Factors associated with intimate partner violence during pregnancy

Material Information

Factors associated with intimate partner violence during pregnancy an evolutionary feminist approach
Silverstein, Meredith N
Publication Date:
Physical Description:
195 leaves : ; 28 cm


Subjects / Keywords:
Pregnant women -- Violence against ( lcsh )
Family violence ( lcsh )
Feminist theory ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 184-195).
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Meredith N. Silverstein.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
51820520 ( OCLC )
LD1190.L566 2002d .S54 ( lcc )

Full Text
Meredith N. Silverstein
B.A., Sarah Lawrence College, 1991
M.S., The American University, 1995
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences

This thesis for the Doctor of Philosophy
degree by
Meredith N. Silverstein
has been approved
h\ 3 \p2

Silverstein, Meredith N. (Ph.D., Health and Behavioral Sciences)
Factors Associated with Intimate Partner Violence During Pregnancy: An
Evolutionary Feminist Approach
Thesis directed by Associate Professor Stacy Zamudio
The American College of Obstetrics and Gynecology and the American
Medical Association have both recognized intimate partner violence (IPV) during
pregnancy as a significant public health problem associated with adverse birth
outcomes; however, there is a paucity of research on factors associated with increased
violence during pregnancy. An integrated evolutionary and feminist theoretical
paradigm (Malamuth, 1996; Gowaty, 1992) was used to test whether a womans
physical proximity to her male batterer prior to pregnancy plays a pivotal role in the
expression of IPV during pregnancy. Two hypotheses were tested: 1) increased time
spent together prior to pregnancy is associated with increased confidence of paternity,
and 2) increased confidence of paternity is associated with a decrease in the
frequency and severity of abuse during pregnancy.
A mixed-methods approach was used. Open-ended interviews were conducted
with 15 pregnant women in womens shelters to increase the content relevance and
construct validity of the survey developed for the quantitative phase and to inform the
statistical findings. Next, self-administered, anonymous surveys were distributed to a
cross sectional sample of women with a confirmed pregnancy, 18 years or older, who
spoke English and/or Spanish presenting at 11 health care sites and domestic violence
Three hundred forty two women responded to the abuse screening questions,
53 women reported physical violence from their current partner at some point in the
relationship prior to and/or during pregnancy. Confidence of paternity was
significantly better explained by a model that included the variables of social isolation
prior to pregnancy and time spent with partner prior to pregnancy than by a model
that included only time spent with partner prior to pregnancy (x2 = 8.23,1 df, g <
0.005, likelihood ratio test). The change in the severity of abuse during pregnancy
relative to before pregnancy was significantly greater among women whose partners
not certain of paternity compared to those who were certain (t = -2.26, 36 df, p =

Restriction of womens access to social contacts by abusive partners is
associated with confidence of paternity. Screening tools and counseling modalities
should consider confidence of paternity as a risk factor for increased severity of abuse
during pregnancy.
This abstract accurately represents the content of the candidates thesis. I recommend
its publication.

This dissertation is dedicated to my greatest teachers, Norman and Keitha
Silverstein, with love and gratitude beyond words or measure; and to the memory of
Frances Silverstein and Elsie Cook, strong women who helped blaze the trail, each in
her own way.

My most sincere thanks go out to the many, many people who helped make
this dissertation possible. First, 1 would like to extend my heartfelt thanks to the
members of my committee, Stacy Zamudio, David A. Young, John Brett, David P.
Tracer, Debbi Main, and Wilson Pace: Stacy, an exceptionally dedicated professor
who has been there for me since (literally) Day One; David Young, who taught me
the true meaning of patience, perseverance, and courage; David Tracer, who always
kept an open door for me; John, who encouraged me to take risks and believe in my
work; Debbi and Wilson, whose support and encouragement helped make data
collection possible and grant applications successful.
I also wish to thank the many people and organizations that participated in this
study: The Department of Family Medicine, University of Colorado Health Sciences
Center with special thanks to Sherry Holcomb; participating CaReNet practices;
Jackie Sievers and the staff of B4 Babies and Beyond; the midwives and staff at
MCPN Jeffco; the midwives, and staff at MCPN Hoffman Heights; Barbara Hughes,
the midwives, and staff at Aurora Nurse Midwives. I would like to extend a special
thank you to the medical assistants at each of these practices who provided invaluable
help and humor. I also owe a large debt of gratitude to the domestic violence shelters
and womens organizations that provided continuous encouragement: the Colorado
Coalition Against Domestic Violence; Jane Pemberton, Linda Mikow, and the staff at
Women in Crisis; Stephanie Wilde and the staff at Alternatives to Family Violence;
Mary Ellen Busdicker and the staff at The Gathering Place. Their unwavering support
and dedication to end violence against women supplied the spirit for this research.
Many, many loving thanks to Mike Ferriter who held my hand the whole way;
to Mary Barry, a true friend who lent her shoulder during the many difficult times and
who always managed to get me laughing again; to Cordt Kassner, Lee Hoffer, Kelly
Causey, and the illustrious members of HBS Cohort 4; and to Norman, Keitha, and
Matthew Silverstein (you know Im still your biggest fan). Thank you, thank you for
helping me to reach my goal.
This dissertation was supported through funding from the A.F. Williams
Foundation, Sigma Xi Grants in Aid of Research, and by grant number R03 HS11269
from the Agency for Healthcare Research and Quality.

1. INTRODUCTION .....................................................1
Hypotheses and Specific Aims ...............................4
Organization of the Dissertation............................6
Theoretical Paradigms.......................................7
Feminist Theory.......................................7
Evolutionary Theoretical Perspectives ...............13
Evolutionary Psychology.......................15
Human Behavioral Ecology......................18
IPV Around the Time of Pregnancy...........................26
Definition of Terms..................................26
Characteristics of IPV Around the Time of Pregnancy .33

Methodological Issues
3. METHODOLOGY.........................................................43
Qualitative Methods: Phenomenology............................44
Procedures for Ensuring Validity.......................46
Qualitative Human Subjects Review .....................50
Sample Selection.......................................50
Qualitative Data Collection............................51
Quantitative Methods: Survey Research.........................52
Survey Pre-Testing.....................................59
Survey Data Collection Locations ......................61
Survey Data Collection.................................66
Quantitative Human Subjects Review ....................69
Methods for Qualitative Analyses .............................69
Methods for Quantitative Analyses ............................71
4. QUALITATIVE ANALYSIS................................................76
Data Analysis.................................................78
Description of the Sample..............................81
Textual Description: What Was Experienced..............83
Structural Description: How IPV Was Experienced........91

Exhaustive Description...............................97
5. QUANTITATIVE DATA ANALYSES........................................99
Survey Refusal Rates ......................................99
Prevalence Data ...........................................101
Statistical Analyses .....................................102
Descriptive Statistics for Subgroups of Abused Women.109
Hypotheses Testing........................................Ill
Results of Hypotheses Tests ........................113
6. DISCUSSION AND CONCLUSIONS.......................................115
Main Findings Compared with Past Studies............125
Study Strengths and Limitations...........................134
Future Research Questions ................................137
THE TIME OF PREGNANCY......................140
INTERVIEWS ................................161
D. HORIZONALIZATION CHART.....................164

F. CLUSTERS OF COMMON THEMES....................181

1.1 Model of Hypotheses .................................................5
2.1 The Wheel of Power and Control ......................................12

3.1 Provider-Driven Data Collection Sites..................................63
3.2 Researcher-Driven Data Collection Sites................................64
3.3 Independent Continuous Variables.......................................73
3.4 Independent Dichotomous and Categorical Variables......................74
4.1 Refusal Rates.........................................................100
4.2 Prevalence of Abuse Around the Time of Pregnancy......................101
4.3 Subject Characteristics...............................................108
4.4 Frequency, Severity of Abuse, Time with Partner, Social Isolation
Prior to/During Pregnancy in Subgroups of Abused Women...............110
4.5 Results of Analyses Pertaining to the Hypotheses......................114

Generally, feminist and evolutionary theories are considered opposing
paradigms, the former allied with the social sciences, the latter with the natural
sciences. In fact, they are synergistic theoretical perspectives (Gowaty, 1992; Smuts,
1992; Malamuth, 1996). When an integrated feminist evolutionary model is applied
to the study of interpersonal partner violence (IPV) around the time of pregnancy this
synergy is revealed. Both theoretical paradigms seek explanations for conflict
between the sexes and focus upon behaviors related to power and control as a source
of that conflict. Both perspectives acknowledge that reproductive biology shapes
behaviors related to power and control, and are somewhat concordant on how the
biology of reproduction impacts social systems (Smuts, 1992; Firestone, 1970). Most
importantly, both theoretical perspectives offer explanations for the phenomenon of
male attempts to control female sexuality, reproduction and mate choice (Clark &
Lewis, 1977; Dworkin, 1974; Echols, 1989; OBrien, 1981; Silverstein, 1998; Smuts,
1992; Malamuth, 1996).
Despite this common ground, these two theories diverge on matters related to
the etiology of violence against women. Feminist theory asserts that violence against

women is rooted in patriarchal societal structures and institutions as well as cultural
practices which serve to maintain mens power and control over women.
Evolutionary approaches are in agreement that societal and cultural factors are
involved; however, evolutionary theory applied to inter-sexual conflict seeks to
establish the origins of power, control and patriarchy through the principles of sexual
and natural selection (Smuts, 1992; Wrangham & Paterson, 1996; Buss & Malamuth,
In light of the substantial evidence that interpersonal male-female violence is
a worldwide phenomenon (Daly & Wilson, 1993; Malamuth, 1996; Walker, 1999),
the research undertaken here proposes that IPV during pregnancy is not only a feature
of patriarchal societies, but also an evolved behavioral trait in human males that
reflects a particular male reproductive strategy employed to assure paternity.
The American College of Obstetrics and Gynecology and the American
Medical Association have both recognized IPV during pregnancy as a significant
public health problem (Petersen et al., 1997). The reported prevalence rates in the
U.S. of physical battery during pregnancy vary, ranging between 0.9% to 20.1%; the
variation being largely due to differing sampling frames. A range of 4% to 8% of IPV
among pregnant women has been reported by most studies (Gazmararian, Lazorick et
al., 1996). Among women with an unintended or unwanted pregnancy, prevalence
rates of 39.5% have been reported (Glander et al., 1998). Physical violence against
pregnant women has been associated with adverse birth outcomes such as low birth

weight (Campbell et al., 1999) and preterm delivery due to placental dysfunction
(Berenson, Stiglich, Wilkinson, & Anderson, 1991, Parker, McFarlane, & Soeken,
1994) as well as increased risk of neonatal death (Webster, 1996). Biomedical
researchers have also reported a connection between IPV during pregnancy and the
batterers lack of paternity (Campbell et al., 1995; 1999). In agreement, data reported
from the fields of evolutionary psychology and ecology indicate a relation between
abuse and paternity when the father of the child is not the biological parent but is
living in the home with the child (Daly & Wilson, 1988; Daly, Singh & Wilson,
1993). No known research to date has examined the link between confidence of
paternity of current partners of abused women and subsequent changes in abuse
during pregnancy.
In feminist-based research, IPV, both in the pregnant and non-pregnant state is
attributed to the perpetrators need to assert power and control their victim (Pence &
Paymar, 1993). Feminist explanations of the origins of power and control rely largely
on tautological reasoning in which patriarchal societies perpetuate the power and
control of men over women; power and control, in turn, exist to reinforce patriarchal
societal structures. This research provides evidence that behaviors commonly
associated with the expression of power and control are also associated with
confidence of paternity, thereby establishing a link between feminist and evolutionary

Hypotheses and Specific Aims
The goal of this study was to test predictions about patterns of IPV during
pregnancy using an integrated evolutionary and feminist theoretical perspective. The
specific aims of this study were:
1. Determine the prevalence of intimate partner violence (IPV) around
the time of pregnancy in an ethnically and socially diverse population.
2. Identify changes in the frequency and severity of battery around the
time of pregnancy and the factors associated with these changes.
From an integrated feminist-evolutionary paradigm, it is hypothesized that a
womans physical proximity to her male batterer prior to pregnancy as measured by
the amount of time the couple spends together in an average week plays a pivotal role
in the expression of physical abuse during pregnancy. The batterers attempts to keep
his female partner within close proximity in essence to guard or watch her by
monitoring her locationis a behavior employed to maintain power and control and
ultimately to assure confidence of paternity.
It was predicted that the extent to which the batterer is successful in keeping
his female partner isolated and away from other social interactions prior to pregnancy
will predict the pattern of physical abuse during pregnancy. Thus, two hypotheses
were tested:
1. Increased time spent together prior to pregnancy will be associated with
increased confidence of paternity.

2. Increased confidence of paternity will be associated with a decrease in the
frequency and severity of abuse during pregnancy.
These hypothesized relationships are schematized in Figure 1.1 below.
Figure 1.1: Model of Hypotheses
A Time spent with partner prior to pregnancy
Confidence of paternity T
Frequency and severity of abuse A
Both qualitative and quantitative methods were used to achieve the specific
aims of this study. In-depth, open-ended interviews were conducted with a sample of
15 pregnant women currently receiving services from domestic violence and
womens shelter organizations. The purposes of the interviews were twofold: first, to
increase the content relevance and construct validity of the survey instrument
developed for use in the quantitative phase of the study; and second, to understand
how women experience intimate partner violence during pregnancy using a
phenomenological approach in order to inform the statistical findings of the study.
Next, a survey instrument was developed to test the hypotheses proposed in this
study. Three hundred sixty five surveys were collected from 11 sites comprised of

primary care medical practices, community health care practices, and domestic
violence shelters.
The combined results shed light on the a prevalence of IPV around the time of
pregnancy in a diverse community-based population and permitted examination of the
factors associated with changes in the frequency and severity of abuse around the
time of pregnancy, as well as the contexts in which these factors exist. These findings
contribute to our understanding of IPV around the time of pregnancy and the risk
factors for increased violence.
Organization of the Dissertation
This introductory chapter has provided an overview of the topic and the issues
addressed by this dissertation, states the research hypothesis and specific aims and
provides the reader with an outline of the dissertation. Chapter two provides a review
of the theoretical perspectives pertinent to this research and previous research
investigating intimate partner violence (IPV) during pregnancy. Chapter three
discusses the methods used in this study. Chapters four and five consist of the
qualitative and quantitative analyses, results and limitations, respectively. Chapter six
provides a summary and interpretations of the findings of the study, its limitations
and strengths, directions for future research and conclusions.

Theoretical Paradigms
Feminist Theory
Feminist theory is a highly diverse collection of schools-of-thought,
ideologies and worldviews that are united by the common goal of exposing and
ending sexist oppression and differential access to power based on gender (Hooks,
1984). Feminist theory is frequently divided into subgroups, e.g. Marxist feminists,
socialist feminists, radical feminists and liberal feminists; however, feminist theory
cannot be confined to these divisions. The lines between these subgroups are often
blurry and there are frequent ideological debates between and within groups (Humm,
Christina Hoff Sommers (1995), an outspoken critic of some branches of
feminism, has divided feminist thinking into two opposing camps: gender feminists
and equity feminists. Equity feminism began with the Womens Movement of the
eighteenth century and embraced the explicitly stated goal of attaining equal and fair
rights for all women under the law. Gender feminism, also known as the Second
Wave of feminism emerged in the mid 1960s with a perspective that differed
markedly from equity feminism. The Second Wave feminists embraced and promoted

an ideological and political agenda which asserted that all women have been
subjected to the rule of and domination by male patriarchy, which has systematically
oppressed women through androcentric culture (Buss, 1996; Silverstein, 1999;
Harway & ONeil, 1999). Regardless of the ideological differences within feminist
camps, feminists have been united in the assertion that violence against women, in all
of its forms, is a means to keep women subordinate to men and is part of both
institutional and ideological control and domination (Humm, 1995). Similarly, as
many evolutionary theorists have noted, the model of a male dominated society and
males dependence on male-male alliances for access to resources, status and females
is so ubiquitous as to be considered a human universal in both the feminist and
evolutionary perspectives (Pratto, 1996; Smuts, 1992, Wrangham & Peterson, 1996).
Both evolutionary theory and feminist theory acknowledge this universal social
pattern and by doing so, illustrate one of the primary links between the two theories.
Intimate partner violence both during pregnancy and in the non pregnant state
has been an issue embraced by feminists who are rightfully credited with bringing it
out from behind closed doors and into the public domain (Harway & ONeil, 1999).
Since 1850, feminist activists in both the United States and England have fought
campaigns to end domestic violence through the establishment of shelters and
publicity campaigns to raise awareness of the issue (Ferraro & Johnson, 1983).
Feminist activists of all perspectives have been instrumental in dispelling popular
myths surrounding domestic violence and other forms of sexual and physical assault;

namely that women ask for or encourage violence against them through appearance
or behavior (Walker, 1979). This work has resulted in sweeping changes in both the
national policy and legal arenas, such as the National Violence Against Women Act,
as well as in public perception and knowledge of the issues. Domestic violence is
indeed on the decline. According to the Department of Justice, Bureau of Justice
Statistics, women reported 900,000 violent victimizations by an intimate partner in
1998, down from 1.1 million in 1993.
In the academic realm, there can be no doubt that feminism has reconstructed
the way in which violence against women is defined and studied (Marin & Russo,
1999). Rather than focus on violence at the individual level in the context of
psychological or psychiatric pathology, feminist theory has instead focused the
analyses of violence against women at the macro-societal level. Through this
approach, feminist theory has contributed toward the development of inclusive
models that reflect the range of violent acts women experience and the diversity of
womens experiences (Marin & Russo, 1999). From this perspective, violence is
constructed in the context of the society in which it is found. All forms of violence
against women are seen as a form of social control exerted by patriarchal societies for
the benefit of maintaining patriarchal societies, an institutionalized systematic
oppression that explicitly and implicitly permeates the structures and systems of
society (Marin& Russo, 1999).

While feminism has offered some powerful proximate explanations for
violence against women which are shared by the evolutionary perspective, feminism
has been unable to consistently and adequately address the etiology and predictors of
intimate partner violence. Feminism asserts that violence against women stems from a
patriarchal social structure whereby men are socialized to develop and maintain
power and control over women in every arena. In addition, violence is used to enforce
culturally and socially sanctioned power and control and to contain womens
sexuality and reproductive roles (Humm, 1995, Harway & ONeil, 1999).
Evolutionary theorists agree (Gowaty, 1992; Smuts, 1994; Malamuth, 1996) but
emphasize the need for a theoretical paradigm and research that sheds light on the
origins of power, control and the use of violence.
Feminist-based methodology could best be described as constructivist, post-
modernist, and process oriented with a strong preference for qualitative methods that
capture subjective experience. The explanatory models developed by feminist
thinkers are primarily descriptive representations of the proximate mechanisms that
shape violence against women (Gowaty, 1992). This is not to say that they are lacking
in meaning and application; however, these models are often powerful educational
tools that gain predictive power when informed by evolutionary models. One such
feminist model is the wheel of power and control (Pence and Paymar, 1993).
Ellen Pence and Michael Paymar of the Duluth Domestic Intervention Abuse
Project developed the Wheel of Power and Control as a conceptual model of the

dynamics and characteristics of abusive relationships (1993). While there are no
statistics kept on the prevalence of the use of this model, it is widely used by
domestic violence organizations throughout the U.S. as an educational and advocacy
tool. As shown in Figure 2.1, this model indicates that the abusive males use of
social isolation and male privilege are behaviors employed to ensure power and
control in the intimate male-female relationship. The use of sexual and physical
violence act as the tires of the wheel thus securing and linking the aforementioned
behaviors as the spokes to the hub of power and control (Pence & Paymar, 1993).

Figure 2.1: Wheel of Power and Control
This model was developed from interviews with women attending educational
classes at a Duluth battered womens shelter and clearly reflects feminist theoretical
perspectives by directing attention to the macro-level roots of violent behavior. The
authors state:
Each of the tactics depicted on the Power and Control Wheel are typical
behaviors used by groups of people who dominate others.. ..Men in particular
are taught these tactics in both their families of origin and through their
experiences in a culture that teaches men to dominate. (Pence & Paymar,
1993, p.2)
According to this model the power and control dynamic is at the heart of
violence against women. All physically violent behaviors as well as psychologically
abusive behaviors such as restriction of access to resources and social contacts stem

from the need to maintain power and control over ones victim. Despite the strength
of this model as a descriptive tool, it falls short as a model from which predictive
hypotheses may be generated since the causal mechanism for violence is
predetermined and environmental variability is not considered. Even so, the Wheel of
Power and Control offers a meaningful conceptual model with which to incorporate
evolutionary theory with feminist thought. It is argued here that the evolutionary
theoretical model of mate guarding (discussed below), when used as a foundation for
the feminist power and control model, illustrates the links between evolutionary and
feminist theoretical perspectives.
Evolutionary Theoretical Perspectives
Darwin recognized the potential application of evolutionary theory to the
study of behaviors and emotions in his book, Expression of Emotion in Man and
Animals (1872). During the next century, the study of human behaviors as evolved
adaptations began to take shape as academic disciplines. In the mid-seventies, E.O.
Wilson published Sociobiology (1975) followed by Richard Dawkins The Selfish
Gene (1989). Since then, evolutionary theory as applied to the study of human
behavior has developed into several (sometimes competing) subfields of study. Most
relevant to this discussion are evolutionary psychology and human behavioral
ecology, descended from the disciplines of psychology and anthropology,
respectively. Each field of study has its defining characteristics which sets it apart

from the other, for example, evolutionary psychologys emphasis on emotions as
evolved mechanisms versus behavioral ecologys focus on the environmental context
of evolved behaviors. Despite these differences, each field of study shares the
underlying evolutionary principles of natural and sexual selection. Frequently these
perspectives are seen as competing viewpoints given their differing areas of emphases
and the academic disciplines from which they descended, although it has been argued
that the advancement of the understanding of human behavior would be better served
if the complementary nature of these fields was emphasized (Smith, 2000).
Gowaty (1992) points out that research in evolutionary theory is conducted
primarily by white, western men in postindustrial societies and thus could benefit
from the inclusion of the other voices offered by the feminist perspective.
Evolutionary psychology has been criticized for focusing on male-male competition
for sexual access to females while virtually ignoring the significance of female mate
choice in the process of sexual selection (Gowaty, 1992). While the study undertaken
for this dissertation research may be guilty of the criticism that it does not focus on
female mate choice, it captures data on changes in intimate partner violence, and
mens controlling behaviors from womens perspectives. Evolutionary theory, though
controversial when applied to human behavior, nevertheless may fill in the gaps in
feminist theory by offering ultimate explanations grounded in scientific methodology.

Evolutionary Psychology. Evolutionary psychology is the study of human
psychology informed by evolutionary biology (Barkow, Cosmides & Tooby, 1992).
Unlike behavioral genetics, which concerns itself primarily with intra-species
variability, evolutionary psychology concentrates on proposed universals of human
behavior and the mechanisms that may underlie them (Segal & Macdonald, 1998).
Evolutionary psychology proposes that certain human behavioral attributes are
adaptations shaped by the forces of natural selection over evolutionary time,
particularly during the Environment of Evolutionary Adaptedness (EEA), which
corresponds roughly with the Pleistocene epoch (Barkow, Cosmides & Tooby, 1992;
Hrdy, 1999). As a result of our evolutionary history, our minds are encoded with
behavioral modules. Each module is specifically adapted to solve problems that
were encountered regularly in the environment of the EEA. This environment is
postulated, with reasonable archaeological support, to have been a hunter-gatherer
way of lifemuch different from the bewildering array of novel environments we
must contend with in modem life. The specific composition and structure of the
mechanisms underlying universal human behavioral attributes has yet to be identified;
however, these mechanisms are fundamental to the study of evolutionary psychology.
It has been suggested that the modules are the result of combinations of genes that
evolved through the forces of natural selection. While natural selection favors
behavioral plasticity, it will nonetheless favor certain types of responses under

particular environmental conditions that were regularly encountered and therefore
extremely common in the EEA (Barkow, Cosmides & Tooby, 1992).
The study of evolutionary psychology is organized around the exploration of
common adaptive problems or issues that humans confronted during the EEA and
how those problems were solved. Behaviors that effectively solved problems resulted
in a survival advantage, were selected for, and have since become a part of our
evolved psychology. The research undertaken here investigates the behavioral
expressions of the evolved psychology of confidence of paternity as an underlying
contributor to IPV during pregnancy. These behavioral expressions include time
monopolization and social isolation and the adaptive problem of fitness
Evolutionary psychology has its critics. One of the more significant criticisms
is directed toward the idea of the EEA, pointing out that it requires that our
psychological evolution was concentrated in a specific time period (Irons, 1998).
Sarah Blaffer Hrdy, an outspoken critic of evolutionary psychology, points out that no
one knows specifically what types of environments (and therefore what selective
pressures) characterized this time period. She writes, Most people assume they were
hunters of the African savannas. However, one might just as well guess that these
were survivors of some ecological perturbation (1999, p. 100). She continues by
saying the EEA should not be used as a boilerplate for a host of unverifiable

assumptions about the lifestyles and living environments of our ancestors (1999, p.
Additional criticism has pointed out that evolutionary psychology relies
heavily on a classic tenet of cognitive psychology, i.e., that highly specialized
cognitive modules or structures guide human behavior (Gleitman, 1991). Yet little
research has been undertaken to date to prove the existence of cognitive modules,
with the notable exception of Cosmides and Toobys work on social exchange and
cheater detection (Smith, 2000).
The criticisms are important to note since they underscore some of the central
debates surrounding the study of evolutionary theory and human behavior: the
conditions under, and the means by which, behaviors were selected for versus the
current environment in which these behaviors currently exist. Like any field of study,
evolutionary theorists may disagree with one another on particular aspects, much like
climatologists who debate aspects of global warming amongst themselves;
nevertheless, the field is unified by the tenets of natural and sexual selection.
Power, control and inter-sexual conflict have been studied by evolutionary
theorists and evolutionary psychologists in particular. Evolutionary theorists
acknowledge similar interests between their perspective and feminism but note that
these two perspectives also operate on different planes (Malamuth & Buss, 1996;
Gowaty, 1992). Feminism focuses on proximate mechanisms for behaviors but carries
an underlying political agenda. On the other hand, behavioral ecology and related

evolutionary based human behavioral theories search for causative explanations for
behaviors while attempting to avoid the ecological fallacy of confusing what is with
what ought to be. Because neither theoretical paradigm fully explains intimate partner
violence from both proximate and causal perspectives, drawing on the similarities and
strengths of each of these perspectives may help us to better understand a more
complete picture of intimate partner violence, particularly during pregnancy.
Human Behavioral Ecology. Behavioral ecology, a field of study applied to
both human and nonhuman animals, is the application of natural selection theory to
the study of adaptation and biological design in an ecological setting (Smith &
Winterhalder, 1992 p. 5). Unlike evolutionary psychology, which is primarily
concerned with psychological mechanisms, behavioral ecology pays particular
attention to the environment of the organism under study at the time it is being
studied (as opposed to the EE A). From this perspective the how of behavior is not
as significant as why a behavior is expressed. Eric Alden Smith argues that
behavioral ecologists take a black box approach to the actual mechanisms involved:
This means taking a calculated risk to ignore the (generally unknown) details
of inheritance (genetic or cultural), cognitive mechanisms, and phylogenetic
history that may pertain to a given decision rule and behavioral domain in
hopes these dont matter in the end result. (Smith, 2000 p. 30)
Flexibility and variability in behaviors are examined within the context of
changes within the local environment at the time the behavior of interest is exhibited.

A major theoretical assumption is that variations in the environment lead to variations
in behavior. As a result, behavioral ecology offers a broad paradigm within which to
study adaptive behaviors and, unlike evolutionary psychology, this paradigm is not
contingent upon some as yet unproven evolved psychological infrastructure. Instead
the focus is on the local environment under current conditions.
The focus on the local environment favored by evolutionary ecologists is a
key element of this research. I hypothesize that violence will escalate or deescalate in
response to the perceived effectiveness of isolation of the female from social contacts
outside the battering relationship, resulting in an increase or decrease in confidence of
paternity. Thus, the abused females local environment of access to social contacts
outside of the primary relationship prior to pregnancy plays a central role in the
expression of battery during pregnancy.
Behavioral ecology relies heavily on modeling, favoring cost-benefit or
decision rule models in order to generate testable hypothesis (Smith, 2000).
Drawing on this tradition, this research posits that the biologically based mate
guarding model is directly applicable to the issue of IPV around the time of
pregnancy. The primary assumption of the mate guarding model is that men act to
maximize reproductive opportunities and subsequent genetic representation in the
next generation while minimizing losses by being cuckolded, or unwittingly investing
resources in offspring that are genetically not his own. Under the environmental
conditions of female polygamy, the male must not only gain access, but also ensure

exclusive sexual access in order to maximize fitness and to reduce the possibility of
being cuckolded. Within this context, IPV around the time of pregnancy is a behavior
designed to maximize exclusive male mating and reproductive opportunities by
restricting female mate choice. While this phenomenon has been studied extensively
in non-human mammalian, avian and insect species (Alcock, 1996), application of
this model to human reproductive behavior is rare (Flinn, 1988). The discussion of
the application of the mate-guarding model to IPV during pregnancy must be
preceded by consideration of the evolutionary theory of sexual selection.
According to the classical theory of natural selection, the local environment
creates selection pressures and the individual members of a species possessing
phenotypes that allow them to buffer stress within that environment tend on average
to survive longer and produce more offspring. The ancillary theory of sexual selection
is a type of natural selection concerned exclusively with reproduction. Characteristics
that give an organism an advantage in securing mating opportunities will be selected
for; those that do not will experience negative selection and may eventually
disappear. In sexual selection two factors create selection pressure: intra- and inter-
sexual selection, sometimes referred to as intra-sexual competition and mate choice
(Gowaty, 1992).
Inter-sexual selection is the result of differential parental investment. In
humans, the number of females a male can successfully inseminate and the number of
offspring that can be raised to successful reproductive status determine human male

reproductive success. Reproductive success in women is limited by the biological
constraints of gestation and post-natal care, primarily the time investment and energy
costs of pregnancy and lactation. Clearly, the intrinsic amount of parental investment
on the part of the female is very large, while that of the male is relatively small.
Within the sexual selection theoretical framework it follows that females must be
particularly selective with whom they choose to mate, given the high-energy costs of
each pregnancy. Males, on the other hand, can never be certain of their paternity, and
thus ensuring their exclusive access to females by whatever means the environment
will favor would be a strategy favored by natural selection (Trivers, 1972).
Inter-sexual selection contributes at least in part to intra-sexual selection. The
higher investing sex (female) becomes, as a result of selective mate choice behaviors,
a valuable reproductive resource. This in turn encourages competition among the
members of the lower investing sex (males). Those that succeed in the competition
are often rewarded with increased mating opportunities while those that fail are often
excluded, although this is not an absolute rule. Social groups in which polygamy is
the normal marriage pattern may serve as an illustration here. In these societies it is
well documented that there is differential reproductive success among males-those
deemed successful by whatever cultural criteria, have a number of wives and many
children, while some males, lacking access to similar resources, have no reproductive
success at all (Daly & Wilson, 1983; Hrdy, 1999). As a result of females selective
mate choice behaviors, males must be concerned about more than simply maximizing

mating opportunities, although under some environmental conditions this is a viable
reproductive strategy. In species where the female is polygamous, males must
contend with other males for mating opportunities and for successful insemination
(Alcock, 1996). This holds true for a wide variety of species from insects to
chimpanzees (Wrangham & Peterson, 1996) and as some researchers suggest, in
humans as well (Wyckoff et al., 2000; Daly & Wilson, 1996; Buss, 1996). Behaviors,
such as mate guarding, that reduce the risk of cuckoldry in an environment where
certainty of paternity is tentative as a result of female polygamy and intra-sexual
competition may have conferred a reproductive advantage and been selected for
(Daly & Wilson, 1983; 1992). While the time and energy investment in mate
guarding are high, these costs are offset by an increase in assurances of paternity.
Research in nonhuman animals has shown that mate-guarding behaviors are
most common when the female is fertile and the possibility of reproduction is high
(Trivers, 1972). For example, in their study of bank swallows, Beecher and Beecher
(1979) found that the male bird closely followed his female mate during the time the
female was fertile. Male American bison will herd their female mating partners
away from other males when the female is fertile. The male will continue to closely
accompany the female after copulation in order to prevent competing males from
mating with the female (Daly & Wilson, 1983).
In one of the rare studies of human mate guarding activity, Flinn (1988) found
that mens time and the amount of attention paid to their female partners was

positively correlated with the reproductive cycles of the women. On the northern
coast of Trinidad where the social environment is characterized by high rates of
infidelity, Flinn found that men spent more time with partners who reported
menstrual cycles than with women who were pregnant or post menopausal (Flinn,
1988). In other words, the males guarding behavior is much more vigilant when his
opportunities for reproduction are highest.
Building on Flinns research, this study proposes that restriction of the
females access to other potentially competing males prior to pregnancy is consistent
with predictions of the mate-guarding model. But whereas the feminist power and
control model asserts IPV is motivated by the proximate need to maintain power and
control over ones victim, the evolutionary based mate guarding model predicts that
power and control are themselves behaviors are motivated by the ultimate need to
maintain power and control in order to maximize fitness and minimize the risk of
cuckoldry. According to the mate-guarding model, guarding efforts should increase or
decrease in relationship to the fertility of the female as well as the risk of cuckoldry in
order to maximize confidence of paternity. Therefore, the batterers efforts at
isolating his partner should increase prior to pregnancy when she is fecund and there
is the possibility of extra-pair copulation. Conversely, mate guarding efforts should
decrease during pregnancy when the risk of cuckoldry is nonexistent. Flence, based
on this evolutionary model, it is anticipated that women who spend less time in
physical proximity to abusive partners prior to pregnancy will experience an increase

in the frequency and/or intensity of battery during pregnancy. Conversely, women
who have been most effectively isolated from other relationships through close
physical proximity to their abusive partners will experience a decrease in battery
during pregnancy.
Across academic disciplines, including evolutionary and feminist theory,
intimate partner violence has been linked cross-culturally and historically with mens
need to control womens reproductive capacity and mate choice through the
restriction of her social interactions (Daly & Wilson, 1982,1988,1996; Hilberman &
Munson, 1978; Humm, 1995; Schechter, 1982; Walker, 1999). The
institutionalization of social norms and cultural beliefs that condone, under certain
circumstances, a mans right to beat his adulterous spouse has been well documented
(Smuts, 1992). As evolutionary psychologist David Buss points out:
In a cross-cultural perspective, the ways in which men attempt to control
womens sexuality is nothing short of bewildering. Veiling a womans face is
an attempt to conceal a womans sexual signals from other men. Placing
women in harems gives a king or emperor exclusive sexual and reproductive
access to those women. Forms of genital mutilation, such as clitoridectomy
and infibulation, seemed designed to dissuade women from seeking sexual
pleasure or pursuing sexual relationships with men other than their
husbands.... Within our own culture, men and especially jealous men,
monopolize all of a womans time, threaten to harm the woman or a rival man
for cues to infidelity and insist that the woman wear possessive
ornamentation. (Buss, 1996, p.298)
Arguably, the cultural traditions and rituals surrounding marriage, even
marriage itself, are likely a form of institutional mate guarding designed to restrict
female mate choice, sexuality and reproduction (Alcock, 1996). For example, in

culturally diverse areas of India, ethnographic investigation reveals, violence against
women is a widely accepted form of behaviour, viewed as a womans due and her
husbands right (Jejeebhoy, 1998, p. 857). Jejeebhoy reports behaving improperly
with outside men is grounds for beating (p. 857). Using grounded theory in the
analysis of qualitative interview data, sociologists found abusive males accusations
of infidelity were a focal point for asserting control over the womans behavior and
social interactions:
Imputations of infidelity were part of the definitional dialogues around the
right to own and control the womens sexuality, as well as many of their
activities. The abusers accusations reflected their culturally assumed claims
as male partners to rights of sexual privilege. By reducing women to sexual
objects, or by making their sexuality an important focal point, the abusive men
rendered every social space a potential arena for sexuality, including gyms,
supermarkets, jobsites, and so on. (Lempert, cited in Strauss & Corbin, 1997)
One of the most effective methods to restrict female mate choice is to keep
ones female partner within close physical proximity to oneself in order to reduce
social interactions that may lead to extra-pair sexual activity. Control over access to
interactions with others outside the relationship is a consistent theme of intimate
partner violence and womens behaviors that challenge this control are associated
with increased violence (Harway & ONeil, 1999; Daly & Wilson, 1988). Physical
acts Of violence themselves may be displays of power intended to ensure compliance
with future efforts at social isolation, or violence may be a means whereby the
offspring with questionable genetic heritage is eliminated.

One must acknowledge from the outset the importance of avoiding the
ecological fallacy in this work. Natural and sexual selection should not be
misinterpreted as forces selecting for behaviors based on their ethical utility. A
behavior may or may not be socially acceptable but, nevertheless, may be the result of
evolved behavioral reactions to particular environmental cues. The degree to which
these behaviors are expressed is highly dependent upon the environment and culture.
Clearly, there are no known genes for violence against women. This research
explicitly denies that there is a 1:1 correlation between genotype and behavior but
draws upon the premise that evolution has favored psychological patterns of
recognition and response (Barkow, Cosmides, & Tooby, 1992). Evolutionary theory
as applied to human behaviors has often been misunderstood and co-opted as
pseudoscience used to further racist and hateful political agendas (Wright, 1994).
This unfortunate history has helped fuel many misconceptions about the application
of evolutionary theory to human behavior. However, the integration of feminist and
evolutionary theories holds the promise of contributing toward our understanding of
intimate partner violence without justifying it.
IPV Around the Time of Pregnancy
Definition of Terms
In response to the growing and often methodologically inconsistent field of
IPV and pregnancy research, the Centers for Disease Control and Prevention (CDC)

published Key Scientific Issues for Research on Violence Occurring around the
Time of Pregnancy in 1997. This report established research guidelines for future
work in this area. The survey developed for this study modified these guidelines in
light of survey design and methods research.
Domestic violence is an umbrella term for a wide range of physically and
emotionally violent behaviors that occur between family members including children
and parents, married and cohabitating couples. Violence against women perpetrated
by their male partners, hereafter referred to as intimate partner violence (IPV),
accounts for the majority of domestic violence. According to the Bureau of Justice
Statistics, females aged 12 or older accounted for 85% of the more than 790,000
victims of IPV in 1999 (Rennison, 2001). Women who are victims of intimate partner
violence are often referred to as battered or abused women, but the terms used
and definitions of this phenomenon are frequently debated across academic
disciplines and social institutions (Walker, 1999). Despite this ongoing debate, this
research will refer to victims of IPV around the time of pregnancy as abused.
According to the CDC, intimate partner violence includes
.. .current and former spouses (legal and common law) and non-marital
partners (boyfriend, girlfriend, same-sex partner, dating partner). Intimate
partners may or may not be cohabitating; the relationship need not involve
sexual activities. (CDC, 1997)
Intimate partner violence takes many forms including physical and
psychological abuse. Psychological abuse can be very difficult to quantify,

operationalize and may be open to cultural influences and biases even more so than
conceptualizations of physical violence. The CDC recommendations offer a clear
definition of physical violence as:
Intentional use of physical force with potential for causing death, injury, or
harm. Physical violence includes, but is not limited to scratching, pushing,
shoving, throwing, grabbing, biting, choking, shaking, poking, hair pulling,
slapping, punching, hitting, burning, and use of restraints or ones body, size
or strength against another person. Physical violence includes the use of a
weapon against a person. (CDC, 1997)
The CDC recommendations make a distinction between violence occurring
around the time of pregnancy and violence during pregnancy. Violence occurring
around the time of pregnancy is defined as,
Physical sexual or psychological/emotional violence or threats of physical or
sexual violence that are inflicted on a woman during the pre-pregnancy,
pregnancy, or post-pregnancy periods or some combination of these periods.
Violence during pregnancy is defined as violence occurring exclusively while
pregnant. This research is concerned with the time period leading up to conception as
well as the pregnancy itself and thus falls into the category of violence around the
time of pregnancy; however, it is important to note that data on IPV during the post-
partum period were not collected in this study.
The CDC defines the pre-pregnancy period as one year prior to pregnancy
while pregnancy itself is defined as consisting of the time since the last menstrual
period until delivery or pregnancy termination. Furthermore, the CDC suggests that
researchers use one year as the period of reference when asking subjects about abuse

prior to pregnancy. However, it is unclear how this time period was chosen given
problems with recall bias. The length of memory retention has been established as a
factor affecting accurate self-report in surveys (Belli, 1998). Belli (1998) cites
research by Cannell, (1965) who found that only 3% of respondents failed to report
hospitalizations when asked within 1 to 10 weeks of the health event, but 42% failed
to report hospitalizations when asked within 1 year of the event. Clearly, length of
time plays a pivotal role in the ability to accurately recall events, even with events
that are particularly traumatic or otherwise memorable. In addition, a one-year time
period prior to pregnancy may confound the results of the survey since women may
report violence from a relationship other than the one that resulted in the index
pregnancy (Ballard et al., 1998). Ballard suggests that women be asked about
violence occurring during a short period before pregnancy (1998, p. 275). The use of
a shorter time period minimizes recall bias and reduces the possibility of recording
violence from a previous relationship. For the purposes of this research, the time
period of interest is immediately prior to pregnancy when the woman would be most
fecund and, from an evolutionary perspective, most likely to be able to conceive from
a competing male if indeed a competing male were available. If isolation is a form of
mate guarding behavior as this research posits, then mate guarding behaviors should
be most fully expressed when the female is fecundprior to conception. In light of
these concerns, three months prior to the index pregnancy was selected as the time
period prior to pregnancy for study purposes. This time frame was used by previous

researchers to define the time period prior to pregnancy (Amaro, Fried, Cabral, &
Zuckerman, 1990, Gazmararian et al., 1995).
While reliable statistics on the worldwide prevalence rates of battery of both
pregnant and non-pregnant women are nonexistent (Fischbach & Herbert, 1997),
violence against women, particularly against female spouses has been documented
across cultures and throughout history. During the fourth United Nations International
Conference on Women held in China in 1995, all countries reported some form of
domestic violence (Walker, 1999). The reported prevalence rates in the U.S. of
physical battery during pregnancy range between 0.9% and 20.1%. In reviewing the
majority of prevalence studies, Petersen et al., (1997) report that a reasonable range
may be between 3.9% and 8.3% of pregnant women in the U.S. Similarly, other
researchers place the figure at 4% to 8% of pregnant women (Gazmararian et al.,
The early 1990s marked a new era in intimate partner violence research.
During the 1970s and 1980s, samples derived from domestic violence shelters
predominated, while in the early 1990s data from population based and clinic based
samples became available (Gazmararian et al., 1996). By increasing the diversity of
sampling frames, a more accurate picture of the prevalence of IPV around the time of
pregnancy came into focus and while it is widely assumed that population-based

studies provide a more accurate prevalence rate, estimates continue to vary given
different study designs and methods (Gazmararian et al., 1996). A summary table of
the prevalence rates, as well as methods and selected findings from previous research,
is presented in Appendix A of this dissertation.
In interpreting prevalence rates, it is also important to point out that domestic
violence advocacy groups assert that intimate partner violence, both during pregnancy
and non-pregnancy, cuts across socioeconomic status (SES) groups. However, how
SES is measured remains inconsistent across the literature. For example, Helton,
McFarlane, & Anderson (1987) looked at race/ethnicity, marital status, employment
and education as markers of SES while Campbell, Poland, Waller & Ager (1992)
looked exclusively at race/ethnicity and eligibility for Medicaid. Most recently,
Bauer, Rodriguez, Eliseo & Perez-Stable (2000) used medical insurance, years of
education, and employment as proxy measures of SES. Many widely-cited studies
sampled women from public clinics, inner-city post-partum wards, and large hospital
settings that serve predominantly lower SES clients and thus these research findings
may be limited to this population (Amaro et al., 1990; Berenson et al., 1991,1992;
Helton et al., 1987, OCampo, Gielen, Faden, & Kass, 1994; Parker et al., 1994,
Bauer, Rodriguez, & Perez-Stable, 2000). Few studies have focused exclusively on
middle or upper SES groups. The exception to this is Sampselle, Petersen, Murtland
& Oakley, (1992). Through a secondary analysis of survey data from a sample of 940
women receiving prenatal care, these authors found that 0.9% reported that they were

currently in an abusive relationshipwell below the commonly reported figures. This
sample was selected from private health care practices, with subjects having an
annual income ranged from $40,000 to $49,000. Nearly 97% of the women surveyed
were high school graduates with a mean of 15 years of completed school. These
authors report these descriptive statistics are consistent with the demographics and
affluence of the surrounding community.
While intimate partner violence before or during pregnancy may indeed exist
across SES groups in the U.S. (possible self-report discrepancies between differing
populations notwithstanding), reasons for the differences in the prevalence of this
phenomenon among different SES groups clearly requires further investigation.
Research on men who batter provide additional data. Gelles & Strauss (1989) report
higher rates of battering behaviors among blue collar workers compared to men with
white collar careers. It is not unreasonable to infer that the female partners of blue
collar workers are of similar SES as their partners and, as a result, experience more
battery at the hands of their partners than their white collar counterparts. This
dissertation research attempted to measure a heterogeneous, diverse population
through diverse health care and human services sites; however, most of the
quantitative data was obtained from community health care centers that serve largely
Hispanic, lower SES populations. As a result, one of the shortcomings of the present
study is that a heterogeneous sample was not obtained.

Characteristics of IPV Around the
Time of Pregnancy
Previous research has shown that the best predictor of abuse during pregnancy
is abuse prior to pregnancy (Helton et al., 1987; Evins & Chesheir, 1996; Webster,
Chanderler, & Battistutta, 1994; McFarlane, Parker, Soeken, Silver, & Reed, 1999;
Martin, Mackie, Kupper, Buescher, & Moracco, 2001). Higher rates of violence
during pregnancy have been reported among women with unintended pregnancies
(OCampo et al., 1994; Glander, Moore, Michielutte, & Parsons, 1998; Goodwin et
al., 2000).
Patterns of battery are variable among pregnant women (Mezey & Bewley,
1997) yet the risk factors for changes in patterns of abuse in both pregnant and non-
pregnant women have not been widely studied. McFarlane et al., (1999) reported no
association between ethnicity and the timing and severity of abuse during pregnancy.
In a qualitative study of 60 non-pregnant battered women, Hilberman and Munson
(1978) found that some women experienced an increase in abuse during pregnancy;
however, others reported that the abuse decreased once their batterers knew they were
pregnant. Amaro et al., (1990) found that of 92 women who reported violence during
pregnancy, more incidents of violence occurred in the first trimester than in the
second or the third trimesters, suggesting that violence may be more frequent in the
earlier stages of pregnancy than in the later stages.

Previous research has found that physical violence during pregnancy is most
severe among younger women and adolescents (Webster et al., 1996; Parker et al.,
1994; Stewart, 1993). Several researchers have found that white non-Hispanic women
are more likely to report any history of abuse than Hispanic or black women (Glander
et al., 1998; Berenson et al., 1991). McFarlane et al., (1999) report that severity and
timing (i.e., when the abuse occurred around the time of pregnancy) did not vary
among African-American, Hispanic and Anglo women. As a result, it remains unclear
whether violence is less prevalent among black or Hispanic women, if violence is
underreported among these groups, or if there are no ethnicity-associated differences
in prevalence.
From womens point of view IPV during pregnancy has been correlated with
perceived lack of social support, housing and economic problems (Campbell et al.,
1992; Martin, 2000). Women who are abused during pregnancy are more likely to be
of higher parity, divorced or separated, have less than a high school education, report
more mental health problems, and to use alcohol, tobacco, illicit drugs and
antidepressant drugs more frequently than non-abused pregnant women (Hillard et al.,
1985; Stewart & Cecutti, 1993; Webster et al., 1996; Campbell et al., 1992).
However, given the aforementioned issues surrounding sampling, i.e., the majority of
data on abused women have been acquired in low SES settings, it should be noted
that these findings might be affected by sample bias.

Explanations for battery during pregnancy range from male sexual frustration
and female mood swings resulting from biochemical changes (Reel, 1992) to
unwanted pregnancy (Gazmararian et al., 1995). In addition, jealousy and anger
directed toward the fetus (Mezey & Bewley, 1997) have been cited as explanations.
Previous research has investigated the link between paternity and IPV around the time
of pregnancy. In concordance with this studys second hypothesis, Campbell et al.,
(1995) reports qualitative findings that suggest that women are at elevated risk for
serious abuse when the womans partner does not believe the child is his biological
offspring. Later research reported an association between abuse and the current
partner not being the father of the pregnancy among full term infants. For example,
among preterm infants, the father of the child not being the mothers current partner
was related to abuse during pregnancy and low birth weight (Campbell et al., 1999).
In 1994, the Institute of Medicine released an interim report that defined
primary health care as
the provision of integrated, accessible health care services by clinicians who
are accountable for addressing a large majority of personal health care needs,
developing a sustained partnership with patients, and practicing in the context
of family and community. (Institute of Medicine, 1994 p.l)
Given their role as providers of health care within the context of family and
community, primary care practitioners are often on the front lines of dealing with
IPV. While family medicine educators have increased the amount of residency
curricular time devoted to domestic violence education and training (Rovi & Mouton,

1999) and medical organizations such as the American College of Obstetricians and
Gynecologists (1989) and the American Academy of Family Physicians (1994) have
provided guidelines for screening and intervention, only 2% to 7% of patients seen in
ambulatory care clinics report that physicians asked them if they had experienced
verbal or physical abuse (Sugg, Thompson, R., Thompson, D., Maiuro, and Rivara,
1999). Primary care physicians have reported that the fear of offending patients
presents a barrier to asking patients about abuse (Sugg, & Inui, 1992).
Other research has shown that providers who do screen for violence are more
likely to do so at the first prenatal visit, rather than at follow up visits (Chamberlain &
Perham-Hester, 2000) thus excluding women whose abuse may begin during their
pregnancy and/or women who may not feel comfortable disclosing abuse at a first
exam but may be more inclined to do so as the relationship between care provider and
patient develops. Several factors that prevent or otherwise affect self-disclosure of
battery to health care providers have been identified. Threats of violence and fear of
retaliation from their partner, embarrassment, concern over police/social services
involvement and a lack of trust all contribute to lack of self-disclosure of IPV
(Rodriguez, Quiroga, & Bauer, 1996). Clearly, there remains a need for continued
health care provider education in this area. In addition, a better understanding of the
risk factors for increased risk of violence around the time of pregnancy must be
identified for the development of efficacious future interventions.

Methodological Issues
Methodological variation is characteristic of research on intimate partner
violence and hence it is difficult to compare results across studies. Throughout the
literature, definitions of abuse and the time periods under study vary widely, as do
methods of data collection. This variability highlights the need for consistent use of
terms across the research as well as the need for theory-based research. For example,
OCampo et al., (1994) permitted both verbal and physical, but not sexual abuse to
qualify as domestic abuse. In contrast, Parker et al., (1994) included physical and
sexual abuse in their sample of both adolescent and adult pregnant women while
Gazmararian et al., (1995) looked exclusively at physical abuse reported in The
Pregnancy Risk Assessment Monitoring System (PRAMS) survey: a population based
survey of recently post-partum women across four states that yielded prevalence rates
of 3.8% to 6.9%.
Previous studies have also varied considerably in the time periods under study
in their samples of pregnant women. Helton et al., (1987) randomly selected pregnant
women from public and private prenatal clinics, 80% of whom were at least 5 months
pregnant: OCampo et al., (1994) asked women in their third trimester about violence
within the last 6 months; Gazmararian et al., (1995) selected women who completed
the PRAMS national survey; Sampselle et al., (1992) and Glander et al., (1998)
included all women who presented for care at any point in their pregnancy while
Berenson et al., (1992) included women who presented for their first prenatal visit.

More recent longitudinal studies have included one year prior to pregnancy,
pregnancy and post-partum periods, although the range of post-partum periods under
study is variable (Campbell et al., 1999; Martin, et al., 2001; McFarlane et al., 1999).
This dissertation research sampled women at any stage in their pregnancy.
The rationale for this strategy was to cast as wide a net as possible over the full
spectrum of pregnant women in order to capture all data about abuse that may be
present at various stages of pregnancy. Parker et al., (1994) found that female victims
of intimate partner violence tend to seek care later in their pregnancy than women
who are not abused. Therefore, inclusion of women at all stages of pregnancy
increases the likelihood of capturing data at all stages of pregnancy and possible
abuse. This sampling criteria has been employed by other researchers (Sampselle et
al., 1992, Glander et al., 1998, Helton et al., 1987).
The ideal research design to track changes in intimate partner violence around
the time of pregnancy would be a prospective cohort design. While this design was
not feasible for the present study given the resources necessary for a successful
prospective cohort design, several prospective studies of intimate partner violence
during pregnancy have been reported. Parker et al., (1994) conducted three structured
interviews with pregnant women. The first interview was conducted during the first
prenatal visit at which time the health care provider/interviewer asked about intimate
partner violence in the past year and since conception. The subject was then re-
interviewed during the second and third trimesters. These authors found that 14.2% of

adult women experienced physical and/or sexual abuse during pregnancy (n = 1203).
The authors also point out that the use of reimbursement for participation may have
contributed to low attrition rates, although the exact amount of reimbursement was
not specified. Similarly, Amaro et al., (1990) used semi-structured interviews to
assess abuse during the prenatal and post-partum periods. These authors also attribute
their low rate of attrition to subject payment. They found that 7% of women reported
physical or sexual assault during their pregnancy and most (94%) knew their assailant
(n = 1243), suggesting that the most common form of assault was IPV. In another
longitudinal study, OCampo et al., (1994) conducted three structured interviews with
358 women throughout their pregnancies and found that 65% of the women in their
study experienced either physical or emotional abuse during pregnancy. The inclusion
of screening for emotional abuse may have contributed to this very high prevalence
Other researchers have conducted single interviews during the first prenatal
visit (Hillard, 1985) or the first post-partum visit (Webster et al., 1994). Both studies
report prevalence rates of intimate partner violence during pregnancy that are
consistent with those previously reported 5.8% (n = 1014, Hillard,1985) and 10.9% (n
= 742, Webster et al., 1994). Evins & Chescheir (1996) also conducted a one-time
interview and found a prevalence rate of intimate partner violence of 7.8% among
pregnant women seeking abortion services.

Glander et al., (1998) distributed an anonymous, self-administered
questionnaire to a cohort of 486 women seeking abortions and found a prevalence
rate of IPV during pregnancy of 39.5%. This is considerably higher than reported by
previous studies and provides support for a connection between pregnancy intention
and abuse. Stewart et al., (1993) also used a self-administered questionnaire to screen
for physical abuse. In this study eligible women were approached by a health care
worker and given information about the proposed survey. Women who consented to
participate were informed that disclosure of their names was optional and given the
survey instruments to complete in a private room. Completed questionnaires were
sealed in an envelope by the participant and returned via locked drop box or returned
to clinic personnel. These authors reported a refusal rate of 2.4%. Of the 548 women
who completed the questionnaires, 6.6% reported abuse. Again, these reports are
consistent with previous findings. It is important to note that in this last study, there
was no evidence that women were offended by the questionnaires and that several
respondents spontaneously wrote positive comments about the survey (p.1262). This
is an important consideration when respondents are asked about such a sensitive,
often embarrassing and private issue.
The protocol used for this dissertation research was similar to the study above,
and to the one developed by McCauley et al., (1995). Their intent was to determine
the prevalence and clinical characteristics of domestic violence in primary care
patients using an anonymous self-administered questionnaire. Office nurses

determined the patients eligibility for participation. If a patient was eligible, the
nurse explained the purpose of the study while the patient was in the privacy of the
examination room. They could complete the survey while awaiting the physician.
Patients then placed completed surveys in locked boxes located in the examination
room. Of 2,392 patients, 1,952 completed the 85-item questionnaire. 108 (5%) of the
respondents reported domestic violence within the past year; 418 (21.4%) reported
physical or sexual abuse at some point in their adult life. While theirs is the largest
study of domestic violence in a primary care adult patient population, this survey
instrument did not focus exclusively on abuse during pregnancy.
A number of researchers have been sensitive to the privacy and safety issues
of participants who were accompanied. Stewart et ah, (1993) reported that women
completed the survey in a private room. Glander et ah, (1998) reported, the male
partner was absent during the administration and completion of the questionnaire
(Glander et ah, 1998, p.1003). McCauley et ah, (1995) reported that women who
were accompanied by a male partner who refused to leave the room were excluded
from the study. Consistent with these concerns, this research did not attempt to recruit
women who were accompanied by a partner into the exam room. Sadly, women who
are most frequently accompanied by their male partner or, as proposed by this
research, are most closely watched may be the very women who are at greatest risk
for IPV or who may be experiencing the most severe IPV.

My research attempted to circumvent this issue by initially proposing to
provide surveys in the bathrooms of participating clinics. This method ultimately
proved unfeasible. However, by recruiting women at all prenatal visits, the possibility
of recruiting a woman who was usually accompanied was increased, as perhaps her
partner would not accompany her on every visit. Recruitment of closely accompanied
women remains an issue in need of further study using innovative, practical
methodological approaches.
The most effective way to screen for abuse and to minimize refusal rates and
response bias is an ongoing issue in research on domestic violence. This study
utilized a design very similar to those used by Stewart et al., (1993) and McCauley et
ah, (1995). The results obtained by these researchers indicate that an anonymous self-
administered questionnaire is a viable method to collect quantitative data on intimate
partner violence during pregnancy while ensuring anonymity.

This research employs both qualitative methods in the form of in-depth
interviews and quantitative methods in the form of survey research. Qualitative and
quantitative methods are often seen as incompatible given their differing
epistemological foundations; however, health research has recently begun to
acknowledge the value and applicability of both methods in scientific inquiry
(Malterud, 2001).
Triangulation of methods, or the use of multiple methods, provides the most
effective and powerful tools with which to study complex social and behavioral issues
(Miles & Huberman, 1994; Lincoln & Guba, 1985). While qualitative and
quantitative procedures operate from different assumptions and seek answers to
questions in different ways, both provide tools necessary to better understand and
develop a more complete picture of the problem of intimate partner violence during
In his book, The Farther Reaches of Human Nature (1971), Abraham H.
Maslow states:

I am convinced that the value-free, value neutral, value avoiding model of
science that we inherited from physics, chemistry, and astronomy, where it
was necessary and desirable to keep the data clean and also to keep the church
out of scientific affairs, is quite unsuitable for the scientific study of life. (p.
This quote is one example of how contemporary epistemological thought has
acknowledged the existence and potential effects of the researchers beliefs on the
outcomes and interpretation of quantitative findings, thereby challenging the
traditional positivist view of a purely neutral investigator (Malterud, 2000; Schweizer,
1996; Nagel, 1986; Maslow, 1971). Clearly, quantitative methodology, while subject
to the strict tenets of statistical theory and procedures, still requires the interpretation
and discussion of data through which the researcher must draw conclusions and
propose recommendations. But this is not to say that all quantitative data are relative
and meaningless if stripped of context (Gross & Levitt, 1994). Both qualitative and
quantitative methods require systematic collection and analysis of data to ensure the'
validity and reliability of the results; however, each approach requires interpretation
of the data in order for the results to be translated into meaningful knowledge. In
doing so, each approach is served better when informed by the other.
Qualitative Methods: Phenomenology
With its foundations in philosophy and psychology, phenomenological
methodology seeks to shed light on the meaning individuals ascribe to their
experiences. It is an approach whereby the essential structure or the fundamental

components of an experience are described through the analysis of the central
meanings people have given to the experience or phenomenon itself.
Phenomenology is rooted in the philosophical reflections of Edmund Husserl
in Germany during the mid-1890s and has since evolved into different schools of
thought, most notably transcendental, existential and psychological phenomenology.
Despite these different branches, there are four fundamental principles every
phenomenological approach shares. First, all phenomenological approaches search for
the critical meanings or essences of the experience. Second, all approaches require
the researcher to suspend or bracket their personal experience, a practice called
epoche (Husserl, 1931), during the analytic process and to rely on the words,
images, gestures and other data provided by informants in order to develop a
complete understanding of the nature of the phenomenon under study. Third, through
the concept of intentionality of consciousness, there is a complete rejection of the
subject-object dichotomy. In other words, an object or event cannot be separated from
the experience that an individual brings to bear on the definition of the object or
event. Finally, all phenomenological approaches require a thorough analysis of the
data in order to identify and define the universal meanings of the experience that are
then distilled into a few summary statements (Creswell, 1998).
The approach I used here is grounded in psychological phenomenology
(Moustakas, 1994) which focuses on the meanings individuals ascribe to their
experiences. In order to increase the content relevance and construct validity of the

survey designed for the quantitative portion of this research, it is necessary to have a
clear understanding of how women who experienced IPV during pregnancy talk
about and make sense of their experience as well as the language and phrases they use
to describe their experience. Moreover, an understanding of how women make sense
of interpersonal violence and the meanings they ascribe to it informs the findings
from statistical data gathered during the quantitative phase of the study. This
approach allows a more complete understanding of how violence may change, as well
as the social and interpersonal factors that shape the way patterns of violence are
perceived, conceptualized and experienced.
Procedures for Ensuring Validity
Like quantitative research, qualitative approaches require a clear purpose and
specific aims to guide the study, careful and methodical data collection, and
systematic analysis of the data (Malterud, 2001). Despite these similarities,
qualitative research has been subject to criticism and regarded with skepticism by
those favoring positivist based research methods. (It is worth noting positivist
approaches have received similar criticisms by those favoring qualitative
perspectives.) Some of the positivists criticisms of qualitative approaches include
accusations of subjectivism and an absence of factual evidence on which qualitative
conclusions are based (Malterud, 2001). Indeed, within the various perspectives of
qualitative research, there exists an ongoing debate on the development and

employment of standards and procedures to ensure the validity and reliability of
qualitative findings. For example, Ely, Anzul, Friedman, Gamer & Steinmetz (1991)
argue that the use of quantitative concepts such as validity and reliability are not
applicable in qualitative research given the philosophically separate epistemologies of
qualitative and quantitative approaches. Flamberg et al., (1994) also claim that
quantitative criteria cannot be applied to qualitative methods given fundamental
differences in the assumptions underlying each approach. In contrast, Lincoln and
Guba (1985) have proposed alternative criteria for establishing measures of validity
and reliability in qualitative research. These criteria include credibility, dependability,
confirmability and transferability. In this scheme, credibility corresponds loosely with
internal validity, confirmability with objectivity and transferability with
generalizability (Malterud, 2001). Building on these criteria, Creswell and Miller
(1997, cited in Creswell, 1998) have offered a classification of procedures for
validity that may be employed regardless of which qualitative method is used. These
procedures include the following:
1. Prolonged engagement/persistent observationThis criterion ensures
the researcher is thoroughly familiar with the culture of the participants and is able to
check for distortions and inaccuracies introduced by the researcher.
2. TriangulationThe use of multiple methods and sources.
3. Peer review/debriefingOffers external checks of the process and
analysis. Peer reviewers play the role of devils advocate, thereby keeping the

researcher honest. This method is also discussed by Ely et al., (1991), Lincoln &
Guba, (1985), and Merriam (1988).
4. Negative case analysis-The process by which the researcher distills
hypotheses as the analysis progresses. The hypotheses are re-evaluated, changed or
modified in light of clarifying or disconfirming evidence.
5. Clarifying researcher biasStated by the researcher at the outset of
the research so the reader knows the researchers position, orientation and biases up
6. Member checksThe solicitation of informants opinions of the
studys findings. This method is also discussed by Ely et al., (1991); Lincoln & Guba,
1985; Miles & Huberman, (1994).
7. Rich, thick descriptionThe research provides in-depth details about
the participants and/or setting under study thereby allowing the reader to determine
the transferability of shared characteristics.
8. External auditsThe process by which an external consultant or
auditor who has no connection with the study examines the process and findings of
the research. The auditor conducts an examination to determine whether or not the
findings or outcomes of the research are indeed supported by the data.
Moustakas (1994) maintains that in the phenomenological tradition, the
researchers perception is the principal criterion for establishing validity, yet he also
discusses the importance of informant feedback (cited in Creswell, 1998), which

complements the criteria set forth by Lincoln and Guba (1985) and Creswell and
Miller (1997).
In keeping with these guiding principles, this analysis meets the criteria for
triangulation, peer review, member checks, and clarification of researcher bias
(statement of assumptions). This exceeds Creswells recommendation that qualitative
researchers meet at least two criteria in any given study (Creswell, 1998). This
analysis underwent peer review by experts in the field of qualitative research. In
addition, member checks by case workers and counselors in the field of domestic
violence who were familiar with my research and the study population were
conducted at regular intervals throughout the analytical process. Ideally, the
participants themselves would have reviewed this analysis. Unfortunately, I was
unable to locate any of the participants due to their personal circumstances and
confidentiality protections assured to the participants. Therefore, member checks
were also conducted with selected women who met the study inclusion criteria but
did not participate in the original study. Clarification of researcher bias is discussed in
the data analysis section of the results. Finally, in order to ensure the internal
validity of each stage of analysis, I continually referred back to the original
transcripts, checking my interpretations against the statements of the participants. By
doing so, it is possible for an external reviewer to refer back to the original transcripts
and original examples that validate each interpretation (Creswell, 1998; Moustakas,

Qualitative Human Subjects Review
Human Subjects Committee approval was received from the University of
Colorado at Denver (Protocol #534). Written informed consent was obtained from all
participants at the beginning of all interviews. In order to maintain the highest degree
of participant confidentiality, participant names or other identifying information were
never used or revealed. Numeric identifiers were used in place of participant names in
data transcription. Signed consent forms and receipts for payments were stored in a
locked container separately from the data. Audiotapes were stored apart from the
signed consent forms and receipts for payment in a separate locked container.
Sample Selection
One-time in-depth interviews using open-ended questions were conducted
with a convenience sample of 15 women drawn from three geographically distinct
Denver area domestic violence and womens services shelters. All interviews were
audio taped and transcribed verbatim for analysis. The sampling frame of domestic
violence and womens shelters was selected to allay concerns over participant
confidentiality and safety inherent in the recruitment and interview process. The
shelter environment also provided participants with a readily accessible source for
counseling and support should the interviews bring up difficult memories and painful
emotions. Participants were recruited through word of mouth, referrals from shelter
staff and through flyers posted in the shelters advertising the study.

Inclusion criteria were women 21 years old or older, who were currently
pregnant (any trimester) and who had experienced intimate partner violence both
prior to and during their current pregnancy.
Informational training sessions with the staff members of the shelters were
conducted to introduce the staff to the research project and familiarize them with the
protocol. I established a contact person at each site who identified and provided
information about the study to potential participants and served as a conduit for
setting up interview appointments. Both participants and shelters were compensated
($20 and $10 respectively) for their time and the use of shelter resources.
Qualitative Data Collection
Interviews were arranged over the phone and set up by appointment at a
mutually convenient time. All interviews were conducted in a private office at the
shelter site. The interview began once written informed consent was obtained and the
participant received compensation. Participants were told they did not have to
participate and could stop at any time. In addition, participants were told there were
no right or wrong answers and were encouraged, should the interview bring up
disturbing memories or feelings, to seek counseling services readily available to them
at the shelter.
Questions were initially developed as a means to help guide the interviewer
through subject area domains. Participants were encouraged to simply tell their

story and talk about what brought them to the shelter and to discuss their relationships
and their pregnancy. The exception to this is a direct question posed by the
interviewer in order to determine the feasibility of using health care settings as a
sampling frame for the quantitative phase of the study. Women were asked, If your
doctor told you that their office was participating in a research study on intimate
partner violence, would you fill out an anonymous survey for the study about your
experiences with violence while you were at the doctors office? When the
participant felt she had completely discussed and described her experiences, the
interview was completed. Interviews lasted from between 25 to 90 minutes.
An experienced, professional transcriptionist with a background in social work
and mental health was hired to transcribe the audiotapes.
Quantitative Methods: Survey Research
An anonymous, self-administered survey was developed specifically to test
the hypotheses proposed by this research. Survey methodology was selected for the
quantitative portion of this study for two reasons. First, and perhaps most importantly,
an anonymous survey was required given the highly sensitive nature of the topic and
the setting in which the data were collected. Previous research has found people are
most likely to disclose information about sensitive issues and behaviors in
anonymous surveys and questionnaires (Czaja & Blair, 1996). Women currently
living in battering situations may feel more comfortable disclosing sensitive

information through anonymous pathways rather than having to confront an
individual face to face. However, it is also important to point out that, regardless of
the method of data collection, many sensitive issues and behaviors are underreported
(Sudman & Bradbum, 1982). Other researchers have effectively employed survey
methods within the area of IPV during pregnancy. Stewart et al., (1993) distributed an
anonymous survey in a health care setting and found no evidence that women were
offended by the questionnaires. In fact, several respondents offered positive
comments about the survey. This is a significant consideration when respondents are
asked about a private and difficult issue. Second, an anonymous survey allowed me to
circumvent the issue of signed informed consent. Under federal regulations, signed
informed consent is not necessary if the survey does not contain any personal
identifiers and cannot in any way be linked back to the individual participant. Given
the healthcare providers legal duty to report in domestic violence situations and the
limited resources of this study, obtaining signed consent would have created
considerable, perhaps insurmountable barriers to the data collection process.
The development of the survey instrument followed modified procedures
established by Czaja & Blair (1996) and included survey design, preliminary planning
and pre-testing. These stages were not mutually exclusive but overlapped and
involved a continual process of development, testing, and revision in order to develop
the final survey instrument.

The framework of the survey instrument was initially developed using the
studys theoretical framework. Brainstorming and consulting sessions with
colleagues, faculty members, committee members, professional counselors, advocates
and administrators in the field of domestic violence prevention and education were
held on a regular basis as the survey revisions progressed. The development of the
survey instrument used in this research followed the procedures established by Czaja
& Blair (1996). Psychometric procedures to establish validity and reliability of the
survey instrument designed for use in this research were not undertaken at this time
since it was not an objective of this research to establish a psychometrically sound
Tapes and transcripts from the qualitative section of this study were reviewed
extensively to hear how women talked about their experiences, what words and
phrases they used to define their relationship(s), their experiences with violence and
their pregnancy. Concepts that were sensitive or that could have multiple meanings
such as self identification of ethnic background, current partner; severity of violence;
and how women labeled and attributed paternity of their pregnancies, were defined
for use in the survey based on the interview data. During the interviews, women
referred to their partner by a variety of terms such as partner, batterer, abuser,
or in some instances, perpetrator, sometimes interchanging terms. If they were
married, women would also make references to their husbands or if not, to their
boyfriends. Partner was used often and did not carry with it any assumptions

about the relationship, gender, or paternity. Therefore, this term was selected as the
appropriate term to use to refer to the respondents index relationship in the survey. In
describing and discussing violence, women talked most often about how bad it was
and hence, this wording was selected for the severity variables. Special care was
taken to discover how to ask questions regarding the paternity of the pregnancy.
Calling the pregnancy the baby could be inappropriate since perhaps not all women
thought of their pregnancy as a baby. In the interviews, women most frequently
referred to their current pregnancy simply as my pregnancy and therefore this term
was used to refer to the pregnancy in the survey.
Women talked about their partners often extreme jealousy, imposition of
social isolation, interference with access to health care, as well as drug and alcohol
abuse which confirms previous research on these attributes as correlates of abuse
(Bullock, McFarlane, Bateman & Miller, 1989; Pence & Paymar, 1993; Harway &
ONeil, 1999). Questions on the severity of jealousy, social isolation and access to
health care were included in the final survey given their ubiquity in the field of
domestic violence research. Social isolation and access to health care were analyzed
as proxy variables for mate guarding behavior.
Previous research in the area of IPV measurement and screening was
reviewed to determine if there were any existing questionnaires with established
validity and reliability that could be used or adapted for this research. No known
previous research has investigated the hypotheses under study here; however, in order

to increase the validity and reliability of the survey, items that identify past and
currently physical abuse (questions 29 and 30) have been adapted and modified from
the Abuse Assessment Screen developed by McFarlane et al., (1992). The Abuse
Assessment Screen is used widely throughout the literature.
Marshall (1992) developed the Severity of Violence Against Women Scales
(SVAWS) to measure the severity of threats of physical violence and acts of physical
violence. This validated scale was used by McFarlane et al., (1999) in a study on
severity of abuse around the time of pregnancy in three different ethnic groups. After
careful review of these scales, I decided not to use them in my survey since the
SVAWS consists of 46 items which would add significant length to an already
lengthy survey. It is worth noting that Marshall used 10 point scales for measures of
severity of violence in the construction of the SVAWS, which is similar to the scales
used in the present study (Marshall, 1992).
The term around the time of pregnancy is recommended by the CDC (1997)
to best define abuse that occurs in pre-pregnancy, pregnancy and post-pregnancy
periods. This term is used in this research in order to meet CDC recommendations;
however, it is important to point out that this study did not collect data from women
post partum.
In addition to establishing the relevant subject domains and evaluating the
qualitative data to ensure content validity, of particular issue was the construction of
the measurement scales for the following variables: time spent with partner, severity

of abuse, jealousy, isolation and access to health care. These variables are subject to
the difficult issue of capturing quantitative data on subjective experiences. Several
early drafts of the survey used visual analog scales (VAS), which were originally
developed to measure psychosocial states (Fridh et al, 1988). Later research found
that VAS are a reliable measure of subjective experiences and are most commonly
used as an index of pain (Scott& Huskisson, 1976). VAS is a cognitive tool consisting
of a straight line, typically 100 millimeters in length. Each end of the line represents
the two extremes of the experience such as none and the most. There are no
numeric markings on the line itself. The respondent is asked to mark along the
continuum where her experience falls. Since, like pain, violence, jealousy, and social
isolation are subjective experiences, it was thought that using a VAS would be the
optimal means to collect this type of data. However, as a result of the findings from
the cognitive testing process, as discussed below, the VAS was replaced with the
more common numeric 10 point scales.
The first stage of the pre-testing process consisted of cognitive testing
sessions conducted with women residing in a domestic violence shelter and with

professional shelter counselors and administrators. These sessions procedure used the
principles of concurrent cognitive testing outlined by Harris-Kojetin, Fowler, Brown,
Schneier, & Sweeny (1999). A think- aloud process whereby the respondent(s) is
asked to talk about their thought processes and how they felt about the questions
while they completed the instrument was used to increase the reliability and validity

of each of the survey items. The VAS were tested, and ultimately replaced with the
10 point numeric scales. During this process, I received some feedback worth noting.
Although the scales were accompanied with instructions and an illustrated
example, participants stated they felt the VAS were confusing, distracting, and took
too long to fill out. One woman residing at the shelter described the scales as scary
and stated that she did not think abused women would fill them out. When prompted
to explain her thinking further, she replied that abused women have had their
confidence so undermined by their batterers that when confronted with something
new and apparently difficult their first impulse is to turn away. When confronted with
the scary looking VAS, she believed many battered women who might otherwise
complete the survey would be intimidated and fail to complete the survey. This
finding confirms previous research that recommends that VAS be administered under
the direction of a trained survey interviewer since many people do not readily grasp
the concept of the scale (Scott & Huskisson, 1976). Much more positive feedback
was received when the VAS were replaced with the traditional 1 to 10 scales. The
participant who referred to the VAS as scary felt more comfortable with the 10
point scales and stated, Thats something Im familiar with. As a result of these
cognitive testing sessions, the VAS were replaced with the numeric scales.
Once finalized, the survey instrument was translated into Spanish by qualified
interpreters following modified guidelines suggested by Guillemin, Bombardier, &
Beeton (1993). The survey was first translated by one translator, then back translated

by a different translator, a native Latina Spanish speaker. This process helped ensure
semantic and conceptual equivalence as well as technical equivalence which includes:
similar or equal cross-cultural and cross-lingual interpretation of measurement scales
and criterion equivalence; similar or equal cross-cultural and cross-lingual
interpretation of criteria of measurement (Brislin, Lonner & Thorndike, 1973;
Guamaccia, 1996).
Survey Pre-Testing
The second stage of the pre-testing process consisted of a pilot study to test and
streamline the data collection procedures. A cross-section of eligible women were
continuously recruited between August 21 and September 11, 2000, at the Salud
Family Health Center in Brighton, Colorado. Prior to the start of the study, a meeting
was held with the staff and clinicians of the Center to discuss the purpose of the
study, the protocol and to answer any questions or receive any feedback they might
have. All staff members were provided with written instructions and a protocol sheet
as well as my contact information.
Medical records staff were asked to place a language appropriate survey into
eligible patients charts. This was done at the end of the day in preparation for the
following days patients. After determining the eligibility of the patient, her health
care provider asked the patient if she was interested in participating in the study. If
she consented, the woman completed the survey alone in the privacy of the exam

room and returned via a survey drop box located in the exam room. If a woman was
accompanied by an adult into the exam room, had already completed the survey, or
refused, the provider circled the appropriate code of A, C or R on the back of
the last page of the survey and return the survey via the drop box. At an early stage of
the development of this research I decided that women who were accompanied by an
adult into the examination room would not be recruited into the study given the lack
of privacy and the possibility that the accompanying adult may be abusive. Women
accompanied by small children and infants were eligible for recruitment. There is no
conclusive evidence suggesting that screening for domestic violence in the presence
of children is detrimental to the patient or her children (Zink, 2000). All surveys were
numbered for tracking purposes. Finally, in an attempt to collect surveys from women
who were consistently accompanied, surveys would be placed in the bathrooms,
thereby ensuring complete privacy for women who elected to fill them out at this
A total of 25 surveys were collected out of 114 eligible OB patient encounters.
As a result of the pre-testing results and feedback, the following revisions were made
to the survey and to the research protocol.
Part of the very low return rate in this phase of the project was attributed to
the fact that there was no appointed staff person on site who was overseeing or
managing the project. Simply providing each staff member with an instruction and
protocol sheet would not provide the management oversight required for successful

data collection. A study coordinator at each site would be necessary in order to
maintain communication between the practice staff and myself, to troubleshoot
problems and answer questions as they arose. At this stage, it was thought that
flexibility and continual communication between the staff and myself would be the
key ingredients for successful provider driven data collection.
Survey Data Collection Locations
Data collection for the quantitative portion of this study initially took place
within the Colorado Research Network (CaReNet), which is based at The Department
of Family Medicine at the University of Colorado Health Sciences Center (CU DFM).
Prior to this study, this network replicated the National Ambulatory Medical Care
Survey (NAMCS), collecting data on over 6,000 patient visits throughout a one-year
period. In doing so, CaReNet had demonstrated that it was a network capable of
meeting the challenges of conducting survey research.
Collectively, CaReNet practices see an average of 3,337 female patients per
month. Of these women, an estimated 8% are confirmed pregnant. It was anticipated
that 267 women would present for prenatal care per month throughout the CaReNet
practices. Thus, CaReNet was selected as an optimal network within which to recruit
a large sample size. Using a conservative median battery during pregnancy rate of
9%, an estimated survey return rate of 60% to 65% and a 10% rate of missed
opportunities to provide patients with surveys, it was expected that it would take

approximately 6 months to collect 98 completed surveys in which women identify
physical violence prior to and during pregnancy. An early power calculation
determined that, in a multiple regression model, assuming a two-sided test with an
alpha level of 0.05, and a cumulative R value of .015 to 0.20, a sample size of 98
battered pregnant women will yield an 80% power to detect an effect size of 0.15
(Munro, 1997).
The demographics of CaReNet patients are very similar to those of the general
population in Colorado. 83.8% of the CaReNet population is white compared to
80.7% in the general statewide population; 19.4% patients are Hispanic compared to
13% for the statewide population. CaReNet was selected as the initial sampling frame
for data collection for these reasons. In addition, at the outset this project focused on
abuse during pregnancy in primary health care settings. However, after the first six
months of data collection, it became clear that this research would benefit by
expanding its scope and sampling frame to include community health care settings,
public health prenatal programs, and domestic violence shelters. Expansion of the
sampling frame beyond primary care ensured the inclusion of women who may not
have access to primary health care and who may seek out care in other arenas. In
addition, inclusion of these additional sites ensured a sufficient sample size could be
obtained within this studys time frame.
The large number of sites that participated in this study and their geographic
distribution made it unfeasible for me to be on site at all the data collection locations.

Therefore, all of the CaReNet sites as well B4 Babies and Beyond (a public health
prenatal program but not a member of the CaReNet network, located in Grand
Junction) used provider-driven data collection methods. That is, staff and health
care providers at these sites were responsible for the recruitment of study participants
and the distribution of surveys. Provider-driven data collection methods are
established as the primary means of data collection throughout the CaReNet network.
The shelter locations, Aurora Nurse Midwives and Metropolitan Community Provider
Network (MCPN) practices used researcher-driven data collection. At these
locations, I was on site recruiting patients and distributing surveys.
Tables 3.1 and 3.2 summarize the sites that participated in data collection and
the dates of their participation. All data collection sites are in Colorado.
Table 3.1: Provider-Driven Data Collection Sites:
Dates of data Total number of
Location collection surveys returned
La Plan de Salud, Brighton 8/21/00-9/11/00 12
La Casa Quigg Newton Family Health Center, Denver 11/6/00-5/1/01 19
Littleton Health and Wellness, Littleton 10/17/00-1/18/00 4
St. Marys Family Practice, Grand Junction 12/18/01-5/5/01 7
La Plan de Salud, Commerce City 1/19/01 5/23/01 26
A.F. Williams Family Medicine, Denver 4/4/01 5/5/01 8
B4 Babies and Beyond, Grand Junction 12/6/00 3/6/01 65

Table 3.2: Researcher-Driven Data Collection Sites:
Dates of data Total number of
Location collection surveys returned
Shelter sites (Denver, Jefferson Co.) 1/25/01 8/27/01 4
MCPN Jeffco, Lakewood 1/22/01 4/20/01 74
MCPN Hoffman Heights, Aurora 5/2/01 6/29/01 72
Aurora Nurse Midwives 7/2/01 8/27/01 51
At the start of the data collection phase, data was collected via the provider-
driven data collection method only. Practices were asked to collect data for a total of
six months. The data collection period was reduced to three months based on
feedback I initially received from participating practices who stated they believed
they would be unable to collect data for this length of time without significant
disruption to their practices. Subsequent return rates after approximately two months
of data collection continued to be lower than anticipated. Therefore, I made two
decisions in conjunction with my committee. First, provider driven data collection
practices that had not completed three months of data collection were contacted. They
were asked to collect 50 surveys instead of collecting surveys for the specified time
period. Again, based on feedback from the practices, I believed that collecting a
specified number of surveys might be considered a more manageable task. Second, in

order for this project to be completed in a timely manner, it would be necessary to
recruit practices with a higher volume of OB visits and for me to remain on site in
order to ensure successful management of the data collection process. Aurora Nurse
Midwives and two geographically diverse MCPN sites in Lakewood and Aurora were
selected for participation given the high volume of OB patients these community
health care providers serve.
Aurora Nurse Midwives and each MCPN site agreed to allow me to distribute
surveys to its OB patients until 100 surveys were collected at each site. 1 fell short of
this sample size since, over time, women who reported they had already been
surveyed outnumbered the women who had not yet been surveyed and therefore, it
was no longer cost efficient to remain at the site.
Inclusion criteria at all data collection sites were women who were currently
pregnant, 18 years or older, who spoke and read English and/or Spanish, who were
unaccompanied into the examination room, who had an appointment to see a care
provider for prenatal care or services, and who were not acutely ill. Eligible women
were recruited via the protocol below.

Survey Data Collection
At shelters: Eligible women were recruited via the same procedures as
recruitment of women for the qualitative portion of this study. Interested women
contacted me by telephone. If women consented to participation, an appointment was
set up at a mutually convenient time to meet at the shelter. After the purpose of the
study was explained and consent was obtained, the participant completed the survey
at the shelter alone in a private room. The survey was then sealed in an envelope and
returned directly to me on site.
At provider-driven data collection sites: Each participating clinic had a
computerized patient tracking system that records the information relevant to the
study. Through this computerized system, the medical records staff identified
potential participants when charts were pulled in preparation for the days visits. As a
part of this process, the medical records staff placed a blank survey in the appropriate
language inside the patients chart for the clinic nurse to access. After the patient had
completed her preliminary exam, the staff nurse or medical assistant (MA) who
conducted the exam recruited participants according to the following protocol.
The practice nurse or MA offered the patient the survey, a brief explanation of
the study, and informed consent to all eligible women. The explanation of the study,
in the form of a cover letter attached to the survey, explained the purpose of the study,
anonymity of the survey, and risks and benefits of participation in accordance with
IRB guidelines. Consent to participate was implied by the completion of the survey.

If she chose to refuse, the staff member checked off the refused box on both the
right and left hand side of the tear sheets attached to each survey.
Each sheet consisted of six perforated tear sheets contained on an 8 lA by 11
piece of paper. Each of the six perforated tear sheets contained the codes originally
included on the back of the survey: A, for accompanied, R for Refused, and C for
already completed. The left side of the tear sheets remained in the chart so the
clinician would know the status of the patient with regard to the survey. The right
hand side of the tear sheet would then be dropped in the survey return box and
returned so that the number of refusals and number of accompanied women could be
tracked. This same procedure was followed if the woman reported she had already
completed a survey, or if the woman was accompanied into the examination room.
Only one data collection site followed this procedure, thereby making the refusal rate
and number of accompanied women impossible to determine in the other study
locations. At the conclusion of the study, all charts were cleared by the staff of any
remaining surveys and tear sheets and these materials were to be destroyed. At no
time did I have access to patients charts or other identifying information.
If the patient consented, she was provided the opportunity to complete the
survey in the privacy of the examination room either before or after her examination.
When each could complete the survey was left to the discretion of the staff based on
patient census and flow for that particular day. After the survey was completed, the

participant dropped the survey into a drop box. Drop boxes were located in each
examination room.
At researcher-driven data collection sites: The practices MA identified
eligible women during their preliminary exam. After the MA had completed the
initial exam, the MA would notify me of the patients eligibility. Prior to the eligible
womans primary exam, I would introduce myself, explain the study and ask if she
would be interested in participating. If she agreed, I reviewed the informed consent
procedure as outlined in the cover letter with her. Patients were assured of anonymity
and were told they did not have to take the survey and could stop taking it at any
time. Consent to participate was implied by completion of the survey. Participants
were told that if they were experiencing domestic violence or had any questions about
domestic violence, they could talk to their care provider or contact local domestic
violence organizations, with names and numbers provided in a pocket sized flyer
attached to each survey. The participant would be instructed to seal the completed
survey in the envelope provided and drop the survey in the survey return box located
in the room with her.
Depending on the schedules of the health care providers, the participant was
asked to complete the survey while she was waiting for her exam, or after their exam
was completed while alone in the examination room. Women who were
accompanied, who had already completed the survey or who refused to participate
were tabulated over the course of the day as they presented for care.

Quantitative Human Subjects Review
Institutional Review Board approval was obtained from the following
institutional committees: The University of Colorado at Denver (UCD) Human
Subjects Committee (Protocol #588); Colorado Multiple Institutional Review Board
(Protocol # 00-506); HealthOne Alliance Institutional Review Board; and the
Saccomanno Research Institute, Grand Junction (Protocol # 2000-007). No signed
informed consent was required. Each survey began with a letter to the participant that
provided all of the information normally addressed in a standard signed consent form.
Participants were told they could take the cover letter with them if they needed to
contact the PI or the practices governing IRB in the future. In addition, each survey
was accompanied with a pocket size flyer provided by the Colorado Coalition Against
Domestic Violence. The flyer listed names and contact information of domestic
violence service agencies across Colorado. The flyer is printed in both English and
Methods for Qualitative Analyses
In order to understand how women experience intimate partner violence
during pregnancy, I used a phenomenological approach. The purpose of this analysis
is to better understand and inform the quantitative results of this study. This analysis
following the approach used by Moustakas (1994) is described below in detail along
with the results.

A fundamental concept underlying phenomenological research is that of
epoche in which the researcher is required to suspend or bracket as far as is
possible all preconceived ideas and experiences which may affect the interpretation
and analysis of the experiences of the study participants (Moustakas, 1994). To do
this it is first necessary to state outright the assumptions and biases the researcher
brings into the analysis. This statement of assumptions also serves to meet the criteria
of validity outlined by Creswell and Miller (1997) by clarifying researcher bias.
I assumed that battery during pregnancy was directly related to the abusive
partners confidence in paternity and that confidence of paternity, in turn, was directly
linked to the ability of the abusive partner to monopolize the time of his victim. I
assumed that frequency of battery was discemable from severity of battery and
psychological abuse discemable from physical abuse. I also assumed, given that
women were pregnant, that they were heterosexual and their abusers were men.
I assumed that all participants were victims of violence and currently
pregnant. I did not ask for proof such as police records, restraining orders or
pregnancy test results. This assumption is not unreasonable given that participants
were currently residing in or receiving services from a shelter and were often visibly
battered and pregnant. Finally, I assumed that women were truthful in their interviews
and retold their stories to the best of their abilities under difficult circumstances.

Methods for Quantitative Analyses
Initial efforts focused on assuring the integrity of the data, with screening of
printouts of the raw data for consistency and accuracy. Inconsistent responses were
evaluated within the context of womens responses to the remainder of the questions
on the survey. For example, there were surveys in which women responded no to
the battery screening questions; however, they then continued with the survey and
responded affirmatively to the remaining items that addressed abuse, specifically
frequency and severity of abuse. Decisions regarding the best course of action for
inconsistent data were made in consultation with the dissertation Chair and
After correction and finalization of the dataset, data were summarized by
calculation of descriptive statistics. The latter translates to sample size, mean and
standard error, median and range, and measures of normality (e.g., skewness,
boxplots) for continuous variables, with frequency of occurrence and percent of total
responses for categorical variables. Overall prevalence of abuse was calculated as the
total number of women reporting abuse around the time of pregnancy divided by the
total number of women who returned a survey. Location-specific prevalence rates
were similarly calculated. Data from domestic violence shelter locations were not
included in these analyses since, due to their function, the prevalence of abuse in the
population they serve is at or very near 100%. Hence inclusion of domestic violence
shelter sites in the overall calculation would overestimate the prevalence of abuse.

Descriptive statistics were also calculated by the following groupings, women
reporting no abuse and women reporting abuse around the time of pregnancy from
their current partner. The latter group was divided into three subgroups: women
reporting abuse prior to pregnancy exclusively (n = 10), women reporting abuse
during pregnancy exclusively (n = 5) and women who reported abuse before and
during pregnancy (n = 38). Two women reported abuse prior to pregnancy but left the
abused during pregnancy screening question blank; these women were not included
in further analyses.
T-tests were used to compare continuous variables selected a priori between
women abused around the time of pregnancy and non-abused women, while chi
square tests were used to compare categorical variables. Odds ratios along with their
95% confidence intervals were calculated to assess the increased risk of abuse around
the time of pregnancy associated with particular dichotomous variables. All t-tests
and chi square tests were two tailed and alpha was set at 0.05. Homogeneity of
variance was tested for continuous variables, and t-tests corrected accordingly.
Spearman and Pearson correlation coefficients were calculated to evaluate
associations between continuous variables. Significant results from the above tests
were used to select variables for entry into a fixed effects linear model aimed at
examining study hypotheses (n = 53, women reporting abuse at any time with her
current partner). While it could be argued that running multiple tests as was done
could raise the issue of multiple comparisons, I defend this approach by stating that

only certain variables were selected for comparison, not all variables, and that this
study was exploratory, and hence sought to point out variables and factors of potential
importance for future study.
The independent, continuous variables used in analyses and the rationale for
their selection are summarized in Table 3.3:
Table 3.3: Independent Continuous Variables
Variable (Survey question number) Rationale for selection in analyses
Age of respondent (Q3)
Age of partner (Q17)
Years of education (Q4)
Number months pregnant at time of
survey (Q10)
Length of current relationship in
months (Q16)
Time spent with partner in an average
week three months prior to pregnancy
Partners jealousy three months prior to
pregnancy (Q34)
Abuse previously reported as most
common among younger women
(Webster et al., 1996; Parker et al., 1994;
Stewart, 1993); possible evolutionary .
significance given increased fertility
among younger women.
Known risk factor for perpetrating
battery (Gelles, 1999)
Proxy of socioeconomic status
Descriptive; to explore whether abused
women presented for care later in
Proxy measurement of time spent
together in abused vs. non-abused women
Independent variable for first hypothesis;
Measure of mate guarding activity
Associate measure of mate guarding

Table 3.3: Independent Continuous Variables (Cont.)
Variable (Survey question number) Rationale for selection in analyses
Partners social isolation of respondent three months prior to pregnancy (Q37) Surrogate variable for first hypothesis; Measure of mate guarding activity
Access to health care three months prior to pregnancy (Q39) Proxy measure of mate guarding activity; indicator of access to others; direct bearing on implications of research for health care providers.
The independent dichotomous and categorical variables used in chi square and
odds ratio analyses and the rationale for their selection are summarized in Table 3.4
Table 3.4: Independent Dichotomous and Categorical Variables
Variable (Survey question number) Rationale for selection in analyses
Language of respondent (language of survey) Measure of ethnicity
Employment during previous 12 months Measure of SES; proxy measure of
(Q6) isolation prior to pregnancy
Number of current sexual partners(Q12) Measure of monogamy
Cohabitation with partner(Q14) Proxy measure of time with partner
Married to partner (Q15) Proxy measure of time with partner
Partners pregnancy intention (Q23) Known link between pregnancy intention, abuse (Goodwin, et al, 2000); female unwilling to bring offspring into harmful environment.

Table 3.4: Independent Dichotomous and Categorical Variables (Cont.)
Variable (Survey question number) Rationale for selection in analyses
Respondents pregnancy intention (Q24) Previous research identified link
between pregnancy intention and
abuse (Goodwin, et al, 2000); proxy
measure of confidence of paternity
insofar as a male may not risk harm
to offspring he intended to have.
Belief of paternity (Q22) Dependent variable in hypothesis 1;
independent variable in hypothesis 2.
To calculate odds ratios and chi squares, partners pregnancy intention (Q23),
and respondents pregnancy intention (Q24) were dichotomized. Response categories
were dichotomized from the original Likert scale so that Yes and Yes, somewhat = 1;
Partner didn t care, Not really and Definitely Not = 0. The same procedure was used
for the respondents pregnancy intention variable. In addition, belief of paternity
(Q22), the primary outcome variable for the first hypothesis, was dichotomized for
chi square and logistic regression analyses. Response categories were dichotomized
from the original Likert scale so that Yes and Yes, somewhat = 1; No and 1 dont know
= 0.

The qualitative phase of this dissertation research was conducted between
September 2000, and April 2001. Open-ended interviews were conducted with 15
women who self identified as 21 years old or older, currently pregnant and in a
violent relationship. They were recruited from Denver metropolitan area domestic
violence and womens services shelters. The qualitative portion of this study achieved
the following specific aims:
1. To increase the content relevance and construct validity of the survey
instrument developed for use in the quantitative phase of the study.
2. To understand how women experience intimate partner violence during
pregnancy using a phenomenological approach.
After signed informed consent was obtained (see Appendix C), women were
asked if they would be willing to participate in a research study by filling out an
anonymous survey about their experiences with violence in a health care setting. This
question prompted some women to offer their experiences with health care.
All women reported that they knew health care is or could be available to
them through a variety of pathways. Participants reported seeing or attempting to see
a provider for care both prior to and during pregnancy. The mitigating factor in

access to health care is not the lack of available care; instead, it is the abusive male
interfering with the womans attempts to seek and receive health care. Interestingly,
the participants who identified as lesbian did not report that their partners interfered
with their ability to seek and receive health care. The abusive male often prevents or
interferes with the womans ability to receive care through threats and intimidation,
control of money and/or transportation. One woman stated, I had a hard time. He
wouldnt even let me, you know, go to my appointments. And if I did, I always had to
have like, proof or something. Another woman said, It was almost like he didnt
want me to go to the doctor.. .1 mean, he threw away my prescription and
everything. Women in this sample who reported the most severe abuse were also
most often accompanied by their partners into the physicians examination room and
their partners remained present throughout the duration of the exam. This denies the
woman the opportunity to talk openly and honestly about the violence she is
experiencing. One woman reported:
He was always there. She [the doctor] was just, she just want to ask me, but I
started crying. But he was always present there....Actually, I would like to talk
to the doctor. I told him, "Please don't come inside. And he said, "No, I have
to come. I'm the baby's father. (Interview 101)
Despite the issues surrounding access to health care, all of the women
interviewed for the pilot study stated that they would be willing to respond to an
anonymous survey that asked them questions about their abuse, provided their
partners were not with them. Certainly these responses may not be representative of

all women given the specialized setting and their willingness to be interviewed.
Nevertheless, these data show that women who are the most closely watched by
their partners may not have the opportunity to participate in practice-based research
because their batterers may be present. These data also have implications for the
quantitative data in this study, as do the findings of previous researchers who also
excluded accompanied women. Taken together, clinicians need to be suspicious when
male partners accompany women on virtually every examination, right into the
examination room. If battery is suspected, the presence of the male partner may be an
indicator of the level of violence in the relationship. One woman in the qualitative
interview sample credits her doctor with saving her life since he was aware of the
constant presence of the male partner, observed his interactions with her and as a
result of what he saw, alerted the police to the situation. Consequently, this woman
was finally able to leave the abusive situation and receive help.
Data Analysis
Verbatim transcripts of the original audiotapes were analyzed using
psychological phenomenological methodology (Creswell, 1998; Moustakas, 1994). It
should be noted that analytical approaches in qualitative research are not always
widely agreed upon; however, in psychological phenomenology the steps for analysis
are generally similar among researchers (Creswell, 1998; Moustakas, 1994). The

approach I used here, described by Creswell, is derived from the analytical approach
developed by Moustakas (1994).
First, all transcripts were read and re-read in order to become completely
familiar with the stories of each of the participants. Next, significant statements were
extracted from the transcripts of each respondent, a process termed
horizonalization. In this step, significant statements were extracted verbatim from
the original transcripts and listed in no particular order so that each statement had
equal value (See Appendix D). Statements were considered significant if they
spoke directly to the experience of intimate partner violence (IPV) during pregnancy
either in terms of actual events surrounding the abuse or the participants thoughts,
actions, emotions or interpretations of the event. Statements were also considered
significant if the participant offered her insight into the phenomenon in general.
From the horizonalization table, I then developed a list of meaning
statements (See Appendix E). The purpose of the meaning statements was to tease
out the significance embedded within the various contexts of the original statements.
In short, the purpose was to distill what the participant was really saying when the
events and context of the narrative were stripped away. The meaning statements are
my interpretations of the participants original statements and were developed based
on the shared meanings that emerged from a process of contemplation, reading, and
re-reading the statements included in the horizonalization table. This process allowed

me to begin to articulate universal meanings based on the shared experience of abuse
during pregnancy.
The meaning statements were then organized into clusters of common themes
(See Appendix F). Each cluster represents a theme of shared experience derived from
the emerging themes of the meaning statements. In order to ensure the validity of
each theme, I checked back with the original transcripts in order to ensure that each
cluster was supported by the original text. Based on these clusters of themes and in
consultation with the original transcripts, a textual description (what was
experienced) and a structural description (how it was experienced) was developed.
This process ensures the validity of the analysis and links together the analytical
transformations of the original text (Creswell, 1998). The textual and structural
descriptions are discussed in detail below.
Finally, through the careful examination and aggregation of the analysis
outlined above, an exhaustive description was developed. As described by Moustakas
(1994) the ultimate purpose of the phenomenological analysis is to boil down the
textual and structural meanings of the phenomenon to a brief but exhaustive
description of the experience. This is a description of the essential structure or
essence of the experience that is shared by all of those who experience it
(Moustakas, 1994). Creswell writes:
.. .this means that all experiences have an underlying structure (grief is the
same whether the loved one is a puppy, a parakeet, or a child). The reader of
the report should come away with the feeling that I understand better what it

is like for someone to experience that. (Polkinghome, cited in Creswell,
1998, p. 55)
The exhaustive description reveals the shared common experiences of women
who experienced IPV during pregnancy. It reveals the nature of their shared
experience regardless of the gender of their abuser, or the various circumstances of
their lives. What follows are the textual and structural descriptions that lead to the
exhaustive description of intimate partner violence during pregnancy.
Description of the Sample
Fifteen women were recruited for this study. After signed consent was
obtained, women were asked how they would describe their ethnic backgrounds. Six
women described themselves as White or Caucasian, three as Native American or
Indian, two as black or African American, two as Hispanic, one as Hindu Indian, and
one as bi-racial.
Two women from this sample self-identified as lesbian. One participant
reported that her pregnancy was the result of a sexual assault while the other
participant self identified as lesbian but engaged in multiple heterosexual affairs, each
of which reportedly resulted in the birth of one child. I was unaware of the
participants sexual preference until it was revealed during the interview. These
participants met the inclusion criteria given that the gender-neutral term partner
was used during recruitment. In the quantitative study no women either abused or

non-abused reported they were lesbian; nevertheless, this finding from the qualitative
data illustrates the need for inclusion of women of all sexual orientations and cautions
against assumptions that sexual orientation may be an excluding factor in research on
pregnancy. To my knowledge, this is the first time data regarding the sexual
orientation of battered pregnant women has been collected.
Fourteen women reported they were monogamous in their relationship with
their abuser while they were in the relationship and they were certain their abuser was
the father of their current pregnancy. One woman reported she was unsure if her
current pregnancy was fathered by her abusive husband or by a recently ended extra-
marital affair. Of the 15 participants, 8 women reported they had been in an abusive
relationship in the past prior to their current abusive relationship. Thirteen of the
participants stated that they had children from a previous relationship. Three
participants were legally married to their batterers; 12 were cohabitating with their
batterers. At the time of the interview, three women were still either living with or in
close contact with their abuser. One of these women was legally married to her
abuser. Twelve women had left the abusive relationship and were in various stages of
recovery at the time of the interview.

Textual Description: What Was Experienced
Description of what women experienced fell into three areas: 1) different
kinds of violence, 2) how that violence changed around the time of their pregnancies,
and 3) their physical reactions to violence.
1. Abuse During Pregnancy
Psychological and physical abuses are frequently linked. Psychological abuse
may precede physical abuse although the two are most often intermixed. Through
open-ended interview questions, it was left up to the participants to decide and define
what they considered abusive. Women who participated in this study described both
physical and psychological abuse experienced before and during their current
pregnancy as well as experiences of abuse with past pregnancies. Episodes of
psychological abuse were sometimes described with the same intensity and fear as
episodes of physical abuse. This finding supports previous research that has shown
the interconnected nature of psychological and physical abuse (Pence & Paymar,
1993; Harway & ONeil, 1999).
Participants reported a range of physical abuse prior to and during pregnancy
that included being punched, hit, kicked, stabbed, having their hair pulled, physical
restraints and attempted poisoning. As a result of these physical attacks, women
reported a range of injuries including cuts, bruises, lacerations, stab wounds, broken
bones, and blunt trauma to their abdomens while pregnant. Incidents of physical

violence were sometimes described without reference to their current or past
pregnancies. In some cases, pregnancy did not appear to be at issue in a particular
incident of abuse and, as discussed below, violence would sometimes occur without
warning. While no one in this sample reported complications of their pregnancies due
to their injuries, some women did report that they believed the physical attacks were
targeted at the fetus. One woman whose abuser did not want the pregnancy reported
incidences of abuse directed at her abdomen:
And he knows I was pregnant. And he started, he was trying to hit me in the
belly. But the first thing that Id done was started holding my belly. Because I
didnt want him to hurt my baby, you know. And he keeps saying that the
babys a mistake and I should get rid of it. (Interview 108, page 4)
In contrast, another woman, whose partner sent conflicting messages about
wanting the pregnancy, reported that her partner would not attack her abdomen:
Well, he would be really good at not hitting me in my stomach. He would hit
me in my head, or pull me around by my hair or hit me in my back and if he
was going to kick me, hed kick me in my back and stuff but mostly it was
like hed yell at me and be mean to me because...and then Id cry and hed
tell me that Im emotionally unstable and that theres no reason why a fucked-
up person like me should be able to have a baby. And I would just like and I
would just go and sit in a comer of the room and just cry. And he would just
be mean and mean and mean and mean. (Interview 115, page 10)
As illustrated by the above quote from interview 115, women also reported
psychological abuse. Psychological abuse ranged from verbal abuse such as
derogatory comments about their abilities in general and as a mother specifically:

He was more, telling me I wasnt a good mother to my kids. Real weird. Like
if [Brian] had a cough, hed go and buy him medicine. And insist that he had
to give it to him because if I did it, it wasnt the right dosage. Or the right
time, you know 1 wasnt doing it at the right intervals. Or, Im like excuse me?
Ive been a mother since I was eighteen. Hello? I know how to take care of
kids. Theyre all 7 years apart. I do know how to take care of kids and he just
insisted I was just a bad mother. (Interview 113, page 9)
To being watched or monitored:
He wanted to know where I was ALL the time. What I was doing ALL the
time. I couldnt even pee in our bathroom without leaving the door open. And
theres no way out of our bathroom.. ..Theres no way I can escape. And I
wasnt even allowed to talk to my mom on a phone in another room. I had to
sit next to him and talk to her to make sure I didnt tell her anything.
(Interview 115, page 8)
To being isolated from resources such as employment:
I felt like I was in jail. He didn't want me to go anywhere. Didn't want me to
talk to anybody, didnt want me to do this, didn't want me to go look for a
job. (Interview 103, page 7)
To being geographically isolated by their abusers from friends and family:
He was always controlling me. He was always scaring me: In this country,
nobodys yours. You are alone. Your parents are away from you. (Interview
101, page 6)
To restrictions on access to health care:
The second time, I had to go back, he didnt want me to go. He told me, Oh,
you dont need to see a doctor. Youre fine. And Im like, You dont seem to
understand. Im pregnant. Theres a life growing inside me. They need to
make sure that the child, the babys okay. And hes like, Youre okay. And
hes like, you know, he started pulling on my hair, and pushing me around.
And hes like, Youre fine. You dont have to go today. (Interview 108,
page 10)

And their choice of health care provider:
Every time Id see a doctor, every time I saw a doctor behind his back, its
like, What did you say? What did you say about me? And that got to the
point where I had to bring him every time I went to the doctor. You know, to
the doctor, or to see the welfare worker. And it had to be a female. It had to be
a female. It could not be a male. (Interview 112, page 8)
To imprisonment:
And I was like, I tried to hold (the key chain) real tight and then push the
door open. And I seen him moving around. And I closed it again and I sat
down on the couch. And he woke up and he looked at me and he goes, Im
hungry. Go cook me something to eat. So I said, Okay. You have to go to
the store for me though, because I need some tortillas and stuff. And he
looked at me and he goes, Okay. Im going to take the keys so I can lock you
in. (Interview 108, page 30)
To threats of physical violence directed specifically at her:
I went into the bedroom. And he just came toward me. And he always likes
to threaten me with a belt. Because he, you know, used to hit me with a belt.
And he was coming towards me and he just like started taking off his belt.. .So
I just started screaming at him. And he asked me, Where were you? I said, I
told you we were going to get groceries....I was standing on that side (of the
windowsill). He came on this side with the belt. He just hit that window real
hard. Just hard and like he was going to do that to me and he didnt. He just
really scared me. (Interview 105, page 11)
To threats of abuse aimed at her children:
(There were) pictures of different women that I didnt even know he had.
There were two of them like completely naked and stuff. And he had like,
panties. And I guess he had like, their initials on the back of their panties or
something. And the cop was looking and he said, Did you know anything
about this? And I was like, No. And then, when they finally got to the
bottom (of the drawer), I seen a panty of my daughters. Thats what scared
me. (Interview 108, page 16)

2. Changes in abuse during pregnancy
Among women in this sample, both physical and psychological violence
during pregnancy were highly variable. Some women believed both psychological
and physical abuse were sometimes directed at or motivated by the pregnancy. At
other times, women reported violence as status quo for the relationship, regardless of
her pregnancy status. It is important to emphasize that women who were interviewed
for this study were at different stages of separation from the abusive relationship.
Some had left their abusers while others had not. As a result women were at different
stages of thinking through the events and dynamics of the relationship. In addition,
women who had left the relationship often had received some counseling and
appeared to be more comfortable articulating their beliefs about the abusers
motivations for violence during pregnancy. Other women who had very recently left
their abuser were still clearly in crisis and while they were eager to talk about their
experiences, the chronology of events was not as clear. One woman, having left her
abuser only several days before the interview, stated: I guess its always been there,
the physical violence (Interview 108, page 4). Another woman reported the
psychological abuse became worse during pregnancy: With this pregnancy here, he
just really keeps an eye on me. And I dont like that... (Interview 105, page 8).
Other women who had left the abusive relationship reported a link between
abuse and pregnancy intention. These women reported less violence (both physical
and psychological) when their abuser wanted the pregnancy and/or felt sure of the

paternity of the pregnancy. Interview 109 compared her current abusive relationship
that resulted in the index pregnancy to her relationship with her abusive ex-husband
that produced two children. She concluded that the perpetration of violence during
pregnancy was highly individual but was also dependent on the fathers belief of
paternity and pregnancy intention. In her current abusive relationship she stated that
the abuser did not want the pregnancy:
I think it depends on the man. I really do. Because I dont think he would
have allowed me to carry this one. I think he would have beat me until I
However, in her past abusive marriage, she stated that her husband would
never hurt his own children since he believed they were his own. This woman
believed there was a differentiation in the violence she experienced and any potential
violence directed at the fetus. She was not seen as his and therefore was subject to
violence. In contrast, his children were seen as his as therefore not the targets of
Because he believed they were his. And when its theirs, theyre real, theyve
got that real possessive thing going...and he wasnt about to hurt what was his.
Period. Now, hed kill me. He wouldnt care. Okay, but he wouldnt hurt his
Another woman reported the happiest and most peaceful time of her
relationship was when her abuser was certain of his paternity:
He quit drinking. He quit smoking pot, whatever. And he stayed home, he
quit drinking. That was probably one of the best times we had together. He
didnt have any question about that [my pregnancy] at all. (Interview 110,
page 14)