometric examination of scales are manifold. Psychometric evaluation of
such scales is traditionally conducted using factor analytic techniques (confirming the presence
of one or more valid unidimensional scales) and/or the calculation of Cronbach’s alpha. In recent
years, however, other more advanced psychometric approaches such as Rasch Analysis have
been developed to improve the precision with which researchers construct instruments, monitor
instrument quality, and compute respondents’ scores. While factor analysis and the calculation of
Cronbach’s alpha can assess traditional psychometric standards of validity and reliability by
confirming the presence of one or more valid unidimensional scales, Rasch techniques can do
this and more.
In addition to testing for unidimensionality, Rasch analyses allow researchers to
intensively examine the extent to which patterns of responses are predictable given what we
know about the trait we are measuring and the sample we have surveyed. In other words,
assuming only one trait is being measured the responses of each person to each item should be
predictable based on their responses to other items within a scale. This information can help
researchers know if they should remove particular items that don’t fit this predictable pattern,
collapse two or more categories of item responses, or add new items to capture aspects of the
trait that are currently missing. In addition to assessing the quality of a measure, other common
problems with survey data may also be addressed within the framework of the Rasch model.
These include complications from ordinal data, unequal difficulties among items, data noise, and
missing data. These problems and the solutions that Rasch measurement offers are detailed
below.

41

Rasch analysis addresses two challenges associated with ordinal survey data: 1)
understanding the degree of difference between responses, and 2) understanding how these
degrees of difference may change throughout the survey. In community psychology, as in many
other specialties, surveys often use self report rating scales (e.g. strongly agree, agree, disagree,
and strongly disagree; never, sometimes, usually, always). The data from these scales are
ordinal, meaning that the researcher only knows the order of the responses (that the response of
“strongly disagree” is a lower level of agreement than “disagree”) but not the degree of
difference (how much lower “strongly disagree” is). With ordinal data, we do not know if the
intervals between the responses (strongly agree-agree, agree-disagree, disagree-strongly
disagree) are equal in size. In other words, we do not know if “agree” is truly halfway between
“disagree” and “strongly agree.” Even if these responses are coded as “3”= agree, “2”=disagree,
and “1”=strongly disagree in a dataset, these numbers are arbitrarily assigned, and do not
represent actual distances.
Not only may the steps between neighboring rating categories be unequal, but the pattern
of steps may differ from question to question. Figure 3 (Boone, 2016, p.2) shows the potential
unequal spacing of rating-scale categories along a line of “agreeableness” for three survey items.
In this figure we can see that the threshold of “agreeableness” in moving from one category to
the next is equal in Question 5, but the space between disagree and agree is much further apart in
Question 8. This suggests it would take a great deal more “agreeableness” to move between these
two responses than to move from strongly disagree to disagree. Conversely, the space between
agree and disagree is much closer in Question 10. This suggests these categories may essentially
be identical, that is, they reflect the same level of agreeableness to the question.

42

Figure 2. “Example survey rating scale. For the Q#5 scale, the “jump” between each of the
ratings is equal. For the second (Q#8) and third (Q#10) scales, the “jump” from each rating to the
next rating is not equal. Furthermore, the way the rating scale functions across the items is not
identical. All that a researcher can assert is that the rating scale is ordinal (SA > A > D > SD) for
each item” (Boone, 2016, p. 2; SA=Strongly Agree; A=Agree; D=Disagree; SD=Strongly
Disagree).

Rasch Analysis can be used to overcome the problems with ordinal data by helping
researchers compute equal interval (linear) measures by essentially transforming the non-equal
interval rating scales. This is important because if parametric tests such as regression and
ANOVA are used on raw data without any transformation, a researcher may be violating
requirements of these tests. Ignoring the parametric requirement of using linear measures, as
researchers often do, can result in incorrect statistical conclusions. For example, a researcher
may think an intervention has not had a significant impact upon clients when it really has
impacted them at a statistical level of significance. Rasch techniques offer a way to avoid these
pitfalls and make use of rating-scale data to compute linear “person measures.” The term “person
measure” is the name of the Rasch scale number that expresses the performance of a scale
respondent, taking into account the measurement nuances described herein.
Another challenge accounted for by Rasch person measures is the existence of unequal
difficulties across all survey items. Just as all test items cannot be assumed to exhibit the same

43

degree of difficulty, all survey items should not be assumed to be equally easy to agree to. For
example, a 3 (agree) in response to item 8 of a survey should not be assumed to indicate the same
level of agreement with the overall scale as answering a 3 (agree) to item 10. Rasch analysis
allows researchers to see how items act on a continuum of “easiest” to agree with, to “hardest” to
agree with, and adjusts items’ weights in participants’ scores accordingly.
To understand this issue, consider an imaginary instrument intended to capture new DV
advocates’ self-efficacy related to their job. One item might be “I work to find better ways to
advocate for my clients,” and another “I am very effective in helping my clients deal with their
legal problems.” It is likely the case that advocates new to the field may have more confidence in
their ability to find better ways to work with their clients than their ability to immediately
perform highly on a complicated task - dealing with legal issues. As such, it may be easier for a
new advocate to answer “4-strongly agree” to the item concerning finding better ways to
advocate for clients in comparison to answering “4-strongly agree” to the item involving legal
problems. If a simple sum score of the items were used to represent advocates’ level of selfefficacy, both questions would be contributing equally to the assessment of this trait without any
regard for the different levels of difficulty. Conversely, if Rasch Analysis was used, the degree to
which each item is easy or more difficult to agree with would be taken into account when the
person measures were calculated. Additionally, the researcher would be able to see the relative
difficulty of each question, which can help to assess if the survey includes a sufficient number of
easy, medium, and difficult to agree with items that can capture respondents at all levels of the
construct.
Rasch techniques also allow for the identification of items that don’t contribute
meaningfully to measurement or that create unnecessary “noise” in the data. Such items might

44

include questions that don’t help differentiate between levels of the construct (e.g. everyone
answered the question the same way), or questions that aren’t working correctly (e.g. an easy
multiple choice question that was answered incorrectly by respondents who otherwise did well
on the test). These items can be identified by reviewing fit statistics for each instrument item; if
an item does not clearly fit, it is often best to remove it from the instrument.
Another challenge aided by Rasch Analysis is missing data. When respondents do not
answer all items on a survey, researchers must make a decision about how to deal with that
missing information. Commonly, researchers choose rather rudimentary solutions, such as
removing the respondent from the study overall, replacing the missing data with the mean for
that item, or inserting a respondent’s most frequent answer to the other scale questions. This can
negatively impact the results of data analyses. Instead of employing these less desirable
techniques, researchers can simply use a Rasch Measurement approach. Rasch measurement
does not require that a respondent have answered all items on a survey to be assigned a person
measure. Rather, the mathematical properties of Rasch Measurement (which are beyond the
scope of the current description) dictate that not all items need to be completed by all
respondents to calculate accurate person measures.
In summary, Rasch techniques involve corrections for a variety of psychometric issues
(e.g., rating scales are ordinal, not all survey items are the same level of difficulty, items are not
useful, data are missing) so that accurate representations of respondents’ scores can be computed
as “person measures.” This process involves a number of activities, which include testing the
assumption of unidimensionality using factor analysis; assessing fit information to examine the
quality of individual items and to look for outlying individual responses; and examining a Wright

45

Map to explore how well the scale targets the sample. A description of a Rasch analysis
conducted in Winsteps (Linacre, 2012) with each study scale is provided below.
Rasch analysis of measures.
Comfort discussing sexuality with clients. The Comfort Discussing Sexuality with
Clients scale was first examined using an Exploratory Factor Analysis (EFA) in SPSS to
determine if it met the Rasch Measurement criteria of examining a singular construct. Principle
axis extraction was used to identify the initial set of uncorrelated factors. Two factors emerged as
having Eigenvalues greater than one; these factors explained approximately 62% of the shared
variance. However, additional evidence supported a one factor solution, including a sharp drop
off in the scree plot between the first and second factors, and a relatively low drop in variance
explained (62% to 51%) in moving from a two to one factor solution. Additionally, in this initial
unrotated solution all items’ largest loadings exceeded 0.59 and were on Factor 1, suggesting a
one factor solution might be sufficient. In order to further aid interpretation, a Promax (oblique)
rotation was conducted. The Promax rotation factor correlation matrix revealed a strong
relationship between the factors (0.65), providing further support for a one factor solution.
Finally, the Promax two factor solution had three items that primarily loaded on Factor 2.
However, these items were originally from two different subscales of the Sexual Health Care
Scale-Attitude (Kim, Kang & Kim, 2011) which was the basis for this measure, suggesting an
atheoretical pattern that might be specific to this sample. Given this collection of evidence, a one
factor solution was retained, suggesting that a Rasch analysis may be undertaken of the singular
construct of Comfort Discussing Sexuality with Clients.
A Rasch Analysis was then undertaken using the Winsteps program (Linacre, 2012). The
quality of data entry and coding were in part verified through a review of the Winsteps item

46

entry table, which suggested the data were ready to be analysed. Prior to statistical analysis of
final person measures, item and person fit and reliability statistics were examined to evaluate
data quality. First, fit statistics were utilized to identify respondents who might have provided
idiosyncratic answers to one or more survey items (outliers). If an individual’s Person Outfit
MNSQ was more extreme than +/- 2, her Person Outfit ZSTD scores were examined. If the
Outfit ZSTD score was more extreme than +/- 3, the individual was marked as a misfitting
respondent. A review revealed a total of six misfitting respondents. These individuals’ responses
to each item were examined; those responses which had a z-residual greater than 3 were removed
from the data set. Doing so improved overall person and item fit, and person and item separation.
An analysis of Item Fit was also conducted. No items were identified that exceeded the
acceptable range for Item Outfit MNSQ or Item Infit MNSQ. Next, a review of the functioning
of rating scale categories was completed. As was expected, each response was “most probable”
for some combination of item difficulty and person measure, and none of the items’ ratings
exhibited disordering in their average person measures. A final review of the overall person and
item fit and person and item separation found that all indices were within acceptable ranges for
statistical analyses (see Table 1).

47

Table 1.
Descriptives, Fit Indices and Separation Indices for Rasch Analysis
Mean
Valid N Person
Measure

Scale

SD

Person
Item
Person
Item Separation
Separation
Outfit
Oufit
(Real,
(Real, Non(MNSQ) (MNSQ) NonExtreme)
Extreme)

Comfort Discussing
325
Sexuality with Clients

2.02

2.19

0.90

0.90

1.97

8.57

Comfort Discussing
Reproductive Health 370
with Clients

2.45

2.35

1.04

1.04

1.55

9.80

Universal
Reproductive
Coercion Practice

362

-2.01

1.83

0.98

1.02

2.33

5.79

Targeted Reproductive
189
Coercion Practice

-1.71

2.13

0.99

0.98

2.21

4.15

RC-Responsive
Organizational
Supports

-1.72

1.90

0.84

0.84

1.41

9.34

338

To evaluate the construct validity of the retained 11-item Comfort Discussing Sexuality
with Clients measurement device, a Wright Map was constructed wherein each item and person
is plotted using a Rasch measure. Lower person measures and lower item difficulties are
presented at the base of the map, and respondents that are more comfortable discussing sexuality
and items that are more difficult to disagree with are presented at the top of the map. Figure 3
presents this map with all items and respondents. An examination of this figure suggests that the
measure may be too easy to agree with, as respondents generally skewed toward the top of the
map. The mean item measure is nearly one standard deviation below the mean person measure,
suggesting most of the people generally agreed with most of the items, and as such were quite
comfortable discussing sexuality according to this tool. This analysis suggests that some changes
48

could be made to improve the precision of this scale in future studies, particularly around the
identification of additional items to fill the measurement gap in the higher person measures.
However for the current examination, these person measures were saved for all respondents for
parametric statistical analyses.

49

Figure 3. Wright Map for Comfort Discussing Sexuality with Clients

50

Comfort discussing reproductive health with clients. The Comfort Discussing
Reproductive Health with Clients scale was likewise first examined using an EFA. Principle axis
extraction identified a single factor with an eigenvalue greater than one, which explained 62% of
the shared variance. This factor structure suggested that a Rasch analysis may be undertaken of
the singular construct of Comfort Discussing Reproductive Health with Clients.
A Rasch Analysis was then undertaken in Winsteps (Linacre, 2012) using the procedures
identified in the previous section. The review of Person Outfit MNSQ and Person Outfit ZSTD
revealed a total of six misfitting respondents. These individuals’ responses to each item were
examined; those responses which had a z-residual greater than 3 were removed from the data set.
Doing so improved overall person and item fit, and person and item separation. No items were
identified that exceeded the acceptable range for Item Outfit MNSQ or Item Infit MNSQ, and the
functioning of rating scale categories was as expected. A final review of the overall person and
item fit and person and item separation found that all indices were within acceptable ranges for
statistical analyses (see Table 1).
To evaluate the construct validity of the retained 6-item Comfort Discussing
Reproductive Health with Clients measurement device, a Wright Map was constructed. Figure 4
presents this map with all items and respondents. An examination of this figure suggests that this
scale, much like the Comfort Discussing Sexuality with Clients scale, may be too easy to agree
with, as respondents generally skewed toward the top of the map. The mean item measure is one
standard deviation below the mean person measure, suggesting most of the people generally
agreed with most of the items, and as such were quite comfortable discussing reproductive health
with clients according to this tool. This analysis suggests that some changes could be made to
improve the precision of this scale in future studies, particularly around the identification of

51

additional items to fill the measurement gap in the higher person measures. However for the
current examination, these person measures were saved for all respondents for parametric
statistical analyses.

Figure 4. Wright map for Comfort Discussing Reproductive Health with Clients

52

Universal reproductive coercion practices. The Universal RC Practice scale was also
first examined using an EFA. Principle axis extraction identified a two factor solution that
explained 66% of the shared variance. However, similar to the Comfort Discussing Sexuality
with Clients scale above, additional evidence supported a one factor solution. This evidence
included a sharp drop off in the scree plot between the first and second factors (eigenvalues of
5.39 and 1.19, respectively), and a relatively low drop in variance explained (66% to 54%) in
moving from a two to one factor solution. Additionally, in this initial unrotated solution all
items’ largest loadings exceeded 0.51 and were on Factor 1, suggesting a one factor solution
might be sufficient. In order to further aid interpretation, a Promax (oblique) rotation was
conducted. The Promax rotation factor correlation matrix revealed a strong relationship between
the factors (0.64), providing further support for a one factor solution. Finally, the Promax two
factor solution had three of the 10 items primarily load on Factor 2—the frequency with which
respondents “Gave out RC brochure/information card,” “Talked about RC as a health concern,”
and “Asked about birth control sabotage.” However, these items do not appear to capture a
theoretical aspect of universal RC practices that is conceptually distinct from the other seven
items, as both potential “factors” include screening and education components related to an array
of RC-relevant topics. This suggests the extraction of a second factor may be the result of an
atheoretical pattern specific to this sample. Given this collection of evidence, a one factor
solution was retained, suggesting that a Rasch analysis may be undertaken of the singular
construct of Universal Reproductive Coercion Practices.
A Rasch Analysis was then undertaken in Winsteps (Linacre, 2012) using the procedures
identified in the previous sections. The review of Person Outfit MNSQ and Person Outfit ZSTD
revealed a total of fourteen misfitting respondents. These individuals’ responses to each item

53

were examined; those responses which had a z-residual greater than 3 were removed from the
data set. Doing so improved overall person and item fit, and person and item separation. Next, a
review of Item Outfit MNSQ and Item Infit MNSQ revealed misfit in Item 9 (Offered a survivor
a free at-home pregnancy test) and Item 10 (Asked a pregnant survivor if she and her partner
agree about what she should do about her pregnancy). Item 9 exceeded the appropriate level of
Outfit MNSQ & ZSTD (1.53, 5.0), and Item 10 exceeded the acceptable level of Infit MNSQ &
ZSTD (1.87, 7.3). These fit statistics suggest underfit, meaning there is too much unexplained
variance in the data. In order to correct this misfit, individual responses to Items 9 and 10 that
had a z residual at or beyond +/- 3 were deleted. This resulted in the deletion of five responses to
Item 9, and eleven responses to Item 10. Following this procedure, a new Rasch Analysis
revealed that Item 10’s Outfit MNSQ was now within an acceptable range (1.15), but that Item
9’s Infit MNSQ and ZSTD were largely unchanged (1.84, 6.9). Consideration was given to
deleting individual responses with a z residual of +/- 2 to further improve Item Fit, but doing so
would have resulted in the deletion of approximately 10% of the sample’s responses, a misfit rate
which is much higher than the 5% rate that would be expected by chance (Boone, Staver, &
Yale, 2014). As such, Item 9 was considered for deletion from the scale.
First, an examination of the substantive significance of the item was undertaken. In
comparison to the other items on the scale, it may be the case that Item 9 (Offered a survivor a
free at-home pregnancy test) captures a conceptually distinct element of RC-responsive practice
that is, in fact, a separate construct. The frequency with which an advocate is able to offer a free
pregnancy test is largely contingent on whether their agency has that relatively expensive
physical resource. This stands in contrast to the majority of other practices on the scale, which
focus on information sharing or screening—practices that don’t require stocking costly materials

54

on site. The exception to this rule is Item 1 (Give out brochure or info card), which also touches
on a physical resource-dependent factor. However, one might argue that 1) the financial cost of
these resources differs greatly, with brochures being much less expensive than pregnancy tests,
and that 2) brochures are a much more common resource for advocates to distribute in their
practice than pharmaceutical devices. Additionally, the presence or absence of such
informational materials might be much more strongly connected with information sharing or
screening practices than pregnancy tests, as such cards are often provided alongside training for
advocates on how to engage in information sharing and screening practices. The data also
provide preliminary support for this conclusion, as most of the other items on the scale have
slightly higher correlations with Item 1 than with Item 9. Given the potentially theoretically
distinct nature of the item, its high misfit values, the threat to measurement presented by item
infit problems, and the difficulty of diagnosing and remedying them (Boone, Staver, & Yale,
2014), the decision was made to exclude this item from the scale.
This resulted in a nine item Universal RC Practices scale. All remaining items fit the
model within an acceptable range. An examination of the functioning of rating scale categories
for each item revealed that Item 1 had an unexpected pattern of mean person measures. The
mean person measure for the response “Almost Always” was lower than for the response
“Sometimes,” which violated the assumption that the mean measures should grow higher with
each response category. However, this was likely the result of the small number of
respondents—only one—who used the “Almost Always” response, and as such this anomaly was
not acted upon. A final review of the overall person and item fit and person and item separation
found that all indices were within acceptable ranges for statistical analyses (see Table 1).

55

To evaluate the construct validity of the new 9-item Universal RC Practice measurement
device, a Wright Map was constructed. Figure 5 presents this map with all items and
respondents. An examination of this figure suggests that this scale may have been difficult, as
respondents generally skewed toward the bottom of the map. The mean item measure is more
than one standard deviation above the mean person measure, suggesting most of the people
generally had low frequencies with which they engaged in most of the Universal RC Practices.
Additionally, the item measures are quite close together, suggesting there may be too many items
oversampling one level of the trait. The small amount of variation that does exist demonstrates
that the least frequent practices were Item 1 (Giving a brochure or info card), and Item 6,
(Giving info about emergency contraception), and the most frequent practices were Item 5
(Asking if survivor wanted to talk to a healthcare worker about reproductive health) and Item 8
(Offering information about getting a pregnancy test elsewhere). This ordering is aligned with
expectations, as advocates would likely be more comfortable referring clients elsewhere for
reproductive health-related discussions and services than offering such education or resources
themselves. This analysis suggests that some changes could be made to improve the precision of
this scale in future studies, particularly around the identification of additional items to fill the
measurement gap in the lower person measures and potentially the deletion of items that seem to
capture similar frequency of practice levels. However for the current examination, these person
measures were saved for all respondents for parametric statistical analyses.

56

Figure 5. Wright map for Frequency of Universal RC Practice.

57

Targeted reproductive coercion practices. The Targeted RC Practices scale was first
examined using an EFA. Principle axis extraction identified a single factor with an eigenvalue
greater than one, which explained 62% of the shared variance. This factor structure suggested
that a Rasch analysis may be undertaken of the singular construct of Targeted RC Practices.
A Rasch Analysis was then undertaken in Winsteps (Linacre, 2012) using the procedures
identified in the first section. The review of Person Outfit MNSQ and Person Outfit ZSTD
revealed a total of seven misfitting respondents. These individuals’ responses to each item were
examined; those responses which had a z-residual greater than 3 were removed from the data set.
Doing so improved overall person and item fit, and person and item separation. Item 3 (Helped a
survivor get a pregnancy test) exceeded the acceptable range for Item Outfit MNSQ and ZSTD
(1.36, 2.7) and Item Infit MNSQ and ZSTD (1.54, 4.0). These fit statistics suggest underfit,
meaning there is too much unexplained variance in the data. In order to correct this misfit,
individual responses to Item 3 that had a z residual at or beyond +/- 3 were deleted. This resulted
in the deletion of three responses to Item 3. This brought Item 3’s Outfit MNSQ within the
acceptable range (1.23), and lowered Infit MNSQ and ZSTD (1.38, 2.9) but did not rectify the
Infit problem entirely. In order to further improve item fit, 13 individual responses with a z
residual of +/- 2 on Item 3 were deleted. Following this procedure, a new Rasch Analysis
revealed that Item 3’s Infit MNSQ was now within an acceptable range (1.00), and that the
functioning of rating scale categories was as expected. A final review of the overall person and
item fit and person and item separation found that all indices were within acceptable ranges for
statistical analyses (see Table 1).
To evaluate the construct validity of the 6-item Targeted RC Practices measurement
device, a Wright Map was constructed. Figure 6 presents this map with all items and

58

respondents. An examination of this figure suggests that this scale may have also have been
difficult, as respondents generally skewed toward the bottom of the map. The mean item measure
is slightly less than one standard deviation above the mean person measure, suggesting most of
the people generally had low frequencies with which they engaged in most of the Targeted RC
Practices, though perhaps slightly more frequently than the items in the Universal RC Practices
scale. This may be the result of a small set of outliers with very high person measures, or a
function of the smaller group of people who completed this scale due to the survey skip pattern.
(Participants were only asked these items if they reported having a client they believed had
experienced RC in the last year.)
Additionally, the item measures are quite close together, suggesting there may be too
many items oversampling one level of the trait. The small amount of variation that does exist
demonstrates that the least frequent practices were Item 4 (Discussing pregnancy termination
options), and Item 3 (Helping survivor get pregnancy test), and the most frequent practices were
Item 6 (Asking pregnancy survivor if she and her partner agree on what to do about pregnancy)
and Item 5 (Incorporating RC into safety planning). This ordering is aligned with expectations,
as pregnancy termination is commonly challenging for people in the U.S. to discuss
(Guttmacher, 2016). Likewise, advocates may expect that pregnant survivors are more likely to
expect questions about their reproductive health than non-pregnant survivors, and may be more
likely to see comprehensive safety planning as part of their role than other RC-responsive
practices. This analysis suggests that some changes could be made to improve the precision of
this scale in future studies, particularly around the identification of additional items to fill the
measurement gap in the lower person measures and potentially the deletion of items that seem to

59

capture similar frequency of practice levels. However for the current examination, these person
measures were saved for all respondents for parametric statistical analyses.

Figure 6. Wright map for Frequency of Targeted Reproductive Coercion Practice.

60

RC-responsive organizational supports. Because the RC-Responsive Organizational
Supports scale uses dichotomous items, an exploratory factor analysis was not appropriate. In
order to determine the extent to which these items may be treated as a single construct, a Rasch
analysis was undertaken.
A Rasch Analysis was completed in Winsteps (Linacre, 2012) using the procedures
identified in the previous sections. The review of Person Outfit MNSQ and Person Outfit ZSTD
revealed a total of nine misfitting respondents. These individuals’ responses to each item were
examined; those responses which had a z-residual greater than 3 were removed from the data set.
Doing so improved overall person and item fit, and person and item separation. Item 15
(Program has pregnancy tests available onsite) exceeded the acceptable range for Item Outfit
MNSQ and ZSTD (1.69, 3.5). These fit statistics suggest underfit, meaning there is too much
unexplained variance in the data. In order to correct this misfit, individual responses to Item 15
that had a z residual at or beyond +/- 3 were deleted. This resulted in the deletion of eleven
responses to Item 15. This brought Item 15’s Outfit MNSQ within the acceptable range (1.07).
Other items exceeded the lower boundary of acceptable Outfit MNSQ, but their Outfit ZSTD
scores did not exceed +/- 2, suggesting no action was required. Lastly, the functioning of rating
scale categories was as expected for all items. A final review of the overall person and item fit
and person and item separation found that, while the measure may be slightly overfit, all indices
were within acceptable ranges for statistical analyses (see Table 1).
To evaluate the construct validity of the 15-item RC-Responsive Organizational Supports
measurement device, a Wright Map was constructed. Figure 7 presents this map with all items
and respondents. An examination of this figure suggests that this scale may have also have been
difficult, as respondents generally skewed toward the bottom of the map. The mean item measure

61

is one standard deviation above the mean person measure, suggesting many of the people did not
have many of these RC-responsive practices present at their organization. The items are more
spread out than in previous scales, suggesting this scale performs better than the previous scales
in sampling from different levels of its trait, organizational RC-responsiveness. However, there
are large gaps, such as between the “easiest’ supports to have in place (Item 1, a private place to
talk, and Item 3, resource list with RSH contacts) and the medium level items. The ordering of
the item is logical. For example, the aforementioned “easy” items are resources that are often
present in DV programs, and the items about formal incorporation of RSH topics into required
forms and processes (a much less common practice) are clumped together toward the more
“difficult” end of the map. Interestingly, Item 4 (required employee training) and Item 5
(leadership bringing up the topic) were located at the same position, suggesting they may sample
a similar facet of RC-responsive organizational practice. This analysis suggests that some
changes could be made to improve the precision of this scale in future studies, particularly
around the identification of additional items to fill the measurement gap between the easiest
items and the medium/hard items. However, for the current examination, these person measures
were saved for all respondents for parametric statistical analyses.

62

Figure 7. Wright map for Reproductive Coercion-Responsive Organizational Supports.
In summary, Rasch techniques were employed 1) to understand how the five study
measures were operating, and 2) to produce “person measures” that correct for common

63

psychometric issues with survey data. This process involved a number of activities, wherein each
measure was determined to be unidimensional; extreme or irregular individual responses were
removed; items were adjusted or removed to ensure the quality of the scales; and the degree to
which the scale adequately captures the trait in the sample was assessed. The person measures
produced herein were used in the study’s parametric analyses, which are described below.
Descriptive and parametric analyses. The study used a combination of statistical
techniques to explore the nature of the sample, advocates’ responses to the survey items, and
relationships among the Rasch person measures. These techniques tested the substantive
hypotheses outlined above by using linear regression, latent class analysis, ANOVA, and
hierarchical regression. Each analysis is described in detail in the Results section.

64

RESULTS
The results of the study are presented next. Following a description of the sample, an
examination of the descriptive statistics for each of the study’s scales was undertaken to describe
the current state of comfort, practice, and organizational supports in the field. Next, advocate
characteristics were examined as potential covariates of the study scales. Finally, the results of
the statistical testing of the study hypotheses are presented.
Sample Characteristics
The final sample included in the analyses was comprised of 371 participants from fortyfour U.S. states and territories. The largest group of respondents from a state or territory was 40
and the smallest was one. The sample was comprised largely of heterosexual (85%), white
(70%), middle-aged (M=40 years, SD=13.96) females (98%) and a majority of respondents
(69%) identified as slightly to extremely politically liberal. On average, participants had worked
or volunteered in the area of IPV services for almost eight years, and had worked at their current
organization for almost three years. Over half of respondents identified their role in their
organization solely as an advocate (58%), 9% identified as a combination of advocate and some
other role (e.g., community education and prevention, outreach, counselor), 6% identified as
Counselors, and 5% identified as Executive Directors. Five percent were volunteering for the
organization; the remainder were paid staff.
Almost 40% of the represented organizations identified as domestic violence victim
service agencies (39.8%), while 40% addressed both DV and sexual assault. Half of the
respondents worked primarily in a communal shelter setting (47%), while 11.6% worked in
Outreach/Satellite/Drop-in Offices and 11.3% worked in Counseling Office settings. Advocates
perceived the political climates in the demographic areas within which they worked to range

65

from liberal (45.2%) to conservative (33.2%; 21.6% moderate) and reported that their
organizations were located in rural (36%), urban (38%), suburban (14%), and mixed/other (12%)
communities. Only 6% of respondents were aware of their organization having a religious
affiliation.
Descriptive Statistics
Advocates’ comfort discussing sexuality and reproductive health. Advocates were
asked a series of items about their level of comfort discussing sexuality and reproductive health
with clients. On average, advocates indicated that they were comfortable discussing sexuality
with their clients (see Table 2). Participants’ reports suggest they generally did not perceive any
discomfort to be rooted in their own misgivings about discussing sexuality. Over 80% of
participants disagreed with the statement “I may be embarrassed if clients broach sexual issues”
(82%). Additionally, approximately three-quarters of respondents reportedly disagreed with the
statements “Discussing sex is a difficult thing to do” (76%) and “I feel uncomfortable discussing
sex with clients” (74%). Conversely, advocates’ greatest discomfort seemed to lie in their fear of
their clients’ reaction. A majority of participants agreed with or were uncertain what they
thought about the statements “I am afraid clients would feel their privacy was invaded if I asked
specific questions about sex” (64%) and “I am afraid clients would be offended if I broached
sexual issues” (59%). This pattern did not hold for all items however; a large proportion of
respondents disagreed with the statement “Clients would be uncomfortable if I broached sexual
issues” (83%).
When asked about their comfort discussing a variety of reproductive health topics with
their clients, responses were more mixed (see Table 3). A majority of advocates reported feeling
very comfortable asking survivors about pregnancy coercion (70%); providing information about

66

concealable birth control methods (67%); and talking to a survivor about emergency
contraception (61%). However, when it came to discussing condom negotiation tactics and
options for putting a child up for adoption, levels of comfort appeared to drop. While 75% or
more participants still reported feeling somewhat or very comfortable in these areas, the degree
of comfort was certainly lessened. Finally, discussing pregnancy termination options with clients
was by far the most contentious topic. 35% of advocates reported feeling very or somewhat
uncomfortable with such conversations.
Organizational supports. Next, advocates were asked about whether or not their
organization had a variety of supports present at their agency (see Table 4). The majority of
respondents indicated that they either did NOT have the support or were not sure if they had the
support on 12 of the 15 items. The most common supports reported were lists including
reproductive and sexual health (RSH) resources (77%); private places to talk with clients about
health concerns (74%); and posters, brochures or other informational materials about RSH
openly displayed (51%). The least common supports were questions on an intake form about
recent unprotected sex (6%); program-wide universal screening of all survivors for RC (9%); and
emergency contraception available on site (9%). Additionally, few participants reported having
any of the RC-responsive questions on their intake forms.
Engagement in RC-responsive practices. Next, survey respondents were asked how
frequently they engaged in a variety of universal and targeted reproductive coercion-responsive
practices when working with a survivor in the last year (see Tables 5 & 6). As was previously
stated, universal RC-responsive practices are those actions that experts in the field believe
advocates should be engaging in with 100% of clients. Interestingly, the sample never exceeded
ten percent of advocates engaging in these practices 75% or more of the time. In fact, the

67

majority of respondents reported engaging in all ten practices less than 25% of the time. This
included three items where the majority never engaged in the practice: offering a survivor a free
at-home pregnancy test (67%); giving information cards about RC to a survivor (60%); and
offering information about emergency contraception (51%).
One hundred eighty-nine advocates reported that they had worked with a survivor they
knew or suspected had experienced RC in the last year. When this group was asked the
frequency with which they engage in certain targeted-RC responsive practices in such situations,
the response was somewhat improved. Although the majority of respondents still reported
engaging in all practices less than 25% of the time, there were no items where a majority had
never taken the action. Additionally, three items exceeded ten percent of advocates engaging in
these practices 75% or more of the time: incorporated RC considerations into safety planning
(16%); asked pregnant survivor if she and her partner agree about pregnancy (13%); and helped
a survivor get a pregnancy test (11%). In short, the distribution of the responses shifted slightly
more positively.

68

Table 2.
Frequencies & Descriptives: Comfort Discussing Sexuality with Clients

Please respond to the following statements by ticking the column which best represents your attitudes.
N

M

SD

Agree

Uncertain

Disagree

1. Discussing sex is a difficult thing to do.

324

2.70

0.57

19
(5.9%)

59
(18.2%)

246
(75.9%)

2. I am not ready to talk about sexual issues
with clients.

322

2.36

0.73

49
(15.2)

109
(33.9)

164
(50.9)

3. I am afraid conversation about sex with
clients would bring about a distance
between me and them.

324

2.50

0.80

63
(19.4)

36
(11.1)

225
(69.4)

4. Clients would be uncomfortable if I
broached sexual issues.

325

2.78

0.51

14
(4.3)

42
(12.9)

269
(82.8)

5. I am reluctant to discuss sex with clients
of the opposite sex.

324

2.28

0.81

74
(22.8)

85
(26.2)

165
(50.9)

6. It is uncomfortable to discuss sexual
issues with clients.

322

2.59

0.67

33
(10.2)

65
(20.2)

224
(69.6)

7. I may be embarrassed if clients broach
sexual issues.

324

2.77

0.52

15
(4.6)

45
(13.9)

264
(81.5)

8. I am afraid clients would feel their
privacy was invaded if I asked specific
questions about sex.

320

2.08

0.79

88
(27.5)

118
(36.9)

114
(35.6)

69

Table 2. (cont’d)
9. I feel uncomfortable discussing specific
sexual activities with clients.

325

2.24

0.77

66
(20.3)

116
(35.7)

143
(44.0)

10. I am afraid clients would be offended if
I broached sexual issues.

324

2.29

0.67

39
(12.0)

153
(47.2)

132
(40.7)

11. I feel uncomfortable discussing sex with
clients.

325

2.69

0.57

18
(5.5)

66
(20.3)

241
(74.2)

70

Table 3.
Frequencies & Descriptives: Comfort Discussing Reproductive Health with Clients

How comfortable do you feel doing the following?
N

M

SD

Very
Uncomfortable

Somewhat
Uncomfortable

Somewhat
Comfortable

Very
Comfortable

1. Asking a survivor if she felt her
partner was trying to get her
pregnant when she did not want
to be

367

3.61

0.70

11
(3.0%)

13
(3.5%)

85
(23.2%)

258
(70.3)

2. Giving a survivor information
about birth control methods that
are easy to conceal or hide from a
partner

359

3.55

0.74

13
(3.6)

15
(4.2)

92
(25.6)

239
(66.6)

3. Talking to a survivor about
emergency contraception
(medications taken after
unprotected intercourse that
prevent pregnancy)

363

3.47

0.78

13
(3.6)

25
(6.9)

105
(28.9)

220
(60.6)

4. Talking to a survivor about how
to negotiate condom use
(convince a male partner to use a
condom)

359

3.24

0.85

17
(4.7)

45
(12.5)

131
(36.5)

166
(46.2)

5. Talking with a pregnant survivor
about her options for putting their
child up for adoption

356

3.11

0.93

24
(6.7)

64
(18.0)

118
(33.1)

150
(42.1)

6. Talking with a survivor about her
options for pregnancy termination

354

2.86

1.08

55
(15.5)

69
(19.5)

100
(28.2)

130
(36.7)

71

Table 4.
Frequencies & Descriptives: RC-Responsive Organizational Supports Items
Now we are going to ask you a few questions about the organization where you currently work/volunteer.
N

Yes

No or
Have No
Idea

1. Does your program have a private place that is regularly used to screen and talk with
clients about health concerns?

338

250
(74%)

88
(26%)

2. Does your program have a particular health educator, nurse, PA or doctor the staff
can contact with questions about reproductive and sexual health?

338

90
(27)

248
(73)

3. Does your program have a resource list(s) that identifies clinical referrals/resources
for survivors who want to access reproductive/sexual health care?

337

258
(77)

79
(23)

4. Do new hires receive training on reproductive/sexual health issues (as related to
DV/SA) during orientation?

336

139
(41)

197
(59)

5. Has anyone in a leadership position brought up reproductive/sexual health-related
issues during meetings with staff (e.g., staff meetings, case reviews) in the last year?

335

133
(40)

202
(60)

6. Are there any posters, brochures, and/or other informational materials about
reproductive/sexual health openly displayed at your organization?

337

170
(51)

167
(49)

7. Does your organization’s intake form include questions asking (directly or
indirectly) if a survivor has recently (within the past 5 days) had unprotected sex?

335

20
(6)

315
(94)

8. Does your organization’s intake form include questions asking (directly or
indirectly) if a survivor is concerned about having an unwanted pregnancy?

335

38
(11)

297
(89)

9. Does your organization’s intake form include questions asking (directly or
indirectly) if a survivor needs access to reproductive/sexual health care?

335

70
(21)

265
(79)

72

Table 4. (cont’d)

10. Does your organization’s intake form include questions asking (directly or
indirectly) if a survivor has experienced birth control sabotage or otherwise had a
partner try to get her pregnant without her consent?

335

38
(11)

297
(89)

11. Does your organization’s intake form include questions asking (directly or
indirectly) if a survivor has had a partner attempt to interfere with her desired
outcome of a pregnancy?

335

45
(13)

290
(87)

12. Does your program universally screen all survivors for reproductive coercion?

335

13. Do staff in your program have scripted tools/instructions on how to assess for the
need for emergency contraception?

335

30
(9)
50
(15)

305
(91)
285
(85)

14. Does your program have emergency contraception available on site?

335

29
(9)

306
(91)

15. Does your program have pregnancy tests available on site?

335

90
(27)

245
(73)

73

Table 5.
Frequencies & Descriptives: Universal Reproductive Coercion (RC) Practice Items

How frequently have you done the following while working with a survivor at your current agency in the last year?
Responses:
■
■
■
■
■
■
■

Never (0%)
Occasionally (1-25% of the time)
Sometimes (26-50% of the time)
Frequently (51-75% of the time)
Almost always (76%-99% of the time)
Always (100% of the time)
[Only on #7: Not applicable- I’ve never worked with a pregnant survivor; Coded as missing]
N

M

SD

Never

Occasionall
y

Sometimes

Frequently

Almost
Always

Always

1. Gave a brochure or
information card about
reproductive coercion to a
survivor

362

1.71

1.05

217
(59.9%)

75
(20.7%)

40
(11.0%)

21
(5.8%)

7
(1.9%)

2
(0.6%)

2. Talked about reproductive
coercion as one of many
health concerns for women

362

2.27

1.16

109
(30.1)

123
(34.0)

71
(19.6)

43
(11.9)

13
(3.6)

3
(3.6)

3. Asked whether a partner
has sabotaged a survivor’s
birth control or otherwise
tried to get her pregnant
when she doesn’t want to
be

362

2.18

1.23

135
(37.3)

107
(29.6)

67
(18.5)

32
(8.8)

15
(4.1)

6
(1.7)

74

Table 5. (cont’d)

4. Asked (directly or
indirectly) if a survivor
may need emergency
contraception

362

2.18

1.26

5. Asked a survivor if she
would like to talk to a
healthcare worker about
her reproductive health
needs

362

2.56

1.38

6. Offered information about
emergency contraception
(instead of waiting for the
survivor to ask)

362

1.94

7. Offered information about
birth control options
(instead of waiting for the
survivor to ask)

362

8. Offered a information
about how to get a
pregnancy test elsewhere

140
(38.7)

104
(28.7)

60
(16.6)

37
(10.2)

14
(3.9)

7
(1.9)

96

109

68

48

29

12

(26.5)

(30.1)

(18.8)

(13.3)

(8.0)

(3.3)

1.24

186
(51.4)

84
(23.2)

46
(12.7)

25
(6.9)

15
(4.1)

6
(1.7)

2.10

1.25

157
(43.4)

93
(25.7)

57
(15.7)

34
(9.4)

16
(4.4)

5
(1.4)

362

2.37

1.39

125
(34.5)

97
(26.8)

71
(19.6)

32
(8.8)

23
(6.4)

14
(3.9)

9. Offered a survivor a free
at-home pregnancy test

362

1.76

1.32

242
(66.9)

42
(11.6)

37
(10.2)

16
(4.4)

13
(3.6)

12
(3.3)

10. Asked a pregnant survivor
if she and her partner
agree about what she
should do about her
pregnancy

343

2.27

1.30

118
(34.4)

111
(32.4)

53
(15.5)

34
(9.9)

19
(5.5)

8
(2.3)

75

Table 6.
Frequencies & Descriptives: Targeted Reproductive Coercion (RC) Practice Items

If you know or suspect a female survivor has experienced [RC], how frequently have you done the following at your current organization
in the last year?
Responses:
■
■
■
■
■
■
■

Never (0%)
Occasionally (1-25% of the time)
Sometimes (26-50% of the time)
Frequently (51-75% of the time)
Almost always (76%-99% of the time)
Always (100% of the time)
[Only on #6: Not applicable- I’ve never worked with a pregnant survivor in this situation; Coded as missing]
N

M

SD

Never

Occasionall
y

Sometime
s

Frequently

Almost
Always

Always

1. Offered information about
emergency contraception (e.g.
that a medication taken after
unprotected intercourse can
prevent pregnancy, how the
medication works, how
someone can get it)

189

2.37

1.40

61
(32.3%)

63
(33.3%)

26
(13.8%)

21
(11.1%)

9
(4.8%)

9
(4.8%)

2. Offered information about
birth control methods that
partners can't interfere with

189

2.44

1.40

62
(32.8)

46
(24.3)

44
(23.3)

17
(9.0)

13
(6.9)

7
(3.7)

3. Helped a survivor get a
pregnancy test

188

2.26

1.49

82
(43.6)

45
(23.9)

22
(11.7)

18
(9.6)

12
(6.4)

9
(4.8)

76

Table 6. (cont’d)
4. Discussed options for
pregnancy termination

189

2.07

1.22

75
(39.7)

64
(33.9)

27
(14.3)

11
(5.8)

8
(4.2)

4
(2.1)

5. Incorporated reproductive
coercion considerations into
safety planning to reduce
survivors' risk of unwanted
pregnancy or sexually
transmitted infection

189

2.73

1.49

47
(24.9)

52
(27.5)

35
(18.5)

24
(12.7)

22
(11.6)

9
(4.8)

6. Asked pregnant survivor if
she & her partner agree about
pregnancy

181

2.66

1.47

47
(26.0)

52
(28.7)

31
(17.1)

27
(14.9)

15
(8.3)

9
(5.0)

77

Demographic and Organizational Covariates of Study Scales
A number of individual as well as organizational level characteristics were examined to
see if they related to advocates’ comfort discussing sexuality, comfort discussing reproductive
health, frequency of universal and targeted RC practices, and RC-responsive organizational
supports. Table 7 displays significant correlations.

Table 7.
Bivariate Correlations Between Advocate Characteristics and Study Scale Person Measures

Scale

Age

Comfort Discussing
Sexuality with Clients

Estimated %
Personal
of Clients who
Years in DV
Years at
Political
Movement Organization Experienced
Conservatism
RC

0.02

0.10

0.04

-0.06

-0.02

0.15**

0.20**

0.14

-0.01

-0.07

Universal Reproductive
0.10
Coercion Practice

0.18**

0.23*

0.16**

0.12*

Targeted Reproductive
0.16*
Coercion Practice

0.12

0.21

0.11

0.06

RC-Responsive
Organizational Supports 0.09

0.12*

-0.01

0.09

0.07

Comfort Discussing
Reproductive Health
with Clients

** Correlation significant at the p < .01 level (two tailed)
* Correlation significant at the p < .05 level (two tailed)

Comfort Discussing Reproductive Health with Clients was significantly correlated with
respondent age and years of participation in the DV movement, suggesting more experienced
and/or older advocates are more comfortable having such conversations. This pattern of age and
experience modestly predicting person measures was found with several scales. Universal RC
Practice person measures were significantly correlated with years in the DV movement and

78

years at one’s organization, Targeted RC Practice person measures were significantly correlated
with respondent age, and RC-Responsive Organizational Supports person measures were
significantly correlated with years in the DV movement. Perception of clients’ experiences of RC
may also influence or be influenced by advocates’ attitudes and practices. Advocates who
reported that they had “worked with a survivor they knew or suspected had experienced RC in
the last year” had significantly higher Comfort Talking about Sex with Clients (F (1, 332)=6.26,
p<.05) and Universal RC Practice (F (1, 359)= 45.45, p< .001) scores than advocates who did
not believe they had worked with a survivor of RC in that time. Likewise, advocates’ estimates
of the percentage of their clients that they believed had experienced RC was significantly related
to Universal RC Practice scores. Finally, personal political conservatism was significantly
correlated with Universal RC Practices, suggesting that advocates who are more politically
conservative may engage in such practices slightly more frequently. No significant correlation
was found between percentage of time spent in direct service with survivors and any of the study
scales, and no group differences were found based on race, gender identity, or sexual identity.
There were no significant differences in the study scales based on the political climate of
advocates’ organization’s service area, their organization’s religious affiliation, their role in their
organization, nor on their main work setting. Likewise, no significant differences existed
between organizations that offered only domestic violence services and organizations that
offered both domestic violence and sexual assault services.
Hypothesis 1
To test whether higher levels of comfort discussing sexuality would be associated with
more frequent universal and targeted RC practice, participants’ person measures for the universal
RC practice and targeted RC practice scales were regressed onto their comfort discussing

79

sexuality with clients person measures. As hypothesized, comfort discussing sexuality with
clients significantly predicted advocates’ frequency of engagement in Universal RC practices, b=
0.294, F (1, 323)= 46.54, p < .001. Level of comfort discussing sexuality with clients explained a
significant proportion of variance in Universal RC practice, R2= 0.13 Comfort discussing
sexuality with clients also significantly predicted advocates’ frequency of engagement in
Targeted RC practices, b= 0.344, F (1, 172)= 24.47, p < .001. Level of comfort discussing
sexuality with clients explained a significant proportion of variance in Targeted RC practice, R2=
0.13. These results support the prediction that advocates who are more comfortable discussing
sexuality with clients would engage in Universal and Targeted RC practices more frequently.
Hypothesis 2
To test whether higher levels of comfort discussing reproductive health would be
associated with more frequent universal and targeted RC practice, participants’ person measures
for the universal RC practice and targeted RC practice scales were regressed onto their comfort
discussing reproductive health with clients person measures. As hypothesized, comfort
discussing reproductive health with clients significantly predicted advocates’ frequency of
engagement in Universal RC practices, b= 0.236, F (1, 360)= 36.99, p < .001. Level of comfort
discussing reproductive health with clients explained a significant proportion of variance in
Universal RC practice, R2= 0.09. Comfort discussing reproductive health with clients also
significantly predicted advocates’ frequency of engagement in Targeted RC practices, b= 0.232,
F (1, 188)= 13.49, p < .001. Level of comfort discussing reproductive health with clients
explained a significant proportion of variance in Targeted RC practice, R2= 0.07. These results
support the prediction that advocates who are more comfortable discussing reproductive health
with clients engage in Universal and Targeted RC practices more frequently.

80

Hypothesis 3
To test whether RC-responsive organizational supports would be associated with more
frequent universal and targeted RC practice, participants’ person measures for the universal RC
practice and targeted RC practice scales were regressed onto their RC organizational supports
person measures. As hypothesized, organizational supports significantly predicted advocates’
frequency of engagement in Universal RC practices, b= 0.40, F (1, 337)= 66.82, p < .001.
Organizational supports explained a significant proportion of variance in Universal RC practice,
R2= 0.17. Organizational supports person measures also significantly predicted advocates’
frequency of engagement in Targeted RC practices, b= 0.463, F (1, 181)= 39.80, p < .001.
Organizational support person measures explained a significant proportion of variance in
Targeted RC practice, R2= 0.18. These results support the prediction that advocates who have
more RC-responsive organizations engage in Universal and Targeted RC practices more
frequently.
Hypothesis 3a
Exploratory Research Question 3a sought to explore if there are certain combinations of
RC-responsive organizational supports that are more influential than other combinations. In
order to undertake the exploration detailed in Hypothesis 3a, a Latent Class Analysis (LCA) was
conducted. LCA aims to uncover unobserved heterogeneity in a population and to find
substantively meaningful groups of people that are similar in their responses to measured
variables (Muthen, 2004). Given the exploratory nature of this study, one-, two-, three-, four-,
and five-class solutions were examined using the 15 RC-responsive organizational support items.
All analyses were conducted using Mplus (Version 7.4, Muthen & Muthen, 2012). The interitem correlations are presented in Table 8.

81

Table 8.
Inter-item Correlations for RC-Responsive Organizational Supports
Item
Private Space

Private
Space

Health
Prof.
Contact

RSH on
New
Leader.
Resource Employee
Bring up
List
Training

IntakeIntakeIntakeIntakeIntakeScript for
Info
Access to
Birth
Universal
Unprotec. Unwanted
Pregnancy
EC
Materials
Screening
RSH
Control
Sex
Pregnancy
Outcome
Assessment
Services Sabotage

EC on
site

Pregnancy
Tests on
Site

1.00

.25

.27

.16

.20

.17

.09

.15

.17

.15

.18

.09

.08

.09

.16

Health
Professional
Contact

.25

1.00

.11

.15

.17

.16

.19

.17

.17

.21

.24

.10

.13

.15

.15

RSH on
Resource List

.27

.11

1.00

.18

.21

.21

.05

.11

.15

.13

.18

.07

.17

.04

.01

New
Employee
Training

.16

.15

.18

1.00

.42

.33

.09

.25

.24

.27

.29

.21

.21

.15

.03

Leadership
Bring up

.20

.17

.21

.42

1.00

.41

.13

.15

.27

.27

.25

.19

.13

.12

.13

Info Materials

.17

.16

.21

.33

.41

1.00

.05

.16

.18

.15

.13

.12

.16

.13

.13

IntakeUnprotected
sex

.09

.19

.05

.09

.13

.05

1.00

.55

.37

.39

.38

.14

.28

.24

.19

IntakeUnwanted
Pregnancy

.15

.17

.11

.25

.15

.16

.55

1.00

.56

.50

.52

.32

.30

.16

.12

Intake- Access
to RSH
.17
Services

.17

.15

.24

.27

.18

.37

.56

1.00

.42

.53

.35

.26

.08

.07

Intake- Birth
Control
Sabotage

.21

.13

.27

.27

.15

.39

.50

.42

1.00

.77

.35

.22

.26

.17

.15

82

Table 8. (cont’d)
IntakePregnancy
Outcome

.18

.24

.18

.29

.25

.13

.38

.52

.53

.77

1.00

.37

.23

.19

.12

Universal
Screening

.09

.10

.07

.21

.19

.12

.14

.32

.35

.35

.37

1.00

.22

.24

.02

Script for EC
Assessment

.08

.13

.17

.21

.13

.16

.28

.30

.26

.22

.23

.22

1.00

.17

.09

EC on Site

.09

.15

.04

.15

.12

.13

.24

.16

.08

.26

.19

.24

.17

1.00

.32

Pregnancy
Tests on

.16

.15

.01

.03

.13

.13

.19

.12

.07

.17

.12

.02

.09

.32

1.00

83

Two statistics obtained with LCA, the Bayesian information criterion (BI) and the
bootstrap likelihood ratio test (BLRT), have been found to be the most effective at identifying
the number of latent classes that should be extracted from the indicator variables (Nylund,
Asparouhov, & Muthen, 2007). Both criteria were considered in selecting from among the five
different solutions. With the BIC, the solution with the smallest value is identified as the optimal
model, whereas the BLRT tests the statistical significance of the improvement in the model when
an additional class is extracted. Additionally, the classification quality of the model was
evaluated according to the entropy criterion, in which the values range from 0 to 1, where values
closer to 1 indicate good classification.
Table 9 shows the fit indices for each model solution. The BIC was smallest for the threeclass solution, indicating three classes should be extracted. The BLRT, however, was statistically
significant for the four-class solution, but not the five-class solution. Interestingly, the four-class
solution extracted three classes with characteristics almost identical to those in the three-class
solution, and one very small class of advocates (n=13).
Table 9.
RC-Responsive Organizational Supports Latent Class Analysis Fit Indices

1 class solution
2 class solution
3 class solution
4 class solution
5 class solution

Free
Parameters
15
31
47
63
79

H0 Value

BIC

Entropy

-2412.06
-2150.31
-2073.06
-2036.99
-2014.37

4911.46
4481.13
4419.80
4440.84
4488.75

0.92
0.79
0.85
0.86

Bootstrapped
LRT p-value
<.001
<.001
<.001
0.06

In both solutions, advocates in the first class (the “Moderate RC-Responsiveness” class)
had the largest number of organizational supports. They overwhelmingly reported having

84

information sharing supports (private spaces, RSH-inclusive resource lists, RSH-related
informational materials), leadership supports (new hires receive training on RSH, and leadership
bringing up RSH at a meeting in the last year), and several procedural supports (questions on the
intake about access to RSH care, birth control sabotage, and pregnancy outcome interference). A
smaller majority of this class reported having a medical contact for RSH-related questions, and
having a question about unwanted pregnancy on their intake forms. This class had a small
prevalence estimate at 11% (3 class solution) and 10% (4 class solution) of the sample.
Advocates were fairly evenly split between the second and third classes (42% and 46%,
respectively in the 3 class solution; 35% and 51% in the 4 class solution). Survey respondents in
the second class (the “Fair RC-Responsiveness” class) were characterized by the presence of
information sharing supports (private spaces, RSH-inclusive resource lists, and RSH-related
informational materials). A small majority of these individuals also reported having leadership
supports (new hires receive training on RSH, and leadership bringing up RSH at a meeting in the
last year). Finally, advocates in the remaining class (the “Poor RC-Responsiveness” class)
reported having few or no organizational supports. These individuals generally only had the
informational supports of Private Spaces and/or RSH-Inclusive Resource Lists, if that.
In the four-class solution, the fourth class (N= 13, 4%) was quite small. While normally
a class of this size would be disregarded as an artifact of over-extraction, the combination of
characteristics of advocates in this class is theoretically intriguing. This class was characterized
by individuals who had the basic informational supports of the poor RC-responsiveness class
(Private Spaces and RSH-Inclusive Resource Lists); did NOT the have leadership supports of the
fair or moderate RC-responsiveness classes (new hires receive training on RSH, and leadership
bringing up RSH at a meeting in the last year); but DID have several procedural supports

85

(questions on the intake about access to RSH care, 84%; unprotected sex, 55%; unwanted
pregnancy, 100%; and scripted tools for assessing need for emergency contraception, 55%). The
latter three of these particular procedural supports were not present for the majority of advocates
in any of the other classes. Finally, very few advocates in this category said their agency had
questions on their intake that were specific to reproductive coercion compared to the moderate
RC-responsiveness class, including questions about birth control sabotage (0% v. 87% in the
moderate class) and pregnancy outcome interference (22% v. 97% in the moderate class). It
could be the case that these agencies have reproductive and sexual health screening supports, but
not reproductive coercion-responsiveness supports. As such, this class is referred to as the “RSH
Screening” class.
There were several types of organizational supports that were not probable for the
majority of members of any class to endorse in either the three- or four-class solution. These
included a procedural support (intake question(s) about universal screening for RC) and medical
supports on site (pregnancy tests, emergency contraception). However, advocates in the
Moderate RC-Responsiveness class had the highest probability of having all three of these
supports.
The final piece of evidence considered in selecting a class solution was the entropy
statistic, which describes the classification quality of an LCA model on a scale from 0 to 1.
Higher levels of entropy are affiliated with better classification, with 0.80 generally considered
to be the lower threshold of acceptability. The entropy was 0.79 for the three-class solution, and
0.85 for the four-class solution, indicating that the latter model was slightly better at assigning
advocates to particular RC-responsive organizational support classes. It may be the case that
avoiding forcing the 13 advocates in Class 4 into one of the three main classifications by

86

providing an additional class may provide better clarity within these classes and allow for a
needed alternative for the advocates who simply don’t fit elsewhere. While replication of a
particular class structure with another data set would provide further insight into the best possible
solution, given the available evidence and the potentially theoretically important nature of the
fourth class, a four class solution was retained. Figure 8 presents a graph of the estimated
probabilities of the occurrence of each support by class, and Table 10 describes the prevailing

Probability that an organizational support is present

characteristics of each class.

1.00

Latent Class
Moderate RCResponsiveness
Fair RC-Responsiveness

0.90

Poor RC-Responsiveness

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

RC-Responsive Organizational Support Number (Corresponds to item number in Table 4)

Figure 8. Estimated Probabilities that Each RC-Responsive Organizational Support is Present for
Advocates in Each Latent Class.

87

88

Table 10.
Characteristics of Each Latent Class of RC-Responsive Organizational Supports (Final Four Class Solution)
Moderate RC-Responsiveness

Fair RC-Responsiveness

Poor RC-Responsiveness

RSH Screening

(10% prevalence)

(35% prevalence)

(51% prevalence)

(4% prevalence)

Information Supports

Information Supports

Information Supports

Private Space

Private Space

Private Space

RSH-inclusive resource lists

RSH-inclusive resource lists

RSH-inclusive resource lists

Information
Supports
Private Space
RSH-inclusive
resource lists

RSH-related informational materials RSH-related informational materials
Medical Contact for Questions
Leadership Supports
New hires receive RSH training
Leadership brings up RSH at
meetings

Leadership Supports
New hires receive RSH training
Leadership brings up RSH at meetings

Procedural Supports

Procedural Supports
Intake Questions
on…
Access to RSH care
Unwanted
pregnancy
Unprotected sex
Scripted tools for
assessing EC needs

Intake Questions on…
Access to RSH care
Unwanted pregnancy
Pregnancy outcome interference
Birth control sabotage
Note: RSH = Reproductive and sexual health; EC= Emergency contraceptive

89

In summary, these results suggest that the sample may indeed be comprised of groups of
advocates from organizations with distinct levels of RC-responsive organizational supports. This
class structure additionally provides support for the use of the organizational support person
measures as a sufficient measure of the construct, and vice versa. As was discussed in the results
of the Rausch Measurement analysis, several organizational supports appear to cluster around a
few levels of “difficulty.” This pattern suggests a step by step movement through these supports
wherein organizations are more able to offer particular services at the same time and before
other, less common services. An examination of the Wright map for this measure reveals that
these “steps” roughly correspond to the three major classes identified in the LCA. First, we see a
clustering of private space and RSH items on resource lists at the bottom of the map,
corresponding with the Poor RC-responsiveness class. Second, we see a clustering of leadership
bringing up RC, training on RC, and RC informational materials toward the bottom/mid portion
of the map, roughly corresponding with the characteristics of the Fair RC Responsiveness class.
Finally, a clustering of the intake and healthcare worker connections toward the mid portion of
the map generally corresponds with the Moderate RC Responsiveness class. The duplication of
these patterns in the two measurement approaches provides flexibility for researchers who may
wish to use one or the other in their analyses.
Latent class probabilities as predictors of practice. To test whether advocates’
probability of being in a particular RC-responsive organizational support class would be
associated with more frequent universal and targeted RC practice, the modified BCH method for
Mplus (BCH; Bakk & Vermunt, 2014, as cited in Asparouhov and Muthen, 2015) was used. The
BCH method allows for the prediction of a distal outcome by an LCA, without the results of the
latent class variable being impacted by the other variable(s). As such, the latent class

90

probabilities are still based on the original class indicators (the dichotomized organizational
supports), despite other variables being included on the model. As described by Asparouhov and
Muthen (2015, p.3),
“The BCH method uses weights wij which reflect the measurement error of the latent
class variable. In the estimation of the auxiliary model, the i-th observation in class/group
j is assigned a weight of wij and the auxiliary model is estimated as a multiple group
model using these weights.”
In order to use this method to predict the person measures for both universal and targeted
RC practice using the LCA, initially the automatic version of the BCH procedure was
implemented in Mplus (Version 7.4, Muthen & Muthen, 2012).. This procedure first estimates
the latent class model using the 15 binary indicator variables, and then evaluates the mean of
universal or targeted practice across the different classes using the approach of Bakk and
Vermunt (2014, as cited in Asparouhov and Muthen, 2015). The results demonstrated that the
mean of the Universal RC-responsive person measures was significantly higher in the Moderate
responsiveness class compared to the Fair responsiveness (p < 0.05) and the Poor responsiveness
classes (p<0.001). Likewise, the Fair responsiveness class’s mean was significantly greater than
the Poor responsiveness class’s mean (p<0.001). Interestingly, the RSH screening class had
significantly higher frequency of Universal Practice compared to the Poor RC-responsiveness
class (p<0.05), but did not significantly differ from either the Fair or Moderate classes (see Table
11 for means and standard deviations).

91

Table 11.
Means and Standard Errors for Universal RC Practices and Targeted RC Practice Person
Measures by Latent Class Probabilities

Universal RC Practice

Moderate
Responsiveness

Fair
Responsiveness

Poor
Responsiveness

RSH Screening

-0.73 (0.24)

-1.43 (0.18)

-2.73 (0.15)

-1.54 (0.59)

Universal RC Practice,
controlling for
Comfort Variables

-1.60 (0.30)

-2.05 (0.21)

-3.20 (0.15)

-2.07 (0.58)

Targeted RC Practice

-0.57 (0.31)

-1.28 (0.25)

-2.57 (0.28)

2.15 (0.57)

In order to test if RC-responsive organizational support class significantly predicted
Universal RC Practice person measures, controlling for Comfort Discussing Sexuality and
Comfort Discussing Reproductive Health, the manual version of the BCH procedure was
implemented in Mplus (Version 7.4, Muthen & Muthen, 2012). In the first step of this procedure,
a four class model was estimated using only the 15 binary indicator variables; the resulting BCH
weights were saved to a data set with the other variables of interest. In step two, several models
were estimated using the BCH weights as training data. As the class specific intercepts show the
influence of the latent class variable on the practice outcome beyond the influence of the
covariate comfort variables, the initial model constrained the intercepts of Universal Practice to
be equal across all classes. The Wald Chi-Square Test of parameter constraints revealed a
significant difference between the constrained and unconstrained models (χ2 = 36.47, df=3,
p<0.001) suggesting there is a difference among classes even when controlling for the other
predictors. Next, separate models were run to constrain two class intercepts of Universal Practice
to be equal at a time (a total of six models) to elucidate which classes differed from one another.
The Wald Chi-Square test was significant in two of the six models, revealing that both Moderate
92

& Fair-Responsiveness Classes had significantly higher Universal RC Practice scores than the
Poor-Responsiveness class, when controlling for the comfort variables. Several significant
differences identified in the previous automatic BCH procedure were no longer present once
these covariates were controlled for. However, it is notable that the model examining the PoorResponsiveness class and the RSH Screening class was nearing significance (p=0.058; See Table
11 for means and standard deviations).
These patterns varied somewhat for Targeted RC practice. Using the automatic version of
the BCH procedure, the RSH screening class had a significantly higher Targeted RC practice
person measure mean than the other three classes (p<0.001). Both the Moderate responsiveness
and the Fair responsiveness classes had significantly higher Targeted RC practice score means
than the Poor responsiveness class (p<0.001), but the Moderate class no longer had significantly
greater scores than the Fair responsiveness class (p=0.08). This may be in part due to the
changes in sample size for the Targeted RC Practice analyses, as participants only had scores for
this measure if they believed they had worked with a survivor of RC in the last year (n=182,
down from n= 338; See Table 11 for means and standard deviations). An attempt to control for
the comfort covariates using the manual version of the BCH procedure described above was
unsuccessful, as the sample covariance of the Comfort Discussing Reproductive Health variable
in the RSH Screening class is singular. As such, a model controlling for the comfort covariates
was not identifiable and could not be estimated.
In summary, these results suggest that the “RC-responsiveness class” to which an
organization belongs significantly impacts the frequency with which its advocates engage in RCrelated practices, regardless of how comfortable they are talking about sex or reproductive
health. Simply having private space and a resource list with RSH services is not sufficient to

93

support advocates in assessing for RC, educating survivors on RC, or incorporating RSH
considerations into their advocacy.
Assigned class as a predictor of practice. A simplified approach to examining group
differences by class is to use assigned class as a stand-in for the more complex probabilities used
in the previous analyses. In this case, “assigned class” refers to a categorical variable with values
(1, 2, 3, or 4) assigned to each advocate based their most likely RC-responsiveness latent class.
While this approach is less precise, it is a desirable stand-in in certain cases. In order to examine
whether or not this simplified measurement approach is reasonable for the field, a test of the
extent to which our results were replicated using assigned class was undertaken.
An ANOVA examining differences in Universal RC practice by assigned class found a
roughly similar mean pattern to that produced by the automatic BCH procedure, but only
identified half of the significant differences. The ANOVA found significant differences between
the Poor Responsiveness class and both the Moderate and Fair Responsiveness classes, but did
not identify significant differences between the Moderate and Fair classes or between the RSH
Screening class and any other class. The Targeted RC Practices ANOVA was likewise impacted.
Again, the pattern in the means was similar, but the mean for the RSH Screening class was much
smaller in the ANOVA compared to the BCH analysis (1.77 v. 2.15). Unsurprisingly, the RSH
Screening class no longer had significantly greater Targeted RC Practice scores than the
Moderate or Fair Responsiveness class, but did retain its significant difference from the Poor
Responsiveness class (p<0.05). The relationships among the other classes remained largely
unchanged with both the Moderate Responsiveness and the Fair Responsiveness classes having
significantly higher Targeted RC practice score means than the Poor Responsiveness class, but
finding no significant differences between the Moderate and Fair Responsiveness classes

94

themselves. These results demonstrate that the assigned class approach, while not as precise a
measure as the weights used in the BCH analysis, may be sufficient for studies examining
differences between “poor responsiveness” organizations and either “fair” or “moderateresponsiveness” organizations. However, it is likely not sensitive enough to test for differences
between “fair” and “moderate” classes, or to examine the RSH Screening class at all.
Hypothesis 4
The fourth hypothesis predicted that the combination of comfort discussing sexuality,
comfort discussing reproductive health, and organizational supports would together significantly
predict both Universal RC Practice and Targeted RC Practice. Prior to conducting the
hierarchical multiple regressions, two power analyses to test the sensitivity of the models were
completed using the “Linear multiple regression: Fixed model, R2 increase” statistical test in
g*power (Faul, Erdfelder, Buchner, & Lang, 2009). For a regression with 3 independent
variables, 2-tailed p< .05, and power of .80, the minimum detectable effect size (f2) for an
independent variable in the Universal RC Practices model (N=324) is 0.02, and for the Targeted
RC Practices model (N=171) is 0.05.
Next, relevant assumptions of hierarchical regression were tested. An examination of
correlations (see Table 12) revealed that none of the independent variables were highly
correlated. Additionally, the Tolerance and VIF statistics were all within acceptable limits,
suggesting the assumption of multicollinearity was met. As all variables used were person
measures resulting from Rasch Analysis, no tests of univariate outliers were conducted. An
examination of the Mahalonobis distance scores indicated there were two multivariate outliers,
which were removed from the analysis. The skewness and kurtosis of the independent variables
were all within an acceptable range, and an examination of scatter plots of the standardized

95

residuals indicated the assumptions of linearity and homoscedasticity were all satisfied.

Table 12.
Bivariate Correlations Between Study Scale Person Measures
Comfort
Discussing
Sexuality

Scale
Comfort Discussing
Sexuality with Clients
Comfort Discussing
Reproductive Health
with Clients

-

Comfort
RCUniversal RC Targeted RC
Discussing
Responsive
Practice
Practice
RH
Org Supports
0.51

0.36

0.35

0.30

-

0.31

0.26

0.19

-

0.74

0.41

-

0.43

Universal Reproductive
Coercion Practice
Targeted Reproductive
Coercion Practice
RC-Responsive
Organizational Supports

-

Note: All correlations were significant at the 0.01 level (2-tailed).

A three stage hierarchical multiple regression was conducted with Universal RC Practices
as the dependent variable. Comfort Discussing Sexuality with Clients was entered at stage one of
the regression. Comfort Discussing Reproductive Health was entered at stage two, and RCResponsive Organizational Support was entered at stage three. The independent variables were
entered in this order to test the extent to which the organizational-level element impacts practice
beyond the individual-level elements. The hierarchical multiple regression revealed that at Stage
One, Comfort Discussing Sexuality significantly predicted Universal RC Practice, F (1, 323) =
46.54 p < .001, R2= 0.13. At Stage Two, Comfort Discussing Reproductive Health significantly
predicted Universal RC Practice above and beyond Comfort Discussing Sexuality, with a
significant change in variance from Stage One (R2 change=0.03, p<0.01). Adding RC-

96

Responsive Organizational Supports to the model in Stage Three likewise revealed that this
variable contributed a significant amount of unique variance beyond Step Two (R2 change= 0.10,
p< 0.001). All three independent variables remained significant predictors of Universal RC
Practice (p< 0.01) in this final stage, and together accounted for 25% of the variance. The
standardized path coefficients are presented in Figure 9.

Comfort
Discussing
Sexuality

Comfort
Discussing
Repro Health

0.17
95% CI [0.05, 0.24]

0.16
95% CI [0.04, 0.21]

Frequency of
Universal RC
Practice

0.34
95% CI [0.22, 0.41]
RCResponsive
Org. Supports

Figure 9. Standardized Regression Path Coefficients for Person Measures as Predictors of
Universal Reproductive Coercion Practice.

A similar three stage hierarchical multiple regression was conducted with Targeted RC
Practices as the dependent variable. Comfort Discussing Sexuality with Clients (person
97

measures) was entered at stage one of the regression, Comfort Discussing Reproductive Health
was entered at stage two, and RC-Responsive Organizational Support was entered at stage three.
The hierarchical multiple regression revealed that at Stage One, Comfort Discussing Sexuality
with Clients significantly predicted Targeted RC Practice, F (1, 171)= 24.47, p < .001. Comfort
discussing sexuality explained a significant proportion of variance in Universal RC practice, R2=
0.17. At Stage Two, Comfort Discussing Reproductive Health significantly predicted Targeted
RC Practice above and beyond Comfort Discussing Sexuality, with a small but significant
change in variance from Stage One (R2 change=0.02, p=0.05). Adding RC-Responsive
Organizational Supports to the model in Stage Three revealed that this variable contributed a
significant amount of unique variance to the outcome variable beyond the predictors in Step Two
(R2 change= 0.11, p< 0.001). All three independent variables remained significant predictors of
Targeted RC Practice (p< 0.01) in this final stage, and together accounted for 25% of the
variance. The standardized path coefficients are presented in Figure 10.

98

Comfort
Discussing
Sexuality

Comfort
Discussing
Repro Health

0.18
95% CI [0.03, 0.33]

0.16
95% CI [0.01, 0.27]

Frequency of
Targeted RC
Practice

0.35
95% CI [0.22, 0.52]
RCResponsive
Org. Supports

Figure 10. Standardized Regression Path Coefficients for Person Measures as Predictors of Targeted
Reproductive Coercion Practice.

99

DISCUSSION
Awareness about the phenomenon of RC and how domestic violence programs should
respond to it slowly growing. However, the prospect of engaging with clients around RSH health
can be daunting for many advocates and organizations. Despite the challenges that may come
with incorporating more RC-responsive practices and policies into services for survivors, the
negative impacts of such victimization on women’s lives demands a robust response from the
field. This is particularly important given the disproportionate experiences of RC among already
marginalized groups, including women of color and sexual minorities (Holliday et al., 2017;
McCauley et al., 2015). As training and technical assistance organizations across the country
begin to formulate and refine their recommended approaches to building RC-responsive capacity
within programs, it is critical that they prioritize interventions that target those barriers and
facilitators that most greatly impact frequency of RC-responsive of universal and targeted RCresponsive practice. By focusing efforts on those factors, DV programs and their staff will
ultimately be better prepared and better able to support survivors of RC in regaining their
reproductive control.
The current study was the first to investigate the effect of three common barriers and
facilitators to RSH practice in the social work and nursing fields in the DV service context—
comfort discussing sexuality with clients, comfort discussing reproductive health with clients,
and the presence of relevant organizational supports. Results of the study demonstrated that each
of these factors was significantly related to frequency of RC-responsive practice. This finding
lends support to the application of these barriers and facilitators from the social work and nursing
literatures into the DV service context.

100

Overall, advocates reported feeling quite comfortable discussing both sexuality and
reproductive health. While they reported fairly high levels of both types of comfort, these
variables were only moderately related, suggesting that each has a distinct place in promoting
advocates’ ability to engage in RC-responsive practice. As such, intervention developers aiming
to improve advocates’ comfort should take both facets into consideration, designing training and
coaching approaches that target the underlying fears, beliefs, and knowledge gaps that may
differentially drive discomfort addressing sexuality and discomfort discussing reproductive
health with clients.
Additionally, the level of comfort discussing reproductive health reported by advocates
varied by topic. For example, participants were less comfortable discussing pregnancy
termination with clients, and were more comfortable discussing emergency contraception. While
discomfort discussing pregnancy termination was to be expected given its politically, morally,
and socially charged nature, advocates’ relative comfort discussing EC was somewhat surprising.
Previous research with social workers has found that these professionals often don’t understand
how EC works (Bell & Rubin, 2013), have misconceptions about it contributing to riskier sexual
behavior (Miller, et al., 2012), and lack knowledge about how to access EC (Bell & Rubin,
2013). Such factors would likely contribute to less comfort discussing this contraception option
with clients. This result could point to a special willingness to discuss EC among often politically
liberal and social justice-oriented DV advocates, or to other fundamental differences between the
social work and domestic violence fields. This finding may also signal a growth in understanding
of EC in general since the previous studies were conducted, or could even be the result of a
sample-specific phenomenon. Further research is needed to determine which explanation holds
true.

101

The presence of RC-responsive organizational supports was the most influential factor on
frequency of RC practice, yet most advocates reported that their agencies did not have the
overwhelming majority of RC supports measured. Indeed, over half of the sample was identified
as having “poor RC-responsiveness” in the class analysis, suggesting that most organizations
studied (and perhaps in the greater field) were not providing adequate supports for their staff to
engage in RC practice. This finding stands in contrast to the very high levels of comfort reported
by many advocates, and suggests that organizational resources may in fact be the area with
greater potential for growth and improvement in the field compared to individual training.
Indeed, interventions focused on increasing the availability of RC-responsive organizational
supports in DV agencies (such as the reorganization or introduction of human, physical, temporal
or financial resources) are likely the next frontier for improving RC-responsiveness among
advocates. While initial guidance exists on best practices therein (Duplessis & Levenson, 2014),
anecdotal reports suggest that interest in such changes lags behind willingness to engage in
individual-level advocate trainings, perhaps due to the more complex nature of organizationallevel change.
This study was the first to examine potential barriers and facilitators to reproductive
coercion-responsive practice on multiple ecological levels. The results of the study support the
assertion that intervention on both the individual level (comfort) and organizational level
(organizational RC-responsive supports) can significantly impact advocates’ practice for the
better, and that a combination of strategies across levels has the greatest potential for successful
change in professionals’ behavior (Seidman & Tseng, 2011). The study supports these points in
two main ways.

102

First, advocates generally reported high levels of comfort discussing sexuality and
reproductive health topics with clients, but very few reported regularly engaging in RC-related
skills with survivors. This dearth of RC practice was even the case when advocates were
working with survivors they believed to be victims of RC. In other words, despite feeling
comfortable discussing these topics with clients, advocates were rarely actually bringing them up
in their practice. These results suggest that something beyond these individual-level factors may
need to be present to more fully support advocates’ RC practice, a proposal that is aligned with
the implementation literature (e.g. Aarons, Hurlburt, & McCue Horwtiz, 2011; Glisson,
Schoenwald, Hemmelgarn, Green, Dukes, Armstrong, & Chapman, 2010; Meyers, Durlak, &
Wandersman, 2012).
Second, the study found that the level of organizational support an advocate had around
RC was more strongly related to how frequently they did RC-related practices than individual
comfort. In fact, no matter how comfortable or uncomfortable an individual advocate was, if she
had certain RC supports in place she was much more likely to be RC-responsive in her practice.
The opposite held true as well—advocates generally engaged in RC practice less frequently if
they didn’t have organizational resources to support them. This finding is aligned with
ecological thinking in the organizational change literature, which asserts that without supportive
organizational structures and leadership, the desired practices are unlikely to be successfully
installed and maintained (Fixsen, et al. 2005).
Identifying Key Levers for Change
While incorporating multiple strategies may be desirable, learning and integrating a large
number of changes can be taxing for both individual practitioners and organizations. Indeed, a
major challenge for implementation specialists is to identify interventions that facilitate

103

implementation of desired practices while minimizing barriers to implementation (Goldman et
al., 2001). In other words, the cure can become the disease, with intervention supports potentially
further hindering incorporation of the desired practice by overloading staff. The study findings
highlight naturalistic combinations of practices currently emerging in the field that may provide
guidance for leadership hoping to identify key changes to implement first.
The study suggests that there may be a minimum level of supports that must be present in
order to promote greater RC-responsive practice among advocates. Simply having a private
space to talk and a resource list that includes RSH-related agencies (the characteristics of the
Poor-responsiveness class) does not appear to be distinctly different than doing nothing at all.
Rather, an organization must additionally have at a minimum the characteristics of the Fairresponsiveness class in order to see preliminary benefits to their advocates’ frequency of RC
practice. In other words, leadership must be willing and able to:
1) require staff to attend trainings on RC specifically and RSH broadly;
2) offer private places to screen and talk with clients about health concerns;
3) provide and openly display informational materials on RC, such as posters,
brochures, and/or cards about reproductive/sexual health;
4) include clinical referrals/resources on their resource list for survivors who want to
access reproductive/sexual health care; and
5) raise RSH-related issues during meetings with staff.
Interestingly, significant differences were not consistently identified between the
Moderate- and Fair-Responsiveness classes. Despite the advocates from organizations in the
Moderate-responsiveness class engaging in RC practice slightly more frequently, these
differences were not large enough to be statistically significant. Further research is needed with

104

more organizations in the Moderate-Responsiveness class to determine if this was a result of the
small class size or not, and to interpret the meaning of the fourth class (RSH Screening).
Lastly, there was no “High-Responsiveness” class identified, as the Moderate class was
characterized by having only two-thirds of the fifteen organizational supports measured. The five
supports that were unlikely to be present were as follows: 1) A question about unprotected sex on
intake form; 2) Universal screening for RC; 3) Scripted tools to assess for EC need; 4) EC
available on site; and 5) Pregnancy tests available on site. These organizational supports
represent two important next steps in RC-responsiveness that innovators in the field are working
to implement: a) Universal screening and b) having over-the-counter reproductive heath tools
available on site. Organizations that implement universal screening incorporate screening and
education tools into their intake process with every survivor they see. This approach is
encouraged for a variety of reasons: survivors may not even recognize that RC is a type of abuse;
survivors might be hesitant to bring it up even though they want to speak about it; and survivors
may take the RC information and share with others in their life who may be experiencing this
type of abuse even if they themselves are not. The more advocates speak about RC as part of
their general practice, the more normalized such discussions will become among staff, clients,
and hopefully among the wider community. Providing over-the-counter reproductive health
tools on site, such as emergency contraception and pregnancy tests, also represents a
normalization of reproductive health concerns among clients. Just as organizations may have
materials present to help survivors respond to common legal, housing, or human services-related
needs, so too should these agencies have reproductive health resources readily available.
Likewise, as EC and pregnancy tests are available over-the-counter, agencies should not
experience any legal problems with making these medications available to survivors engaging

105

with their program. Only when organizations are able to incorporate these elements in to their
institutional supports will they reach the point of “High-Responsiveness.”
The study has potential implications for informing existing approaches to promoting RCresponsive capacity among organizations and advocates. As was described earlier, the results
suggest that a multi-level approach to such interventions may provide the most leverage for
changing the behavior of advocates. As such, agencies invested in responding to the needs of
survivors of RC may not be providing sufficient resources for their advocates by solely offering
trainings focused on improving comfort or other individual-level factors. The study findings
provide guidance for leadership hoping to identify key internal changes to tackle first. Results
suggest leadership must be able and willing to at a minimum those practices present in the “Fair
RC-Responsiveness” class described above. Organizations offering RC training and technical
assistance should share this minimum service threshold with their clients to ensure that the
services they are providing will make as significant an impact as possible.
Training organizations and the DV agencies implementing changes should also be
encouraged that additional supports above and beyond the basic necessities outlined above do
appear to improve advocates’ frequency of RC, although the degree to which this is the case
remains to be seen. Additional supports that may encourage practice include the incorporation of
RC-specific questions on intake forms, and the formation of relationships with a health educator,
nurse, PA or doctor that staff can contact with questions about reproductive and sexual health.
Such multi-level organizational changes to processes and structures could both facilitate and
reinforce the alterations in individual comfort that training and coaching may elicit for
employees, paving the way to higher levels of practitioner fidelity to the desired practice (Fixsen,

106

et al 2005). Additional research is needed to identify the maximum threshold of new supports
after which staff may become overloaded, impairing implementation.
Study Limitations
There are a number of methodological limitations that must be considered when
interpreting the results of the study, including the recruitment strategy, the relatively
homogenous sample, and the nature and administration of the measures. While these limitations
do raise some questions about the representativeness of the sample and strength of the study’s
conclusions, they are well within the scope of limitations common to preliminary studies into a
new line of inquiry.
Although the sample was relatively large and nationwide, it was a convenience sample
and not nationally representative. While a variety of states and organizational settings were
represented, the settings may not have been representative of DV service settings as a whole.
Advocates were invited to participate through their state domestic violence coalition, and some
coalitions may have been more effective at recruiting than others. In addition, not all domestic
violence organizations are members of their state coalitions and therefore might not have been
invited to participate. Additionally, it is possible that advocates from the same agency may have
participated in the survey, but we do not know the extent to which that occurred. (Advocates
were not asked which agency they represented in order to maintain anonymity.) This could have
resulted in unaddressed interdependence among those observations, and could inflate the
organizational measures. Future studies should attempt to account for this possibility by
collecting agency-level data in a manner that is responsive to participants’ need for anonymity
when completing these potentially sensitive measures, or by collecting a larger sample that
would allow for statistical corrections for this possibility.

107

This study also lacks diversity among advocate respondents. The sample consisted
primarily of heterosexual, White, middle-aged females, which—while representative of many
domestic violence staff across the country—did preclude examining differences in attitude or
behavior by singular racial groups, ethnicity, gender, or sexual orientation. Expanding upon this
study’s findings should include specific efforts to broaden sample diversity. Furthermore,
advocates were intentionally not told the topic of the survey a priori, and while that is a strength
in reducing bias, it is unclear whether this would have hampered or helped recruitment.
The study relied on retrospective self-report measures that were developed by the
research team, as there are no existing validated scales in the area of comfort, practice, or
organizational supports related to RC service provision, and no scales measuring attitudes toward
reproductive and sexual health service provision that have been validated for DV advocates.
While self-report measures have a variety of potential limitations, including the possibility of
bias from participants’ inaccurate memory or understanding of their beliefs or behaviors, or
pressure to respond in a socially desirable manner (Bachrach, et al., 1999), these are common
weaknesses of studies that seek to extend the borders of our working knowledge. Further
refinement of these measures, as well as triangulation of the self-report data using client reports
or reviews of agency policies, could improve future explorations in this arena.
This survey was conducted online—requiring participants to have a minimum level of
computer literacy, as well as internet access. In addition, this survey was not translated into
additional languages beyond English—which may have eliminated non-English-speaking
advocates from participating. Taking steps to improve the accessibility of the survey to better
incorporate these groups—whether by providing translation or allowing survey administration
via telephone—could break down these barriers to participation. Despite these limitations, this

108

study offers invaluable preliminary insights into DV advocates’ perspectives, practices, and
resources so that programs can offer the best possible services to survivors of RC.
Future Directions
The results of this study provide a strong foundation upon which to build, but further
research is needed in order to support DV service organizations and staff in effectively
integrating RC-responsive practices. While comfort discussing sexuality and reproductive health,
and the presence of RC-responsive organizational supports each significantly predicted
frequency of universal and targeted RC-responsive practice, the combined model only explained
a quarter of the variance in each of these outcomes. This suggests that further research is needed
to understand other barriers and supports that may impact advocates’ behaviors. Frameworks for
implementing evidence-based practices in public service sectors (e.g. Aarons, Hurlburt, &
McCue Horwitz) could point researchers toward other elements to examine that typically impact
the successful uptake of an intervention, including other “inner context” elements within the
intrapersonal, interpersonal and organizational ecological levels of DV agencies (e.g. advocates’
RC-related knowledge, survivors’ responses to RC-related discussions, or other organizationallevel barriers to practice) and “outer context” elements in the ecological levels of the broader
community and world (e.g. sex-phobic community attitudes, collaboration or competition with
other service providers, or funder-level barriers). Likewise, research on how the barriers and
facilitators impact each other, particularly across ecological levels, could further develop our
understanding of how these factors interact. Such an understanding would inform the
development of future interventions with an intended ripple effect across the whole ecological
system (Kelly, 1971).

109

Future research could also include interviews or focus groups with advocates to explore
the results of this study in depth. Greater context to explain why providers are hesitant to address
RC, as well as additional information on other barriers that advocates face, would further enrich
the survey findings and inform future quantitative efforts. Likewise studies could also examine
the extent to which advocate training on RC promotes individuals’ RC-responsive practices and
the degree to which such practices promote improved survivor outcomes. This study could be
employed across a variety of settings with differing organizational supports and barriers to
explore the extent to which such institutional-level factors impact the effectiveness of training or
practice in supporting survivors.
Finally, while existing IPV-specific intervention materials provide a helpful introduction
to the topic and a brief set of practice guidelines, further development therein is needed to ensure
all critical barriers are addressed. In addition to drawing on best practices in organizational
change, developers should likewise actively involve survivors and IPV providers in this
intervention development process in order to find and build upon successful indigenous
approaches (Miller & Shinn, 2005). Finally, further testing of these approaches is necessary to
demonstrate their effectiveness. Given the risks that survivors of RC face to their RSH and the
unique position of IPV service providers to intervene therein, the need for evidence-informed
approaches that will help agencies make the greatest possible impact is great.
Conclusion
Reproductive coercion is a newly acknowledged but commonly experienced form of
intimate partner violence with serious potential consequences for women’s health and wellbeing. By normalizing and attending to clients’ experiences of RC and connecting them with
needed services and resources, DV advocates can provide much-needed support for survivors of

110

this type of abuse. Despite having the unique opportunity to intervene in this area as part of their
routine work with clients, few advocates report regularly engaging in RC-related practices. In
order to better understand the factors that may be impeding advocates’ RC-responsiveness, the
study sought to identify critical barriers and facilitators to RC-responsive practice on the
intrapersonal and organizational ecological levels of DV organizations.
The findings of this study have implications for the leaders of DV service agencies.
While intrapersonal factors (levels of comfort discussing sexuality and comfort discussing
reproductive health) influenced advocates’ frequency of universal and targeted RC practice, the
level of RC-responsive supports provided by their organizations was actually much more
impactful. While the ideal maximum level of organizational supports is still up for debate, the
current study outlined a minimum level of supports that may be necessary for improved
frequency of RC practice. This guidance may be useful for agencies looking to use limited
resources strategically in future RC-responsiveness efforts. Additionally, the study results
reinforce the validity of employing an ecological approach to understanding barriers and
facilitators to RC-responsiveness, suggesting that taking a multi-level change approach may
indeed be most beneficial for understanding change in individual behavior. Finally, the study laid
the groundwork for future research in the area of RC-responsive practices in the DV service
setting. While much additional work is needed to understand the factors that contribute to more
frequent RC practice and the degree to which such practices impact survivors’ health and wellbeing outcomes, the present study lays the groundwork for a myriad of exciting future
examinations.

111

REFERENCES

112

REFERENCES

Aarons, G. A., Hurlburt, M., & Horwitz, S. M. (2011). Advancing a conceptual model of
evidence-based practice implementation in public service sectors. Administration and
Policy in Mental Health and Mental Health Services Research, 38(1), 4-23.
Akers, A.Y., Gold, M.A., Borrero, S., Santucci, A., & Schwarz, E.B. (2010). Providers'
perspectives on challenges to contraceptive counseling in primary care settings. Journal
of Women's Health, 19(6), 1163-1170.
Akinci, A.C. (2011). The comfort levels of nurses during clinical experiences which include
sexual topics. Sexuality and Disability, 29(3), 239-250.
Algier, L. & Kav, S. (2008). Nurses’ approach to sexuality-related issues in patients receiving
cancer treatment. Turkish Journal of Cancer, 38(3), 135–141.
Alzate, M. (2009). The role of sexual and reproductive rights in social work practice. Affilia:
Journal of Women and Social Work, 24(2), 108-119.
Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture modeling: Using the BCH
method in Mplus to estimate a distal outcome model and an arbitrary secondary
model. Mplus Web Notes, 21(2).
Ayaz, S. (2013). Sexuality and nursing process: a literature review. Sexuality and Disability,
31(1), 3-12.
Bachrach, C. A., Cain, V. S., Jobe, J. B., Kurtzman, H. S., Stone, A. A., & Turkkan, J. S. (1999).
The Science of Self-Report: Implications for Research and Practice. LEA, Mahwah, NJ,
29-47.
Bal, M.D. & Sahiner, N.C. (2015). Turkish nursing students’ attitudes and beliefs regarding
sexual health. Sexuality and Disability, 33(2), 223-231.
Begun, S., Bird, M., Ramseyer Winter, V., Massey Combs, K., & McKay, K. (2016). Correlates
of Social Work Students' Abortion Knowledge and Attitudes: Implications for Education
and Research. Social Work in Public Health, 31(4), 276-287.
Bell, M. & Rubin, D. (2013). Social Workers and Family Planning Knowledge: Are We Still
Gatekeepers?. Journal of Baccalaureate Social Work, 18( 1-17).
Bennett, C.C., Anderson, L.S., Cooper, S., Hassol, L., Klein, D.C., & Rosenblum, G. (1966).
Community psychology: A report of the Boston Conference on the education of
psychologists for Community Mental Health (pp. 1-30). Boston: Boston University.

113

Bergmann, J. & Stockman, J. (2015). How does intimate partner violence affect condom and oral
contraceptive Use in the United States? Contraception, 438-455.
Black, M.C., Basile, K.C., Breiding, M.J., Smith, S.G., Walters, M.L., Merrick, M.T., Chen, J.,
& Stevens, M.R. (2011). The National Intimate Partner and Sexual Violence Survey
(NISVS): 2010 Summary Report. Atlanta, GA: National Center for Injury Prevention and
Control, Centers for Disease Control and Prevention.
Boone, W. J. (2016). Rasch Analysis for Instrument Development: Why, When, and
How?. CBE-Life Sciences Education, 15(4), rm4.
Boone, W. J., Staver, J. R., & Yale, M. S. (2014). Rasch analysis in the human sciences (pp. 6991). Dordrecht: Springer.
Borrero, S., Nikolajski, C., Steinberg, J.R., Freedman, L., Akers, A.Y., Ibrahim, S., & Schwarz,
E.B. (2015). It just happens: A qualitative study exploring low-income women’s
perspectives on pregnancy intention and planning. Contraception, 91, 150–156.
Bouffard, S., & Little, P. (2004). Promoting Quality through Professional Development: A
Framework for Evaluation. Issues and Opportunities in Out-of-School Time Evaluation.
Number 8. Harvard University Harvard Family Research Project.
Bulow, S.M. (2012). Initiation of sexual discussion in social work practice with cancer patients:
A nationwide survey of social workers. (Doctoral dissertation). Retrieved from ProQuest
Dissertations & Theses Global. (Order No. 3541259).
Burton, C.W. & Carlyle, K.E. (2015). Screening and intervening: evaluating a training program
on intimate partner violence and reproductive coercion for family planning and home
visiting providers. Family & Community Health, 38(3), 227-239.
Cattaneo, L.B. & Goodman, L.A. (2015). What is empowerment anyway? A model for domestic
violence practice, research, and evaluation. Psychology of Violence, 5, 84-94.
Chamberlain, L. & Levenson, R. (2012). Addressing Intimate Partner Violence, Reproductive
and Sexual Coercion: A Guide for Obstetric, Gynecologic and Reproductive Health Care
Settings, Second Edition. San Francisco, CA: Futures Without Violence [Online].
Accessed on August, 27, 2012.
Chrisler, J.C. (2014). A reproductive justice approach to women's health. Analyses of Social
Issues and Public Policy, 14(1), 205-209.
Clark, L.E., Allen, R.H., Goyal, V., Raker, C., & Gottlieb, A.S. (2014). Reproductive coercion
and co-occurring intimate partner violence in obstetrics and gynecology
patients. American Journal of Obstetrics and Gynecology, 210(1), 42-e1
Constantine, W.L., Jerman, P., & Constantine, N.A. (2009). Sex education and reproductive
health needs of foster and transitioning youth in three California counties. Center for
Research on Adolescent Health and Development, Public Health Institute. Retrieved on

114

November 10, 2016 from
http://www.researchandevaluationsystems.com/reports/4_full_report_fty_2_ 11_2009.pdf
Dattilo, J. & Brewer, M.K. (2005). Assessing clients’ sexual health as a component of holistic
nursing practice: Senior nursing students share their experiences. Journal of Holistic
Nursing, 23(2), 208-219.
Duplessis, V. & Levenson, R. (2014) Integrating health services into domestic violence
programs: Tools for advocates. San Francisco, CA: Futures Without Violence
Dyer, K. & das Nair, R. (2013). Why don't healthcare professionals talk about sex? A systematic
review of recent qualitative studies conducted in the United Kingdom. The Journal of
Sexual Medicine, 10(11), 2658-2670.
Dyer, K., Aubeeluck, A., Yates, N.L., & Nair, R.D. (2015). A multiple timepoint pre-post
evaluation of a ‘Sexual Respect’ DVD to improve competence in discussing sex with
patients with disability. Sexuality and Disability, 33(3), 385-397.
El-Bassel, N., Gilbert, L., Wu, E., Go, H., & Hill, J. (2005). HIV and intimate partner violence
among methadone-maintained women in New York City. Social Science &
Medicine, 61(1), 171-183.
Ely, G., Flaherty, C., Akers, L.S., & Noland, T.B. (2012). Social work student attitues towards
the social work perspective on abortion. Journal of Social Work Values and Ethics, 9(2),
34–35.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using
G*Power 3.1: Tests for correlation and regression analyses. Behavior Research
Methods, 41, 1149-1160.
Faúndes, A., Duarte, G.A., & Osis, M.J.D. (2013). Conscientious objection or fear of social
stigma and unawareness of ethical obligations. International Journal of Gynecology &
Obstetrics, 123, S57-S59.
Fixsen, D., Blase, K., Metz, A., & Van Dyke, M. (2015). Implementation science. International
encyclopedia of the social and behavioral sciences, 695-702.
Flaherty, C., Ely, G.E., Akers, L.S., Dignan, M., & Bonistall, N.T. (2012). Social work student
attitudes toward contraception and the HPV vaccine. Social work in Health Care, 51(4),
361-381.
Foster, J., Núñez, A., Spencer, S., Wolf, J., & Robertson-James, C. (2015). Sexual safety
planning as an HIV prevention strategy for survivors of domestic violence. Journal of
Women's Health, 25(6), 623-629.
Glisson, C., Schoenwald, S.K., Hemmelgarn, A., Green, P., Dukes, D., Armstrong, K.S., &
Chapman, J.E. (2010). Randomized trial of MST and ARC in a two-level evidence- based

115

treatment implementation strategy. Journal of Consulting and Clinical Psychology, 78
(4), 537–550.
Golden, S. D., & Earp, J. A. L. (2012). Social ecological approaches to individuals and their
contexts twenty years of health education & behavior health promotion
interventions. Health Education & Behavior, 39(3), 364-372.
Goldman, H.H. Ganju, V., Dranke, R.E., Gorman, P., Hogan, M., Hyde, P.S., et al. (2001).
Policy implications for implementing evidence based practices. Psychiatric Services, 52,
1591-1597.
Goodman, L.A. & Epstein, D. (2008). Listening to battered women: A survivor-centered
approach to advocacy, mental health, and justice. Washington, DC: American
Psychological Association.
Gott, M., Galena, E., Hinchliff, S., & Elford, H. (2004). “Opening a can of worms”: GP and
practice nurse barriers to talking about sexual health in primary care. Family
Practice, 21(5), 528-536.
Grace, K.T., & Anderson, J.C. (2016). Reproductive coercion: A systematic review. Trauma,
Violence, & Abuse, 1-20.
Gregory, A., Henry, D. B., Schoeny, M. E., & Metropolitan Area Child Study Research Group.
(2007). School climate and implementation of a preventive intervention. American
journal of community psychology, 40(3-4), 250-260.
Guttmacher Institute. (2016). State policies in brief: An overview of abortion laws. Retrieved on
December 5 from https://www.guttmacher. org/state-policy/explore/overview-abortionlaws
Haboubi, N.H.J. & Lincoln, N. (2003). Views of health professionals on discussing sexual issues
with patients. Disability and Rehabilitation, 25(6), 291-296.
Harris, L.H., Debbink, M., Martin, L., & Hassinger, J. (2011). Dynamics of stigma in abortion
work: findings from a pilot study of the Providers Share Workshop. Social Science &
Medicine, 73(7), 1062-1070.
Hautamäki, K., Miettinen, M., Kellokumpu-Lehtinen, P.L., Aalto, P., & Lehto, J. (2007).
Opening communication with cancer patients about sexuality-related issues. Cancer
Nursing, 30(5), 399-404.
Hess, J. & Rueb, J. (2005). Attitudes toward abortion, religion, and party affiliation among
college students. Current Psychology, 24(1), 24–42.
Higgins, A., Barker, P., & Begley, C.M. (2008). ‘Veiling sexualities’: a grounded theory of
mental health nurses responses to issues of sexuality. Journal of Advanced
Nursing, 62(3), 307-317.

116

Holliday, C., McCauley, H., Silverman, J., Ricci, E., Decker, M., Tancredi, D., Burke, JG,
Documet, P., Borrero, S., & Miller, E. (2017). Racial/ethnic differences in women’s
experiences of reproductive coercion, intimate partner violence, and unintended
pregnancy. Journal of Women’s Health.
Iantaffi, A. (2015). Having “the talk” across disciplines. Sexual and Relationship Therapy, 30(2),
195-195.
Jaarsma, T., Strömberg, A., Fridlund, B., De Geest, S., Mårtensson, J., Moons, P., ... &
Thompson, D.R. (2010). Sexual counselling of cardiac patients: nurses' perception of
practice, responsibility and confidence. European Journal of Cardiovascular
Nursing, 9(1), 24-29.
Joffe, C. (2010). Dispatches from the abortion wars: The costs of fanaticism to doctors, patients,
and the rest of us. Beacon Press.
Jones, K.A., Cornelius, M.D., Silverman, J.G., Tancredi, D.J., Decker, M.R., Haggerty, C.L….&
Miller, E. (2016). Abusive experiences and young women’s sexual health outcomes: Is
condom negotiation self-efficacy a mediator? Perspectives on Sexual and Reproductive
Health, 48(2), 57–64.
Jones, R.K. & Jerman, J. (2014). Abortion incidence and service availability in the United States,
2011. Perspectives on Sexual and Reproductive Health, 46(1), 3-14.
Kazmerski, T., McCauley, H.L., Jones, K., Borrero, S., Silverman, J.G., Decker, M.R.,…&
Miller, E. (2015). Use of reproductive and sexual health services among female family
planning clinic clients exposed to partner violence and reproductive coercion. Maternal
Child Health Journal, 19(7), 1490-1496.
Kim, S., Kang, H.S., & Kim, J.H. (2011). A sexual health care attitude scale for nurses:
Development and psychometric evaluation. International Journal of Nursing Studies,
48(12), 1522-1532.
King, L., Miree, C.A., Wilson, C., & Clayton, H. (2008). Oncology nurses' perceptions of
barriers to discussion of fertility preservation with patients with cancer. Clinical Journal
of Oncology Nursing, 12 (3), 467.
Kotronoulas, G., Papadopoulou, C., & Patiraki, E. (2009). Nurses’ knowledge, attitudes, and
practices regarding provision of sexual health care in patients with cancer: critical review
of the evidence. Supportive Care in Cancer, 17(5), 479-501.
Krebs, L.U. (2007). Sexual assessment: research and clinical. Nursing Clinics of North America,
42(4), 515-529.
Lavie-Ajayi, M. (2016). ‘It Will Continue to Embarrass Me on Some Level, and I Think that’s
OK’: Conceptualising Embarrassment in Discussions about Sex between Social Workers
and Service Users. British Journal of Social Work, 0, 1-18.

117

Lavin, M., & Hyde, A. (2006). Sexuality as an aspect of nursing care for women receiving
chemotherapy for breast cancer in an Irish context. European Journal of Oncology
Nursing, 10(1), 10-18.
Linacre, J. M. (2012). Winsteps®(Version 3.75. 0)[Computer Software]. Beaverton, Oregon:
Winsteps. com. Retrieved January 1, 2012.
Magnan, M.A., & Norris, D.M. (2008). Nursing students' perceptions of barriers to addressing
patient sexuality concerns. The Journal of Nursing Education, 47(6), 260-268.
Magnan, M.E., Reynolds, K.E., & Galvin, E.A. (2005). Barriers to addressing patient sexuality
in nursing practice. Medsurg Nursing, 14(5), 282.
Mahieu, L., Van Elssen, K., & Gastmans, C. (2011). Nurses’ perceptions of sexuality in
institutionalized elderly: A literature review. International Journal of Nursing
Studies, 48(9), 1140-1154.
McCauley, H. L., Silverman, J. G., Jones, K. A., Tancredi, D. J., Decker, M. R., McCormick, M.
C., ... & Miller, E. (2017). Psychometric properties and refinement of the Reproductive
Coercion Scale. Contraception, 95(3), 292-298.
McCauley, H.L., Dick, R.N., Tancredi, D.J., Goldstein, S., Blackburn, S., Silverman, J. G., ... &
Miller, E. (2014). Differences by sexual minority status in relationship abuse and sexual
and reproductive health among adolescent females. Journal of Adolescent Health, 55(5),
652-658.
McCauley, H.L., Silverman, J.G., Decker, M.R., Agénor, M., Borrero, S., Tancredi, D.J., ... &
Miller, E. (2015). Sexual and reproductive health indicators and intimate partner violence
victimization among female family planning clinic patients who have sex with women
and men. Journal of Women's Health, 24(8), 621-628.
McKay, K. (2015). Sexuality education within masters of social work programs. (Doctoral
dissertation). Retrieved from ProQuest Dissertations & Theses Global. (Order No.
3712681).
Meadows, D. H., & Wright, D. (2008). Thinking in systems: A primer. chelsea green publishing.
Meyers, D. C., Durlak, J. A., & Wandersman, A. (2012). The quality implementation framework:
A synthesis of critical steps in the implementation process. American journal of
community psychology, 50(3-4), 462-480.
Miller E., Decker M.R., Reed E., Raj A., Hathaway J.E., & Silverman J.G. (2007). Male partner
pregnancy-promoting behaviors and adolescent partner violence: Findings from a
qualitative study with adolescent females. Ambulatory Pediatrics, 7(5), 360-366.
Miller, E., & Silverman, J. (2010). Reproductive coercion and partner violence: Implications for
clinical assessment of unintended pregnancy. Expert Review of Obstetrics &
Gynecology, 5(5), 511-515.

118

Miller, E., Decker, M., McCauley, H., Tancredi, D., Levenson, R., Waldman, J., . . . Silverman,
J. (2010). Pregnancy coercion, intimate partner violence and unintended
pregnancy. Contraception, 81(4), 316-322.
Miller, E., Decker, M.R., McCauley, H.L., Tancredi, D.J., Levenson, R.R., Waldman, J., ... &
Silverman, J.G. (2011). A family planning clinic partner violence intervention to reduce
risk associated with reproductive coercion. Contraception, 83(3), 274-280.
Miller, E., McCauley, H.L., Tancredi, D.J., Decker, M.R., Anderson, H., & Silverman, J. G.
(2014). Recent reproductive coercion and unintended pregnancy among female family
planning clients. Contraception, 89(2), 122-128.
Miller, E., Tancredi, D. J., Decker, M. R., McCauley, H. L., Jones, K. A., Anderson, H., ... &
Silverman, J. G. (2016). A family planning clinic-based intervention to address
reproductive coercion: a cluster randomized controlled trial. Contraception, 94(1), 58-67.
Miller, M.K., Plantz, D.M., Dowd, D.M., Mollen, C.J., Reed, J., Vaughn, L., & Gold, M. A.
(2011). Pediatric emergency health care providers’ knowledge, attitudes, and experiences
regarding emergency contraception. Academic Emergency Medicine, 18(6), 605-612.
Miller, R. L., & Shinn, M. (2005). Learning from communities: Overcoming difficulties in
dissemination of prevention and promotion efforts. American journal of community
psychology, 35(3-4), 169-183.
Mittal, M., Senn, T. E., & Carey, M. P. (2013). Fear of violent consequences and condom use
among women attending an STD clinic. Women & Health, 53(8), 795-807.
Moore, A. M., Frohwirth, L., & Miller, E. (2010). Male reproductive control of women who
have experienced intimate partner violence in the United States. Social Science &
Medicine, 70(11), 1737-1744.
Muthén, B. (2004). Latent variable analysis. The Sage handbook of quantitative methodology for
the social sciences. Thousand Oaks, CA: Sage Publications, 345-68.
Muthén, L. K., & Muthén, B. O. (2012). Mplus statistical modeling software: Release 7.0. Los
Angeles, CA: Muthén & Muthén.
Nagel, K. & Neal, M. (2008). Discussions regarding sperm banking with adolescent and young
adult males who have cancer. Journal of Pediatric Oncology Nursing, 25(2), 102-106.
Nakopoulou, E., Papaharitou, S., & Hatzichristou, D. (2009). Patients' sexual health: A
qualitative research approach on Greek nurses' perceptions. The Journal of Sexual
Medicine, 6(8), 2124-2132.
National Domestic Violence Hotline. (2011). 1 in 4 Callers to the National Domestic Violence
Hotline report birth control sabotage and pregnancy coercion. Retrieved May 20, 2014
from http://www.thehotline.org/2011/02/1-in-4-callers-to-the-national-domesticviolence-hotline-report-birth-control-sabotage-and-pregnancy-coercion/

119

National Network to End Domestic Violence, NNEDV. (2015). Domestic Violence Counts 2015:
A 24-hour census of domestic violence shelters and services. Retrieved July 20, 2016
from http://nnedv.org/downloads/Census/DVCounts
2015/DVCounts 15_NatlReport.compressed.pdf
Nikolajski, C., Miller, E., McCauley, H.L., Akers, A., Schwarz, E.B., Freedman, L., ... &
Borrero, S. (2015). Race and reproductive coercion: A qualitative assessment. Women's
Health Issues, 25(3), 216-223.
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in
latent class analysis and growth mixture modeling: A Monte Carlo simulation
study. Structural equation modeling, 14(4), 535-569.
Panzano, P.C., Seffrin, B., Chaney-Jones, S., Roth, D., Crane-Ross, D., Massatti, R., Carstens,
C., 2004. The innovation diffusion and adoption research project (IDARP). In: Roth, D.,
Lutz, W. (Eds.), New Research in Mental Health, vol. 16. Ohio Department of Mental
Health Office of Program Evaluation and Research, Columbus, OH, pp. 78–89.
Phillips, S.J., Bennett, A.H., Hacker, M.R., & Gold, M. (2016). Reproductive coercion: an underrecognized challenge for primary care patients. Family Practice, 33(3), 286-289.
Quinn, B. (2003). Sexual health in cancer care. Nursing Times, 99(4), 32-34.
Quinn, C. & Happell, B. (2012). Getting BETTER: Breaking the ice and warming to the
inclusion of sexuality in mental health nursing care. International Journal of Mental
Health Nursing, 21(2), 154-162.
Quinn, C., Happell, B., & Browne, G. (2011). Talking or avoiding? Mental health nurses' views
about discussing sexual health with consumers. International Journal of Mental Health
Nursing, 20(1), 21-28.
Raj, A., Silverman, J.G., & Amaro, H. (2004). Abused women report greater male partner risk
and gender-based risk for HIV: Findings from a community-based study with Hispanic
women. AIDS Care, 16(4), 519-529.
Reebals, J.F., Brown, R., & Buckner, E.B. (2006). Nurse practice issues regarding sperm
banking in adolescent male cancer patients. Journal of Pediatric Oncology
Nursing, 23(4), 182-188.
Reed, J.L., Vaughn, L.M., & Pomerantz, W.J. (2012). Attitudes and knowledge regarding
emergency contraception among emergency department adolescents and
providers. Pediatric Emergency Care, 28(8), 775-779.
Reynolds, K. & Magnan, M.A. (2005). Nursing attitudes and beliefs towards human sexuality:
Collaborative research promoting evidence based practice. Clinical Nurse Specialist,
19(5), 255-260.

120

Rosenheck, R.A. (2001). Organizational process: A missing link between research and practice.
Psychiatric Services, 52, 1607-1612.
Ruane, J. & Hayter, M. (2008). Nurses’ attitudes towards sexual relationships between patients
in high security psychiatric hospitals in England: An exploratory qualitative
study. International Journal of Nursing Studies, 45(12), 1731-1741.
Sack, S., Drabant, B., & Perrin, E. (2002). Communicating about sexuality: an initiative across
the core clerkships. Academic Medicine, 77(11), 1159-1160.
Saunamäki, N. & Engström, M. (2014). Registered nurses' reflections on discussing sexuality
with patients: responsibilities, doubts and fears. Journal of Clinical Nursing, 23(3-4),
531-540.
Saunamäki, N., Andersson, M., & Engström, M. (2010). Discussing sexuality with patients:
nurses’ attitudes and beliefs. Journal of Advanced Nursing, 66(6), 1308-1316.
Sedgh, G., Singh, S., Shah, I.H., Åhman, E., Henshaw, S.K., & Bankole, A. (2012). Induced
abortion: incidence and trends worldwide from 1995 to 2008. The Lancet, 379(9816),
625-632.
Seidman, E. & Tseng, V. (2011). Changing social settings: A framework for action. In Aber, M.
S., Maton, K. I., & Seidman, E. (eds), Empowering settings and voices for social change.
Oxford;New York: Oxford University Press, 12-37.
Seth, P., Raiford, J.L., Robinson, L.S., Wingood, G.M., & DiClemente, R.J. (2010). Intimate
partner violence and other partner-related factors: Correlates of sexually transmissible
infections and risky sexual behaviours among young adult African American women.
Sexual Health, 7(1), 25–30.
Sohocki, J. (2010). Nurses' sexual attitudes, beliefs, and education: The threats to sexual health
and holistic nursing practice. (Doctoral dissertation). Retrieved from ProQuest
Dissertations & Theses Global. (Order No. AAI3452694).
Stead, M.L., Brown, J.M., Fallowfield, L., & Selby, P. (2003). Lack of communication between
healthcare professionals and women with ovarian cancer about sexual issues. British
Journal of Cancer, 88(5), 666-671.
Strawgate-Kanefsky, L. (2000). A national survey of clinical social workers’ knowledge,
attitudes, and practice regarding sexuality. (Doctoral dissertation). Retrieved from
ProQuest Dissertations & Theses Databases. (Order No. 9963439).
Sullivan, C.M. (2016). Examining the work of domestic violence programs within a “Social and
Emotional Well-being Promotion” conceptual framework. Harrisburg, PA: National
Resource Center on Domestic Violence at www.dvevidenceproject.org.

121

Sung, S.C. & Lin, Y.C. (2013). Effectiveness of the sexual healthcare education in nursing
students' knowledge, attitude, and self-efficacy on sexual healthcare. Nurse Education
Today, 33(5), 498-503.
Sutherland, M.A., Fantasia, H.C., & Fontenot, H. (2015). Reproductive coercion and partner
violence among college women. Journal of Obstetric, Gynecologic, & Neonatal
Nursing, 44(2), 218-227.
Teitelman, A.M., Tennille, J., Bohinski, J.M., Jemmott, L.S., & Jemmott III, J.B. (2011).
Unwanted unprotected sex: Condom coercion by male partners and self-silencing of
condom negotiation among adolescent girls. Advances in Nursing Science, 34(3), 243259.
Teng C.I., Hsiao F.J., & Chou T.I. (2010) Nurse-perceived time pressure and patient-perceived
care quality. Journal of Nursing Management, 18(3), 275–284.
Thaller, J. & Messing, J.T. (2016). Reproductive coercion by an intimate partner: Occurrence,
associations, and interference with sexual health decision making. Health & Social Work,
4(1), e11-e19.
Thiel de Bocanegra, H., Rostovtseva, D.P., Khera, S., & Godhwani, N. (2010). Birth control
sabotage and forced sex: Experiences reported by women in domestic violence shelters.
Violence Against Women, 16(5), 601–612.
Trotter, J., Brogatzki, L., Duggan, L., Foster, E., & Levie, J. (2006). Revealing disagreement and
discomfort through auto-ethnography and personal narrative: Sexuality in social work
education and practice. Qualitative Social Work, 5 (3), 369-388.
Tsai, Y.F. (2004) Nurses’ facilitators and barriers for taking a sexual history in Taiwan. Applied
Nursing Research 17(4), 257–264.
United States Conference of Catholic Bishops. (2009). Ethical and religious directives for
Catholic health care services. Retrieved on March 4, 2016 at http://
www.usccb.org/issues-and-action/human-life-and-dignity/health-care/ upload/EthicalReligious-Directives-Catholic-Health-Care-Services- fifth-edition-2009.pdf.
Upadhyay, U.D., Dworkin, S.L., Weitz, T.A., & Foster, D.G. (2014). Development and
validation of a reproductive autonomy scale. Studies in Family Planning, 45(1), 19–41.
Ussher, J.M., Perz, J., Gilbert, E., Wong, W.T., Mason, C., Hobbs, K. and Kirsten, L. (2013)
Talking about sex after cancer: A discourse analytic study of health care professional
accounts of sexual communication with patients. Psychology & Health, 28(12), 1370–90.
Vassiliadou, A., Stamatopoulou, E., Triantafyllou, G., Gerodimou, E., Toulia, G., & Pistolas, D.
(2008). The role of nurses in the sexual counseling of patients after myocardial
infarction. Health Sciences Journal, 2(2), 111-118.

122

Weerakoon, P., Sitharthan, G., & Skowronski, D. (2008). Online sexuality education and health
professional students' comfort in dealing with sexual issues. Sexual and Relationship
Therapy, 23(3), 247-257.
Whitaker, T. & Arrington, P. (2008). Social workers at work. NASW Membership Workforce
Study. Washington, DC: National Association of Social Workers. Retrieved from
http://workforce.socialworkers.org/studies/SWatWork.pdf
White, I. (2002). Facilitating sexual expression: Challenges for contemporary practice. In H.
Heath & I. White (Eds.), The challenge of sexuality in health care (pp. 243-263). Malden,
MA: Blackwell Science, Inc.
Wingood, G.M. & DiClemente, R.J. (2000). Application of the theory of gender and power to
examine HIV-related exposures, risk factors, and effective interventions for
women. Health Education & Behavior, 27(5), 539-565.
Wright, D. & Pugnaire-Gros, C. (2010). Let’s talk about sex: Promoting staff dialogue on a
mental health nursing unit. Journal for Nurses in Staff Development, 26 (6), 250–255.
Yıldız, H. & Dereli, E. (2012). The views and attitudes of nurses on the topic of sexual
counseling. Sexuality and Disability, 30(1), 89-95.

123