UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Domestic violence research in Canada : an assessment and re-direction Brimner, Tammy Lanette 1997

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1997-0509.pdf [ 10.42MB ]
Metadata
JSON: 831-1.0087794.json
JSON-LD: 831-1.0087794-ld.json
RDF/XML (Pretty): 831-1.0087794-rdf.xml
RDF/JSON: 831-1.0087794-rdf.json
Turtle: 831-1.0087794-turtle.txt
N-Triples: 831-1.0087794-rdf-ntriples.txt
Original Record: 831-1.0087794-source.json
Full Text
831-1.0087794-fulltext.txt
Citation
831-1.0087794.ris

Full Text

DOMESTIC VIOLENCE RESEARCH IN CANADA: AN ASSESSMENT AND RE-DIRECTION by TAMMY LANETTE BRIMNER B.A., Trinity Western University, 1992 A THESIS SUBMITTED TN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Department of Anthropology and Sociology) We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA October 1997 © Tammy Lanette Brimner, 1997  In presenting this thesis in partial fulfilment  of the requirements  for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly, purposes may be granted by the head of my department  or  by  his or  her  representatives.  It  is understood  that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department of  An-bhr-opologV cxcvzk S o c i o l o g y  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  QoVobsY" I 4 ,  ffl7  ABSTRACT  Domestic violence has been identified as a problem for over twenty-five years. However, a number of reviews have demonstrated the lack of research in this area, especially in Canada. In order to encourage and provide direction for future research, past studies must be assessed. Attention must be given to both methodological and theoretical issues and the implications for domestic violence research. The focus of this thesis is the critique of three Canadian surveys. The purpose is: (1) to offer a critique of the methodology and the theoretical assumptions underlying each of these surveys, and (2) to propose revisions to advance the empirical research of domestic violence. The objectives are accomplished through a literature review of domestic violence and through a detailed assessment of the strengths and weaknesses of each of the surveys. This assessment revealed the need for more domestic violence research in Canada, making theoretical and methodological revisions to increase the reliability of these data and the validity of surveys as effective research tools. Theoretically, researchers need to consider a synthesis of existing domestic violence perspectives, emphasizing psychological and interactional theories, in order to provide a fuller understanding of the complexity of domestic violence. Methodologically, future research should focus on both men and women as victims and offenders, creating measurement tools tailored to each gender. Researchers need to further explore the perceptions of men and women about their intimate relationships and how domestic violence affects their lives.  TABLE OF CONTENTS ABSTRACT  ii  TABLE OF CONTENTS  iii  LIST OF TABLES  v  ACKNOWLEDGEMENTS  vi  INTRODUCTION  1  Purpose of the Study  5  CHAPTER ONE: A Theoretical Review of Domestic Violence  8  1.0 1.1 1.2 1.3  Introduction Criteria for Evaluating Theory The Theories Conclusion  8 9 10 27  CHAPTER TWO: A Literature Review of Domestic Violence  31  2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8  31 32 33 36 47 56 57 59 61  Introduction Overview of History of Domestic Violence Prevalence of Domestic Violence Domestic Violence Research Methodology The Findings of Domestic Violence Research Sexual Violence Dating Violence The Debate of the Battered Husband Conclusion  CHAPTER THREE: An Assessment of the 1987 All Alberta Study  65  3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7  65 66 68 87 88 90 91 94  Introduction Background of the Survey and its Objectives The Methodology of the Survey Ethical Issues The Findings The Quality of Measurement The Theoretical Assumptions Conclusion  C H A P T E R FOUR: An Assessment of The Study of Male-Female Dating Relationships in Canadian Universities and Colleges 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7  Introduction Background of the Survey and its Objectives The Methodology of the Survey Ethical Issues The Findings The Quality of Measurement The Theoretical Assumptions Conclusion  C H A P T E R FIVE: A n Assessment of the Statistics Canada Violence Against Women Survey 5.0 5.1 5.2 5.3 5.4 5.5 5.6 5.7  Introduction Background of the Survey and its Objectives The Methodology of the Survey Ethical Issues The Findings The Quality of Measurement The Theoretical Assumptions Conclusion  96 96 97 99 117 118 122 131 134  138 138 139 140 151 153 155 167 170  CHAPTER SIX: Proposed Revisions to Domestic Violence Surveys... 173 6.0 Introduction 6.1 The 1987 A l l Alberta Study 6.2 The Study of Male-Female Dating Relationships in Canadian Universities and Colleges 6.3 The Statistics Canada Violence Against Women Survey 6.4 Similarities Between The Three Surveys 6.5 Conclusion  173 174 180 185 188 194  C H A P T E R S E V E N : Conclusion  196  BIBLIOGRAPHY  206  V  LIST OF TABLES Table 1. Husband-to-Wife Violence: Comparison of Specific Acts for 1985 U.S. and 1987 (Canada) Alberta Surveys 89 Table 2. Incidence and Prevalence of Sexual, Physical, and Psychological Abuse 119 Table 3. Number and Percentage of Women 18 Years of Age and Over Who Have Experienced Violence, by Relationship of Perpetrator 154 Table 4. Number of Ever-Married Women 18 Years and Over Who Reported Violence by a Marital Partner, by Number of Occurrences, Canada, 1993 154  vi ACKNOWLEDGEMENTS I would like to thank my committee, Fiona Kay, Ken Stoddart and David Tindall. To my friends who have gone before me, Sara and Anne, for our early morning breakfasts. To my friend who follows, Chantelle, may the force be with you. To my friends and family, for all of your support. To Morticia and Wednesday, for helping me believe that it is possible to have it all — I can be a wife, have a family and a career. And lastly, but most importantly, to Bill, for being my best friend: sic gorgiamus alles subjectus nunc!  1  INTRODUCTION  Domestic violence, or more specifically, violence against women has been identified as a social problem for over twenty-five years (McCall and Shields 1986:98). However, women struggled to move the problem of such violence into the social and political arenas decades earlier. More specifically, wife abuse, stranger violence and dating violence have become topics of research and debate both inside and outside academia. Within each of these areas, varying levels of violence have been differentiated. Physical, emotional, and sexual abuse by men are all factors constituting violence against women. At this point, the term "domestic violence" should be defined. For the purposes of this thesis, "domestic violence" is interpreted as physical, sexual, and psychological abuse between intimates in a relationship, whether dating, married or cohabiting. Some scholars have objected to the term domestic because they feel that such terminology obscures the dimensions of gender and power that are fundamental to understanding the abuse of women (McHugh 1993:58; Breines and Gordon 1983; Schechter 1982). On the other hand, terms such as wife abuse and wife battering are not sufficiently inclusive because many women are battered by intimates in nonmarital relationships (McHugh 1993:58). My use of the term "domestic violence" is not to diminish the reality of violence against women but to acknowledge the existence of violence against men. I argue that it is only through this acknowledgment that a fuller understanding of domestic violence can be attained.  2  It is also important to differentiate between domestic violence and other types of violence, such as stranger violence. There are a number of characteristics of domestic violence which set it apart from other violence. O'Leary (1988a:36) has outlined six differences: First, domestic violence occurs within the social group known as the family, which is markedly different from other small groups. In the family, statuses and roles are assigned on the basis of age and sex rather than interest and competence which is the case in most social groups. Second, although there are conflicting normative expectations within a relationship about violence, there is a long-standing right or obligation to use force on a wife or girlfriend by her husband or boyfriend. Thirdly, relationships carry considerable commitment, making it difficult to leave the situation if violence occurs. Commitments, such as legal, moral, financial, and affective commitments, make it especially hard for individuals to leave the relationship even though violence has occurred. Fourthly, the emotional involvement which characterizes most intimate relationships may actually be a partial cause of some of the domestic violence. Fifthly, until very recently domestic violence was seen as a private matter, resulting in little research or legal attention. Lastly, intimate relationships are highly stressful units because of the inherently unstable nature of the groups (O'Leary 1988a:36; see also Straus 1977a). For the past twenty-five years, researchers, under the guidance of a number of different theories, have been making progress in understanding violence against women. Although progress has been made, there has been neither consensus on the propositions nor consistency in the findings. There is still much more research required to fill in the gaps for a complete picture of this problem (DeKeseredy 1993:10; Johnson and Sacco  3 1995:282). In order to fulfill this requirement, research methods and questions need to be revised and new ideas explored. Past studies must be assessed in order to generate and direct future research. M y interest in violence against women began in my post-secondary education when I was confronted with both academic and public discussions on the subject. However, the focus of my interests changed when I discovered that domestic violence was not always the case of violence against women, as demonstrated by the details of my brother's marriage.  1  I was not aware of my brother's failed marriage until he announced that his wife had moved out and that they planned to divorce. My brother has always been a man of few words, refusing, or perhaps unable, to speak about his problems. After his wife of eighteen months had left, he began to talk, sharing with me the details of their stressful relationship. She had withdrawn sexually and emotionally after being married for only a few days, although she had been involved in all respects prior to their wedding day. She refused to help with housework in their apartment, refused to get a job and continued to spend recklessly the little money they had. My brother cleaned their home, cooked their meals, did the laundry and worked long hours to pay for their expenses. His requests for her help were answered with degradation; my brother was continuously humiliated in front of friends and family by the woman he loved. In her anger she would slap and punch him, leaving physical and emotional bruises. M y brother's attempts to ask her to stop were futile. He began to threaten to hit her if she did not stop hitting him. Even the threats were not effective. The day he  slapped her arm to end her violence, she left. She called a shelter and was given a room. She was given free counselling, government funding for an apartment, and employment assistance. When my brother was at work, she cleared out their apartment, leaving him with his clothing and little furniture. She began calling him from the shelter, accusing him of abusing her, the severity increasing with each call. Out of concern for my brother, I began contacting organizations on his behalf, disheartened to realize the lack of resources and sympathy for my brother. My brother had threatened to hit her and had slapped her in the arm, which are signs of woman abuse. However, the knowledge that his actions were a direct response to her verbal and physical abuse brought a different perspective to the rally cries against woman abuse. Her actions were, perhaps, in response to her abuse as a child, not to the actions of my brother. I empathize with this young woman whose childhood was stolen from her; the love my brother bestowed upon her was not enough to help her recover from those wounds. My heart still goes out to my brother who is healing from the emotional wounds and he is learning to love and trust again. Now my mind has begun the search for understanding in domestic violence, with the realization that what is known today is still not enough to fully understand the complex dynamics of domestic violence.  5 Purpose of the Study This thesis presents an assessment of Canadian research on domestic violence, and more specifically, violence against women. In particular, three Canadian surveys will be discussed: Statistics Canada's Violence Against Women Survey (1993); Kennedy and the Population Research Laboratory's All Alberta Study (1987); and DeKeseredy and Kelly's study of problems in male-female dating relationships in Canadian colleges and universities (1992). These studies were selected for the following reasons. All three surveys investigate domestic violence in Canada. All three surveys focus on violence against women, although two of the surveys begin to explore men's experiences with domestic violence. Each survey represents a different population, from a national population of women to a provincial population of men and women to a national college and university population of men and women. Each survey uses different survey modes, including telephone interviews, face-to-face interviews and self-administered questionnaires. The lack of domestic violence research in Canada has been noted by comprehensive reviews, such as DeKeseredy (1993), DeKeseredy and Hinch (1991) and Smith (1987, 1990a). These reviews have revealed the need for much more empirical work because so few studies have been conducted. Due to this deficiency, the choice was limited, however, the three surveys chosen represent the typical methods used within this field of research and also offer variation in sample populations. For each survey, the strengths and weaknesses will be identified, in order to propose revisions for future research. Prior to this assessment, the theories of domestic violence will be outlined and discussed in the first chapter. This discussion will address the need for the revision and synthesis of current theories. Chapter Two will consist of a  6 literature review of domestic and dating violence research in Canada and the United States. This review will discuss the current use of self-reports as the only methodology in domestic violence research and highlights the various findings. The literature review will draw attention to the need for further research, to build a more reliable and more valid data base. Each of the three surveys will be assessed in Chapters Three through Five, critiquing the methodology, outlining the findings, discussing the theoretical assumptions and proposing directions for future surveys. Chapter Six will include proposed methodological and theoretical revisions, as evidenced by a comparative analysis of the three surveys. The final chapter, Chapter Seven, will conclude with further proposed revisions, indirectly related to the three surveys, for future research on domestic violence. In sum, this thesis provides a sociological analysis of the strengths and weaknesses of Canadian research on domestic violence. Because there is limited research on domestic violence in Canada, this thesis provides methodological direction for further studies through the analysis of the social survey as an effective research tool and the consideration of variations of research methodologies. Problems of the current dominant theories and the consequences for building our knowledge of domestic violence are highlighted. The theoretical review shows that most perspectives contribute to our understanding of domestic violence, but that none have been proven to be sufficient to provide an effective explanation of domestic violence (Carden 1994:563). The range of theoretical applications will be expanded through the exploration of possible syntheses of different theories. Throughout these methodological and theoretical discussions, the need for a theoretical synthesis and an improved conceptualization of definitions and ideas will be demonstrated. The revisions suggested will provide further directives for the  7  sociological study of domestic violence not only in Canada, but also in other countries and in other disciplines. 1 have included the story of my brother in order to acknowledge my personal bias to my readers. I acknowledge my prejudice as a starting point, knowing that "[prejudice is not arbitrary understanding but, rather, a knowledge of what our prejudices are, an understanding that involves and entails critique" (Hekman 1990:15). 1  8 CHAPTER ONE: A THEORETICAL REVIEW OF DOMESTIC VIOLENCE  1.0 Introduction There are a number of different theories used to explain domestic violence. Within the domestic violence literature, there are ongoing debates regarding the legitimacy of each theory. Each theory provides an explanation of at least one aspect of domestic violence, but none provides an encompassing perspective. There is also disagreement regarding the classification of these theories (Bersani and Chen 1986; Carden 1994; Straus 1994). The boundaries between different theories are often blurred; it becomes difficult to distinguish one theory from another. One explanation for the lack of distinctions is the underlying feminist perspectives, which can be identified in many discussions of domestic violence, especially violence against women by men. Feminist perspectives have influenced all theories of domestic violence, making connections between the theories and thus blurring their boundaries. In identifying gender as a social category, feminist perspectives explore many areas, including organizations, health, race and ethnicity, social stratification, and education (Lengermann and Niebrugge-Brantley 1990:319). The feminist perspective will be discussed later in this chapter, but it is important to note that feminism plays a significant role in domestic violence theories and research. Due to the scope of this review, not all theories will be addressed. According to Lewis Okun (1986), there are approximately thirty-five theoretical perspectives of domestic violence. However, the selection of the following theories is more appropriate  9 to discuss the different perspectives of domestic violence. This review is a reflection of the various debates in domestic violence literature, and also a preview of the theoretical assumptions underlying the three surveys under study. Each theory will be described by identifying its major tenants in relation to domestic violence. The theories are as follows: 1. 2. 3. 4. 5. 6.  Feminist Theories Sociological/Structural Theories Interactional Theories Learning Theories Alcohol Abuse Theories Other Theories  1.1 Criteria for Evaluating Theory  In order to fairly and objectively assess theories, it is necessary to have a set of criteria for the evaluation of those theories. There is still much debate over the best strategy for building theories but there is some consensus about the characteristics of a good theory. According to Hergenhahn (1982:21), theories should have the following characteristics: 1. A theory synthesizes a number of observations. 2. A good theory is heuristic, that is, generates new research. 3. A good theory must generate hypotheses that can be empirically verified. If such hypotheses are confirmed, the theory gains strength; if not, the theory is weakened and must be revised or abandoned. 4. A theory is a tool and as such cannot be right or wrong: it is either useful or not useful. 5. Theories are chosen in accordance with the law of parsimony: of two equally effective theories, the simpler of the two must be chosen. 6. Theories contain abstractions, such as numbers or words, which contain the formal aspects of the theory. 7. All theories are attempts to explain empirical events and they must therefore, start with empirical observations.  10 Each theory will be reviewed and assessed using this set of criteria.  1.2 The Theories 1.2.1 Feminist Theories  According to Gelles (1993:41), feminist theory is becoming the dominant perspective in explanations of violence toward women, as evidenced by the three surveys to be discussed in the following chapters. Domestic violence is viewed by feminists as an "expression of social power" (McCall and Shields 1986:107). It is considered a typical result of the male socialization process, in which domination of women is reinforced in our society (Dobash and Dobash 1979:6). Our society has been defined as patriarchal, a "system of social relationships that privileges men while disadvantaging women, although the precise form of power inequities varies historically and cross-culturally" (Creese and Strong-Boag 1992:2). Feminists empirically observe the legal, social, religious and medical means that support the inequality of women in society (Davidson 1978; Dobash and Dobash 1992; O'Leary 1993; Yllo 1993). They also observe that many women have not achieved the political, economic, and social independence that would empower women to leave violent relationships. Feminists criticize the mass media for romanticizing violence, male dominance, and female submissiveness (Carden 1994; Dines 1992; Jarvie 1991; Kuhn 1985). As already noted, feminist perspectives synthesize a number of empirical observations, and uses terms such as "patriarchal society," "social power," "inequality," "male domination" and "female submission." One of its strengths is the use and promotion of qualitative methods, which are primarily identified with feminist theory  11 within domestic violence research (Breines and Gordon 1983; Murphy and O'Leary 1994; Yllo and Bograd 1988). The use of such methods have been integral in making feminist theories heuristic. Feminist theories have contributed a great deal to domestic violence research. For example, feminists have encouraged the use of women's experiences as new empirical and theoretical resources, providing social science with a new purpose, and also the location of the researcher in the same critical plane as the overt subject matter (1987:6-8). Feminists have also added to domestic violence research by making victims visible in the research and by making distinctions based on gender (Renzetti 1994:196). These innovations have been responsible for providing a wealth of knowledge about domestic violence, especially violence against women, which had previously been either ignored or unknown. Feminist theories which provide a perspective on domestic violence are espoused by researchers such as Clifton (1982), Dobash and Dobash (1983), Dobash, Dobash and Cavanagh (1985), Gondolf (1985), Greenblat (1983), Kincaid (1982), Pressman (1983), and Washburne (1983). One criteria of a good theory is that is must generate hypotheses which can be empirically verified. If such hypotheses are confirmed, the theory gains strength as a result. According to Gelles (1993:41), feminist theories have generated diverse, yet consistent, empirical support for the proposition that gender inequality contributes to the violence toward women. Gelles refers to a number of recent studies such as Morley (1994) and Straus (1994) by researchers who used different methodological approaches to study diverse populations. These studies have all found that gender inequality explains variations in the incidence and rates of violence toward women (Gelles 1993:41).  12  However, feminist theories have also generated hypotheses which cannot be empirically verified, indicating weakness in the theory. For example, feminist theories propose the hypothesis that patriarchal values encourage men to abuse women, especially within relationships. However, O'Leary et al. (1989) demonstrate that there are young married men holding patriarchal values who do not abuse their wives. There are also a number of studies which suggest, or directly indicate, that physical aggression against women is viewed as socially undesirable (Arias and Johnson 1989; O'Leary 1993:23; Riggs, Murphy and O'Leary 1989). Gelles (1993) states that the main problem with the feminist perspective is that it uses a single variable, patriarchy, to explain the existence of wife abuse, while excluding, for example, other salient and important aspects of social structures and social institutions. Such exclusions render feminist theories ineffectual in the explanation of other areas of domestic violence, such as husband battering and lesbian and gay battering. Since husbands are viewed as the dominant member of marital relationships, feminist theories fail to address relationships in which the woman is the dominant member of the relationship (McCall and Shields 1986:108; Gelles 1993:43) or in which a relationship is composed of same-sex partners. In defense of feminist theories, researchers and theorists have been almost exclusively concerned with those violent relationships in which the perpetrator is male and the victim is female (McCall and Shields 1986:108). Feminist theory is an analysis of only one type of violence or victimization (Gelles 1993:43), and as such, must be given that allowance. Problems arise, however, when theorists focus on the use of violence by men in general because it will necessarily emphasize broader social forces that  13  differentiate men from women and will de-emphasize differences among men (Dutton 1994a: 168). Dutton (1994a: 168) argues that the focus on violence by men in general has manifested itself in a feminist focus on patriarchy, male concern for power (Walker 1989), and larger social factors, rather than on personal factors that might differentiate one male from another. O'Leary (1993:23) argues, in his own analysis of the problem of wife abuse, that "a patriarchal society is a critical but not a sufficient risk factor in the development of spouse abuse." Gelles (1993:43) also argues that although feminist perspectives are politically attractive and amenable to broad social action, it does not provide a useful theory to explain the complex nature of domestic violence. Dutton (1994a: 177) argues this point by posing the question, "if patriarchy 'causes' violence, how can we hold men individually responsible for their violence?" In light of these conclusions, if feminist theories are to continue to advance our understanding of domestic violence, then revisions must be made.  1.2.2 Sociological Theories According to Gelles (1993:31), the underlying assumption of sociological theories is that social structures affect people and their behavior. In the case of domestic violence, the structure of the modern family as a social institution has a strong presiding influence on the occurrence of family violence. Sociologists focus on organizational features to provide evidence of the family as a social institution, such as: high time at risk (because family members spend so much time together, they are at substantial risk of becoming victims of violence); a wide range of activities and interests  14 among members of the family; great intensity of involvement with family members; competing and infringing activities of various family members; family membership rights to influence; age and sex discrepancies leading to conflict; ascribed roles; family privacy; involuntary memberships; high levels of stress; and extensive knowledge of social biographies (Hotaling and Straus 1980:15-17). The family has been studied as an institution since the late 1800s and early 1900s, demonstrating the impact upon the family of such issues as poverty, child labor, divorce, desertion and women's suffrage (Bradbury and Fincham 1990:40). Social structure is defined by the "relations between and among social categories of individuals. To the extent that these relations define inequality, subordination, and superordination, the aspect of social stratification figures prominently in any account of social structure" (McCall and Shields 1986:100). Sociological theories of domestic violence are espoused by researchers such as Murray Straus (1973, 1977a, 1977b, 1980a, 1980b, 1980c, 1980e), Richard Gelles (1973, 1983 1993), and Suzanne Steinmetz and J. S.Lucca (1988).  1  As mentioned earlier, there is a link between feminism and other theories of domestic violence. The link between feminism and sociological theories is especially strong when considering that the major focus of feminists is the identification of patriarchy as the primary force behind violence against women. Feminist sociologists have created a distinctive view of social organization currently anchored in five ideas: (1) that gender, like class, is a pervasive system of stratification; (2) that stratification in any society arises from the organization of social production;  15 (3) that social production has to be broadly construed as a multifaceted, hierarchically organized process that reproduces and maintains social life; (4) that analyzing this process effectively requires suspension of the convention of categorizing phenomena as micro- or macro-social, and the creation of a vocabulary to describe the fluidity of the relation between interactional and structural arrangements; and (5) that ideology masks the reality of social production while reproducing stratification in the intricacies of personality and social interaction (Lengermann and Niebrugge-Brantley 1990:324). According to feminist views that most violence, at least most serious violence, is perpetrated by men against women, violence emanates from the patriarchal organization of society, in which men dominate and control women. Men's use of violence reflects their greater relative power, authority, and status in society (McCall and Shields 1986:100).  2  Besides feminism, there are a number of other theories within sociology, such as culture, social institutions, social systems, resource, and social roles theories, that place domestic violence within a sociological context. Culture theories have been used by analysts, such as Gil (1970), Huggins and Straus (1980), Steinmetz (1977), and Straus (1980a), who argue that our culture legitimates, inspires and reinforces the use of violence in the family (McCall and Shields 1986:100). Social institution theories identify the family as a social institution in which the reproductive, sexual, and nurturant relations between individuals and generations are regulated (McCall and Shields 1986:101). Domestic violence, taking place within the family across societies, is therefore an area of research to be addressed by social institution theories.  16 Another set of sociological theories identifies the family not as a social institution, but as a system within the larger system of society (McCall and Shields 1986:102). According to systems theory, a system is "an interdependent group of components, with a distinguishable boundary" (Giles-Sims 1983). These theories are based on the observations that the family is affected by the larger society, such as culture and religion, and as such constitutes an open system (McCall and Shields 1986:102). Thus, domestic violence taking place within the system of the family is influenced by the systems of society. Resource and exchange theories are a more specialized social-systems account given by domestic violence researchers such as Brown (1980), and Gelles (1974, 1983).  3  These theories portray violence as a resource used to restore a dominant status which is perceived to be threatened. For example, in the case of woman abuse, when a man does not receive the rewards he expects (e.g. socio-economic), or he feels that his status is threatened, he resorts to the use of violence (McCall and Shields 1986:102). Resource and exchange theories focus on the status and resources of a husband and wife and attribute wife assault to changes in relative amounts of status and power, whereby the wife's status and resources are greater than those of her husband or to the wife's lack of resources and economic dependence upon her husband (Frankel-Howard 1989:63). Such theories are based on observations that women have traditionally been economically poorer than men, but have been challenging this tradition by entering the workforce, gaining in social stature and independence and placing themselves at odds with men (see Calzavara 1988).  17 Another set of sociological theories involves social roles constituted by behaviours expected of persons by virtue of their occupying a particular social position and how these persons powerfully influence the orderliness of social life (McCall and Shields 1986:103). There are a number of applications of social role theories in relation to domestic violence. One of the most common applications is sex-role theory in which traditional childhood sex-role socialization prepares boys to become aggressive and violent, and girls to become submissive and victims of men (Fox 1980; Walker 1979). A second application is family-role theory which focuses on failures of role performance within the family (McCall and Shields 1986:104); family-role theory has been identified by batterers as a factor in domestic violence. Another application involves the argument that if a person is victimized, an identity and role of the victim is likely to develop, making this person more vulnerable to further victimization (MacFarlane 1978:82; Shields and Hanneke 1984). Sociological theories have gained strength through research and debate. Gelles (1993) argues that sociological perspectives provides the widest and most inclusive perspective from which to understand and explain domestic violence, without excluding or diminishing the contributions of psychological or social psychological variables (p. 43). He argues that the sociological perspective places these other variables within a wider explanatory framework that considers the impact of social institutions and social structures on social behavior. According to Gelles (1993:43), sociological theories offer a more complex explanation of domestic violence and is applicable to a wider range of victimization than is feminist theory.  18 However, the fact that sociological theory offers a more complex formulation leads to one of its weaknesses, identified by Gelles himself: because sociological perspectives do not focus on a single characteristic of social life (e.g., personality or gender inequality), sociological theories are by definition complex (1993:43). If the theories are complex, the practical value of such theories is lessened because they do not lead to simple solutions, either in clinical settings or in terms of social policy. He also offers that social factors are often considered irrelevant, or not as relevant as other factors, such as psychological factors, in explaining domestic violence because "family violence can be found in all social groups and in all income levels" (p. 40). Anecdotal examples of violence in wealthy and professional families are also offered as evidence that social factors play only a minimal causal role in domestic violence (Gelles 1993:40). Although domestic violence studies continue to base research questions on sociological assumptions, as demonstrated by the three surveys to be assessed in the following chapters, researchers must be aware of the weakness of sociological theories. Another set of theories which attempt to provide an explanation of domestic violence is interactional or conflict theories, which will be discussed in the next section.  1.2.3. Interactional Theories  Interactional theories or symbolic interactionism view the family as a context for interaction among its members (Bradbury and Fincham 1990:40). Men and women learn to mutually read each other, to anticipate each other's responses, and to adjust to each other. George Herbert Mead termed this capacity 'taking the role of the other,' or role taking, which is the ability to see the other's attitudes and dispositions to act (Turner  19 1991:393). Mead took this abstraction to investigate interpersonal action problems, such as those common to intimate relationships, in which one person is himself or herself a source of stimuli for the partner. This person must, therefore, be attentive to his or her actions, since those actions elicit reactions from the partner and also determine his or her continuing actions (Joas 1987:91).  4  Using the interactional model, some research focuses on how conflict arises out of social interaction. In applying this model to domestic violence, conflict is inevitable in social relationships, and the more the conflict, the more likely violence will be used as a problem-solving tactic (Goode 1971:632; Hepburn 1973:420). One group of interactionists focus on the violence arising from conflict and emphasize the importance of meaning to both parties involved in the violent interaction, such as how people decide when a line of conduct is violent or how threats are perceived (Chatterton 1976:28; Ferraro 1984). Other research, however, focuses on "communication for the purpose of solving problems of collective concern", which becomes an essential condition of social order (Joas 1987:92), especially within intimate relationships.  5  Feminism is also linked to interactional theories. For example, Pagelow (1981:65) made note of theories which view women as having superior verbal skills and men as being frustrated and angry because they cannot match them verbally. Such "wars" are another indicator of men struggling to maintain the patriarchal hierarchy, the basis of feminist theories. Another example involves the "nagging" wife which has been traditionally viewed as deserving of punishment and also cited as a reason and justification for wife assault (Frankel-Howard 1989:62). MacLeod (1980:24) describes female victims in violent relationships as verbal tormentors who nag, argue over drinking  20 and gambling and criticize the sexual performance of their aggressor, until the aggressor is deprived of self-control and reacts violently towards his tormentor. Interactional theories provide direction for domestic violence research through the focus on verbal and nonverbal communication which occurs prior to and throughout the continuation of the violence. The difficulties in recording such interaction poses a challenge to researchers both methodologically and ethically — to know of domestic violence and to neglect to offer help in the desire to obtain data is extremely unethical. Thus researchers must be creative in their methodological choices, such as third party observations which may reveal communication patterns in violent relationships. Interactional theories still have the potential to contribute to the knowledge of violence between intimates. Researchers are better able to analyze the similarities and differences between men and women in communication which lead to violence, using interactional theories. According to Cantos, Neidig and O'Leary (1994) and Cascardi, Langhinrichsen and Vivian (1992), there are differential impacts of the physical and verbal abuse of men and women which need to be studied from all perspectives, including the interactional model. Such theories underlie the Conflict Tactics Scale used in the three surveys to be discussed in the following chapters. However, more assumptions can be made from these theories in future domestic violence research.  1.2.4. Learning Theories Domestic violence is also studied from the perspective of learning theories. Learning theories, or more specifically, social learning theory, were originally discussed by Albert Bandura (1969) in Principles of Behavior Modification and later given more  21 complete form in his book, Social Learning Theory (1977). Bandura provided the first major theoretical exposition of learning theory which incorporated "three major conditioning, classical condition and cognitive mediational processes" (O'Leary 1988a:32). According to social learning theory, most human behavior is learned through modeling (p. 33). Terms used in such theories include "modeling" and "conditioning."  6  Learning theories help to explain the origin of violence, and are linked to feminist theories. Using these theories, researchers attempt to show how violence is learned in the home, as children watch their parents engage in violent behaviour, and is then a factor in these children's own development as adults. This theory has also been referred to as the "intergenerational transmission theory of family violence" and stresses that family violence is learned through modeling, in which one person emulates the behaviours of someone he or she admires and/or respects, such as a father or mother (McCall and Shields 1986:104). Admiration and respect are not, however, the only factors encouraging behaviour emulation. In domestic violence research, such emulation may occur when violent behaviour is the only method of anger and conflict resolution demonstrated by the parents. These theories can also be expanded to include the transfer of attitudes. In relation to the study of domestic violence, especially violence against women, findings suggest that physical aggression is much more likely when attitudes of an individual are shaped that support the use of physical force against a woman (O'Leary 1993:24). The strength of learning theories arises out of a substantial body of research which has established a strong relationship between being abused by parents and/or witnessing inter-parental violence as a child, and being violent toward an intimate partner as an adult  22 (Carden 1994:531). O'Leary (1988a:34) stipulates the conditions under which learning 7  theories are effective in explaining behaviour, creating boundaries for these theories: people will model behavior if they attend to the modeled event, if they remember that event, if they can physically perform the specific behavior, and if they believe that engaging in such behavior will have positive consequences for them. Learning theories such as intergenerational transmission theory have been successful in generating considerable empirical research on many forms of family violence. However, some argue that its explanatory value is not yet clear. Although a substantial body of research has established a strong relationship between suffering abuse or witnessing interparental violence as a child, and being violent toward an intimate partner as an adult, there are many individuals who see their parents hit each other and do not engage in spousal aggression (O'Leary 1988a:44; Tyree, Malone and O'Leary 1987). Straus (1980c) adds that although physical punishment is very widespread, not all physically punished children grow up to be abusive toward their partners. These theories appear to be more explanatory of attitudes that encourage violence against women than of the assault itself (Frankel-Howard 1989:60). Others point out that statistical associations between behaviors in one generation and the next within a given family do not support an assumption of direct or inevitable causality. There is ample empirical evidence that violence in a man's family of origin is neither a sufficient nor a necessary condition for his violence in his intimate relationships (Hastings and Hamberger 1988; Kalmuss and Seltzer 1986; Schuerger and Reigle 1989; Straus and Gelles 1986; Straus, Gelles and Steinmetz 1980). These results strongly suggest the need for multiple predictors of domestic violence, not just those proposed by learning theories (O'Leary 1988a:44;  23 Tyree, Malone and O'Leary 1987). Learning theories are directly relevant to the three surveys to be assessed.  1.2.5. Alcohol Abuse Theories Other theories explaining domestic violence use alcohol abuse as a predictor. Alcohol abuse theories arise out of the observation that women are often the victim of violence when their perpetrator has been abusing alcohol (Bard and Zacker 1974; Hanks and Rosenbaum 1977; Brisson 1981; Bern 1985; Flanzer 1987). The belief that alcohol abuse by men is a major cause of wife beating has been around at least since the Temperance movement (Kantor and Straus 1990:203; see also Bordin 1981; Pleck 1987). There are two main theories of alcohol abuse. Disinhibition theory, as its name indicates, suggests that disinhibition, caused by alcohol abuse, is the cause of domestic violence. Disinhibition is regarded as a "complex process resulting from alcohol's pharmacologically induced 'cognitive disruption' (Leonard 1984:79) and mood-altering effects (Blum 1981) interacting with varying individual expectancies about alcohol's powers" (Kantor and Straus 1990:204; see also Marlatt and Rohsenow 1980; Sher 1985). The second theory is an offspring of social learning theories. Social Learning and Deviance Disavowal theory has been suggested as another possible factor in domestic violence. MacAndrew and Edgerton (1969:90) argue that drunken behaviour is learned and may take the form of a 'time-out.' Individuals who take time outs by drinking are exempted from the usual behavioral constraints associated with sobriety (Kantor and Straus 1990:04).  8  24 There appear to be no larger theoretical explanation in alcohol abuse theories beyond the correlation of domestic violence and the presence of alcohol abuse. O'Farrell and Murphy (1995) report de-elevated levels of violence after effective alcohol abuse treatment and elevated levels after non-effective treatment. The complexity of the association between alcohol and violence has been demonstrated by identifying some of the factors which may intervene to determine alcohol-violence outcomes: "The symbolic meaning attached to alcohol use and expectancies about alcohol's effects; contextual factors present in the setting or the interaction of individuals; and perceptual and cognitive changes produced by alcohol" (Kantor and Straus 1990:205). With so many interrelationships, researchers are faced with a major difficulty in attempting to disentangle these relationships (Kantor and Straus 1990:205). Current studies report finding that, while alcohol or drugs may be used by an abuser, their function is to only facilitate or excuse the violence rather than cause it. Alcohol and drugs affect the nature, timing and extent of the violence, but not the propensity to resort to violence (FrankelHoward 1989:13). Although there is little supporting evidence to support alcohol abuse theories, they assist theorists in identifying the role played by alcohol in situations of domestic violence. Research continues to ask questions about the role of alcohol in domestic violence, as shown by two of the surveys to be discussed. There are other theories that warrant discussion, although they are not explicitly integrated in the three surveys to be discussed.  25 1.2.6. O t h e r Theories  One further set of theories which warrants discussion is psychological theories. Psychological theories have also been used to interpret domestic violence. Proponents of these theories believe that perpetrators of violence in relationships suffer psychological problems that play an important role in their abusive behavior (O'Leary 1993:24). These theories rely on observations of violent actions which fall outside the boundaries of acceptable behaviour; for example, it is a assumed that people who are violent toward their spouses or children must be mentally ill (Frankel-Howard 1989:12). The theories are proven useful and supported by studies, such as Hamberger and Hastings (1986a:338), which report that men in a domestic violence treatment program exhibit three personality profiles, as assessed with the Millon Clinical Multiaxial Inventory: schizoid/borderline, narcissistic/antisocial, and possessive/ dependent/ compulsive. Such studies allow for the identification of men who are more likely to be violent toward their female partners in order to plan prevention and precaution. Other studies, such as those on closed-head injuries, show how damage to the frontal and temporal lobes tends to significantly affect personality and social adjustment (Kolb and Whishaw 1989:139): Despite residual capacities that are often considerable, one combination or another of impaired initiative and apathy, lack of critical capacity, defective social judgment, childishness and egocentricity, inability to plan or sustain activity, impulsivity, irritability, and low frustration tolerance is likely to render these patients unemployable or only marginally employable. These same qualities also make a person who has sustained moderate to severe head injuries at best a nuisance at home, at worst a terror. By virtue of these qualities, and again despite their residual capacities, these patients are rarely able to form or maintain close relationships, so that those who have not been rendered silly and  26 euphoric or apathetic by their injuries, are often lonely and depressed as well (Lezak 1983:170). Psychological theories have provided researchers with direction for future research, with the testing of new hypotheses regarding psychological profiles not yet refined and the retesting of established hypotheses to strengthen the theories. However, these theories are often excluded in domestic violence research. Psychological theories have influenced studies, based on self-reports, observations by prison officials and tests by therapists, which report that as the level of physical violence increases, the greater the likelihood that some personality style, trait, or disorder will be associated with that violence (O'Leary 1993:25; see also Dutton 1994b; Dutton and Hart 1992a, 1992b; Hastings and Hamberger 1988). Dutton characterizes this psychopathology typically with extreme scores on borderline personality organization, narcissism, antisocial behavior, and aggressive-sadistic personality (Dutton 1994a: 176). In studies of assaultive males, the majority of both court-referred and self-referred men exhibited diagnosable psychopathology, typically personality disorders (Dutton 1994a:176).  9  This finding allows for new hypotheses about severe violent offenders and comparing these offenders to, for example, occasional violent offenders which may be characteristic of most or some violent relationships. O'Leary (1993:26) has suggested that at the severe end of the continuum of physical violence, the percentage of men who have alcohol abuse problems and/or personality disorders is much higher than that found in general populations or in maritally discordant populations. The principal theory of causation in family violence was, for many years, psychopathology, because of the belief that people who hurt their spouses and/or children  27 must have psychological problems. Feminist researchers rejected this theory on the grounds that these violent offenders were not mentally ill, but rather were seeking to maintain their power in the family or relationship (Gillespie 1971; O'Leary and Curley 1986). Although violent offenders can be both mentally ill and seeking to maintain their power status, feminists ignore the impact of mental illness on domestic violence. This belief has been and continues to be reiterated in the literature on family and domestic violence. Others point to psychological explanations which minimize the contributions of social and structural factors to the occurrence and persistence of violence and abuse in family and intimate relationships (Gelles 1993:40). O'Leary (1993:26) maintains that, even at lower, almost normative levels of physical aggression, the role of psychopathology or personality traits is small but often statistically significant. While psychological theories propose exclusively psychological factors, researchers have made consistent findings in their studies testing these theories, which prevents their dismissal and warrants their inclusion in future sociological research.  1.3 Conclusion In this chapter, I have outlined five dominant theories and have discussed these theories in terms of their application to domestic violence research. It is these theories, especially feminist, sociological and social learning theories, which provide the foundation for the three surveys in the following chapters. Explanations range from identifying woman abuse as a problem in a patriarchal society (feminists and sociologists) to identifying communication and behaviour patterns (interactionists and learning  28 theorists) to acknowledging the effects of alcohol abuse. Each theory has identified one factor or set of factors to help expand our knowledge of domestic violence. I argue that while these perspectives appear to be necessary, not one has been demonstrated to be sufficient in and of itself to support a level of understanding which will inform an effective response (Carden 1994:563). With this knowledge, it is then necessary to create new hypotheses or revise old ones, in order to re-direct discussions about domestic violence. A number of analysts such as Finkelhor (1979, 1984), Gelles (1978, 1990, 1993), Gil (1970) and Straus, Gelles and Steinmetz (1980) argue that violence in the home is multidimensional and requires multifactor causal models to explain such violence (McCall and Shields 1986:106). These multifactor causal models are created by combining different perspectives. Such models are relatively new to domestic violence research, although this concept has been discussed to some extent in the literature, such as Carden (1994) and those analysts mentioned above. An inclusive integration of theories or interdisciplinary perspectives would also prove beneficial. One example of a multifactor causal model is Dutton's (1985) ecologically nested theory of male violence toward female intimates, which consists of a combination of sociopolitical and psychological perspectives. His theory proposes that violence is determined by an intricate network of variables within the individual, the family, the community, the culture, and the species, and that each of these is nested within one another. Dutton argues that a combination of factors from each level produce violence against women by men (1985:404). Although a number of domestic violence theorists and researchers have suggested the use of multifactor causal models, few have actually proposed new theories as Dutton  29 (1985) did. Fewer still have actually tested these proposed theories, such as McKenry, Julian and Gavazzi (1995) who tested a biopsychosocial model of domestic violence. But if our knowledge and understanding of domestic violence is to expand, new theories need to be advanced. According to theorist Jonathan Turner (1991), new theories are necessary in order to eliminate all possible lines of inquiry, thus advancing science. This advancement is the result of what Turner calls "good theoretical statements" which are created to be proven wrong; if a theory, in principle, cannot be proven wrong, than it is not very useful. Such theories are then a "self-sustaining dogma that is accepted on faith. By successively eliminating incorrect statements, those that survive attempts at refutation offer, for the present at least, the most accurate picture of the real world. Although having one's theory refuted may cause professional stigma, refutations are critical to theory building" (Turner 1991:23). In the following chapters, the theories discussed in this chapter will be analyzed in order to show how theory informs research, and how theory can be used to advance future research.  Notes: Terms such as "societies," "cultures," "roles," "status," "institutions" and "social systems" are used in sociological theories. For examples of feminist sociologists, see Dobash and Dobash (1979), Russell (1982), Straus (1980b), and Walker (1981). " For further examples of resource and exchange theories, see Hornung, McCullough and Sugimoto (1981), O'Brien (1974), Straus, Gelles and Steinmetz (1980), and Whitehurst (1974). There are a number of researchers who have used this perspective to identify dyadic patterns in relationships with physical aggression and abuse, such as Margolin, Sibner and Gleberman (1988), O'Leary (1988a), O'Leary and Vivian (1990), and Rosenbaum and Maiuro (1990). Other interactional theorists include Gelles (1974), Hotaling (1980), Pagelow (1981), Reid, Taplin and Lorber (1981), and Straus (1973). 1  2  4  5  30  Researchers using learning theories include Gayford (1975), Gelles (1976), Gil (1970), Gondolf (1985), Herrenkohl, Herrenkohl and Toedter (1983), MacLeod and Cadieux (1980), Pressman (1983), Sears, Maccoby and Levin (1957), Steinmetz (1977), and Straus, Gelles and Steinmetz (1980). For further research on learning theories, see Barnett and Hamberger (1992), Beasley and Stoltenberg (1992), Brisson (1981), Caesar (1988), Cappel and Heiner (1990), Carter, Stacey and Shupe (1988), Corvo (1992), Hastings and Hamberger (1988), Kalmuss (1984), Rosenbaum and O'Leary (1981), Rouse (1984), and Telch and Lindquist (1984). Other researchers who have studied domestic violence with the help of the alcohol abuse perspective include Coleman and Straus (1983), Gelles (1974), Gottheil et al. (1983), Pernanen (1976) and Taylor and Leonard (1983). For further studies of assaultive males in which men displayed personality disorders, see Dutton (1994b), Dutton and Starzomski (1994), Hamberger and Hastings (1986a, 1989), Hart, Dutton and Newlove (1993), Hastings and Hamberger (1988), and Saunders (1992). 6  7  9  31 CHAPTER TWO: A LITERATURE REVIEW OF DOMESTIC VIOLENCE  2.0 Introduction A comprehensive knowledge of domestic violence research literature is essential to gain the direction and guidance required in future research. The following section will review domestic violence literature, beginning with a brief overview of the history of domestic violence and the prevalence of domestic violence in both Canada and the U.S. A discussion of domestic violence research methodology will focus on the use of selfreports, particularly self-administered questionnaires, and telephone and face-to-face interviews. A number of methodological concerns will also be highlighted, including: reporting biases; the use of the Conflict Tactics Scale; the use of different definitions; labels and measures; the problem with the sex composition of samples; and the use of clinical versus community samples. A review of the findings of domestic violence research will follow, outlining the use, initiation and outcome of physical violence by men and women, the issue of self-defense, the identification of risk markers, and psychological abuse. The issues of sexual violence, dating violence, and the debate of the battered husband will also be discussed. This review will identify major deficiencies in domestic violence literature and also discuss the key variables and relevant knowledge in the study of domestic violence.  32  2.1 Overview of history of domestic violence Violence within intimate relationships is not a recent phenomenon. Historical analyses demonstrate that domestic violence has long been a characteristic of relationships, and at times has been tolerated and sanctioned by society. Breines and Gordon (1983:490) note that until the last few decades, terms such as child abuse, wife beating and incest would have been understood, but not perceived as serious social problems. Although such violence was tolerated, there was also the belief that domestic violence was unusual, deviant behaviour perpetrated by individuals who were mentally ill. In most recent years, however, domestic violence research has portrayed the violence as not only normative, but as culturally accepted. The family is no longer thought of as a happy, harmonious and peaceful refuge from the outside world and its problems (FrankelHoward 1989:9). The growing influence of the women's movement, beginning in the 1960s, was primarily responsible for alerting society to the problems and consequences of domestic violence, and specifically, wife assault. Domestic violence, as a result, is now being treated as a serious problem by numerous groups. Politicians, the media, social scientists, law enforcement agencies, health and social service professionals and concerned citizens are helping to develop policies, establish programs and undertake studies to deal with the problem (Frankel-Howard 1989:9). The women's movement has also had a tremendous impact on the study of victimology. In the United States, the victims' movement began in the 1970s (Viano 1992:1). Since then, victimology has been criticized for lacking a theoretical perspective and foundation. However, it can be best seen as representing a different focus of  33 application of relevant theoretical insights already developed by other disciplines (Viano 1992:4). Domestic violence research became focused on the study of victimology in order to make victims of domestic violence visible. Many criminologists and the sociologists of deviance had tended somehow to "obliterate the victim for a very long while, failing to see what, in retrospect, should probably have been evident all along" (Rock 1994:xi). Victimology is responsible for identifying topics of research such as factors which make people vulnerable to victimization, the impact of crime on victims, perceptions of crime seriousness, and fear of crime (Viano 1990:3). For example, Garofalo (1979:96) concluded that fear of crime is often independent of the risk of and experience with personal victimization, but dependent on social role expectations. Other research by Sacco (1990:485) has made connections between gender, fear and victimization by encouraging the use of the power-control theory which focuses on the study of socialization processes within the context of family class relations. These topics are an important aspect of domestic violence research and literature.  2.2 Prevalence of domestic violence 2.2.1 Within the United States'  The prevalence of domestic violence in the United States was first determined in the 1970s. Straus and Gelles, researchers from the University of New Hampshire and the University of Rhode Island, respectively, conducted personal interviews with 2,143 families. Their findings show that 28% of those families involved at least one incident of physical assault, including punches, kicks, bites, hits with an object, "beats up," uses or attempts to use a gun or knife (Straus and Gelles 1986:470).  34 Other studies reporting incidence include Schulman (1979) who conducted the Kentucky Survey, which sampled 1,792 women who were married or living with a partner. The overall levels of violence were 21%, with urban areas reporting 16% and rural areas 8% (Kennedy and Dutton 1989:44-5; see also Schulman 1979). In the early 1980s, Arias and Beach (1987:143) conducted a survey of 82 married men of which 32% reported engaging in physical aggression (as measured by the CTS) toward their wives, while 39% of the women reported engaging is such aggression against their husbands. In a study of violence and 272 engaged couples, (O'Leary et al. 1989:265) reported aggression rates of 41% for women and 34% for men, which included throwing objects, pushing, grabbing, shoving, slapping, kicking, biting, hitting with a fist or an object, beating up and threatening with a knife or gun. The U.S. Department of Justice also helps provide more general rates of violence by conducting national crime victimization surveys. For example, in 1990 alone, of 6,008,790 crimes of violence against persons age 12 and older reported in a national crime victimization survey (U.S. Department of Justice 1992), 39% were victimized by a member of the family or by a person in a relationship with the victim. Fifty-eight percent of those incidents reported involved the spouse or ex-spouse of the victim. Others, such as Gelles and Straus (1988), Langan and Innes (1986), and Straus and Gelles (1990), report that violent crimes occur more frequently within families than among strangers.  2.2.2 Within Canada  The original prevalence rates of domestic violence were estimated from general victimization surveys such as the Canadian Urban Victimization Survey (CUVS)  35 conducted in 1982, and the third cycle of the General Social Survey (GSS) conducted in 1988. Neither survey was specifically designed to measure the incidence of wife abuse or domestic violence, but rather to measure violence in general. On behalf of the CUVS, the Research and Statistics Group reported the incidence of wife abuse to be less than 1 % (DeKeseredy and Hinch 1991:19; see also Research and Statistics Group 1984). These data from GSS were analyzed for estimates of wife abuse and reported incidence rates of 1.5%, which were criticized for greatly underestimating the amount of wife abuse in Canada (DeKeseredy and Hinch 1991:20). The questions used in the GSS did not reflect women's experiences with violence and abuse. Outside of the general victimization surveys, the prevalence of domestic violence in Canada was estimated from smaller nonrepresentative surveys, such as the city-wide representative samples in Brinkerhoff and Lupri (1988) and Smith (1985, 1986, 1987). Smith's research revealed that 13% of married or cohabiting women and 31% of divorced and separated respondents reported having been abused in the past year (Smith 1990a:48). Most data available came from populations of specific communities, such as universities and battered women shelters, which are not representative of the Canadian population of women. From questioning community samples about all types of victimization to small non-representative samples and specific domestic violence questions, the need for larger and more comprehensive research projects was acknowledged by other Canadian researchers. The first national Canadian sample study was conducted by Lupri (1990). This survey of 1,834 men and women aged 18 and over was conducted by Decima Research Ltd. in 1986. However, only eight of the questions were about domestic violence. From these questions, 18% of married or cohabiting men reported committing at least one  36 violent act, with the figure rising to 30% among divorced or separated respondents (Lupri 1990:171). The Statistics Canada Violence Against Women Survey (1993) provided national prevalence rates of domestic violence or more specifically, violence against women. The findings showed that 29% of ever-married or cohabiting women have been assaulted by a spouse or live-in partner (Johnson and Sacco 1995: 296).  2.3 Domestic Violence Research Methodology  Research on domestic violence has been limited in its choice of methodologies due to the nature of this violence. Most intimate violence occurs in the privacy of the home, making it difficult to conduct experiments or quasi-experiments in the laboratory, or to make observations in a field study. Thus, most research is based on self-reports by the victim and/or the perpetrator of the violence, through either interviews or selfadministered questionnaires. Few researchers will argue with this limited choice, but they will argue with how, and with which questions, the self-reports are made. The use of self-reports, as well as methodological concerns, will be discussed later in this chapter, but first the sorts of questions asked need to be discussed in terms of quantitative and the qualitative research. Quantitative research refers to studies of subjects in which these data can be handled numerically (Vogt 1993:1.84). Qualitative research, dealing with studies of subjects in which behaviours and experiences are difficult to quantify (p. 183), uses variables which are designated by words or labels in discrete nonnumerical categories (Singleton et al. 1988:73). For example, with the variable "gender," consisting of the discrete categories male and female, categorical, but not numerical, distinctions can be made between people of different genders (p. 74).  37 However, those within these discrete gender categories can be analyzed quantitatively. In terms of domestic violence, earlier research had been quantitative in an attempt to report the prevalence of violence in relationships, but feminist methods encourage the addition of qualitative research to explore the precursors and the effects of the violence (see McHugh 1993:55). Dobash and Dobash (1983), among the earliest domestic violence researchers in North America, note that they had originally approached their research from the positivist tradition which typifies most social science research. The positivist method seeks to abstract personal experience into theories and models of behaviour, usually based on quantitative research methodologies; this method is in contrast with the feminist research postulate that "the personal is political" (Grant 1993:33). The use of positivist methods is an attempt by social scientists to gain scientific legitimacy for their work, comparable to that accorded to the natural sciences. However, Dobash and Dobash point out that the utilization of such methods as experimentation, surveys, abstracted measurement devices and statistical manipulations has resulted in "intellectual blinkers and mindless adherence to a sterile sophistication" (Dobash and Dobash 1983:263). Their later research, as does recent domestic violence research, emphasizes the importance of historical understanding and the consideration of personal, social, institutional and political contexts (Dobash and Dobash 1983:265). Breines and Gordon (1983) support Dobash and Dobash by pointing out that domestic violence involves "social and cultural factors which cannot always be quantified; qualitative and theoretical analysis must complement the positivist method (Frankel-Howard 1989:18). Both qualitative and quantitative methods are included in the self-reports used in  38 domestic violence research, which will be discussed in the next section.  2.3.1 The Use of Self-Reports Verbal or self-reports are the most common form of social measurement because they are the most pragmatic (Arias and Beach 1987:147). Most research on domestic violence has been reliant on the self-reports of the men and women involved through replies to direct questions in self-administered questionnaires, telephone interviews and face to face interviews, which will be discussed in the next sections. This method of research provides simple and generally accurate measures of background variables such as age, gender, marital status, and education, as well as measures of subjective experiences, such as knowledge, beliefs, attitudes, feelings, and opinions (Singleton et al. 1988:101). The first type of self-reports to be discussed are self-administered questionnaires.  2.3.1.1 Self-administered Questionnaires A self-administered questionnaire is a set of questions in print form in which the survey respondent answers the questions on paper without the assistance of an interviewer. Such surveys use, for the most part, close-ended questions which can be answered by simply checking a box or circling the proper response from a set provided by the researcher. One reason for using close-ended questions is that asking respondents to answer questions in their own words increases the difficulty of the task, which will affect the rate of nonresponse for many respondents. Another reason is that open answers on a self-administered questionnaire may not produce useful data. "With no interviewer  present to probe incomplete answers for clarity and for meeting consistent question objectives, the answers will not be comparable across respondents, and they will be difficult to code" (Fowler 1988:64). In domestic violence research, self-administered questionnaires can help researchers overcome the challenge of encouraging respondents, especially the perpetrators of violence, to admit socially undesirable or negatively valued characteristics or behaviours (Fowler 1988:65).  2.3.1.2 Interviews Interviews have been rated the same as or higher than self-administered questionnaires as an effective research tool (Denzin 1989:29). Interviews consist of a pre-set questionnaire which is read by an interviewer, answered verbally by the respondent and recorded by the interviewer. Interviews overcome a number of problems faced by self-administered questionnaires; interviewers are able to motivate the respondent to participate, respond to feedback from the respondent, elaborate a question or statement, probe for further information, and guide the respondent through complex measurements (Gray and Guppy 1994:97). There are two types of interviews: face-toface interviews and telephone interviews. Face-to-face interviews involve the presence of the interviewer and the interviewee in the same place, often the home of the interviewee. Such interviews are composed of elements of visual communication as well as a preinterview acquaintance period which allow interviewers to establish the legitimacy of the interview and also the image of themselves as a pleasant, understanding, and safe person with whom to interact. "The interviewer's physical presence permits the communication of attention to, interest  40 in, and acceptance of what the respondent has to say, through nonverbal as well as verbal indicators" (Miller and Cannell 1989:306). Some argue that face-to-face interviews are most effective for asking sensitive questions, such as those about domestic violence, because the interviewer has an opportunity to build rapport and to establish the kind of trust that is needed for interviewees to report potentially sensitive information (Fowler 1988:65). Telephone interviews involve interviewers who conduct the interview via the telephone, allowing more respondents to be contacted in a shorter time period than faceto-face interviews. Interviewers can work from their home or at a centralized location with the help of computer assisted telephone interview systems. Such interviews require different communication patterns and respondent motives than face-to-face interviews. They also lack visual cues which assist in face-to-face interviews (Miller and Cannell 1989:321). Due to these differences, others have argued that telephone interviews may appear impersonal which should help people report negative events or behaviors, such as participating in domestic violence (Fowler 1988:65).  2.3.2 Methodological Concerns  There are a number of methodological concerns with the use of self-reports, whether with self-administered questionnaires or telephone or face-to-face interviews.  2.3.2.1 Overreports, Underreports and Nonreports  The problem with relying on the self-reports of men and women in situations of domestic violence is two-fold: non-reports and assessing the accuracy of the reports.  41  Non-reporting is a challenge for researchers relying on self-reports. Due to the sensitive nature of the content of the questionnaires, respondents may be inclined to not report their actions or victimization without considerable encouragement or unless the respondent feels strongly about the topic him or herself. Assessing the accuracy of the reports is also difficult because of the inherent private nature of domestic violence; how does a researcher find the truth if only the two individuals involved in domestic violence were there and know the truth (see Arias and Beach 1987)? In assessing the accuracy of the reports, researchers are faced with the problems of both underreporting and overreporting. Knowing when this problem arises increases in difficulty when only one member of the couple is asked to report on the violence; this has been a major criticism of domestic violence research (McHugh 1993:57). When corroborating evidence has been available, the results indicate low to moderate agreement between partners.  Underreporting and overreporting have been discovered in the  reporting of domestic violence by both men and women. The problems of overreporting and underreporting in self-reports provide researchers with the challenge of improving questionnaires and interviewing methods in order to increase the accuracy of the findings. For example, corroborating evidence often provides low to moderate agreement between partners involved in domestic violence. Researchers need to seriously question the accuracy of the reports of both partners, and perhaps should seek third-party corroboration (McHugh 1993:58). More research is required to identify possible influences, such as different perceptions and definitions. For example, some men may perceive violence in relationships as normal, and thus perceive the relationship as mutually violent or do not  42 perceive the relationship as violent at all. Some women, on the other hand, might overreport their victimization because they are influenced by the way domestic disputes are handled by police, counselors and shelter workers: men are more often arrested and placed in batterer programs, whereas women are placed in safe houses and given counselling for their victimization. One example of this is provided by Kelly (1988) who noted that in her study, over 60% of the women did not initially define their experiences as a form of sexual violence, but about 70% of the women changed their definitions of their experiences over time, almost always in the direction of relabeling an incident as abuse (p. 119). Although Kelly's report was originally intended to demonstrate women's enlightenment with knowledge about domestic violence, her findings also demonstrate that there are other influences on overreporting that have previously not been addressed in the literature. Researchers also need to assess their respondents' definitions of violence. For example, Browning and Dutton (1986) point out that "husbands tend to see the relationship as mutually violent, whereas the wives view their relationship as husbandviolent" (p. 377). Perhaps these husbands define violence solely in terms of violent acts whereas women define violence in terms of the injurious outcome for themselves; i.e. a man views his slap as the same as his partner's slap, but his partner views his slap as more violent because of the possible injurious outcome. Checks must be placed in the 4  questionnaire to identify the respondents' definitions and perceptions of domestic violence in order to determine the accuracy of the reporting.  43  2.3.2.2 The Use of the Conflict Tactics Scale Most of the research on domestic violence, especially in Canada, has relied on the use of the Conflict Tactics Scale (CTS) designed by Murray Straus (1979). This scale is a quantitative instrument consisting of eighteen items which measure three different ways of dealing with interpersonal conflict: reasoning, verbal aggression and physical violence. The eighteen items are categorized on a continuum from least to most severe, with the first ten items describing non-violent tactics and the last eight items describing violent strategies (DeKeseredy and MacLean 1990:20). It was used extensively by Straus and his colleagues in studies referred to in Straus (1980d), Straus and Gelles (1986), and Straus, Gelles and Steinmetz (1980). Straus and his colleagues argue that the CTS has high reliability and validity (Straus, Gelles and Steinmetz 1980). The use of the CTS has provided estimates of 5  violence against women, which has been a useful contribution to the limited Canadian data base, as well as improved upon the accuracy of previous estimates reported by other well-publicized Canadian studies such as MacLeod (1980, 1987) and the Research and Statistics Group (1984) (DeKeseredy and MacLean 1990:20). The use of the CTS is important in that it provides comparability of results, but it also perpetuates inadequacies. Such inadequacies include ignoring the gendered power imbalances that can exist within marriage, and excluding crucial details about motives, intentions and consequences (Johnson and Sacco 1995:291). The questions in the scale 6  are also set in the context of conflict situations, which may not accurately reflect specific violent situations (McHugh 1993:57). Others argue that the continuum of least to most severe conflict tactics is problematic. For example, it assumes that physical violence is  44 worse than psychological abuse. This assumption has been refuted by women victimized by both physical and psychological abuse (DeKeseredy and MacLean 1990:20; see also Breines and Gordon 1983; Walker 1979). Another problem with the CTS is that the ranking of the violent tactics does not realistically reflect the severity of each of the violent tactics; for example, the CTS places slapping at the lower end of the continuum although such actions can easily result in bleeding or broken teeth, similar to the results of a higher end act of kicking or using a knife (Dobash and Dobash 1988:59; Smith 1987:178). There is also concern that the scale does not accurately reflect the reality of both men and women through either its measures or its use of terms; for example, Cascardi and Vivian (1995:283) found that there were husbands who admitted to being aggressive during the interview but did not report any aggression on the CTS. The CTS requires further examination to determine if there are differences in reporting due to men's and women's reaction to the scale itself, similar to the suggestion that the Sexual Experiences Scale was not perceived or conceptualized by some men as intended by the researchers (Koss, Gidycz and Wisniewski 1987:162-70). Although the CTS has a number of deficiencies, it can only provide further challenges to domestic violence researchers in their attempt to increase the quality of their surveys. For example, researchers can study the perceptions of men and women and their definitions of terms used in the CTS (see also Dutton 1994a; Y116 andBograd 1988).  2.3.2.3 The Use of Different Definitions, Labels and Measures The use of different definitions, labels and measures of abuse and violence in  45 domestic violence studies has been criticized by theorists and researchers. For example, some have argued that the "interchangeable use of terms such as assaultive, abusive, aggressive and violent has led to conceptual ambiguity and confusion" (McHugh 1993:58; see also Browne 1989). Others have objected to the terms domestic violence, family violence and spouse abuse because they feel that such terminology obscures the dimensions of gender and power that are fundamental to understanding the abuse of women (Breines and Gordon 1983:492; see also Schechter 1982:33). The term battered woman has also been debated with some arguing that a battered woman endures a long period of physical and psychological abuse while others consider one act of violence sufficient to constitute a woman who has been battered (Saunders 1988:94). Without agreed upon definitions, labels and measures of abuse and violence, there can be no validity or reliability granted to studies; each study uses a different definition, label or measure, making comparisons impossible (DeKeseredy and Hinch 1991:9). It is essential that the field of domestic violence research adequately validate the large number of measures that have already been generated (Schumm 1990:32). With the validity of the measures established, the definitions and labels will also follow suit.  2.3.2.4 Sex Composition of the Sample The sex composition of study samples has also been a methodological concern because many domestic violence surveys include only women in the samples, while others include only men. As long as the survey is intended to only study men or women, these samples are valid (for example, Hamberger and Hastings 1986b; Jones 1993; Saunders 1992). The problem arises, however, when the surveys are intended to record  46 the actions of both partners and the sample consists of only one partner (for example, see Labell 1979; Smith 1987, 1990b; Walker 1978, 1979, 1984). Mainstream research has been criticized for its sex bias in studying male samples, and drawing conclusions about all humans (Yllo 1988:33). Domestic violence research has corrected the problem of studying only male samples by replacing the men with women. However, such research can also be criticized for studying samples of women and drawing conclusions about both women and men; just as men cannot report women's experiences, women cannot report men's experiences. Until women are allowed to speak for women and men for men alone, sex biases will be found in domestic violence research and its findings will be less reliable and less valid. Another problem with the sex composition of study samples includes the lack of corroborating findings in samples of only one partner. There are no checks on the reliability of those findings. Samples of both men and women help overcome this problem, but only if the men and women are asked corresponding questions of victimization and perpetration.  7  2.3.2.5 Clinical vs. Community Samples  The last methodological concern to be discussed involves the use of clinical and community samples. Clinical samples refer to samples of persons receiving assistance or treatment for a problem (Straus 1990:86). Community samples refer to samples of persons not receiving assistance or treatment for a problem.  9  In domestic violence research, clinical samples refer to women in shelters for battered women, transition homes, and/or in counseling and to men who are attending  47 treatment programs for violence or counseling, often court-ordered; in all cases of clinical samples, the violence and abuse has been reported to someone. The problem with using such samples is that the findings, which are unrepresentative, are often used to generalize to community samples. Sociologists sometimes use the term clinical fallacy to describe research, based on "clinical" samples, which may have limited applicability because those who seek or receive "treatment" are often not representative of the entire population manifesting the problem (Straus 1990:86). For example, it has been noted that studies noting gender differences in reporting marital violence are generally based on clinical samples, while studies that report no gender differences are based on community samples (Stets and Straus 1990:151). Another term used is the "representative sample fallacy" which refers to the "danger inherent in attempting to generalize from the characteristics and experiences of the total population who manifest a certain problem (such as assaults on wives), to populations receiving assistance for the problem (such as users of a battered women's shelter)" (Straus 1990:86). The obvious solution to improper generalizations is to clearly state that the findings are only representative of clinical or community samples. Another solution is to use both samples in a study for the purpose of contrast and comparison, such as Hamberger and Hastings (1991) who studied personality correlates of men who batter and nonviolent men and noted some continuities and discontinuities. Such studies can 10  reveal the different natures of violence in relationships.  2.4 The Findings of Domestic Violence Research In domestic violence research, a number of different findings have emerged.  48 Some of the studies have reported consistent findings, while other studies show discrepancies. Variations in research findings are discussed in this section in order to demonstrate the need for further research, but also the need for an improved conceptualization of definitions and measurement tools. The following discussion looks at the use, initiation and outcome of physical violence by men and women, the use of psychological abuse by men and women, as well as the issue of self-defense and risk markers of domestic violence.  2.4.1 Use of Physical Violence by Men and Women  Domestic violence research, such as Ben-David (1993), Pagelow (1981), Saunders (1986), and Straus, Gelles and Steinmetz (1980), has discovered that men and women in intimate relationships use physical violence. However, there have been differing reports of which group uses violence more, or if the violence is mutual. Some reports found that men use violence more than women (Bograd 1990:133; see also Dobash and Dobash 1988). For example, in studies of male samples from treatment programs, almost half of the men acknowledged that they were the more violent partner (at least in terms of frequency of violence), although the other half perceived the relationship as mutually violent or wife-violent (Browning and Dutton 1986:378). Other studies report that women use violence more. For example, gender differences were reported in a study which found that, in cases of non-reciprocal violence, males are more likely than females to be victims, with men reporting such experiences three times more often than women (Deal and Wampler 1986:468). Other studies such as Arias, Samios and O'Leary (1987), Bernard and Bernard (1983), and Henton et al. (1983)  49 found that women are slightly more likely to report using violent tactics than are men (Bethke and DeJoy 1993:37). Although these studies could be subject to men underreporting or women overreporting, such findings indicate the need to further study men's and women's involvement in domestic violence.  11  Other reports indicate that men and women are involved in mutually violent relationships. Stets and Straus (1989:46) reported that the most frequent pattern for the use of violence is for both partners to be violent. Stets and Straus also reported that female-only violence is a more frequent pattern than male-only violence. In the National Survey of Families and Households, no significant difference was reported between wives and husbands in these respondents' reports of hitting, shoving, or throwing things in the 12  current relationship (Brush 1990:61).  2.4.2 The Initiation of Physical Violence by Men and Women  One important discussion in domestic violence research is the initiation of physical violence, whether by men or by women. Researchers have focused on this issue in an attempt to ascertain the cause of the violence, and incidentally, the guilty party, which, for the most part, has been men. For example, in a study of men and women who killed their partners, two-thirds of the men admitted to initiating the physical violence (Jurik and Winn 1990:235). In another study, a high percentage of men were found to blame themselves for the initiation of violence; as the violence continued, men were found to be as likely to blame themselves as they were to blame their wives (Cantos, Neidig and O'Leary 1993:297). Although the focus has been on men, women have also reported their own  50 initiation of violence. For example, in a study of non-Hispanic whites and Mexican Americans, women reported that they were more likely to have struck first and to have struck first more than once (Sorenson and Telles 1991:11). Others report that women initiate the physical violence at rates roughly equal to male-to-female violence (Straus and Gelles 1986:470).  13  Others have reported a mutual escalation of violence (Deschner 1984; Neidig and Friedman 1984). Cascardi and Vivian (1995:282) have reported that in most cases, marital violence may be an outgrowth of conflict between both partners in which each actively contributes, albeit not necessarily symmetrically, to the escalation of violence. These findings demonstrate the need for more research, and especially for the growth and refinement of theories regarding the role of men and especially women in the initiation of violence. Some argue that women initiate the violence in order to control when the violence occurs, rather than waiting for the man to do so. For example, Walker (1979) presents the cycle of abuse, involving tension-building, explosion and reward, which implies that in a violent relationship, a woman may initiate the violence simply to complete the cycle, especially if she is unable to avoid episodes of violence (Sorenson and Telles 1991:12). However, the belief that women do not initiate violence still needs to be challenged from a level of acknowledging women's aggression. Theories, such as those found in Dougherty (1993), which explain away women's violence continue to restrict women to common stereotypes which characterize all women as passive victims. These same theories characterize all men as aggressors. These stereotypes need to be challenged to allow men and women to be characterized not as homogeneous groups, but rather as groups with variety and contrast. The group can include such men and women  51 who have always had aggressive tendencies or have been aggressive only in certain situations, as well as those who have always been passive or only in certain situations.  2.4.3 The Issue of Self Defense Another important issue that has received a lot of attention from researchers is the issue of self defense. This follows the issue of the initiation of violence, because if the men initiate the violence, then how do women respond, or vice versa? Stets and Straus (1990:156) report that both men and women respond to violence with violence. In fact, women were found to respond to violence with violence more often than men, with men responding to violence through other avenues. However, many researchers argue that women use violence only in self-defense or in retaliation (Marshall and Rose 1990; Hamberger and Potente 1994), and that violence used in self-defense can be in response to immediate danger or to a prolonged period of abuse (Walker 1993:133). Feminist theorists like Walker (1993) argue that to not accept self-defense is to blame the victim, who in most cases is a woman, due to the size and strength difference between men and women involved in physical altercations (Dobash and Dobash 1979). For example, Cascardi and Vivian (1995:283) report that when women admitted to being severely aggressive, they showed a tendency to perceive their violence as due to self defense more often than severely aggressive partners.  1  Other research has reported that women do not always use violence in self-defense (Bland and Orn 1986; Stets and Straus 1990). For example, Mann (1988:48) suggests that, although 58.9% of women who kill in domestic encounters claim self-defense as a rationale for slaying their partners, individual case analyses suggests that women may be  52 victors in the domestic fight. Premeditation (56.3%), and the offender's prior violent arrest histories (30%), indicate that the self-defense motive is suspect (Mann 1988:131). However, an overall pattern of self-defense has yet to be identified. From the above findings, evidence is provided for other avenues of research in the field of domestic violence. For example, the need to discriminate between community and clinical samples becomes evident when some women in clinical samples have been exposed to a severe batterer and responded to violence in self-defense, such as those referred to in Walker (1995), as opposed to women in community samples who may be mutually involved in the violence and less likely to use violence in self-defense, such as those referred to in Johnson (1995a). Studies using community samples "reach only populations in which violence is a relative isolated reaction conflict," whereas studies using clinical samples "reach primarily victims of violent, but multi-faceted, strategies of control" (Johnson 1995a:288). Theory and research need to incorporate this knowledge. There is also the need to examine men's and women's perceptions or descriptions of their experiences with violence. For example, when a woman does commit murder and claims self-defense, there is the chance that self-defense, from a legal point of view, is a statement encouraged by her lawyer. A n argument of self-defense is now more likely to be believed in court, as is evidenced by Canada's latest efforts to reassess women's spousalmurder charges to allow them the chance to plead self-defense (Canadian Press Newswire 3 March 1997). Jurik and Winn (1990:235) report that when women kill their partners, their victims frequently initiated the fatal physical altercation. Considering that the deed has been done, there is no set of checks and balances to determine what the true circumstances were, especially when the other party is unable to speak on his own behalf  53 and the act took place in the privacy of their own home, as reported by Wilson and Daly (1994). But contrary to common belief, some women have the strength and size, or the ability to use a weapon, to harm, kill or threaten to kill their partners, and not in self defense (see Freeman 1979).  2.4.4 The Outcome of Men's and Women's Physical Violence The physical consequences of violence is also an important area of domestic violence research. One of the main arguments for focusing on violence against women by men is that because of men's size and strength, women are more likely to suffer injuries at the hands of men than men are to suffer at the hands of women. However, this does not account for strong, large women or for physically aggressive women, whether large or small. Studies such as Hamberger (1991a, 1991b) and Vivian and LanghinrichsenRohling (1994) have reported that women are involved in domestic violence, often at levels of severity similar to that of men.  There are, however, others studies which found  that women are more likely to report being injured in domestically violent situations (Brush 1990:64). Although an overall pattern appears to exist in which women are more likely to be injured, this is still an unclear area of dispute. There are future research avenues suggested by this area of study, especially in terms of which injuries are being reported and those which are not being reported by both men and women. For example, men are less likely to report their injuries because of the risk of humiliation (Freeman 1979); it may also be that men's injuries are more psychological than physical. Brush (1990:61) noted that men perceived injuries of equal frequency for both themselves and their partners, whereas women reported being injured  54 2.75 times more than men. There may be a difference for men and women, in that, for example, a scratch for a women may be reported as an injury but a scratch for a man may not be considered an injury. There is a need for understanding men's and women's perceptions and definitions of injuries, and what factors influence their reporting.  2.4.5 The Risk Markers of Domestic Violence One of the objectives of domestic violence research is to identify risk markers. A risk marker refers to an "attribute or characteristic that is associated with an increased probability to either the use of husband to wife violence, or the risk of being victimized by husband to wife violence," and is not necessarily a causal factor (Hotaling and Sugarman 1986:102). In the search for a profile of domestic violence victims and their aggressors, a number of risk markers have been identified. More importantly, researchers, such as Hotaling and Sugarman (1986) who have identified 97 possible indicators, have been searching for the most consistent risk markers, those which are most often found in domestic violence situations. With so many factors to consider however, the ability to provide a profile of victims and aggressors is hindered not only by a challenging task, but also by methodological concerns such as a lack of comparison groups and single subject versus participating couples (Hotaling and Sugarman 1986:117). Risk markers of domestic violence include sociological or demographic factors, such as youth, low income, and divorce or separated status (Smith 1990b:45). Personality factors, such as anger and hostility, have also been added to the list of the most consistent risk markers (Dutton 1988; Maiuro, Cahn and Vitaliano 1986, 1988; Shields, McCall and  55 Hanneke 1988). Risk markers also include psychiatric disorders for both victims and aggressors (Rounsaville 1978; Gleason 1993).  14  Risk markers, such as personality factors, can be further determined through the study of psychological abuse by both men and women.  2.4.6 The Psychological Abuse by Men and Women Psychological abuse often precedes and/or accompanies physical abuse as a result of the violence, but it can also be a result of verbal aggression or nonverbal actions not considered violent. Psychological abuse has been a major issue for feminists in supporting battered women because of the psychological injuries incurred, but not always recognized. Battered women claim that physical injuries heal, but the effects of psychological abuse last much longer and with more perceived damage (Murphy and Cascardi 1993:107). This discovery gives researchers other areas to study within the field of domestic violence. The focus of research on psychological abuse has been primarily on women victimized by their partners; this is because of evidence presented which attests to women victims suffering more psychological injury than male counterparts, in terms of "psychosomatic symptoms, stress, and possibly more depressive symptomatology" (Stets and Straus 1990:158-60). Women victims appear to be characterized by low self-esteem, low self-sufficiency, psychological symptomatology such as depression and anxiety, drug and/or alcohol abuse or dependency, and suicide attempts (Arias and Beach 1987:139; see also Sedlak et al. 1985). These women express more fear of their partners, and have also been diagnosed with posttraumatic stress disorder (Cascardi et al. 1995).  56 Psychological abuse has been the domain of men, according to much research, but there is evidence that this domain belongs to both men and women, such as DeKeseredy and Kelly (1993b), Dutton (1995), Hyden (1995) and Loring (1994). Most research on 15  domestic violence does not consider women's verbal aggressiveness to be on par with men's verbal aggressiveness, but without any explanation for such a consideration (see Breines and Gordon (1983), Greenblat (1983), Saunders (1986)). Men's verbal aggression is considered psychological abuse, but little attention has been given to women's verbal aggression which may play a part in the physical violence. For example, a study of 33 men in a batterer program reported that over half of the men blamed their partner's verbal aggressiveness for provoking the physical violence (Coleman 1980:208). While not condoning men's violence, the study does allude to the effects of psychological abuse caused by verbal aggression. In the next section, sexual violence will be discussed, which involves both physical and psychological abuse.  2.5 Sexual Violence  Sexual violence involves "forcing a female intimate partner, through the use of verbal or physical threats or intimidation, to participate in sexual activities against her will" (Carden 1994:547). Motivating factors of sexual violence include anger, the need to dominate and occasionally sexual obsession. Such violence decreases women's selfesteem and negatively impacts their attitudes toward heterosexual behaviour and men that are even more severe than the psychologically debilitating effects of emotional abuse suffered by these women (Bowker 1983:348). In domestic violence research, sexual  57 violence is rarely addressed and often without justification. DeKeseredy and MacLean (1990:23) point to the use of the Conflict Tactics Scale as part of the reason for the ignorance of sexual assault, especially in Canadian surveys. The omission of sexual violence in these studies may silently acknowledge men's domain in sexual violence, as little evidence can be provided for women's sexual violence. Others, such as Bowker (1983:348), have suggested that sexual violence may constitute a distinct syndrome apart from nonsexual violence, therefore condoning separate studies of sexual and nonsexual violence. However, others argue that physical violence cannot be discussed without including sexual violence because there is physical violence which is not sexual but there is no sexual violence which is not physical (see Browne 1987:95; Dobash and Dobash 1979). Sexual violence was included in the questionnaires of two of the surveys to be assessed.  2.6 Dating Violence Physical violence in dating relationships has been a subject of investigation since the early 1980s with studies such as Bernard and Bernard (1983) and Laner and Thompson (1982). However, recognition of dating and sexual exploitation came years 16  earlier with studies such as Kanin (1957) and Kirkpatrick and Kanin (1959) (Bethke and DeJoy 1993:36). Bethke and DeJoy (1993) provide a brief overview of findings from dating violence research (see Cate et al. 1982; Henton et al. 1983; Laner 1983; Laner and Thompson 1982; Roscoe and Benaske 1985). Dating violence appears to begin at about the age of fifteen or sixteen (see Durst 1987; Henton et al. 1983). Less severe acts of violence are more frequent than more severe acts (Makepeace 1981; Stets and Pirog-  58 Good 1989), and recurring and escalating episodes of violence in a relationship are common if the relationship is not terminated (Bethke and DeJoy 1993:37). Studies of dating violence consistently report the use of violence by both men and women, with women being slightly more likely to report using violence than men (Arias, Samios and O'Leary 1987; Bernard and Bernard 1983; Henton et al. 1983). However, women report greater feelings of victimization (Bernard and Bernard 1983; Makepeace 1981), and report experiencing greater physical and emotional harm (Stets and Pirog-Good 1987, 1989). Bethke and DeJoy (1993) conclude that in the majority of violent dating relationships, violence does not occur until some degree of commitment has been established (Burke, Stets and Pirog-Good 1989; Cate et al. 1982; Henton et al. 1983; Laner and Thompson 1982; Stets and Pirog-Good 1987), or until the relationship has been sustained for a considerable period of time (Arias, Samios and O'Leary 1987). Dating violence has been studied separately from domestic violence with the use of specific samples of young people from high schools, colleges and universities, but the similarities between dating and domestic violence are apparent. Studies have noted high prevalence rates ranging from 10% to 60% among dating couples, as is common among married couples (Arias, Samios and O'Leary 1987; Barling et al. 1987). As mentioned above, a level of commitment appears to be necessary for the violence to begin, which is already present in a marriage or a marriage-like relationship. The social, legal, emotional and economics bonds thought to maintain violent marriages were not expected to be binding in dating relationships (Arias and Johnson 1989:298; see also Levinger 1976). However, it becomes obvious that similar bonds do exist within dating relationships; although less likely to be economic and legal, social and emotional bonds are as  important in dating relationships. The inclusion of dating violence in domestic violence research is essential for a fuller understanding of violence between intimates, especially when studies such as Smith (1990a) consistently show youth and low income to be consistent risk factors in domestic violence (DeKeseredy and Hinch 1991:28). Dating violence was included in the questionnaires of only one of the three surveys to be assessed.  2.7 The Debate of the Battered Husband As indicated by the various studies mentioned in this chapter, men and women experience physical and psychological abuse in relationships. Some researchers acknowledge the existence of husband-beating and have even suggested the possibility of a battered husband syndrome. However, the battered husband syndrome has been rejected by feminist sociologists such as Berk et al. (1983), Dobash et al. (1992) and Walker 1984 and husband-beating is rarely addressed in journals. The lack of published empirical data regarding the injurious outcomes of domestic violence for men has led many to believe that husband-beating should not be the focus of research, although there is evidence of such injury. There are some sociological studies (Shupe, Stacey and Hazlewood 1987) and a number of medical journals which have reported case examples of husband-beating (Alani 1976; Everett 1980; Scott 1980). These authors have expressed dismay at the lack of discussion of men abused by women. Saunders (1988:91), who is one of the advocates of self-defense, has conceded that there are still conceptual problems in discussions of physical violence between men and women as to whether it is wife abuse, husband abuse, or mutual combat, although the  60 reported overall distribution reveals husband abuse to comprise the majority of domestic violence. He lists three reasons for the lack of resolution: First, little research has addressed the motives behind the use of physical violence, the consequences of violence by men and women or the contexts in which these men and women choose to use violence. Second, domestic violence research is plagued by conceptual confusion; for example, does retaliation constitute self defense? And thirdly, social science researchers, theorists and activists have been polarizing into opposing camps, "which makes it difficult to resolve this controversy with scientific and moral integrity" (Saunders 1988:91). More research into domestic violence is required, with attention paid to violence against both men and women, in both heterosexual and homosexual relationships. This means not only that there needs to be more research, but that research on men needs to be tailored for men's experiences as research on women is tailored for women's experiences. The focus on women as victims and men as aggressors has made it difficult for men to recount their own experiences as victims. There are other possible reasons for men's non-reporting or under-reporting, such as the suggestion that men deny their use of violence or that men have different perceptions than women (Stets and Straus 1990:152). However, until research explores the experiences of both men and women, our understanding of domestic violence will only be one dimensional. Discussions of husband abuse are not meant to discount women's experiences or to deny the prevalence of violence against women. Even in a family where the woman may be as violent as the man, their mutual violence does not cancel out one another. If women's role in the violence is not understood, a problem will still exist with no realistic solution (Shupe, Stacey and Hazlewood 1987). Studies of husband abuse allow the field  61  of domestic violence research to expand in order to rule out alternative explanations or to provide new ones.  2.8 Conclusion  In this literature review, I have outlined omissions and weaknesses in the existing domestic violence literature. For example, this review demonstrates the need for more domestic violence research in Canada. I have also provided evidence suggesting the need for methodological (sampling and measurement) improvement in surveys as an effective research tool. This includes determining the effects and prevention of reporting biases, finding consensus in the use of definitions, labels and measures such as the CTS, as well as the need for men and women in both clinical and community samples. My review of the findings encourages the recognition of men's and women's use of physical violence and psychological abuse and the need for more research on issues such as self-defense and perceptions of violent experiences. The various findings also indicate the need for new or revised perspectives on domestic violence. This can begin by delineating different types of violence used by men and women. This leads to a consideration of different types of relationships, such as relationships characterized respectively by male-only violence, female-only violence or mutual violence. Such delineations allow studies to help explain the different types of relationships, rather than trying to explain all situations of domestic violence. The acknowledgment of different types and processes of relationships will also allow for the dissipation of stereotypes which plague domestic violence research (Crenshaw 1995). Domestic violence research can be improved by assessing past research and  62 improving on its weaknesses, and secondly, by triangulation of different methods. Assessing past research begins with a set of criteria for proper survey methodology. Gray and Guppy (1992:23) list questions which must be answered in order to conduct a good survey (see also Fowler 1995; Mangione 1995). (1) Who paid for the results?; (2) When did data collection begin and end?; (3) How were these data collected? The researcher must report relevant details of these data collection which affect the error level of these data (Fowler 1993:142); (4) Of what group is these data representative? It is important to know which population is to be studied (Fowler 1993:142); (5) How was a representative sample acquired? It is essential to know the alternate ways in which a list of the population can be obtained and how adequately each of these ways reflects that population (Mangione 1995:107); (6) Exactly how many people participated? The sample size should be large enough to determine the required estimates (Mangione 1995:107); (7) How many of the people contacted refused to participate? Those who participate in the study may be different from those who refuse to participate so it is important to create strategies to increase the response rate as much as possible (Mangione 1995:108); (8) What was the exact wording of the questions? It is important to gather information about respondents' level of comprehension and ability to answer the questions in order to evaluate the effectiveness of the questions (Fowler 1988:103); (9) How accurate are these data? For example, pretesting is an essential factor in minimizing measurement error (Mangione 1995:108); and (10) What ethical issues were considered (Gray and Guppy 1994)? Respondents and interviewers must be informed and protected by the researchers, but they must also benefit from their participation in the study (Fowler 1988:132-35).  63 By having this set of criteria, past surveys can be assessed fairly, therefore, providing a basis for improved methodology in the future, as will be demonstrated in the following chapters.  Notes: The prevalence of domestic violence in the United States is included here for comparison with rates in Canada, to provide evidence of reliability and validity. Other studies suggesting low to moderate agreement between partners include Cascardi and Vivian (1995), Christensen and Nies (1980), Jacobson and Moore (1981), Jouriles and O'Leary (1985) and O'Leary and Arias (1988). " Underreporting of victimization and perpetration has been discovered in both men's and women's reporting of domestic violence. Reasons for underreporting victimization include fear of reprisal by one's partner, protecting the partner from looking bad, and protection of one's self because of the concern that others would view him or her negatively if the victimization was tolerated repeatedly (O'Leary 1988b:223). Reasons for underreporting one's own violent actions include the social undesirability of abusing one's partner, and the fact that memory for negative events is often poorer than memory for positive events (O'Leary 1988b:223). Other studies which suggest underreporting include Browne and Dutton (1990), Carden (1994), Edleson and Brygger (1986), Jouriles and O'Leary (1985), McHugh (1993), McNeely and Mann (1990), McNeely and Robinson-Simpson (1987), Riggs, Murphy and O'Leary (1989), Sonkin and Durphy (1985), Stets and Straus (1990) and Szinovacz (1983). Overreporting of victimization and perpetration has also been found in both men's and women's reporting of domestic violence. Reasons for overreporting victimization can include a desire to make the partner look bad in order to increase alimony payments if a divorce proceeding is imminent, dramatizing to the therapist how bad the relationship is and how strongly professional help is needed, and exaggerating negative events because of depression and paranoia (O'Leary 1988b:223). According to O'Leary (1988b), there are cases of overreporting one's own violent acts. He suggests that reasons for overreporting aggression against one's partner include the desire to be strong and in control of the relationship, and an egocentric bias, which is a tendency to report both more positive and more negative events than one's partner reports about an individual (O'Leary 1988b:223). Other studies providing evidence of overreporting include BenDavid (1993), Browne and Dutton (1990), Marshall and Rose (1987) and Sonkin and Durphy (1985). 1  Other studies reporting the possibility of different definitions of violence between respondents include Arias and Johnson (1989), Cascardi and Vivian (1995), Marshall and Rose (1990) and Szinovacz (1983). Reliability is defined as the ability of a question to consistently provide the same answer in similar circumstances (Mangione 1995:28). Validity is a term used to describe  4  5  64  whether a measurement instrument measures what it is supposed to measure (Vogt 1993:240). For further discussion, see Breines and Gordon (1983), Browning and Dutton (1986), Brush (1990), DeKeseredy (1989), DeKeseredy and MacLean (1990), Dobash et al. (1992) , and Saunders (1988, 1989). For examples of studies asking both men and women questions of victimization and perpetration, see Arias and Beach (1987); Cantos, Neidig and O'Leary (1994); Jouriles and O'Leary (1985); Stets and Straus (1990); Szinovacz (1983). 6  7  Q  For examples of clinical samples, see Cantos, Neidig and O'Leary (1994, Dobash and Dobash (1979), Dutton (1995), Dutton et al. (1994), Giles-Sims (1983), Gondolf (1988), MacLeod (1980, 1987), Pagelow (1981), Saunders (1986, 1992) and Walker (1978, 1979, 1984). Examples of community samples are found in Straus, Gelles and Steinmetz (1980) who relied on data from the National Family Violence Survey conducted in 1975, as well as Arias and Beach (1987), Smith (1990a), Stets and Straus (1990) and Szinovacz (1983). Examples of studies which use both community and clinical samples include Cascardi et al. (1995), Holtzworth-Munroe and Anglin (1991), and Julian and McKenry (1993). Other studies reporting that women use violence more than men include Ben-David (1993) , Cate et al. (1982), Sorenson and Telles (1991) and Yllo and Straus (1981). 9  10  11  12  Other studies reporting mutual violence with no significant gender differences include Arias and Johnson (1989), Arias, Samios and O'Leary (1987), Lane and Gwartney-Gibbs (1985), Mahlstedt, Falcone and Rice-Spring (1993), Riggs (1993) and White and Koss (1991). Other studies reporting similar findings include Ansberry (1988), Ben-David (1993), McNeely and Mann (1990), Solomon (1980), Stets and Straus (1990), Straus (1980d) and Vesterdal (1980). For further discussion on risk markers, see Hotaling and Sugarman (1986, 1990). See also Bersani et al. (1992), Coleman and Straus (1986), Douglas and Der Ovanesian (1983), Dutton and Hart (1992a), Faulk (1974), Hotaling and Sugarman (1986), Margolin (1988), Rosenbaum and O'Leary (1981), Roy (1982), Straus, Gelles and Steinmetz (1980), Telch and Lindquist (1984), Walker (1984), and White and Humphrey (1994). Dutton (1995) makes reference to an "anger diary" to demonstrate anger control but his example of such a diary provides evidence of psychological abuse of the man by his wife, who is drunk and/or violent (258-9). For further studies in dating violence, see Bernard, Bernard and Bernard (1985), Billingham and Sack (1987), Caulfield and Riggs (1992), Ferraro and Johnson (1984), Henton et al. (1983), Lane and Gwartney-Gibbs (1985), Laner (1983), Makepeace (1981, 1983, 1987), Matthews (1984), O'Leary and Arias (1988), Riggs, O'Leary and Breslin (1990), Sack, Keller and Howard (1982), and Sigelman, Berry and Wiles (1984). 13  14  15  16  65 CHAPTER THREE: AN ASSESSMENT OF T H E 1987 A L L ALBERTA STUDY  3.0 Introduction The 1987 A l l Alberta Study was one of the first surveys in Canada to report community findings of wife assault. It was also the first survey in Canada to use the Conflict Tactics Scales (CTS) to determine the incidence of wife assault (Kennedy and Dutton 1989:46). Until then, only surveys conducted in the United States used the CTS. Through its use, the CTS allowed for important comparisons between urban and rural regions in Canada and the United States. However, the 1987 A l l Alberta Study was not specifically about wife assault in Alberta. Rather, it was an omnibus survey with various questions of actions, attitudes, and beliefs. The survey was a stepping stone for future research in the area of domestic violence in Canada. It also had methodological drawbacks. In this chapter, the methodological advantages and disadvantages of the survey will be discussed. Assessing this survey requires a set of criteria for proper survey methodology, including the specifically issues: 1. Background and Objectives 2. The Methodology 3. Development 4. Timing 5. Method of Data Collection 6. Population 7. Sample Frame 8. Sample Size 9. Response Rate 10. Questions 11. Ethical Issues  66 A report of its initial findings and a discussion of the quality of measurement and its theoretical assumptions will follow. Throughout this chapter a number of revisions will be suggested based on the weaknesses highlighted. The revisions will provide a basis for improved methodology in the future.  3.1 Background of the Survey and its Objectives The Population Research Laboratory within the University of Alberta's Department of Sociology was responsible for collecting data about the population of Edmonton. Previous to the 1987 A l l Alberta Study, the Population Research Laboratory had conducted an annual survey project, using face to face interviews, for the past ten years, thus increasing the reliability of the study. The questionnaires were of the amalgam type, representing different questions pertaining to specific research interests of university researchers and outside agencies, with questions about both the respondent's actions and values regarding several issues (Kinzel 1987:17). There were similar questions in each of the surveys, but the survey also included questions about certain issues which were pertinent to that year in Edmonton. Such studies allowed university researchers and local community agencies to investigate and explore social issues, of importance to the population of Alberta in an on-going research framework (Kinzel 1987:16). The 1987 A l l Alberta Study is the eleventh study of this series. The principal investigator was Leslie Kennedy from the Department of Sociology, University of Alberta. The survey sample was selected from the area in and surrounding Edmonton, as were the previous surveys, but it was expanded to include all of Alberta, and comprised  67 both face-to-face and telephone interviewing (Kinzel 1987:16). Although there was no designated special topic per se, a pertinent issue in this survey, which was not previously addressed, was spousal violence. Spousal violence, in this study, is defined as any action from the Conflict Tactics Scale used by one spouse to resolve conflict with the other (Kennedy and Dutton 1989:42). The purpose of the study was to specifically collect survey data on domestic violence and wife assault for both urban and rural areas in Alberta and to compare these data with U.S. results from a variety of geographic areas (Kennedy and Dutton 1989:46). As mentioned above, it was the first survey in Canada to use the CTS to determine the incidence of wife assault in a community sample. This study also provided data on the incidence of wife assault at a regional level, which would help begin to solve the problem of a lack of empirical data at national and regional levels (Kennedy and Dutton 1989:40). Previous to this survey, there were no reliable rates of wife assault in Canada. Rates that had been used in the media were based on rough estimates taken from workers in shelters and transition homes for battered women. The importance of the use of the CTS lies in the provision for the comparison of incidence rates of domestic violence between Canada and the United States. As mentioned above, many of the American surveys of domestic violence used the CTS or versions of it. Since its creation, the CTS has received and continues to receive criticism, because of its exclusion of the context of the acts and the definitions used. Each version 1  of the CTS has attempted to improve the validity of its findings, based on these criticisms.  68 3.2 The Methodology of the Survey The Population Research Laboratory has used social surveys to collect data for the Edmonton and All Alberta studies. Although surveys handle time order poorly and are subject to the problems of testing effects and instrumentation changes which may make it highly susceptible to interviewer bias (Denzin 1989:29; Gray and Guppy 1994:50), surveys are considered to be the best method available to collect original data describing a population too large to observe directly (Babbie 1989:237). The survey conducted was a cross-sectional survey, in which data are collected at one point in time from a sample selected to describe some larger population at that time (Babbie 1973:62). However, this cross-sectional survey is part of a larger longitudinal survey, in which data are collected at different points in time, allowing the researcher to report changes over time. Longitudinal designs include trend studies, cohort studies, and panel studies (Babbie 1973:63). Trend studies involve a given, general population sampled and studied at different points in time (p. 63). Cohort studies focus on the same specific population each time data are collected, although the sample studies may be different, whereas panel studies involve the collection of data over time from the same sample of respondents (Babbie 1973:64). This survey is best described as a cohort study, but also exhibits characteristics of trend studies, except that trend studies are often conducted via secondary data, rather than having these data collected regularly by the same organization. One concern with this study in particular is that the questions specific to domestic violence were new and were not repeated in subsequent surveys by the Population Research Laboratory, defeating the purpose of following trends or comparing  69 cohorts. The nonrepetition also allows for the questioning of the reliability of the results of this study; these data cannot be compared to another similar study in Alberta.  3.2.1 Development of the Survey Little information was provided about the development of the survey, except that both questionnaires were thoroughly pretested on 59 Edmonton households (Kennedy and Dutton 1989:46). This pretest exceeded the average test of 20-50 cases which is usually sufficient to discover the major flaws in a questionnaire before they damage the main study (Sudman 1983:181). As the questionnaires were based on or replicated the studies used in the last ten years by the Population Research Laboratory, the development of this survey did not require extensive discussion, except to mention the inclusion of the Conflict Tactics Scale. However, the researchers should have consulted with relevant interest groups regarding the wording and context of the domestic violence questions. The researchers could also have used three techniques to evaluate these questions. The first technique involves asking test subjects to rephrase the questions in their own words to discern their interpretations. Another technique is to conduct a double interview, in which the second interview entails asking the subject to explain how he or she arrived at that answer. The last technique involves asking the test subjects to think aloud as they answer each question to identify areas of difficulties (Foddy 1993:186-7).  3.2.2 Timing of the Survey The importance of timing involves noting the start and end date of the administration of the survey. If the time period is too long in a cross-sectional survey  70 design, other unanticipated variables may affect responses. Unidentified variables cannot be controlled for by the researchers, thus negatively impacting the reliability of the survey data (Gray and Guppy 1992:43; Singleton et al. 1988:353-4). The interviews in this study were conducted between January and March, 1987. In Edmonton, the interviews began on February 5 and continued for six weeks (Kinzel 1987:8), while in all other areas, they began at the end of January and were completed by the end of February (Kinzel 1987:9). Because the time period was relatively short effects of unanticipated variables will not be related to the timing of the survey.  3.2.3 Method of Data Collection This survey consisted of interviews and questionnaires. The Population Research Laboratory used two techniques of interview research for this study. For the Edmonton sample, face-to-face interviews were conducted. For Calgary and all other areas, telephone interviews were conducted. The use of two interview techniques allowed for a unique comparison of the two techniques. There were concerns that the use of both interview techniques would yield inconsistencies in these data. For example, respondents may disclose more in a telephone interview which allows for anonymity, than in a faceto-face interview. However, as far as the survey in general goes, underreporting was expected due to the nature of the survey topic, but it was no worse than in other telephone surveys. Also, there was no significant drop in responses as from less to more sensitive questions were asked in the face to face interviews (Kennedy and Dutton 1989:42). The Population Research Laboratory conducted its own test in the following year to detect any noticeable differences between using face-to-face and telephone interviews. When the  71 questions were controversial and/or very long, some differences were found; but overall, there were no radical differences between face-to-face and telephone interviews (Lalu 1988:9). The questionnaire required careful consideration in this dual method study; in fact, two questionnaires were used, one for the telephone interview and one for the faceto-face interview (Kinzel 1987:16; Kennedy and Dutton 1989:46). The telephone survey involved questions taken from the face-to-face survey, and was about half the length of the face to face interview (Kinzel 1987:16; Kennedy and Dutton 1989:46). It is important that the questionnaire for the telephone survey be similar to the questionnaire for the faceto-face survey for comparability and reliability of the results; when variations between respondents do appear, they can be attributed to actual differences in response, not to the instrument (Denzin 1989:104; Fowler 1988:75). However, it is also important that the telephone survey questions be adjusted to accommodate the method. Three approaches have been suggested for such an adjustment. First, response scales can be limited to 2-4 categories. Second, longer lists should be read slowly, and then repeated so that the respondent can choose the correct category when it is read out aloud. This suggestion, however, has not demonstrated that answers obtained in this way are identical to those given to a visual list of categories. For some kinds of questions, the answers are affected by the order in which responses are read (Richardson, Dohrenwend and Klein 1965:43). Thirdly, complex questions can be broken down into a set of two or more simpler questions (Fowler 1988:64). In comparing the questionnaires used for both the telephone and face-to-face interviews, the questions  72 were identical except for a few which had additional wording for suitable administration over the telephone (Kinzel 1987:16; Kennedy and Dutton 1989:46). In order to administer the survey, interviewers were required for both the telephone and face-to-face questionnaires. The face-to-face data collection in Edmonton required 36 trained interviewers. The telephone survey required fewer interviewers; only seven interviewers were used in Calgary and five for the remainder of Alberta. All faceto-face interviews were conducted at the respondents' homes. The interviewers in Calgary conducted the telephone interviews from their homes, while the interviewers for the remainder of Alberta used a private office in the Population Research Laboratory. From what is known about the training and supervision of these interviewers, there were two concerns. First, the telephone interviewers in Calgary were allowed to work from their homes unsupervised. The necessary monitoring of interviews mentioned above was not given careful consideration by the researchers, and could have influenced the quality of data collected. Second, both male and female interviewers conducted the interviews for both male and female respondents, which also will have had an effect on these data. Interviewer effects, for men interviewing women and women interviewing men, have been reported in surveys, especially ones with sensitive questions about domestic violence and one's participation in that violence. If both men and women conduct the interviews, then the respondents are not being presented with the same set of stimuli, and therefore affects the reliability of these data (Denzin 1989:116). However, there is no published discussion of the extent of bias.  73 3.2.4. The Population The target population, or the population which the researcher would like to study (Frankel 1983:24), in this case consists of individuals within households in the province of Alberta. The 1987 Edmonton population was designated as all persons 18 years of age or older who at the time of the survey were living in a dwelling unit that was enumerated during the spring of 1986 City of Edmonton Civic Census (Kinzel 1987:3-4). The population designated for telephone interviewing in Calgary and all other areas was all persons 18 years of age or older who at the time of the survey were living in a dwelling unit in Alberta outside of the City of Edmonton, that could be contacted by direct-dialing. From this population two samples were drawn to cover the area: the first one represented the City of Calgary and the second represented the remainder of the province (Kinzel 1987:5).  3.2.5 The Sample Frame The main features of the sampling design included the delineation of the province of Alberta into three areas and a two-stage selection process. The three areas of the province, as mentioned above, included the cities of Edmonton and Calgary, and the remainder of the province. The two stage selection process involved the selection of households and the subsequent selection of a respondent within each household (Kinzel 1987:17). The selection of households began with sample frames or with other alternatives not requiring a sample frame. A sample frame is represented by the list or other record of the population from which the sample are chosen (Vogt 1993:202). Researchers often  74 find locating an available list of the general population difficult, if not impossible, because of the constant change in migration, deaths, and aging into adult status (Henry 1990:85). In the Edmonton sample, the researchers relied upon a computerized list of addresses compiled by the city of Edmonton from the most recent enumeration in 1986. The researchers randomly selected 620 addresses for personal interviews. Then, nursing homes, military and temporary residences were deleted from the sample. Within each household, one eligible respondent was selected for the one hour interview (Kinzel 1987:18). The Calgary sampling procedure did not require a sampling frame due to the use of random digit dialing (Henry 1990:85). The only reliance on a list for this procedure was the assembly of the three-digit prefixes of the exchanges in Calgary in order to generate seven-digit telephone numbers for the city of Calgary. This procedure was dual staged with an exchange being randomly selected in the first stage. In the second stage, a four-digit random number between 0000 and 9999 was selected and added to the exchange prefix. This process was completed until a sufficient number of telephone numbers had been collected. Businesses and non-existent numbers were eliminated by using the most recent edition of the "Calgary Numerical and Street Address Directory" (Kinzel 1987:19). This procedure excludes new home constructions which would not be included in the directory, homes without telephones (albeit few), or the increased chance of nonresponse for homes with unlisted telephone numbers (Fowler 1988:68). The sampling procedure for the remainder of Alberta telephone interviews was different from the Calgary telephone interviews sampling procedure because no reverse directory existed to screen out the large amount of non-existent numbers which would be  75 generated. This sampling procedure involved using eighteen directories from Alberta Government Telephones and selecting ten percent of the telephone numbers to be processed. The numbers were determined by selecting random pages from the directories and then recording ten percent of the total number of the columns into a computer. The last two digits of each number were stripped to allow for a random selection of a number between 00 and 99 to generate each telephone number. Duplicate numbers were identified and discarded (Kinzel 1987:19). For this sample, as well as the Edmonton sample, the sampling frames did not truly include all elements that their names might imply. For example, in Edmonton, only dwelling units enumerated in the 1986 census were included in the target population. All dwelling units inhabited or built after that census were excluded. Telephone directories are subject to regular changes, ignoring unlisted numbers as well as those who do not have a telephone. The researcher needed to assess the extent of these omissions and correct them, unless the number of omissions was small (Babbie 1973:91). In this study, the researchers corrected their original samples which were plagued with non-residential or ineligible units, numbers not in service, vacant or demolished dwellings or nonprimary residences. For example, the original sample size for the remainder of Alberta was 858 households but was corrected to 430 (Kinzel 1987:11). The selection of a respondent from each household was the second stage of the sampling design. The respondents for both face-to-face and telephone interviews were chosen following the same selection procedure guidelines: a) The dwelling unit must be the person's usual place of residence and he/she must be 18 years of age or older; b) If an adult male answers and is willing to be interviewed,  76 he is the respondent; c) If an adult female answers and there is an adult male present who is willing to be interviewed, interview the male. If the male is not willing to be interviewed, and the female is willing, interview the female; d) If an adult female answers and there is no adult male present, choose her as the respondent (Kinzel 1987:18). The rationale for these guidelines, which may appear to be problematic because such methods can bias the results (Henry 1990:92), was that some women, such as housewives and mothers, are in the home more often than men. Therefore, the interviewers needed to seek a male respondent more ardently than female respondents; "interviewer discretion, respondent discretion, and availability (which is related to working status, life-style and age) would affect who turned out to be the respondent" (Fowler 1988:33). These guidelines enabled the interviewers to interview equal numbers of men and women. In the remainder of Alberta sample, the final sex breakdown was 49.3% male and 50.7% female. In the Calgary sample, the final sex breakdown was 47.5% male and 52.5% female. The Edmonton sample consisted of 50.7% male and 49.3% female respondents (Kinzel 1987:4).  3.2.6 The Sample Size  Sample size is the most potent method of achieving precise and reliable estimates that are sufficient for policy decisions or scientific inquiry (Henry 1990:117). One common misconception is that the adequacy of a sample depends heavily on the fraction of the population included in that sample (Fowler 1988:40); however, there is no uniform standard of quality that must always be reached by every sample. The quality of the  77 sample depends entirely on the stage of the research and how the information will be used (Sudman 1983:146). A number of factors influenced the sample size in this study. The Edmonton study required a sample size of 400 or more in order to retain the comparability of the Edmonton samples in past Edmonton Area Survey studies. The adequacy of the sample size is determined by the minimum sample sizes that can be tolerated for the subgroups within the total population for which separate estimates are required (Fowler 1993:35). The sample sizes for Calgary and the remainder of Alberta were determined by budgetary constraints, but a minimum sample size was still sought to permit separate analysis for these two areas to increase the reliability of these data (Kinzel 1987:3). The researchers used a corrected sample size of 595 households in Edmonton, 318 households in Calgary and 430 in the remainder of Alberta. By the completion of the surveys, the sample size had decreased to 454 respondents in Edmonton, 244 in Calgary and 347 in the remainder of Alberta. The Population Research Laboratory noted that the sample obtained for each area was not proportional to the actual population. Therefore, weighting was necessary in order to combine the samples for a provincial sample (Henry 1990:57). Limits were placed on the sample size by budgetary constraints but the sampling of the entire province strengthened the generalizability of the sample and contributes to a favourable assessment of the sampling method.  3.2.7 The Response Rate The response rates were reported to be 76.3% in Edmonton, 76.7% in Calgary and 80.7% for the remainder of Alberta (Kinzel 1987:12). These rates were calculated based 3  78 on the corrected sample. Nonresponse is due to no contact or answer, respondent 4  refusals and language difficulties (Fowler 1988:45). The largest factor of nonresponse involved respondents' refusals to answer the questions, ranging from 9% to as high as 18%. This can create nonsampling bias because a portion of the population is underrepresented in the sample (Henry 1990:124); however, if most of those selected provide data, as evidenced by the high response rates in this study, sample estimates will be very good even if the nonrespondents are distinctive (Fowler 1988:48). There was no evidence of systematic bias in the nonresponse rate.  3.2.8 T h e Questions  A reasonable set of standards for survey questions has been established as the following: (a) respondents must understand the questions as intended by the researcher; (b) respondents must have access to the information needed to answer the questions; (c) respondents can recall or retrieve the information needed to provide the answers; (d) respondents can transform that information into the form required to answer the questions; and (e) respondents must be willing to give accurate answers (Fowler and Cannell 1996:15). Researchers are able to apply three main procedures to help design and evaluate questions which include focus groups, intensive cognitive interviews, and field pretests (Fowler and Cannell 1996:16). These procedures also increase the validity of the questions, ensuring that these questions measure what they are intended to measure (Carmines and Zeller 1979:17). There are three types of validity: A question must have criterion-related validity, being able to estimate abuse and violence (p. 17), by specifying the full domain of content  79 relevant to those behaviours (content validity) (p. 20). It must also have construct validity by relating to other measures of abuse and violence (p. 23). As mentioned previously, the questions were pretested on an Edmonton sample and the researchers were confident that the questions followed the above set of standards. Close-ended questions were mostly used in both the face-to-face and telephone interviews. Close-ended questions are usually a more satisfactory way of obtaining data because respondents can more reliably perform the task of answering questions when response alternatives are given (Fowler 1988:87). For example, Question #46 asks, "Which of the following describes what you did as a result: hit back or threw something, cried, yelled or cursed, ran to another room, ran out of the house, called a friend or relative, called the police?" The recording of the answers is simplified for the interviewers with the use of such questions (Gray and Guppy 1994:84). However, a disadvantage to using such a question is that if a close-ended question has not been sufficiently pretested, the researcher risks inadequate response categories and decreases the validity of that question as a measurement tool (Gray and Guppy 1994:84-5; Mangione 1995:21); the respondents will not be able to answer the question accurately if a particular response category is not given. Whether the respondent is forced to choose an inappropriate response or chooses an inappropriate response with little recognizance, these data collected from that question is unreliable. The likeliness of a negative effect from inappropriate responses decreases with the desire of the respondent to contribute to the study. The other questions which were used may be considered open-ended; the researcher cannot anticipate the various ways in which people are likely to respond to a  80 question, so questions are asked in order to "stimulate free thought, solicit suggestions, probe people's memories, and clarify positions" (Dillman 1978:87). In this study, some questions were open-ended. There were no response categories given to the respondent, but the questions were specific enough to not warrant such categories, so as not to affect the coding of these data. These questions help add to the respondents' enjoyment (Gray and Guppy 1994:82) and allow the respondents to provide answers in their own words (Fowler 1988:87). An example is one part of a four part question which asks, "What did the police do?" (Population Research Laboratory 1987:29). The researchers anticipated some of the possible responses, such as "the police took down names and gave the victim a ride to the hospital," "jailed the offender for one day," or "jailed the offender and went to court" (p.29). The respondent, in this situation, is allowed to recount the actions of the police in his or her own words. It is then the responsibility of the interviewer to fit the response into the categories, provided by the researcher, but not given to the respondent. This becomes problematic because the response categories provided may not be sufficient for this question. The above question should have included other response categories such as "the police taking down names only," or "jailed offender but the case never went to court," in order to increase the content validity (Carmines and Zeller 1979:20). These responses were not given and therefore must somehow fit into what is given, thus providing data which does not accurately reflect reality for that respondent. This not only portrays an inaccurate reality but it also introduces interviewer bias. If the respondent gives an unanticipated response, the interviewer may have to encourage the respondent to rephrase his or her answer in order to fit into one of the response categories (Gray and  81  Guppy 1994:82-3). Interviewers should never be given any options about what questions to read, or how to read them (Fowler 1988:78) to avoid interviewer biases. A comprehensive list of possible responses is essential for standardized coding and the analysis of these data (Schwarz et al. 1989:166; see also Sudman and Bradburn 1982:21) or the list should allow for unanticipated responses. Open-ended questions should allow the researcher to obtain answers which were unanticipated and give the interviewer space on the questionnaire to record such answers (Fowler 1988:87). One solution is to use partially close-ended questions which allow the respondent to choose from provided response categories or to create their own responses in order to prevent "forcing people into boxes in which they clearly do not fit" (Dillman 1978:92). The survey instruments contain questions regarding demographic characteristics of the population, health, smoking, abortion attitudes, the Alberta economy, family violence, crime, police, employment, and television viewing. There were also questions regarding views of the school system, seatbelts, small business, handgun use, fertility, automated machines (such as automated bank tellers), political party preference, and personal finances (Kinzel 1987:18-19). The questions of family violence made inquiry into the actions of offenders and experiences of victims for both men and women which was an important step for domestic violence researchers to take in Canada. Such questions help to prevent placing men on the defensive when answering questions about domestic violence; if questions are demeaning, embarrassing or upsetting, respondents, may terminate the interview or falsify their answers (Sudman and Bradburn 1982:5; Dillman 1978:101). The questions also allow men and women to identify their own experiences as victims.  82 The questions also incorporated an abbreviated version of the CTS to only include physical violence. This was the first Canadian survey to use the CTS and allowed for comparisons with U.S. findings from studies using the CTS and similar definitions. Because the questions from the CTS have been asked of other U.S. samples, collecting comparable data in Canada added to the generalizability of domestic violence research and the reliability of these data (Fowler 1988:101). The focus of this abbreviated CTS is on physical violence only, even though psychological abuse is an important factor, because physical violence is more likely to receive a social policy response (Kennedy and Dutton 1989:41). Sexual violence is also excluded without justification, which means that the emotional damage which accompanies psychological and sexual abuse is ignored. The scale included the physical acts of throwing objects, pushing, grabbing, shoving, slapping, kicking, biting, hitting with a fist, hitting or trying to hit with an object, choking, threatening with a knife or gun, or using a knife or gun. Because of the threatening nature of these questions, a context was established in the introduction of this scale in order to make it easier for respondents to admit that they have engaged in a behaviour which they see as undesirable (Sudman andBradburn 1982:78). In order to provide that context, the introduction read as follows: "No matter how well a couple gets along with one another, there are times when they disagree, and they may use many different way of trying to settle their disagreements" (Population Research Laboratory 1987:27). The introduction refers to different ways of settling disagreements but the respondent is only allowed to respond to items of physical violence. The introduction should have alluded to that fact to enhance the criterion-related validity of  83 this version of the CTS as a measurement tool (Carmines and Zeller 1979:17). The CTS must be able to estimate the incidence of physical violence as it is intended. Respondents who do not use violence may feel that their reality is not important to the researcher because they are not allowed to tell the interviewer how they do settle disagreements. It also may have affected the responses given to these questions, since this introduction is intended for the full scale. In the original CTS, verbal and physical abuse appear on the questionnaire only after such conflict management strategies as discussion or reasoning are mentioned, which lends an air of "conflict as a last resort" that serves to legitimize the reporting of the use of violence (Shehan and Lee 1990:436). If other strategies are not presented, there is concern that the introduction will not have its intended effect. Also, as is often the case with the use of the CTS, there were no questions of context, motives, what the conflict was about or even if there was a conflict involved (DeKeseredy and Hinch 1991:23-5). The CTS is helpful in gaining quantitative data but other measurement tools are necessary to determine the context and motives, which need to be measured qualitatively and quantitatively. Future research has attempted to use both quantitative and qualitative measures, further developing our knowledge of domestic violence. This, however, can not be accomplished without the guidance of further theoretical development. Another problem with the use of the CTS is the lack of checks on the accuracy of reporting, especially when it involves lifetime or annual rates, as in this study. Even with monthly rates there is a concern with a decrease in the reporting of events as time elapses, affecting the reliability of these data (Cannell, Marquis and Laurent 1977:7; see also Bachman and O'Malley 1989:182); there is also a concern with situations in which some  84 respondents may actually overestimate their recollections of events during the past few months, as the events may appear to be more severe than in actuality (Bachman and O'Malley 1989:183). Event recall can be extremely tedious and taxing for respondents; however, despite these problems, asking questions of past events is important for some survey research projects (Gray and Guppy 1994:89), such as this one because there are no other alternatives to obtain data required to test hypotheses regarding domestic violence. This survey is also subject to the criticisms often made of omnibus victimization surveys. These surveys do not directly address the issue of domestic violence. However, the All Alberta study introduced such questions into Canadian research. These sensitive questions about domestic violence were presented in the middle of the questionnaire in order to cushion their threatening characteristic by other issues which were nonthreatening (Fowler 1988:102). This survey included questions which were rare, but important, in domestic violence research. One particular question regarding the initiation of most serious physical violence is the following: "Let's talk about the most serious time in the last year when you and your (spouse/partner/former spouse/partner) got into a physical fight. In that particular instance, who started the physical conflict, you or your (spouse/partner)?" (Population Research Laboratory 1987:29). The codebook allowed for the following possible responses: "you," "spouse/partner," "both," "neither," "not sure," or "not applicable" (p.29). This is an important question to help identify the process of violence and its initiators in immediate situations. The process of violence has been known to begin with psychological abuse, sometimes years prior to the violence (Dobash and Dobash 1979:97; Sabourin 1996:199). This demonstrates the need for including  85 questions about psychological abuse in domestic violence research. However, it is also important to identify the actions and reactions of the victim and aggressor immediately prior to the violence. This question also avoids asking respondents to summarize past events, which are difficult to report with accuracy, by asking them to recall actual events or episodes in order to gather events or episode history information (Gray and Guppy 1994:90). If the respondent reported that the spouse or partner was the initiator, then the following question was asked: "Which of the following describes what you did as a result: "hit back or threw something," "cried, yelled or cursed (him/her)," "ran to another room," "ran out of the room," "called a friend or relative," or "called the police?" (Population Research Laboratory 1987:29). Although this question does not allow for the recording of successive events, it does help identify the violence process. Ideally, the question would ask what happened following the initiation of violence; the interviewer would then be required to record the events that followed in order. If the respondent reported being the initiator of violence, he or she was read the following: "In family life, sometimes conflict and tension occur. Let us consider the hypothetical situation where as a consequence of this conflict you hit your (spouse/partner). I am going to read a list of things which might happen as a result of this action. Please rate the chances of each result on a '0 to 10' scale. You should give a '01 if you think this has no chance at all of happening, and a '10' if you think this is sure to happen. And you can use any number between '0 to 10"' (Population Research Laboratory 1987:31). A hypothetical question, such as this, converts a behavioral question into an attitudinal question. This is only acceptable if such a change in the kind of information being  86 gathered is acceptable to the researcher (Dillman 1978:114). In this situation, the use of a hypothetical situation is intended to test the argument that wife assault remains prevalent in our society because it is viewed as acceptable behaviour (Dobash and Dobash 1979:104-5; Straus 1977b:68). The hypothetical questions allowed the researcher to tap the attitudes of respondents about wife assault. However, there is the risk that social desirability will influence respondents' reporting. Respondents may report only what they think the interviewer wants to hear (Denzin 1989:117) or what social norms dictate as appropriate attitudes (Gray and Guppy 1994:75). Researchers must use such questions with considerable caution, especially if respondents are likely to have limited direct experiences on which to base their answers (Fowler 1995:80). It is also important to realize that there is no objective way to validate these responses except through correlations with other responses to related questions (Fowler 1993:80). Other options include using hypothetical actors in hypothetical situations, rather than placing the respondents in hypothetical situations, in order to decrease any personal meaning given to the questions which might bias the respondents' answers and decrease the reliability of these data. The above question also presents another problem with the use of a '0 to 10' scale. Many of the questions used in the family violence section contained 11 digit response categories. Such response categories are too arbitrary for respondents who prefer to use answers already provided such as "strongly disagree," "disagree somewhat," "agree somewhat," "strongly agree," or "neutral feelings/no opinion." The respondents are less likely to answer truthfully to a numerical scale question that if given a descriptive list of possible responses (Dillman 1978:206).  87 Another problem with the questions on family violence involves the lack of definition or clarification of the word "hitting." Such a word will vary in interpretation for both men and women as gender groups and as individuals, as well as for the researchers. For example, a woman may report hitting, but a man will not report experiencing that hit because he did not perceive that hit as injurious or threatening, as well as a man may not report hitting a woman because he did not perceive it to be injurious or threatening as perceived by the victim. No checks and balances were conducted in this survey for differences in definitions. "Since the same word can conjure significantly different meanings to different respondents, it is not surprising that different words designed to tap the same object or feeling state can actually serve as significantly different stimuli and trigger different response patterns" (Smith 1989:100). Either the questions should have been more specific in the answers provided or a definition should have been provided for the benefit of the respondents (Babbie 1989:141; Gray and Guppy 1994:72; Sudman and Bradburn 1982:1) in order to define and clarify the word "hitting." Ensuring that all respondents have the same understanding of what is to be reported is one of the most difficult tasks for researchers, and failure to accomplish this properly is a major source of error in survey research (Fowler 1995:13).  3.3 Ethical Issues As the primary investigator, Kennedy faced the ethical dilemma of weighing the benefits of more reliable crime statistics against the risks of rekindling personal anguish for the respondents (Gray and Guppy 1994:25) by asking questions about experiences as victim or perpetrator of spousal violence. However, there is no discussion available  88 regarding any precautions taken to deal with these ethical issues. In order to deal with such issues, Kennedy should have used trained interviewers who had the skills to help respondents cope with the after-effects of questioning. Kennedy should have provided respondents with information about domestic violence after the interview, as well as follow-up counselling with therapists (p. 25). Researchers of sensitive topics such as spousal violence must accept responsibility for the consequences of asking people related questions (p. 25).  3.4 The Findings The findings of this study in comparison to U.S. findings were virtually identical for overall violence rates, but serious rates were less in Alberta than in the U.S. Table 1 presents a comparison of specific acts of husband-to-wife violence for 1985 U.S. and 1987 Alberta surveys. The overall violence rates were 11.3% in the U.S. samples and 11.2% in Alberta for husband-to-wife violence (Kennedy and Dutton 1989:47). The items including "beat up" and "threatened with knife or gun" in Alberta were both 25% of the corresponding U.S. rates (p. 48). The initial analyses of these data were reported by Kennedy and Dutton (1989) who focused entirely on wife assault. The highest rates of wife assault at 19.5% were found in the age range of 18 to 34 years. Marital status also seemed to be a factor in the highest rates of wife assault, especially for those women who were single, separated, divorced or had been living together in the last year. Wife assault rates were also high (20.5%) for young people in school (p. 49). High rates were  89  Table 1. Husband-to-Wife Violence: Comparison of Specific Acts for 1985 U.S. and 1987 Alberta (Canada) Surveys Rate Per 100 Couples Type of Violence U.S.(1985) Alberta (1987) A . Minor Violence Acts 1. Threw something 2.8 4.2 9.2 2. Pushed/grabbed/shoved 9.3 1.7 3. Slapped 2.9 B. Severe Violence Acts 4. Kicked/bit/hit with fist 1.5 0.9 5. Hit, tried to hit with something 1.7 1.6 6. Beat up 0.8 0.2 0.4 7. Threatened with gun or knife 0.1 0.2 8. Used gun or knife 0.0 Number of cases 3,520 708 (Source: Kennedy and Dutton 1989:49)  also found in lower incomes and larger centres, such as Calgary and Edmonton. For example, the rate of wife assault in small towns was 8.3%, compared with 14.1% in Edmonton. The rate of wife assault in lower income families was 13.8% compared with 7.5% in higher incomes (p. 50). The use of three areas within the province of Alberta allows for the distinction between rural and urban area findings. This is important as sample survey findings can be taken only as representative of the sum of elements that comprise the sampling frame (Babbie 1973:91). The findings are presented appropriately. To my knowledge, there are only three reports of the findings, including Kennedy and Dutton (1989). The other reports include an unpublished doctoral dissertation by Avakame (1993) at the University of Alberta and a highly technical report for the Edmonton Area Series by TanVergert, Gillespie and Kingma (1988). Kennedy and Dutton (1989) chose to report only on wife assault because of the serious nature of the  90 injuries incurred by women with male partners. The basis for this observation comes from a report on domestic dispute calls of which the majority of injuries were incurred by women (Berk et al. 1983:204). A report of the findings on the violent experiences of the male respondents was not published, although these data were available through interviews conducted with both men and women.  3.5 The Quality of Measurement This discussion of the quality of measurement will focus on the issue of replication of studies; other issues have been discussed in terms of the quality of measurement in previous sections. The Edmonton Area Studies used annual rates as opposed to lifetime rates, because the studies were conducted every year, replicating the questions. However, the questions of wife assault were only asked in the 1987 All Alberta Study; these questions were never replicated. Because the questions were never replicated, no steps were taken to minimize researcher bias by reproducing the results of this study (Gray and Guppy 1994:23). The only replication that took place in this study involved the use of the CTS in the Canadian provincial study replicating U.S. national studies. Caution should be taken given this study only used a revised CTS, as opposed to the full CTS used in the U.S. This may be problematic because the shorter version of the CTS may prevent the scale from measuring what it is intended to measure. However, this study allows for comparisons of rates of physical violence in Alberta and the U.S.; if the same results can be observed in several locations with widely different populations, researchers have a great deal more confidence in the generality of the results than if the sample is only of a single location (Sudman 1983:154). These studies should be  91 replicated in many more areas to increase the reliability of measures. Until then, the results from this study will provide a baseline of domestic violence in Alberta, Canada.  3.6 The Theoretical Assumptions  Before any research project begins, one or more theories lay the groundwork for the study. These theories set out the problems for the research project, staking out new objects for examination and directing empirical inquiry (Blumer 1954:3). The research project provides the fundamental test of those theories and their verification. Through the use of research methods, new observations are brought forth to modify, verify, and change the theory under examination (Denzin 1989:64-5). In this study, there are four theories or perspectives which provide the framework (see also Chapter One). The first theoretical perspective is the feminist perspective. These theories explain, through studies such as Dobash and Dobash (1979), that men who physically assault their spouses do so because their partners have violated, or are perceived as violating, the ideals of familial patriarchy (DeKeseredy and Kelly 1993a: 138), with the assumption that men abuse their partners more often than women and with graver consequences. Such themes are evident in the questions specifically referring to wife abuse, such as the following: "Have you ever personally known of a woman who has experienced wife abuse?" (Population Research Laboratory 1987:33). The use of a modified version of the CTS and further questions about domestic violence are also evident of the influence of feminist perspectives. These data did reveal an overall rate of violence to be 11.2% in the province of Alberta. Further findings revealed that wife assault is viewed as having extremely serious negative consequences for the victim by all  92 respondents (Kennedy and Dutton 1989:49). Although feminist perspectives are responsible for bringing domestic violence to the focus of research, these data did not provide evidence of patriarchal values encouraging wife assault. More research is required to determine the theories' effectiveness in explaining domestic violence. Interactional theories allowed for the consideration of physical violence as a strategy for conflict resolution, in which conflict is seen as a part of social relationships, and the more the conflict, the more likely it is that violence will be used as a problemsolving tactic (Goode 1971:632; Hepburn 1973:420). However, in this case, this consideration appears only to be used in an introduction to encourage respondents to report their violent behaviour: "No matter how well a couple gets along with one another, there are times when they disagree, and they may use many different ways of trying to settle their disagreements" (Population Research Laboratory 1987:27). No other strategies of conflict resolution are discussed. This, however, is necessarily linked to issues of power and control in feminist theories — how violence is used, especially by men, to maintain or regain the power and control in a relationship. Such violence is viewed by feminists as an "expression of social power" (McCall and Shields 1986:107), and may or may not be an issue of conflict or disagreement except over who holds the power and control. Another theory underlying the assumptions of this study is the social learning theory. Social learning theory explains that human behaviour is learned through modeling (O'Leary 1988a:33), and that behaviour is supported by others' attitudes reinforcing that behaviour. Findings have been presented to suggest that when an individual's attitudes are shaped to support the use of physical force against a woman, the  93 likelihood of physical aggression increases (O'Leary 1993:24). For example, question #49 addresses a hypothetical situation in which the respondent hit his or her partner and then asks what are the chances that, e.g., "your friends or relatives disapprove or lose respect for you" or "you lose respect for yourself (Population Research Laboratory 1987:31). These attitudes are influenced by the attitudes of those around the respondent and may encourage or discourage violence. The findings actually revealed that the overwhelming majority of both men and women clearly view wife assault to be unacceptable behaviour, although the incidence rates reveal otherwise (Kennedy and Dutton 1989:50). More research will be needed to determine the connection between beliefs and actions. Questions about the respondent's age, income, marital status and the location of residence are guided by sociological theories which attempt to identify risk factors of wife abuse. These factors constitute social structure and within this structure, there is inequality, subordination and superordination, and thus social stratification (McCall and Shields 1986:100). The problems within this social structure affect the family, one of society's sub-structures, and thus helps identify risk factors of violence in intimate relationships such as low income, young age, common-law relationships and highly populated urban areas, as revealed by the findings. The researchers did not explicitly acknowledge the influence of these theories, although the questions provide evidence of such underlying assumptions. However, this study was strengthened through the use of these theories which allow for the possible testing of numerous hypotheses, making the study more worthwhile. However, there are  94 a number of other theories which also contribute to our understanding of domestic violence which should be addressed in future research.  3.7 Conclusion The 1987 All Alberta Study is important because it was the first Canadian survey to address the problem of domestic violence. It was also the first Canadian survey to use the CTS to determine the incidence of wife assault and the first to use a community sample. This study also provided data at a regional level to begin to solve the problem of a lack of empirical data at national and regional levels (Kennedy and Dutton 1989:40). The use of the CTS, although deserved of its many criticisms, was also important in providing for the comparison of incidence rates of domestic violence between Canada and the United States. This study was also one of the first to use both telephone and face-to-face interviews. The researchers paid careful attention to these differences, and responded appropriately with two questionnaires, which were carefully pretested before the administration of the survey. The research team was rewarded with a high response rate, indicating further the appropriateness of the questionnaires. The survey is also important because it included questions about violence as a process, rather than a static event. However, this study fell short of determining the reliability of its findings in Canada because the study was not repeated in Alberta or anywhere else. Although other studies support its findings, the Population Research Laboratory should have repeated its questions, on domestic violence in future A l l Alberta studies. There is also concern with the validity of these data because the interviewers in Calgary were left unsupervised,  95 allowing for unidentified biases affecting these data, along with the use of both male and female interviewers which may have influenced respondents or introduced other biases. Some of the questions and measurement tools also contained weaknesses which can be improved in future research. Despite its weaknesses, the Population Research Laboratory provided more valid and reliable rates of incidence and prevalence of wife abuse than were available previously. It also encouraged others to continue to ask questions about violence in intimate relationships. This survey took the first step into domestic violence research in Canada, and in so doing, opened the door for the development of alternative positions of hypotheses for future research (Gray and Guppy 1994:23). The findings should help guide questions in future studies, such as why do high rates of violence exist in commonlaw relationships involving young low income men and women in urban areas. This study encouraged future researchers to advance survey instruments in the study of domestic violence. For example, Statistics Canada created a victimization survey that allowed women to speak about their experiences of violence. It is to be hoped that it will continue to encourage researchers to ask questions about violence of both men and women, and also to analyze data for a fuller understanding of domestic violence that affects so many relationships. Notes: See Chapter Two for a full discussion of the CTS. See Chapter Two for further discussion of face-to-face and telephone interviews. The response rate is simply the number of people interviewed divided by the number of people sampled (Fowler 1988:46). Although error-free and clean sampling frames should yield the 'corrected sample' without adjustments, the perfect sampling frame is rarely found when using public documents that are time-dated (Kinzel 1987:11). 1  2 3  4  96  CHAPTER FOUR: AN ASSESSMENT OF THE STUDY OF MALE-FEMALE DATING RELATIONSHIPS IN CANADIAN UNIVERSITIES AND COLLEGES 4.0 Introduction This study is the first national representative sample survey on the sexual, physical, and psychological victimization of Canadian women in university and college dating relationships (DeKeseredy and Kelly 1993a: 137). Prior to this study, only much smaller non-probability samples were used in Canadian studies of dating violence (see Barnes, Greenwood and Sommer 1991; DeKeseredy 1988a; DeKeseredy, Kelly and Baklid 1992; Elliot, Odynak and Krahn 1992; Finkelman 1992). These studies, as well as domestic violence studies, demonstrate that young people are at the highest risk for violence. Studying violence in both dating and domestic contexts will help us to better understand the dynamics within relationships and better prepare policy-makers for the prevention and treatment of violence. This study provided greatly needed estimates of dating violence against women students. However, it did not adequately address men's experiences although men participated in the study. The survey accomplished many of its objectives, but was also flawed by several methodological drawbacks. In this chapter, the methodological advantages and disadvantages of the survey will be discussed. Assessing this survey requires a set of criteria for proper survey methodology, including specifically the issues: 1. Background and Objectives 2. 3. 4. 5.  The Methodology Development Timing Method of Data Collection  6. Population 7. Sample Frame 8. Sample Size 9. Response Rate 10. Questions 11. Ethical Issues A report of its initial findings and a discussion of the quality of measurement and its theoretical assumptions will follow. Throughout this chapter a number of revisions will be suggested based on the weaknesses highlighted. These revisions will provide a basis for future improvements in research design and methodology.  4.1 Background of the Survey and its Objectives This survey project was sponsored by a grant from Health and Welfare Canada's Family Violence Prevention Division. The study was conducted by Walter DeKeseredy and Katharine Kelly from the Department of Sociology and Anthropology at Carleton University (DeKeseredy and Kelly 1993a: 137). The primary objective of this study was to yield estimates of woman abuse that were representative of undergraduate and community college students across Canada (p. 139). Prior to this study, very few researchers had conducted comparable Canadian studies. These studies provided data on the extent of female victimization in post-secondary school dating relationships. However, the findings were derivedfrom non-probability samples of university and college students in Ontario, New Brunswick, and western Canada (p. 138). The lack of estimates of the incidence and prevalence of abused women across Canadian university and college dating relationships revealed a major research gap.  98 A secondary objective of the survey was to provide a direct test of a feminist perspective on dating violence, which had previously been tested only in a domestic violence context. Scholars such as Lamanna and Reidman (1985) and Laner and Thompson (1982) contend that many men in college and university dating relationships espouse a set of attitudes and beliefs supportive of familial patriarchy similar to men in marital or common law relationships. More specifically, DeKeseredy and Kelly predicted that "males who espouse familial patriarchal attitudes and beliefs are more likely to abuse their female dating partners than those who do not adhere to such an ideology" (1993b:26). Another objective of the survey was to study more closely the connection between dating violence and male peer support. DeKeseredy and Kelly also argued that "men who hold patriarchal attitudes and beliefs and who are supported by their male peers, are most likely to sexually, physically, and psychologically victimize their dating partners" (1993b:26). Woman abuse, in this study, refers to "any intentional physical, sexual, or psychological assault on a female by a male dating partner" (DeKeseredy and Kelly 1993a: 146). The term "abuse" was used instead of "violence" or "battering" in order to present these women as victims of a wide range of assaultive behaviours in a variety of social contexts (Okun 1986:xix). Incidence refers to the percentage of those abused and those who were abusive in the past twelve months, while prevalence is the percentage of those abused and those who were abusive since they left high school (1993a: 138).  99 4.2 The Methodology of the Survey According to Babbie (1989:237), surveys are considered to be the best method available to collect original data describing a population too large to observe directly. Therefore, the survey is the best choice for attempting to provide estimates of the incidence and prevalence of woman abuse in dating relationships across Canadian universities and colleges, especially since this abuse is more likely to occur in privacy. The researchers chose a cross-sectional survey, in which data can be "collected at one point in time from a sample selected to describe some larger population at that time" (Babbie 1973:62). This cross-sectional survey also simultaneously allows for both descriptive and analytical research, gleaning estimates of the prevalence and incidence of woman abuse while testing hypotheses and examining relationships between variables such as peer support, familial patriarchal values, and woman abuse (Henry 1990:48). The only disadvantage of such a survey is that the researchers "cannot establish or measure the temporal order of the independent and dependent variable" (Gray and Guppy 1994:50), thus only describing associations and correlations. To determine both incidence and prevalence of woman abuse in dating relationships, respondents were asked about their experiences in the last 12 months, and then again for the time period since leaving high school including the last 12 months. DeKeseredy and Kelly (1993a) chose to conduct the survey via self-administered questionnaires in a group setting with a research team present. Group administration is often not possible because few survey populations meet in groups, but students are one exception (Gray and Guppy 1994:132). The research team distributed the questionnaires in the classroom for two reasons. First, consistent with Russell (1986), the researchers  100 felt that it was important to be present to offer emotional and informational support (e.g., referral to a woman's centre or rape crisis centre) to any respondents who might be traumatized or upset by the subject matter or the recollection of their past experiences. Additionally, the investigators' presence ensures a higher completion rate and encourages respondents to answer all of the questions (DeKeseredy and Kelly 1993a: 144). Self-administered questionnaires should follow six principles. One guiding principle is the restricted use of close-ended questions which can be answered simply by checking a box or circling the proper response. Close-ended questions also produce useful data, comparable across respondents, and easy to code (Fowler 1988:64). The second principle is that self-administered questionnaires should be self-explanatory as instructions will not be read consistently (Fowler 1988:103). The questionnaires used in this study did include a number of instructions; however, the pretest did not indicate any misreading. Thirdly, the question forms in a self-administered questionnaire should be few in number; if the questionnaire contains similar kinds of tasks and questions to answer, respondents will less likely become confused (Fowler 1988:103). The questionnaires used in this study contained the same kinds of tasks and questions, lessening the chance of confusing the respondents. Fourthly, the questionnaire should be typed and laid out in a way that seems clear and uncluttered (Fowler 1988:103), as demonstrated by both questionnaires in this study. As well, researchers should keep skip patterns to a minimum. Although these questionnaires used skip patterns, no arrows or boxes were used to communicate skips; the respondents were responsible for reading and understanding these questions on their own. Lastly, self-administered questionnaires should contain redundant information to  101 prevent respondents' confusion (Fowler 1988:103), as demonstrated by this study's questionnaires which repeats much of its information between questions, reminding respondents that this survey was about problems in heterosexual dating relationships. If a self-administered questionnaire follows the above principles, the survey tool should be effective and valid. Such questionnaires are also effective because, as mentioned in Chapter Two, respondents do not have to directly admit a socially undesirable characteristic or behavior (Fowler 1988:65), such as physical, sexual or psychological abuse. Researchers, however, must recognize the importance of predetermining their own interaction with the respondents, as there are no interviewers administering the questionnaire (Gray and Guppy 1994:97). DeKeseredy and Kelly created the self-administered questionnaires by following the six guiding principles, thus increasing the effectiveness of their survey tool.  4.2.1 Development of the Survey The development of the questionnaires included the assistance, comments and criticisms of a number of internal and external persons including John Pollard and his colleagues at the Institute for Social Research in Toronto, the Ottawa Regional Coordinating Committee to End Violence Against Women, Holly Johnson of Statistics Canada, Dawn Currie of the University of British Columbia and Brian MacLean of Kwantlen College (DeKeseredy and Kelly 1993a: 137). The researchers devoted "substantial effort" to considering the various measures used by researchers in the field of dating and domestic violence in the past. This allowed for the collection of comparable data, which would add to the generalizability of the research and increased reliability of  102 these data (Fowler 1988:101). DeKeseredy and Kelly (1993a:154) wanted to balance the need to repeat previous research projects with the necessity of avoiding their methodological problems. They selected the best available measures and where necessary, made modifications to address known difficulties. A pilot survey, using the questionnaires, was conducted the previous year with a sample of 285 men and women (Globe and Mail 8 February 1993). This field pretest allowed the researchers to help evaluate and redesign the questions used to ensure the validity of those questions as measurement tools (Fowler and Cannell 1996:126; Carmines and Zeller 1979:17). Both the consultation and pretesting strengthened the survey as an effective research tool.  4.2.2 Timing of the Survey The importance of timing involves noting the start and end date of the administration of the survey. If the time period is too long in a cross-sectional survey design, other variables may affect responses which were unanticipated. Researchers cannot control for unidentified variables (Gray and Guppy 1992:43; Singleton et al. 1988:353-4), thus negatively impacting the validity and reliability of the survey data. The research team in this study administered the questionnaires in the fall semester of 1992, and completed the project before January 1993 (DeKeseredy and Kelly 1993a: 143). There were no other variables identified by the researchers after these data collection relating to any timing effects; however, it is unclear if this was considered by the researchers.  103 4.3.3 Method of Data Collection  As previously mentioned, the method of data collection involved the selfadministration of questionnaires in a group setting with a research team present. There were two questionnaires, one for men and one for women. Prior to each administration, students were asked to participate in a study on problems in male-female dating relationships. These students were explicitly informed about the voluntary and confidential nature of their participation, both by the research team and by the cover page of the questionnaires (DeKeseredy and Kelly 1993a: 144). Each questionnaire contains two measurement tools. One measurement tool is a modified version of the Conflict Tactics Scale (CTS) used to measure psychological and physical abuse. The CTS was modified by adding two new items, employed previously 1  by Statistics Canada in their pretest for a national Canadian telephone study on violence against women (to be discussed in Chapter Five); the modification increased the content validity of the CTS. The two measures are "put her (you) down in front of family" and "accused her (you) of having affairs or flirting with other men." Prior work by Smith (1990b:258) report that these items are also related to physical violence in marital relationships, thus improving the ability of the CTS to specify the full domain of content relevant to physical violence. The modified version of the CTS also includes three questions asking male and female respondents to explain why they engaged in dating violence since they left high school. These questions are modified versions of those developed by Saunders (1988). They are as follows:  104 1. On items...what percentage of these times overall do you estimate that in doing these actions...you were primarily motivated by acting in self-defence, that is protecting yourself from immediate physical harm? 2. On items...what percentage of these times overall do you estimate that in doing these actions...you were trying to fight back in a situation where you were not the first to use these or similar tactics? 3. On items...what percentage of these times overall do you estimate that in doing these actions...you used these actions on your dating partners before they actually attacked you or threatened to attack you? (DeKeseredy and Kelly 1993a: 148). There were several other close- and open-ended supplementary questions included in the questionnaires. These questions were included to respond to criticisms of the CTS that it is a simple count of abuse with no sense of the context, meaning, or motives for being violent (DeKeseredy and Kelly 1993a: 147; see also Breines and Gordon 1983; DeKeseredy and MacLean 1990; Dobash et al. 1992). The CTS also excludes many abusive behaviours such as suffocating, scratching, squeezing, sexual assault, and spanking (DeKeseredy 1994:76; see also Smith 1987:180-81). However, the CTS as a tool for estimating abuse of women has been a useful contribution to a limited Canadian data base. The CTS has also improved upon the accuracy of estimates reported in other Canadian studies such as MacLeod (1980, 1987). The second measurement tool used in this study is a slightly reworded version of the Sexual Experiences Survey (SES), used by Koss, Gidycz and Wisniewski (1987). The SES is employed to operationalize various forms of sexual assault, covering a range of unwanted sexual experiences. These data from this measurement tool allow the researchers to fill in some of the gaps left by the CTS and increase its validity (DeKeseredy 1994:76). The use of the SES allowed DeKeseredy and Kelly to replicate  105 previous work and to compare Canadian results with American data (1993a: 148). The use of both the SES and CTS was supported by the wide use of both tools and their reliability and validity (DeKeseredy and Kelly 1993a: 148; see also Koss and Gidycz 1985; Smith 1987; Straus, Gelles and Steinmetz 1980).  4.3.4. The Population Researchers must identify two populations in any survey: the target population and the survey population. The target population in this study, the collection of elements to be studied (Frankel 1983:24), is all men and women students enroled in a Canadian college or university in the fall of 1992. The survey population is the population that is actually sampled and for which data may be obtained (Frankel 1983:24). The determination of the survey population for this study began with the division of Canada into six regions for the purpose of making regional comparisons. The six regions were as follows: Atlantic Canada, including Newfoundland, Prince Edward Island, Nova Scotia, and New Brunswick; Quebec (French-speaking schools); Ontario; the Prairies, consisting of Manitoba, Saskatchewan, and Alberta; British Columbia; and a language crossover stratum which included both English-language institutions in Quebec and Frenchlanguage schools outside this province (e.g., in Ontario and New Brunswick) (DeKeseredy and Kelly 1993a: 139). The Institute for Social Research prepared a listing of all universities and colleges for each region. Universities with fewer than 500 students and colleges with less than 100 students were excluded (p. 139). Further description of the survey population will be discussed in the next section.  106 4.3.5 The Sample Frame According to Babbie (1973), the "ultimate purpose of surveys is to select a set of elements from a population in such a way that descriptions of those elements (statistics) accurately describe the total population from which they are selected" (p. 83). This can be accomplished through the method of probability sampling, using techniques such as a sample frame. A sample frame is the actual list of units from which the sample is selected (Babbie 1973:81). In this study a sample frame is not necessary for all of the stages in the sample selection. There are a number of alternatives that can be used to create a representative sample without a sampling frame, such as random digit dialing, cluster or multistage sampling, and systematic sampling (Henry 1990:85). For this study, the research team developed a multi-stage, systematic sampling strategy, with the assistance of York University's Institute for Social Research (DeKeseredy and Kelly 1993a: 139). This sampling strategy is used to increase the representativeness of the sample as well as to increase the cost-effectiveness and the efficiency of the research at the same time (Hay 1993:54). Systematic sampling involves the systematic selection of a sample by first determining the size of the sample and the selection interval. Then, after obtaining a listing or physical representation of the survey population, a random start is chosen to select the first member of the sample. Once the random start is determined, then the rest of the sample is selected at intervals previously determined until the list is exhausted (Henry 1990:98). Researchers consider this method of sampling to be superior to simple random sampling because it is more convenient and mechanically easier to draw (Babbie 1973:93; Fowler 1988:23). The one concern with systematic sampling is the ordering of  107 units which may systematically exclude or include a group of units with similar characteristics; this may bias the results and decrease the reliability of those results. However, the researcher can usually remedy this problem quite easily (Babbie 1973:93). DeKeseredy and Kelly (1993a) provide a detailed description of the different stages in their systematic sampling design. As previously mentioned, six regions were designated and the smaller populated universities and colleges were excluded. The number of schools selected in each region was based on the regional distribution of the Canadian student population as documented by Statistics Canada (1992a; 1992b). The schools included were determined by random numbers, and by each school's population relative to the overall regional student population. The sample plan required the selection of 27 universities and 21 community colleges for a total of 48 schools. The researchers randomly picked four schools twice, adjusting the total number of schools chosen to 44 (DeKeseredy and Kelly 1993a: 139). Once the schools were chosen, the sample was also stratified by program of study, in order to ascertain whether disciplines vary in their conduciveness to woman abuse. This part of the sampling procedure required a sample frame, prepared by the Institute of Social Research. The faculties in each school and all of the subjects taught within each faculty were listed respectively. This information for universities was derived from the 1991 Corpus Almanac and Canadian Source Book (Southam Business Information and  Communications Groups 1991:7.10-46). The school calendars provided statistics on the community colleges (DeKeseredy and Kelly 1993a: 143). It is important to remember that in using sampling frames, lists do not truly include all elements (Babbie 1973:91); the use of lists from earlier years will exclude new faculties and classes and include defunct ones.  108 There were no reports that these exclusions were addressed. If at all possible, researchers should be aware of these deficiencies and provide a remedy where possible (Mangione 1995:54). At this point in the sample selection, nonprobability sampling methods, which are selected based on the judgment of the researchers to achieve particular objectives of their research were used (Henry 1990:17). The main program of study or faculty to be selected in each participating school was directly related to the percentage of students enroled in each faculty. The classes were then chosen from that main program of study or faculty, based on statistics by Statistics Canada (1992a; 1992b). This resulted in students enroled in larger faculties, such as Arts, having a greater chance of being selected. This method of sample selection ignores the requirement of a representative sample, in which each unit has an equal chance of being selected. This method, instead, involves a quota, where samples are selected that yield the same proportions as the population proportions on easily identified variables (Henry 1990:18), such as percentage of students enroled in a faculty. Probability sampling was used again once the main program of study was picked. For example, if the researchers chose engineering as the main program of study, all of the subjects taught under this rubric were given random numbers and a particular subject, such as civil engineering, was chosen (DeKeseredy and Kelly 1993a: 143). The researchers anticipated different responses from students who attended university or college for various lengths of time, and thus further divided the sample into junior and senior segments. Two classes were selected from each institution (four from the institutions selected twice), resulting in a total of 96 classes (DeKeseredy and Kelly 1993a: 143). In the universities chosen, the eligible classes had to have enrolments of not  109 less than 35 students. In the colleges chosen, the classes were required to have a minimum of 20 students. This again reflects nonprobability sampling in which the classes were selected on the basis of a required quota and convenience in surveying as many students as possible. In the end, 95 of the projected 96 classes were surveyed (DeKeseredy and Kelly 1993a: 143). The students in each of these classes were selected using nonprobability sampling, not just because of quotas but because of convenience. The students who participated in the study were chosen based on their availability for the study (Henry 1990:18). Therefore, the sample was chosen using both probability and nonprobability sampling methods. Prior to administration, in the summer of 1992, the Institute for Social Research contacted the chairs of the college and university department that had been randomly selected to participate. The initial contact was made by telephone, in which the purpose of the study was outlined, questions were answered and an attempt was made to gain initial approval to administer the survey. Once verbal approval was given, letters were sent to confirm the details of the methodology and to determine the specifics of the intended visit (DeKeseredy and Kelly 1993a: 143).  4.3.6 The Sample Size  According to Henry (1990:117), sample size is the most potent method of achieving precise and reliable estimates that are sufficient for policy decisions or scientific inquiry, In this study, the sample consisted of 3,142 people, including 1,835 women and 1,307 men (DeKeseredy and Kelly 1993a: 145). The median age of the  110 female respondents was 20 and the median for the men was 21. The majority of the respondents were junior students and never married. A large portion of students (42.2% of the women and 26.9% of the men) were enroled in Arts programs (DeKeseredy and Kelly 1993a: 145). One concern with the size of the sample is the overrepresentation of women in the student population, and the lack of justification for this overrepresentation. The selection of the sample was based on both nonprobability and probability methods which may rule out the possibility that the overrepresentation was random. Such overrepresentation brings the reliability of these data into question when making gender comparisons, because these data may not be representative of the actual population. The extent of the impact of this overrepresentation on the study is unknown.  4.3.7 The Response Rate One of the strengths of group-administered surveys is the high rate of response, usually limited only by absenteeism or scheduling problems (Fowler 1988:66). If nonresponse does occur, its effect on survey estimates depends on the percentage not responding and the extent to which those not responding are biased or, in other words, systematically different from the whole population. If the response rate is high, sample estimates will be very good even if the nonrespondents are distinctive (Fowler 1988:48). According to DeKeseredy and Kelly (1993b), the response rate in this study was very high with fewer than one percent of the participants refusing to answer (p. 29). This indicates that the sample estimates should be very good; however, the actual response rate is unknown. With the sampling procedure used, it is difficult to clearly establish the number of eligible responding units selected (Hay 1993:58).  Ill  In this study, there was the possibility of nonresponse in all stages of the sample selection. Institutions could refuse to participate, as well as departments, instructors and students, each affecting the overall sample size and representativeness. For example, one of the 48 schools originally selected chose not to participate because the institution did not have a policy on research involving human subjects. Another school originally selected was simply not agreeable to the study for reasons unknown (DeKeseredy and Kelly 1993:142). Nonresponse in this study could have been affected by the professors who obtained their students' consent before allowing the research team to survey their classes. Although all professors should have obtained their students' consent, those students who were aware may not have come to class. These students may have had a number of reasons for not attending class, including their refusal to participate in the study or simply their lack of desire to attend class, as admitted by DeKeseredy and Kelly (1993a:144). The precise number of students who did not participate because of this influence is not known (p. 144). However, the rate of nonresponse could have been more precisely determined by obtaining class lists for each of the selected classes (Hay 1993:58). One suggestion to the researchers to help combat such influence would be to offer an incentive such as a cash reward or prize draw for completion of the survey in order to encourage the students to attend class. Potential nonresponse could also be avoided in student samples by considering when classes are scheduled, noting both the time of day and the day of the week. For example, Monday mornings and Friday afternoons often have lower attendance rates compared to other times. Thus, in order to reduce nonresponse due to lack of availability,  112 it is best to avoid such times. It is also very useful to determine at the end of an interview which questions were considered threatening or hard to understand, and for which respondents. Respondents encountering such questions are more likely to underreport than other respondents (Sudman and Bradburn 1982:68).  4.3.8 The Questions As previously mentioned, the majority of the questions in both questionnaires were close-ended, providing the respondents with a set of responses from which to choose. The advantage of close-ended questions lies in the fact that respondents can consider alternatives and give uniform responses, while not having much demand placed on them. However, close-ended questions can provide inadequate response categories and encourage superficial responses (Gray and Guppy 1994:84). To help counter the disadvantages of close-ended questions, there were a few open-ended questions. These open-ended questions allow respondents to use their own words to answer the questions (Fowler 1988:87). For example, the last question in the women's questionnaire read as follows: 27. We realize that the topics covered in this survey are sensitive and that many women are reluctant to talk about their own dating experiences. But we're also a bit worried that we haven't asked the right questions. So now that you have had a chance to think about the topics, have you had any (any other) experiences in which you were physically and /or sexually harmed by your dating partners and/or boyfriends while you attended college or university? This was also the second to last question in the questionnaire administered to male respondents, except that it asked the male respondents about any other experiences in  113 which they had physically and/or sexually harmed their dating partners and/or girlfriends. The men's final question was as follows: 29. Again, now that you have had a chance to think about the topics, have you had any (any other) experiences in which you were physically and/or sexually harmed by your dating partners and /or girlfriends while you attended college or university? Unfortunately this last question is the only opportunity for the male respondents to speak of their own experiences as victim; the majority of the questions asked the men about their abusive actions only. These three questions allowed the respondents to fill in any gaps left open by the researchers, such as unanticipated answers, or a fuller description of the real views of the respondent (Fowler 1988:87). However, the researchers need to be aware of the disadvantages associated with the use of open-ended questions, such as vagueness, lack of comparability, and the increased difficulty in coding and summarizing (Gray and Guppy 1994:82-3). The extent to which this problem plagued these data is unknown because there have been no reports of these data analyses from these questions. The two questionnaires contained some identical items but the wording was changed to ensure that the proper gender was identified as the dating partner (DeKeseredy and Kelly 1993a: 144). The identical items include questions about the respondents' dating history, biographical information, involvement in fraternities and sororities, violent actions toward their dating partner and their reasons for using such actions, use of drugs and alcohol, and beliefs and attitudes about familial patriarchy and dating violence. There are also some different items used in each questionnaire. For example, in the male questionnaire, Question #13 read as follows: "The next questions are about your participation in various recreational, school, or work activities with other men. How  114 many times in a typical month in the past year have you engaged in each activity with other men? Only include those events which were all-male." Another question only in the male questionnaire asks about involvement in social clubs or community organizations that only allow male members. These questions were important for the researchers' interest in the connection between peer support and woman abuse. However, these questions are not asked of the female respondents, although such data would have been effective in determining if women's activities and peer groups render her more susceptible to being a victim of abuse. The researchers instead ask women, in Question #15, about how safe they do, or would, feel alone in public places after dark, such as walking, riding a bus or subway, walking to a car in a parking lot, waiting for public transportation, walking past unknown men and being at home. There are a number of concerns that need to be addressed in regard to this question. For example, this question is not related to dating relationships. However, although this question is not directly related to dating relationships, it does provide data to test a hypothesis that physical, sexual and psychological abuse contribute to a "generalized fear of crime" (Kelly and DeKeseredy 1994:17). Other concerns include a couple of problematic omissions, such as questions of a woman's perception of safety during the day. If the researchers are to ask questions about fear and safety, then the questions should be comprehensive, including perceptions of safety both before and after dark. The questions should also have included perceptions of safety while with friends, family and acquaintances, while not just being at home alone after dark in order to increase the content validity of these questions. It is also interesting that "being at home alone after dark" is included in this question when it is not a "public  115 place" as stated in the introduction. However, it is important in determining the connection between a woman's fear of being at home alone and her victimization. Statistics continue to show that a woman has a greater chance of being abused by someone she knows than a stranger (Brinkerhoff and Lupri 1988; DeKeseredy 1988a; Kennedy and Dutton 1989; Lupri 1990; Smith 1987, 1988; Stets and Henderson 1991; Straus and Gelles 1986; and Straus, Gelles and Steinmetz 1980) and yet survey questions such as this one continue to ignore this area. Another problem with the questions about fear and safety involves the combination of how a female respondent would feel in these situations with how she does feel; there may be great discrepancy between how a woman does feel and how a woman thinks she might feel, decreasing the criterion-related validity of these questions. One suggestion is to ask women if they have ever participated in the above actions and then ask about their perceptions of safety, as in Statistics Canada's Violence Against Women Survey (1993a). There are studies noting the differences between women's perceptions of their personal safety, their actual experiences with safety, and the experiences of others (Hindelang, Gottfredson and Garofalo 1978; Maxfield 1984; Skogan and Maxfield 1981; Research and Statistics Group 1984), giving this concern credibility. Explanations of these connections consider fear to be related to assumptions about vulnerability or to being socialized to be fearful. Fear may also be related to the undercounting of women's victimization by crime surveys because a variety of abusive male behaviours are ignored (Kelly and DeKeseredy 1994:18). The male respondents are not asked about their own perception of safety, although such data would allow for interesting gender comparisons.  116 Other dissimilar items included in the questionnaires ask the male respondents about the information current male friends may have given them concerning how to deal with problems in male-female dating relationships, current male friends' dating relationships and their use of physical, sexual and psychological abuse, and their perceptions of their male friends' approval of slapping their girlfriends. The questionnaire for female respondents asks women about their most recent incident of physical, sexual or emotional hurt by a boyfriend or dating partner to discover what sources of support are used by these women and their satisfaction with that support. This questionnaire also asked women if they had ever been raped, sexually assaulted, physically assaulted or sexually harassed by someone other than a dating partner. The questionnaires also include similar items such as a version of the CTS. The modified scale is "tailored to elicit women's reports of their victimization and the other [is] designed to elicit men's accounts of their abusive behaviour" (DeKeseredy and Kelly 1993a: 147). Furthermore, the SES is tailored to ask men of their perpetrated abuse and women of their victimization. Other, questions are similarly designed to obtain data of physical, sexual and psychological abuse in dating relationships in elementary and high school. Another example includes a question about the upsetting of a woman by dating partners or boyfriends trying to get her to do what they had seen in pornographic pictures, movies or books. Due to the sensitive nature of the questions in both questionnaires, a "proper context" was established to encourage respondents, particularly the male respondents, to answer the questions truthfully. Such a context makes it easier for respondents to admit that they have engaged in a behaviour they see as undesirable (Sudman and Bradburn  117 1982:78). This context is provided in the introduction to the CTS questions for both the men and women: No matter how well a dating couple gets along, there are times when they disagree, get annoyed with the other person, or just have spats or fights because they're in a bad mood or tired or for some other reason. They also use may different ways to settle their differences. Such an introduction establishes a context in which abuse in dating relationships might be seen as acceptable or normal, demonstrating sensitivity when the questions required sensitivity. The questions from the SES are introduced to both the men and the women with the single statement that "sexual relations are common in dating relationships." No context is provided for these question items in the introduction, but was included in the wording of each question.  4.4 Ethical Issues As the primary investigators, DeKeseredy and Kelly faced the ethical dilemma of "weighing the benefits of more reliable crime statistics... against the risks of rekindling personal anguish for respondents" (Gray and Guppy 1994:25) by asking questions about experiences as victim or perpetrator of spousal violence. Prior to each administration of the survey, the students were asked for their voluntary and confidential participation. Although the professors in each class consented to their students' participation, not all of the students were asked for consent prior to the arrival of the investigators. Following each administration, the respondents were debriefed with a discussion of the reasons for the research. The respondents also received the existing information on the frequency and severity of dating violence, and the role that peers play in the process. Each  118 participant was provided with a list of local (on- and off-campus) support services for survivors and abusers. The respondents were also encouraged to ask the research team any questions or to discuss the survey with them after completion (DeKeseredy and Kelly 1993a: 144). The debriefing techniques are similar to those used in Koss, Gidycz and Wisniewski's (1987) national sexual assault study. These researchers of dating violence have accepted the responsibility for the consequences of asking people related questions.  4.5 The Findings DeKeseredy and Kelly's initial release of these data reported that men are more likely than women to report having engaged in less lethal forms of abuse, and women are more likely than men to report having been victimized by such behaviour (DeKeseredy and Kelly 1993a: 137). The incidence and prevalence of abuse is presented in Table 2. Also, every type of physical violence was used by at least one respondent, and that less lethal forms of assault were reported more often than more lethal forms (DeKeseredy and Kelly 1993a: 152). The findings also reveal that the global incidence rate of the range from unwanted sexual contact, to attempted rape and rape for female victims was 27.8%. Approximately 11 % of the male respondents report having victimized a female dating partner in this way in the past year. The prevalence figures are considerably higher, with 45.1% of the women stating that they had been victimized since leaving high school, and 19.5% of the men reporting at least one abusive incident in the same time period  119  T A B L E 2. Incidence and Prevalence of Sexual, Physical and Psychological Abuse  :  Yes No Missing  Yes No Missing  Sexual Abuse (Incidence) 27.4% (N=435) 72.6% (N=1150) (N=235)  Physical Abuse (Incidence) 22.2% (N=354) 77.8% (N=1243) (N=223)  Psychological Abuse (Incidence) 79.3% (N=1264) 20.7% (N=331) (N=225)  Sexual Abuse (Prevalence) 44.5% (N=725) 55.5% (N=904) (N=191)  Physical Abuse (Prevalence) 34.5% (N=567) 65.6% (N=1082) (N=171)  Psychological Abuse (Prevalence) 86.2% (N=1418) 13.8% (N=227) (N=175)  Note: The survey oversampled some regions of the country and the results were weighed to yield these percentages.  (Source: Kelly and DeKeseredy 1994:24)  (DeKeseredy and Kelly 1993a: 148). The men in the study were more likely to report using less severe forms of coercion to get women to engage in sexual activities, including arguments and pressure, and the use of alcohol. The women's reports of the types of coercion used to engage in sexual activities concur with the male responses (p. 151). The findings also reveal that similar accounts of psychological abuse were provided by both men and women. The proportion of men who reported having been psychologically abusive was 74.1%; 79.1% of the female respondents indicated having been a victim of such mistreatment (DeKeseredy and Kelly 1993a: 153). Such congruency in reporting suggests that there is a perception on the part of abusers that these occurrences are part of the "common currency" of dating relationships (DeKeseredy and Kelly 1993a: 154).  120 Many of the findings reveal large gender differences in reporting the incidence of abuse. The reporting gaps widen for the prevalence data. Many researchers argue that socially desirable reporting is more common among perpetrators than victims (Arias and Beach 1987:140; Dutton and Hemphill 1992:29). For example, the findings indicate that women are seven to eight times more likely to report sex play, attempted intercourse, and sexual intercourse involving some degree of force than men when response differences were standardized using women's figures as the base. DeKeseredy and Kelly suggest that social desirability was probably shaping men's responses, but that the response differences on four other items were also large. The four other items included giving in to sex play, or to sexual intercourse, due to continual arguments and pressure, and attempted sexual intercourse, or actual sexual intercourse, when a woman was too drunk or high to resist. Women were 6 to 6.5 times more likely to have reported these actions, which focus on the negotiations between men and women over sexual activity. These differences in reporting rates suggest considerable miscommunication between men and women, not just the influence of social desirability in reporting differences (DeKeseredy and Kelly 1993a: 151). The influence of social desirability would have been decreased through the confidential and anonymous nature of the study and also through the use of a self-administered questionnaire (Fowler 1993:90). Researchers need to study the definitions of these activities for men and women to improve measures for these activities. Social desirability and miscommunication play a role in differences in reporting. However, the differences in reporting may also be attributed to the lack of dating relationships between the male and female respondents. For example, the men reported  121 physical abuse at the incidence rate of 13.7%, whereas the women reported a rate of 22.3%. Almost 35% of the women reported having been physically assaulted and 17.8% of the men stated ever having used physical abuse since leaving high school (DeKeseredy and Kelly 1993a: 152). The researchers need to consider that these women may have dated men outside of the generalizable sample, such as men who joined the work force right after high school, or men who were much older; the universities and colleges are not closed systems in which the majority of the dating partners are other university or college students (Hay 1993:62). This is demonstrated by the results which show that men were more likely to report having used a weapon than women were to state having been victimized by this form of abuse. Also, more men reported threatening a date with a weapon than women reported being threatened (DeKeseredy and Kelly 1993a: 153). To refer to the differences in reporting as a gender gap is not justified until couples or ex-couples have been included. No checks were made for such inclusions; future research must determine if the respondent dates students or others outside of colleges and universities (Hay 1993:63). Therefore, these data must be accepted as rates of incidence and prevalence only; these data do not allow gender comparisons because the sample population consisted of men and women whose reports revealed, not a gender gap, but a nonrepresentative sample of a dating population. DeKeseredy and Kelly produced results comparable to past studies in the United States, which helps fulfill a major gap in dating violence research. Future research should produce comparable studies to cohorts in the workforce to identify if rates of dating violence are higher in university and college settings or in the workforce. Comparisons  122 should also be made between students who live in residences on campus, students who live alone or with friends off campus, and students who live with family. For example, the lack of parental supervision, extended periods of time spent alone with dating partners and the constant presence of peer pressure may be reflected in higher rates of dating violence in populations found in campus residences. The associations with different groups could be revealing in the types and frequency of victimization and perpetration by women and men. In every study there will always be concerns about the findings, whether it involves which data are released or not released, which data are gathered, or who funded the study. It is well known that if a project is funded by government sources or university sources, there is an obligation to produce results. The implications include the exclusion of certain areas of study, which may not be publicly acknowledged as a problem, and the reporting of only certain findings, particularly the positive ones (Denzin 1989:249). DeKeseredy and Kelly excluded areas such as the psychological abuse of men by their dating partners or girlfriends, which is a problem as evidenced by other research. But, in further discussions, Kelly (1994) reveals how these data were prematurely released to the media, tending to distort the integrity and credibility of the results of this particular study (p. 83). However, this can only be truly assessed after a discussion of the quality of measurement.  4.6 The Quality of Measurement  Throughout the previous discussions, a number of strengths regarding the methodology have been highlighted. In this next section, a number of general concerns  123  will be raised, as well as specific concerns in regard to the measurement tools. One such general issue is the correspondence between the introduction of the survey and the actual questions, bringing into question the criterion-related validity of the survey as an effective measurement tool (Carmines and Zeller 1979:17). For example, the survey is introduced to the respondents in the context of problems in heterosexual dating relationships. However, male respondents are asked about their own abusive actions and their friends' support of such behaviour, not about their own experiences or perceptions of the relationships. This must be viewed as a problem because such questions can place men on the defensive, affecting the reliability of these data and validity of reports. If questions are demeaning, embarrassing, or upsetting, respondents may terminate the interview or falsify their answers (Sudman and Bradburn 1982:5). For the female respondents, many of the questions relate to experiences outside of their dating relationships, such as their perception of personal safety in public places after dark. These questions appear to be inconsistent with the purpose of the study. Also, for both male and female respondents, the questions are limited to abuse and violence. Problems in dating relationships are affected on other levels, such as intimacy and level of commitment. Another concern with the quality of measurement involves the use of two different questionnaires for men and women. The use of both male and female respondents is encouraged in surveys of male-female dating relationships, particularly since they are both involved. However, it is disappointing to find that the questionnaires were, to a great extent, different. One problem with this difference is that the questions which were identical were not asked in the same order; the identical items were interspersed with other questions not asked of each respondent. Research on order effects  124 reveal how people's responses to one item in a survey are consciously or unconsciously influenced by their responses to a previous item (Fowler 1993:6; Mangione 1995:32; Sigelman 1989:144). If the questions had been asked in a different order, their responses might be different. There is also the assumption that the context of each question must be identical if its meaning is to be identical for each respondent and, since all preceding questions are part of the context, the sequence of the questions must also be identical (Richardson, Dohrenwend and Klein 1965:43). In this study, the men and women were asked identical questions in a different order, which may have affected the responses. There were no sets of checks and balances in place to deal with this effect, therefore, jeopardizing the comparability of results from the two questionnaires and the reliability of these data. Another problem with the use of two different questionnaires is that gaps were left in these data and subsequently in the research project as a whole. For example, the last question for the male respondents asked if the respondent had "any (any other) experiences in which you were physically and/or sexually harmed by your dating partners and /or girlfriends while you attended college or university?" The researchers did not ask the male respondents about their experiences as victims of psychological abuse. The women were also not asked about their psychologically abusive actions toward their dating partners and/or boyfriends. However, the women were asked about their experience as victims of psychological abuse and the men were asked of their perpetration of such abuse. The researchers do not provide justification for this omission; either a justification for the omission or the questions themselves should have been included. This demonstrates a lack of sensitivity where sensitivity is required, and may  125 decrease the reliability of these data provided by these questions. Unfortunately, reports analyzing these data provided by these questions have not yet been released. Another concern with the quality of measurement involves the gender of the researchers, who administered the questionnaires and whether or not there were any biases due to their gender. The research team consisted of the principal investigators, Walter DeKeseredy and Katherine Kelly. According to Denzin (1989), interviews are "gendered, interactional productions" affecting these data constrained by the "gendered identities that are enacted in the interview encounter. Gender filters knowledge" (p. 116). The gender of the research team may have affected the responses given by the men and women who participated in the survey, and decreased the reliability of these data. For example, DeKeseredy's presence may have discouraged female respondents from answering the sensitive questions truthfully, or Kelly's presence may have placed male respondents on the defensive, also affecting the accuracy of their reports. Although the effect of the gender of the interviewers was never determined, their presence was probably beneficial in increasing the overall response rate (Hay 1993:61; see also DeKeseredy and Kelly 1992:9). The quality of measurement could also have been affected by the lack of definitions and vague wording used in the questions. For example, one question in the female questionnaire (#24) asks if the woman has ever been sexually harassed. No explanation of sexual harassment is given nor do the researchers attempt to determine what sexual harassment means to the respondents. The information gathered from this question is not very useful when the above is considered. Another example is found in Question #18 in the female questionnaire, which refers to emotional hurt. This term is  126 also very subjective, with no definition provided or checks to determine what emotional hurt means to the respondents. The researchers may have elicited emotional hurt that may not be the equivalent of psychological abuse; a distinction between emotional hurt and psychological abuse can be made. For example, a girlfriend may feel emotionally hurt when her boyfriend chooses to spend an evening with his friends instead of with her, but this should not be considered psychological abuse. The researchers must define terms, such as "emotional hurt," in order for these data to be useful. Without definitions, each respondent will approach these terms from "a unique perspective that reflects past experiences, personal idiosyncrasies, and current mood" (Denzin 1989:236). Because these terms can hold significantly different meanings for different respondents, "it is not surprising that different words designed to tap the same object or feeling state can actually serve as significantly different stimuli and trigger different response patterns" (Smith 1989:100). Ensuring that all respondents have the same understanding of what is to be reported is one of the most difficult tasks for researchers, and failure to accomplish this properly is a major source of error in survey research (Fowler 1995:13). Another concern in determining the quality of measurement involves the potential for memory biases, whether this results in underreporting, overreporting or changes in perceptions of past events. The information obtained in this survey consists of reports about the respondents' behaviours and experiences or more specifically, the respondents' beliefs about those behaviours and experiences (Dillman 1978:83). Detailed retrospective questions about past events are extremely difficult to answer because most people tend to recall only general information (Gray and Guppy 1994:89; see also Bachman and O'Malley 1981; Cannell, Marquis and Laurent 1977). Over time, values  127 and norms often affect how respondents choose to recall past events; details are often forgotten or distorted (Gray and Guppy 1994:89). A number of studies have provided evidence that respondents are often unreliable and inaccurate in answering direct and objective survey questions (Bickart and Felcher 1996:115; see also Bradburn 1983; Burton and Blair 1991; Felcher and Calder 1991; Means and Loftus 1991; Schuman and Presser 1981; Sudman and Bradburn 1974; Thompson and Mingay 1991; Tournageau and Rasinski 1988). The event must be important and salient in order for the respondent to remember; it may be easier for victims to remember an event than it is for the perpetrators (Cannell, Marquis and Laurent 1977; Fowler 1988:66; Sudman and Bradburn 1982:42). One way to avoid such memory bias is to not ask respondents to summarize past events, but rather to ask them to recall actual events or episodes (Gray and Guppy 1994:90). Although the questions in both questionnaires ask the respondents for summaries of past events, the female questionnaire asked the women to recall the most recent incident of emotional, physical or sexual abuse (Question #18). However, the purpose of this question was only to determine the women's assessment of the support she received from various people and groups in disclosing that most recent incident. Despite these concerns, questions about past events are important for survey projects such as this one, although researchers must make every effort to decrease the ill effects. The quality of measurement is also determined by the use of valid tools. DeKeseredy and Kelly (1993a) define woman abuse as "any intentional physical, sexual, or psychological assault on a female by a male dating partner" (p. 146). Their use of the CTS and SES has been criticized by scholars such as Gartner (1993). She argues that DeKeseredy and Kelly have used both instruments to simply measure "woman abuse,"  128 when they are designed to measure conflict tactics typically used in relationships (in the case of the CTS) and coercive sexual contacts (in the case of the SES). Gartner believes that DeKeseredy and Kelly hold the assumption that "any one of the items from either instrument is a valid measure of 'abuse,' and that each also measures an instance of 'assault' and 'victimization.'" This assumption is reflected in their conclusion that 86% of the women have been psychologically abused, and 45% have been sexually abused since leaving high school (Gartner 1993:315). Many researchers have criticized the CTS for its inability to determine the "meaning of the acts to the persons involved, the sequence of the acts in a wider pattern, and the context of the acts in a particular relationship" (Gartner 1993:315). DeKeseredy and Kelly anticipated this criticism of their study, using a set of questions discussed below. However, there is still concern that a respondent will admit to being a victim of certain actions, but may not describe those actions as abusive herself, decreasing the criterion-related validity of these questions as measurement tools (Carmines and Zeller 1979:17). For example, spiteful statements, insults, or swearing will not be considered abuse by all women, but the researchers still labeled it as abuse. DeKeseredy (1994) argued that the inclusion of these items is based on qualitative feminist research and interviews with students, activists, friends and family members. The research and interviews reveal the importance of including potentially psychologically abusive items, which may be more injurious than physical assaults (p. 77). Gartner (1993) and Fox (1993) believe that DeKeseredy and Kelly violate a fundamental principle of feminist research, which is listening to women's subjective interpretation of their experiences by not determining the respondents' perceptions of their experiences with violence and  129 abuse. However, the CTS encourages researchers to create other measurement tools to determine context, meaning and motives for the abuse to increase the construct validity of the CTS by relating it to these other measurement tools (Carmines and Zeller 1979:23). As Kelly (1994) agrees, it is "virtually impossible to understand the complexities of human interactions using survey instruments," and that requires in-depth face-to-face interviews with dating couples are required (p. 81). Although the CTS is flawed, it is better than no measurement at all in some situations; researchers need to recognize when these measurements are flawed and report their findings accordingly (DeVellis 1991:10). The three questions asking male and female respondents to explain why they engaged in dating violence since they left high school are an example of tools used to determine context, meaning, and motives for abuse. These questions are modified versions of those developed by Saunders (1988). They are as follows: 1. On items...what percentage of these times overall do you estimate that in doing these actions...you were primarily motivated by acting in self-defence, that is protecting yourself from immediate physical harm? 2. you were trying to fight back in a situation where you were not the first to use these or similar tactics? 3. you used these actions on your dating partners before they actually attacked you or threatened to attack you? (DeKeseredy and Kelly 1993a: 148). Following each question is a line with 0% and 100% at either end and percentage points spaced equally and numbered by tens. Although the question is evidence of good intention, it requires too much thinking on the part of the respondent: the responses are many in number, difficult to choose and too time consuming to answer accurately (Mangione 1995:12). The researchers have not yet released analyses of these questions.  130 The use of the CTS requires the asking of questions in a variety of ways and at different times throughout the investigation, in order to improve the reliability of measurement and overcome random errors (Mangione 1995:29). Researchers have discovered that respondents make many reports of violence while answering seemingly unrelated questions. Respondents also make reports of violence toward the end of interviews. In order to determine accurate incidence and prevalence rates of woman abuse, surveys must use "carefully crafted measures designed to create more than one opportunity for respondents to disclose experiences related to abusive incidents" (Hippensteele, Chesney-Lind and Veniegas 1993:69). DeKeseredy and Kelly provide their respondents with numerous opportunities to disclose events, using the CTS, the SES, as well as other questions. These questions allow the respondent to qualify responses and to express exact shades of meaning, rather than forcing him or her to choose from a number of possibly threatening alternatives (Smith 1994:11). The other measurement tool in this study is the SES. Researchers have generally accepted the SES as a valid psychometric index of unwanted sexual experiences, except that is it generally used in conjunction with an interview. There is potential that, on its own, the SES may not be the most accurate measure of recent sexual experiences, since victims may tend to disclose less-recent incidents more readily because they are not as painful to remember (Hippensteele, Chesney-Lind and Veniegas 1993:69). There is no evidence proving its validity in this study nor its inaccuracy; however, it is the only measurement tool available to measure sexual experiences at this time.  131 4.7 The Theoretical Assumptions Before any research project begins, one or more theories lay the groundwork for the study. These theories set out the problems for the research project, staking out new objects for examination and directing empirical inquiry (Blumer 1954:3). The research project provides the fundamental test of those theories and their verification. Through the use of research methods new observations are brought forth to modify, verify, and change the theory under examination (Denzin 1989:64-5). There are four theories or perspectives that provide the framework for this study, which were discussed in Chapter One. The first theoretical perspective is the feminist perspective. These theories explain, through studies such as Dobash and Dobash (1979), that men who physically assault their spouses do so because their partners have violated, or are perceived as violating, the ideals of familial patriarchy (DeKeseredy and Kelly 1993a: 138). DeKeseredy's study (1988b) also reported similar results. The feminist perspective highlights relevant themes such as an insistence upon women's obedience, respect, loyalty, dependency, sexual access, and sexual fidelity (DeKeseredy and Kelly 1993a: 138). Such themes are evident in the questions about the beliefs and attitudes held by men and women. For example, respondents were asked about their beliefs regarding who has control over the family income, the sharing of household chores, and the man's right to decide if his wife should go out in the evening with her friends. They were also asked about their beliefs regarding the man as the head of the house, rape victims being partly to blame, the man's right to have sex whenever he wants and the man's right to hit his wife if he has lost his temper. Another example involves the question asking about the respondent's attitudes regarding when he or she would approve of a man hitting his  132 dating partner or girlfriend. The testing of this hypothesis reveals that "males who report abusing their dating partners are more likely to espouse the ideology of familial patriarchy than those who do not report abusive behaviour" (DeKeseredy and Kelly 1993b:45). However, this finding also reveals that woman abuse is more likely to occur when there is a lack of consensus about familial patriarchal values, not just with the existence of those values (p. 27-28). Another theory underlying the assumptions of this study is the social learning theory. Social learning theory explains that human behaviour is learned through modeling (O'Leary 1988a:33), and that behaviour is supported by other's attitudes reinforcing that behaviour. In relation to violence against women, "when attitudes of an individual are shaped that support the use of physical force against a woman, physical aggression is much more likely" (O'Leary 1993:24). Women who reject or fail to live up to the ideals of familial or courtship patriarchy are also regarded as "appropriate targets for physical, sexual, and psychological abuse by some of the male friends of these men" (DeKeseredy 1994:78). Both questionnaires include the above questions of beliefs and attitudes, but the male questionnaire also includes a question about their friends' beliefs and attitudes. These data from these questions allow the researchers to determine if male peer support contributes to violence against women. The testing of this theory reveals that "male peer support does contribute to the relationship between patriarchal ideology and female victimization in university and college courtship" (DeKeseredy and Kelly 1993b:45).  3  The researchers also considered theories of alcohol and substance abuse. Although these theories have not demonstrated that alcohol and drugs cause violence, the  133 theories have pointed to a correlation between alcohol and drugs and violence (O'Farrell and Murphy 1995; Frankel-Howard 1989). In this study, both questionnaires include questions about the respondents' use of alcohol and drugs in the last 12 months, and whether it occurred with their girlfriend, boyfriend or dating partner. Although no connection can be made to violence and use of alcohol and drugs at the same time, it provides evidence of one being more likely to be violent or be victimized if alcohol and/or drugs are used. DeKeseredy and Kelly (1993b) identified the number of times a man and his dating partner were at parties where alcohol was being consumed as a situational variable, which may help explain variations in the types of woman abuse examined in this study (p. 47). This study also contributes to victimology research, which has helped make victims of dating and domestic violence more visible (Rock 1994). The contribution helps identify the impact of crime on women and their fear of crime (Viano 1990:3). The researchers contributed to this identification by asking female respondents about their perceptions of safety in public places and in the home after dark. These data allowed the researchers to test a hypothesis about a woman's victimization and her perception of safety in public places after dark. Although this hypothesis was tested and not supported by these data, it did lead to the creation and examination of another hypothesis assessing the relationship between fear in private places such as the home and abuse by male dating partners. The re-examination of these data revealed that women victimized by sexual and psychological abuse by a male dating partner are more fearful in the home than nonvictims (Kelly and DeKeseredy 1994:26).  134 4.8 Conclusion  DeKeseredy and Kelly have provided the field of dating violence research with more accurate estimates of the incidence and prevalence of sexual, psychological and physical victimization of Canadian women in universities and colleges. These data also allowed for the testing of a number of hypotheses based on a number of different theories, and for the consideration of similarities between dating and domestic violence. These data also allowed for the exploration of the extent of dating violence from a broader international perspective, previously confined to the United States (Hippensteele, Chesney-Lind and Veniegas 1993:67). These contributions have provided a better understanding of the dynamics within violent relationships. During the development of the survey, DeKeseredy and Kelly consulted with a number of groups in order to improve on the various measures previously used in domestic violence research. The questionnaires contained a modified version of the CTS, including other measures of abuse such as the SES, to address criticisms about the lack of context and motive in these data produced by the CTS. The researchers also strengthened the study by considering the ethical issues faced in asking respondents about their victimization or perpetration of violence. The study was also strengthened by the number of underlying theoretical assumptions, which encouraged a wider range of questions. DeKeseredy and Kelly have faced a number of criticisms for their study on malefemale dating relationships in Canadian universities and colleges. Some of the criticisms are valid, while others are not. The valid criticisms allow researchers to develop alternative positions or hypotheses for future research (Gray and Guppy 1994:23). For example, this study encourages future research to empirically discern the major "risk  135 markers" (Hotaling and Sugarman 1986) associated with assaults on female university/college students, such as level of intimacy, ethnicity, and educational status (DeKeseredy and Kelly 1993a: 155). Other risk markers to be discerned include the amount of time spent together, level of commitment, and differences between the activities of students living in residences and those living off campus, either with family, friends or alone. Past history of violence, parental familial patriarchal values and norms, the structural fact of male dominance with the family, absence of deterrence and a closer look at memberships in social groups such as fraternities are also possible factors for researchers to consider in future research (1993b:47-8). Such information will help identify those men and women who are at the greatest risk of being abused or being abusive, while also assisting in theory development (DeKeseredy and Kelly 1993a: 155). There were also a number of other weaknesses identified, such as the concern with the overrepresentation of women in the sample, the problem of nonresponse in the multistage systematic sampling design, and omissions in the questions. The quality of the measurement tools was also identified as problematic in such areas as the correspondence between the introduction of the survey and the actual questions, the use of two different questionnaires for the men and women respondents, interviewer effects, a lack of definitions and vague wording used in the questions and the concern about reporting biases. Criticisms, however, do not necessarily destroy the integrity and credibility of the study. Sometimes they are only evidence of the necessity of making choices between the numerous possibilities. For example, Kelly (1994) discusses the importance of doing indepth interviews with dating couples (p. 81). However, that would take a lot of time and  136 money. In order to make the best of the time and money they did have, DeKeseredy and Kelly opted for a survey. This method allowed them to gather data representative of the Canadian reality for young women in universities and colleges. Although DeKeseredy and Kelly have been widely criticized for their use of the CTS, it was the only measurement tool available. The CTS also helps minimize researcher bias by allowing for the replication of research findings (Gray and Guppy 1994:23). However, DeKeseredy and Kelly (1993a: 148) admit that their findings of incidence and prevalence "cannot be compared to any other studies because of methodological differences. For example, Finkelman (1992) used the same measures and time period, but he does not provide data on gender variations in victimization. The researchers admit that such problems in interpretation make comparisons meaningless (DeKeseredy and Kelly 1993a: 151). This calls for the creation and use of different measures in future research in order to obtain comparable data. Particularly in the case of dating and domestic violence, there is no clear agreement on what constitutes such violence and how to measure it. It is, therefore, important to measure it using several different tools and methods and to also ensure that researchers measure all of its different dimensions (Babbie 1989:126). For example, researchers require different tools to determine the context of abuse, and to determine the respondents' views about relationships, sex, and boundaries. Although I have raised several criticisms and concerns, it is important to note that researchers are not always able to provide sufficient details about the actual planning and conduct of research to adequately address all of the methodological questions or criticisms that might arise (Hay 1993:54). While there are certain methodological  137 difficulties associated with this study, these are problems which any survey might encounter (p. 53). Although perfect surveys are not possible, striving for perfection should always be the goal of the researcher. Another study which encountered similar methodological problems but also made numerous improvements over past surveys, is Statistics Canada's survey on violence against women, which will be discussed in the following chapter. Notes: See Chapter Two for a discussion of the CTS. For a fuller description of the stages of the systematic sampling design, see Pollard (1993). However, the researchers admitted that some "apparently anomalous finding" in which peers espousing familial patriarchal attitudes actually predict lower values for sexual, physical and psychological abuse of women, is due to ambiguity in the way in which the questions were formulated in this index (DeKeseredy and Kelly 1993b:47). There is no discussion of how this affects their conclusions. 1  2  3  138 CHAPTER FIVE: AN ASSESSMENT OF THE STATISTICS CANADA VIOLENCE AGAINST WOMEN SURVEY  5.0 Introduction The Violence Against Women Survey (1993) by Statistics Canada was selected for evaluation because it is the largest, most recent national victimization survey of women in Canada. It is also the first survey in the world to focus on women's victimization by men at the national level. Many have credited Statistics Canada for conducting the survey which provides evidence of the need to support shelters and programs for battered women. The survey also made a number of improvements over past surveys researching violence against women. But others have criticized Statistics Canada for problems in its methodology. In this chapter, the methodological advantages and disadvantages of the survey will be discussed, specifically the following: 1. Background and Objectives 2. The Methodology 3. Development 4. Timing 5. Method of Data Col lection 6. Population 7. Sample Frame 8. Sample Size 9. Response Rate 10. Questions 11. Ethical Issues A report of its initial findings and a discussion of the quality of measurement and its theoretical assumptions will follow. Throughout this chapter a number of revisions will be suggested based on the weaknesses highlighted.  139 5.1  B a c k g r o u n d of the Survey a n d its Objectives  The federal government began its $40 million Family Violence Initiative in 1988 in order to work actively with other organizations to prevent family violence. The funding was allocated among six federal government departments: Health and Welfare Canada, the Solicitor General for Canada, Justice Canada, Secretary of State, Indian Affairs and Northern Development, and the Canada Mortgage and Housing Corporation. Their stated goals included involving all Canadians in this initiative and mobilizing community action, strengthening Canada's legal framework, establishing services in Indian reserves and Inuit communities, strengthening Canada's ability to help victims and stop offenders, providing more housing for abused women and their children, providing better national information on the extent and nature of family violence and sharing information and solutions with all involved (Government of Canada 1990:4-5). The Violence Against Women Survey was identified as a priority by the federal government and conducted on behalf of Health Canada under this initiative. The primary objective was to provide reliable estimates of the nature and extent of wife assault in Canada in order to address the need for improved information on domestic violence. Secondary objectives involved the provision of data on violent victimization of women by non-intimates and women's fear of victimization in the larger community. Underlying these objectives was the need to: a) provide governments and service providers with accurate information about wife assault victims and offenders to ensure appropriate allocation of resources; b) examine links between violence against women generally, women's fear of victimization, and wife assault; and  140 c) inform the development of appropriate policies and legislation designed to assist victims of wife assault and address issues related to women's fear (Johnson, Norris andPaton 1994:1). The general data requirements were to provide national and regional level estimates of the nature and extent of wife assault for breakdowns of the population such as age, personal and household income, number of children, and marital status (p. 3). Before any further discussion, a number of terms should be defined. The term "violence" in this survey is defined as "experiences of physical or sexual assault that are consistent with legal definitions of these offenses and could be acted upon by a police officer" (The Daily 18 November 1993:2). These behaviours include physical assaults ranging from threat of imminent attack to attack with serious injury (p. 2), and sexual assaults including unwanted sexual touching up to and including violent sexual attacks resulting in wounding, maiming or endangering the life of the victim (p. 3). Physical and sexual assaults by marital partners are defined as "wife assault" (p. 2).  5.2 The Methodology of the Survey Statistics Canada chose a survey to conduct their study. Although surveys handle time order poorly, and are subject to the problems of testing effects, and instrumentation changes which may make them highly susceptible to interviewer bias (Denzin 1989:29; Gray and Guppy 1994:50), surveys are probably the best method available to the researcher collecting original data for describing a population too large to observe directly (Babbie 1989:237). The survey conducted was a cross-sectional survey, in which data are  141  collected at one point in time from a sample selected to describe some larger population at that time (Babbie 1973:62). Due to budget constraints, telephone rather than face-to-face interviews were conducted, allowing Statistics Canada to access a large sample of women across Canada. Interviewers were required to administer the telephone survey in a central location, where the researcher could supervise the interviewers (Fowler 1988:107).' The interviewers used a computerized questionnaire, called Computer Assisted Telephone Interviewing (CATI), from which they read introductions and questions. The questions were a mixture of open- and close-ended questions; however, some open-ended questions had possible responses pre-programmed. Interviewers who received responses outside of those programmed were required to clarify the responses for measurement reliability (Johnson and Sacco 1995:288). There are a number of advantages to using the CATI system, which reduce the effects of the interviewers. First, the computer can follow complex question patterns that are difficult for interviewers to follow in a paper-and-pencil version of a survey. Second, information from previous questions, or even previous interviews, can be taken into account in question wording or the sequence of questions asked. Thirdly, the CATI system allows that if inconsistent data are given, the interviewer can correct them immediately. Fourthly, interview data can be added to a data file ready for analysis immediately after the interview (Fowler 1988:133). Fifthly, the use of the CATI system also makes telephone surveys the least expensive survey method. The concern regarding interviewer effects in Chapter Two becomes less important because the CATI system controls the flow and questioning sequences of the interview  142  (Groves and Magilavy 1989:301). Although the CATI system allows for the interviewers to make their calls from the same room under the watchful eye of supervisors, it cannot prevent interviewer effects. However, the CATI system does offer the potential for greatly reducing them (Dillman 1978:251), by making it more difficult for interviewers to change the interviews or to conduct them in any way other than that which is prescribed by the researcher (p. 252). There are some disadvantages to using the CATI system, such as the lead time needed to program the CATI interview. The program must also be errorfree if it is to be useful (Fowler 1988:133). In addition, there is no control over these data entry or any coding decisions made by the interviewers, except to make sure that the entries are "legal codes and internally consistent" (p. 133). However, in the final analysis, the CATI system allowed for a number of improvements in the administration of surveys.  5.2.1 Development of the Survey  Holly Johnson was the primary investigator of this $3.2 million survey (Johnson, Norris and Paton 1994:1). During the development of the survey, Johnson sought advice about the methodology from many groups: a sub-committee of the Canadian Association of Chiefs of Police; the Atlantic Institute of Criminology; University of Alberta Centre for Criminological Research; federal and provincial government Departments of Justice, Health, Social Services, Status of Women; transition house workers and residents; sexual assault centres; provincial and national associations; and victims of violence seeking support from these agencies (Johnson 1995b: 148). These groups gave suggestions on the content of the questionnaire, question wording, and innovative approaches that would  143 offer respondents options as to when and where they would participate (Johnson and Sacco 1995:287). The groups consulted wholly supported this initial step in the development of the survey. The study represented an improvement from previous surveys of a similar nature. However, the consultation of certain groups, holding clearly defined political agendas, might have also introduced sources of bias in the construction and implementation of the survey. For example, the consultation process might have influenced the decision to exclude questions of certain areas of domestic violence, such as women's role in the violence. Questions like this would have provided a fuller understanding of the dynamics of domestic violence. The decision to focus the survey exclusively on the victimization of women demonstrates how the agenda of the researchers or more specifically their funding source was shaped in a particular direction. Johnson responded that, "as a survey on violence against women, it was critical to involve women who had experienced violence, as well as crisis counselors and other experts on the effects and impact of violence, in order to develop appropriate question wording and response categories" (1995b: 149). Those groups consulted have a political agenda which may be incorporated into the survey, affecting the reporting of the findings, although consultation does not necessarily indicate incorporation of those political agendas. An interesting task would be to assess the recommendations of these groups to discover which suggestions were taken into account and which were ignored, in order to identify the impact of political agendas. Statistics Canada could also have consulted with psychologists, counselors and therapists of men, especially those labeled "batterers."  144 In addition to extensive consultation, pretests of the question wording were also conducted to help ensure the content of the survey was appropriate and that the results would be both reliable and valid. Pretests are also useful for uncovering aspects of questions that will cause interviewers to have difficulty (Foddy 1993:185). This was accomplished through focus group testing of question wording, one-on-one interviews with drafts of the questionnaire, as well as two large field tests with random samples of women over the phone (Johnson and Sacco 1995:289). The number of cases in the pretest was not published in the reports released.  5.2.2 Timing of the Survey Another consideration in the development of the survey involves the timing of the survey. The importance of timing involves noting the start and end date of the administration of the survey. If the time period is too long in a cross-sectional survey design, unanticipated variables may affect the responses. Unidentified variables, which cannot be controlled by the researchers, negatively impact the validity and reliability of the survey data (Gray and Guppy 1992:43; Singleton et al. 1988:353-4). Johnson and her research team conducted the survey between February and June, 1993. As the time period was not too long, any unanticipated variables were most likely not related to a timing effect. This time period also minimizes "the number of trained interviewers required, reduces 'burn-out' and is expected to yield improved data quality" (Johnson, Norris and Paton 1994:10).  5.2.3 Method of Data Collection  The method of data collection introduced a number of improvements over past surveys researching violence against women. One improvement was the use of a reconstructed victimization survey. Traditional victimization surveys are proficient at measuring property offenses and perceptions of crime, but are not designed to measure more sensitive victimizations, which primarily affect women. These victimization surveys are designed to measure a wide variety of crimes; they are not designed for a detailed analysis of all types of violence and threats to women, the emotional and physical consequences, women's satisfaction with support services, and other detailed information that is necessary to the development of public policy around this issue (Johnson 1995b: 126). The Statistics Canada survey evolved out of the traditional victimization survey, in which a random sample of the population is interviewed about their perceptions of crime and their experiences of victimization, and whose responses are then weighted to represent the population at large (Johnson 1995b: 126). Another extension of this victimization survey was to include victimization prior to the twelve months before the time of the interview, allowing for lifetime prevalence rates. Traditional crime surveys emphasize annual rates of victimization which avoid problems of memory recall and are useful for tracking crime trends. However, Smith (1994:109) argues that these traditional surveys undercount the rate of victimization in the population and may obscure the scope of the problem. Statistics Canada also treated the selection and training of interviewers as critical factors in enabling a relationship of trust to develop between interviewers and respondents; it was hoped this relationship of trust would allow respondents to feel safe  146  and comfortable discussing their experiences. Only women were selected for the job of interviewing because of concerns that many female respondents would not feel comfortable discussing these personal experiences with male interviewers, negatively affecting participation rates (Johnson and Sacco 1995:288).  5.2.4 The Population There are two populations to be identified in any survey: the target population and the survey population. The target population in this study, the "collection of elements that the researcher would like to study" (Frankel 1983:24), is all Englishspeaking and French-speaking women aged 18 to 64 years old in households in the ten provinces in Canada. The survey population is "the population that is actually sampled and for which data may be obtained" (Frankel 1983:24). The survey population included women from across Canada, excluding the Yukon and the Northwest Territories. Women who resided in homes without telephones (2.4%) and/or did not speak either English or French (3%) were also excluded, as well as new immigrants. Younger women, such as those aged 15 to 17 years were excluded because of their "limited experience with intimate relationships" (Johnson, Norris and Paton 1994:5). As discussed in Chapter Four, the literature on dating violence indicates otherwise. Elderly women aged 65 and older were also excluded because the methodology, sample frame and content of this survey were considered to be inappropriate (p. 5). Lengthy telephone interviews cause fatigue in elderly respondents, who often experience hearing difficulties (Groves and Magilavy 1989:301). Also, those women in shelters and transition homes for battered women and women in university and college residences were not included in the  147 sampling, in order to avoid the "clinical fallacy" discussed earlier in Chapter Two (Straus 1991:23). The exclusion of women in shelters and transition homes for battered women is considered to be an improvement over past research. This survey uses a community sample which only included women outside of these shelters and transition homes. Most knowledge about women victimized by men has been restricted to those battered women in shelters and transition homes, and to those who have sought help from the police or hospitals. Because few women use shelters, other crisis services or hospitals, or report violent events to the police, these statistics undercount the incidence of violence against women. These statistics are also based on the reports of battered women who are not representative of all abused women; battered women are likely to be more severely injured, and to have been abused on many occasions. This survey was designed to address the need for information about the victimization of women which is not reported to police, hospitals, or shelters. The design also allows for further comparisons of clinical and community samples.  5.2.5 The Sample Frame  In order to obtain a sample from this population of women, the sample frame or an alternative must be identified. According to Babbie (1973), the "ultimate purpose of surveys is to select a set of elements from a population in such a way that descriptions of those elements (statistics) accurately describe the total population from which they are selected" (p. 83). This can be accomplished through the method of probability sampling, using techniques such as a sample frame, or the "actual list of sampling units from which  148 the sample, or some stage of the sample, is selected" (Babbie 1973:81). In this study a sample frame is not necessary for the sample selection. There are a number of alternatives that can be used to create a representative sample without a sampling frame, such as random digit dialing, cluster or multistage sampling, and systematic sampling (Henry 1990:85). Johnson and her research team chose random digit dialing as the method to be used to select the households. Eligible respondents were randomly selected within the households contacted. Random-digit dialing provides the option of having an anonymous survey procedure, since the name or location of the respondent is never known by the interviewer (Fowler 1988:65). Other advantages include "all households with telephones have an equal probability of being contacted, the need for directories is eliminated and response rates are comparable to the more costly method of personal interviewing" (Johnson, Norris and Paton 1994:9).  5.2.6 The Sample Size Once the sampling method is identified, the sample size must be determined. According to Henry (1990), sample size is the most potent method of achieving precise and reliable estimates that are sufficient for policy decisions or scientific inquiry (p. 117). The base sample size was 20,000 women. In the end, approximately 12,300 women, 18 years of age and older were interviewed in depth by telephone (The Daily, November 18, 1993:2). The adequacy of the sample size is determined by the minimum sample sizes that can be tolerated for the subgroups within the total population for which separate estimates are required (Fowler 1993:35). Johnson determined the minimum sample size  149 for the different provincial subgroups, thereby increasing the reliability of the survey's estimates.  5.2.7 The Response Rate The response rate is simply the number of people interviewed divided by the number of persons sampled (Fowler 1988:46). As reported in the previous section, the number of people interviewed was 12,300, now divided by the number sampled which was 20,000, giving a response rate of 64%. According to Fowler (1993:144), if the response rate drops below 70%, it would be more beneficial for the researcher to reduce the sample size and devote their resources to obtain responses from a higher percentage of the sample. A higher response rate would have produced more accurate and useful estimates. Nonresponse occurred for a variety of reasons including refusals, language difficulties, and unavailability of the woman selected for the interview. Such variations in levels of response can affect the existence and magnitude of relationships observed in these data (Traugott, Groves and Lepkowski 1989:95). Among those households where a respondent was contacted (13,500), the response rate was 91% (The Daily 18 November 1993:9). Unfortunately, the use of random digit dialing is distinctive in that no advance notice to respondents is possible since addresses are not known. This has led to lower response rates than those obtained by personal interviews (Fowler 1988:67; Groves and Kahn 1979); however, telephone interview response rates are higher than those obtained by mail out surveys (Fowler 1988:49). With that weakness in mind, the only concern for the researchers should be the effect that nonresponse has on these data depends on the  150 percentage not responding and the extent to which those not responding are biased or systematically different from the whole population (Fowler 1988:48). Although no discussion of this effect has been given, I would argue that nonresponse could have negatively affected the quality of measurement because the language difficulties could have'been due to new immigrant women who have experienced violence but were not able to participate in the study. Refusals could have been due to women who have not been victimized, thus biasing the nonresponse or to women who were contacted when their husband or partner was in the same room or house, making it difficult or impossible to answer the questions. Mothers with young children or extremely busy schedules would also be more likely to refuse to participate, again possibly biasing the survey data. Each of these groups of women may have been biased or systematically different from the whole population, thus affecting the quality of data. The next step in assessing the advantages and disadvantages of this survey is to discuss the questions asked of respondents.  5.2.8 The Questions As previously mentioned, both close- and open-ended questions were used in this study. The majority of questions were close-ended, allowing for ease in recording these data, using the CATI system. For example, a version of the Conflict Tactics Scale was used in this survey, which contains a series of close-ended questions about women's experiences of violence. Seven of the questions used were open-ended. The open-ended questions allowed the respondents to add information not already covered in the questions and to provide their opinions and advice to other women in similar situations. For  151 example, Question #A24 asks, "Is there anything else you do to increase your personal safety that I have not already mentioned?" Such questions allow the respondents to use their own words and to reveal unanticipated responses (Fowler 1988:87). The questions addressed the following issues: perceptions of personal safety and measures taken to reduce the risk of violent victimization; sexual harassment; experiences of physical and sexual assault by strangers, dates, boyfriends, husbands and common-law partners, and other known men; power/control and emotional abuse by husbands/ partners; spouse abuse in family of origin; the impact of the experience on women who reported violence; who they turn to for help; and involvement and satisfaction with the criminal justice system (Statistics Canada 1994:1). The inclusion of issues of harassment and the perception of fear have been ignored for the most part in past surveys, and thus, added to the importance of this survey. These issues were introduced in the beginning of the survey, followed by questions of violence and threats by dates and strangers, as well as some demographic requests. Questions of marital abuse and violence constituted the larger part of the survey, closing with a wrap-up section with open-ended questions. Specific questions will be analyzed in the section on the quality of measurement. In order to conduct a good survey, it is important to consider the ethical issues involved, as discussed in the next section.  5.3 Ethical Issues As the primary investigator, Johnson faced the ethical dilemma of "weighing the benefits of more reliable crime statistics... against the risks of rekindling personal anguish for respondents" (Gray and Guppy 1994:25) by asking questions about experiences as  152  victim or perpetrator of spousal violence. However, Johnson never lost sight of "the possibility that the respondent could be living with an abusive man and that her safety could be jeopardized should he learn of the content of the survey" (Johnson and Sacco 1995:286). In order to deal with such issues, special training was given to the female interviewers to enable them to help respondents cope with the after-effects of questioning (p. 283). The interviewers were trained to detect whether respondents had privacy and were able to speak freely; if the respondent was not comfortable proceeding at that time, the interview was rescheduled (Johnson and Sacco 1995:287). The interviewers also had a computerized list of shelters and other services for abused and sexually assaulted women across the country. Using the CATI, the interviewer only had to activate a special computer key to access services in the respondent's geographic area. The ability to refer women to support services in the community addressed the need of the interviewers to respond to participants' distress without compromising their role or the objective of collecting statistical data (Johnson and Sacco 1995:288). Another important aspect of the survey was the opportunity for the respondents to choose when and where they wished to participate. A toll-free telephone number was given to each respondent, in case she was unable to complete the survey for reasons of safety or time constraints (Johnson 1995b: 128). The interviewers never called back, giving the respondents ultimate control over their participation (Johnson and Sacco 1995:287). Therefore, Johnson accepted the responsibility for the consequences of asking people sensitive and threatening questions (Gray and Guppy 1994:25).  153 Johnson also accepted responsibility for the consequences of having interviewers ask respondents about experiences of victimization. The interviewers may be distressed by the personal stories they hear, either as a result of particular interviews or an accumulation of interviews (Johnson and Sacco 1995:288). Because of the intense nature of these conversations, a clinical psychologist was present throughout the training and the actual interviews in order to help the interviewers deal with stress and any other concerns which regarding their work on the survey (Johnson and Sacco 1995:289). The computerized list of services and the toll-free telephone number, as well as the presence of a clinical psychologist, provided the interviewers with further tools to lessen any personal stress resulting from the interviews. These steps were improvements over past research on domestic violence.  5.4 The Findings Statistics Canada released the initial findings of the survey on November 18, 1993 in its newsletter, The Daily. The highlights were as follows: • • •  •  • •  One-half of all Canadian women have experienced at least one incident of violence since the age of 16. Almost one-half of women reported violence by men known to them and one-quarter reported violence by a stranger (see Table 3). One-quarter of all women have experienced violence at the hands of a current or past marital partner (includes common-law unions) (see Table 4). One-in-six currently married women reported violence by their spouses; one-half of women with previous marriages reported violence by a previous spouse. More than one-in-ten women who reported violence in a current marriage have at some point felt their lives were in danger. Six-in-ten Canadian women who walk alone in their own area after dark feel "very" or "somewhat" worried doing so.  154 •  Women with violent fathers-in-law are at three times the risk of assault by their partners than are women with non-violent fathers-in-law.  Other findings were revealed in Statistics Canada's Juristat and numerous press reports including the report that the highest rates of wife assault were found among young women and men and among marital partnerships of less than two years. Children were reported to have witnessed violence against their mother in almost 40% of marriages with violence. In many of these cases, the children witnessed very serious forms of violence (Rodgers 1994:1). Police were informed about 26% of wife assault Table 3. Number and Percentage of Wornen 18 Years of Age and Over Who Have Experienced Violence, by Relationship of 'erpetrator Relationship Number in millions % adult lifetime % 12 months Total women 5.38 51 10 reporting violence Spouse or ex-spouse 2.65 29 3 Date/boyfriend ' 1.72 16 2 Other known man 2.46 4 23 Stranger 2.46 23 4 Based on the number of women who have ever been married or lived with a man in a common-law relationship. Figures do not add to totals because of multiple responses (Johnson and Sacco 1995:285). 1  1  Table 4. Number of Ever-Married Women 18 Years and Over who Reported Violence by a Marital Partner, by Number of Occurrences, Canada, 1993 Marital Partner Number of Ever-married women Current partner Previous partner occurrences No. % No. % No. % Numbers in thousands) Total 2652 100 1020 100 1781 100 women victimized One 922 604 35 24 59 435 2-5 593 22 224 22 387 22 6-10 242 9 73 7 11 188 11 or more 842 32 103 734 41 10 Not stated 54 2 2 38 (Source: Statistics Canada 1993b)  155 cases reported by the respondents, and 28% of those cases involved charges being laid. These data also reveals that 24% of the women who reported being abused used a social service. Of all respondents reporting abuse, 13% left their partner to stay at a transition house (p. 2). The majority of women reporting abuse relied on the support of family and friends, but many of those who experienced serious violence had never told anyone before disclosing it to an interviewer (p. 20-1). These data gathered by this survey revealed numerous impressive findings; however, the reliability and validity of these findings can only be determined after the quality of measurement is discussed.  5.5 The Quality of Measurement Throughout the previous discussions, several strengths regarding the methodology in general have been highlighted. In this section, a number of concerns will be raised in regard to the quality of the measurement (see Lenton 1995). These criticisms include concerns about: 1. the single sex survey; 2. the controls for reporting biases; 3. the problem with the wording of questions; 4. the problem of equating researchers' with respondents' perceptions; and 5. the use of the Conflict Tactics Scale.  5.5.1 The Single Sex Survey One criticism of this survey is the decision to only interview women about their victimization by men, rather than interviewing men about their experiences also, which would have provided a fuller understanding of domestic violence. As stated earlier,  156 Health Canada requested that a survey of this nature be undertaken for the federal government's Family Violence Initiative. At the commencement of this Initiative, the federal departments involved undertook extensive consultations with all levels of government and community groups across the country. Through this consultation, it was determined that elderly people, children, and women should be the focus of government efforts to understand the prevalence and nature of family violence (Johnson 1995b: 150). Their objective for this survey was to determine the prevalence and nature of violence against women. The researchers reported that their objectives were met by studying only women and their experiences. Although the focus on women was intentional, the perspectives of the men involved should have been included to determine the true nature of violence against women. Also, the federal government's decision to attain only numbers regarding violence against women indicates a clearly defined political agenda, one that does not include the experiences of men which are needed for a fuller understanding of violence against women. Johnson believed that many women may have felt uncomfortable relating experiences of violence on traditional crime victim surveys, which cover a wide variety of both personal and property crimes. Such traditional victimization surveys cannot address the complexity or details of violence against women in order to test theories or devise prevention strategies (Johnson and Sacco 1995:283). Statistics Canada decided that a survey for women only would allow for a more reliable reporting of violence and a higher quality of data. The survey also added information about threats, intimidation, and sexual harassment which would place women's experiences of violence into a broader social  157 context and would significantly enhance discussions about the correlates of women's fear (Johnson and Sacco 1995:283). However, the inattention to the offender or the accused as the respondent is a methodological concern which has been a problem for numerous studies in the past, as well as for the Statistics Canada survey (Browning and Dutton 1986; Rosenbaum and O'Leary 1981; Dutton and Browning 1983). The decision to focus only on women's victimization reveals the lack of intention to comprehensively study domestic violence. There are a number of reasons for the need for gender comparison within studies. One includes the logic of elementary alternative explanations (Cohen 1989:274). If there is no comparison between the experiences of men and women as victims, researchers can never claim that one is different from the other. Without the experiences of men as victims, researchers cannot ignore men's experiences, as an important factor, out of their programs. Another reason is that the meaning of many observation statements, such as those offered by these survey data, depends on comparison (Cohen 1989:273). Without the comparison of the experiences of both men and women, the statements regarding women's experiences in this survey do not provide as much information as they possibly could. Without such comparisons, the study fails to ask whether the situation has different implications for men and women.  4  Anthony Doob from the Centre of Criminology at the University of Toronto, one of those consulted by Statistics Canada, supported Johnson's decision to conduct such a survey because "the issue of violence against women is, quite simply, a different question from violence against men" (1995:160). Doob argues that although violence against men certainly is an issue, the victimization of women is more of an issue because of the  158 serious nature of the outcome for women (Doob 1995:162). However, the victimization of men by women and the nature of their injuries has not been examined to the same depth. This statement raises the issue of violence against men, which is highly contestable among researchers of violence against women; there is much debate regarding whether the violence against men by women is defensive or aggressive. However, only asking women about their victimization to test theories and devise prevention strategies in all of their complexity and detail brings this decision into question. In their consideration of the often exclusion of women from survey research, Eichler and Lapointe (1985) point out that certain research questions habitually asked only of one gender lead to a distortion of reality and to inappropriate interpretations (p. 10). Inappropriate interpretations can negatively affect the secondary objectives of the survey. One such objective was to use these data to encourage the development of federal programs and policies. Programs and policies based solely on knowledge of victims and their perceptions of their victimization will be useful, but although knowledge of the victim is important to designing prevention strategies, it is not enough. The victimization of women by men includes the activities of both men and women, and the research must reflect both. Research on women, "needs to be placed within the context of female-male relations. As long as the woman remains a passive object of desire, never the initiator of action but always acted upon, not much more than appearance has changed" (Eichler and Lapointe 1985:12). As Lenton writes, "we cannot continue to focus exclusively on the victims (no matter who they are) unless we believe that the etiology of the behaviour is to be found within the victim" (1995:321). The perspective of the victim has been traditionally  159 excluded, but the study must be expanded to include the behaviour of the offender. The interest in which characteristics are risk markers of abuse becomes more of an issue because Lenton's literature review suggests that the characteristics of the perpetrator and relational factors are consistent risk markers of abuse (Lenton 1995:321). However, the case may be made in support of researching both the characteristics of men and women involved in violence. Anita Roberts, a creator of a local B.C. assertiveness and self-defense program for women, instructs women to be aware that they can prevent an attack by being assertive, both verbally and physically, by using a firm tone of voice, and by maintaining eye contact and an erect posture; low self-esteem and fear are identifiable through poor posture, fidgeting hands, and lack of eye contact, which are characteristic of victims of attacks (Roberts 1993). Also, the findings of the Statistics Canada survey show that while 51% of all women have experienced violence, 49% have not had such experiences. Although the statistics are high for women who have experienced violence, it is still important to study those who have not had these experiences. In all studies, it is important to have this control group to identify the variables which may interact to produce abusive experiences, in order to reveal the full continuum of women's victimization. It is reasonable to conclude that there may be differences in characteristics between those who identify themselves as victims and those who do not. If this is true, then attention should be given to both the characteristics of men and women involved in violent activities, and to the characteristics of men and women not involved in violence. To include both characteristics, it is necessary to conduct surveys which allow both men and women to recount their experiences with violence. Only then will such violence be placed within a broader social context.  160  It is essential that researchers design future studies which incorporate the experiences of both men and women. For example, studies should focus on couple pairings, allowing reports of the perspectives of each partner. Couples known to be involved in domestic violence, through police reports of domestic disputes, counseling and therapy participants and reports from shelters and transition houses, can be used. Studies should also be conducted for each gender, creating survey instruments, some specifically tailored for men and others for women. These studies would be most effective with community samples to tap into the experiences of the general population. The problems encountered in single sex surveys also include the reliability of these data. With only women's responses, there are fewer controls for reporting biases, which affect the reliability of these data, as will be discussed in the next chapter.  5.5.2 The Controls for Reporting Biases Another concern faced in determining the quality of measurement involves the potential for memory biases, whether this results in underreporting, overreporting or changes in perceptions of past events. The information obtained in this survey consists of reports about the respondents' behaviours and experiences or more specifically, the respondents' beliefs about those behaviours and experiences (Dillman 1978:83).  5  One way to avoid such memory bias is to not ask respondents to summarize past events, but to ask them to recall actual events or episodes (Gray and Guppy 1994:90). This questionnaire includes two sections (V and W) in which the respondent is asked to recall the most recent or any other incident of victimization by a male stranger, acquaintance, boyfriend or husband/partner. Specific questions were asked about this  161  incidence including the relationship of the perpetrator to the respondent, where it happened, if the perpetrator had been drinking, the respondent's injuries, the effects of the experience, use of weapons, burglary, and the respondent's response to the incident. The questionnaire also includes numerous questions which ask the respondent to summarize past events. For example, Question #J20 asks, "How many different times did these things happen?" Such questions will introduce memory bias, allowing for both underreporting and overreporting. These questions require the respondent to specifically remember a number of different events and to know how many times these events occurred in a lifetime, a task difficult for the average respondent in even remembering yesterday's events. The major concern for Johnson and her research team regarding reporting bias was underreporting because of the sensitive nature of the questions. The use of telephone interviews has raised questions about the ability of the respondents to speak openly and honestly while being interviewed by telephone in a household setting. Johnson was aware of the risks of using such a method, such as those articulated by Smith (1994): a woman may feel that her experiences are too personal or painful to discuss or she may be embarrassed or ashamed about it; she may fear further violence from her abuser should he find out or she may have forgotten about if it was minor or happened a long time ago (Smith 1994:109). In order to be sensitive to these constraints, the survey designers discussed these issues with women who had been victims of violence and designed what they felt was the appropriate approach. The designers felt strongly that this approach in the survey would take account of the many realities of women responding, and would allow the respondents to speak openly and honestly (Johnson 1995a: 127).  162 Part of this appropriate approach was the selection and training of interviewers who could develop a relationship of trust with the respondents. This would produce a climate in which the respondents would feel safe and comfortable discussing their experiences (Johnson 1995a: 128). This is a positive factor because of the concern with the effect of male interviewers dealing with women who might be victims of male violence, or when there is an obvious or perceived power differential. However, throughout the testing phase and the actual survey, the researchers found that talking about experiences of this nature to an anonymous interviewer over the telephone, even though she was another woman, was something that many women found very difficult (Johnson 1995b: 155). Researchers still face the challenge of developing ethical strategies to encourage respondents to speak openly and honestly about experience with violence. The respondents may have found the questions difficult to answer because of the requirement of revealing sensitive information about the respondent herself. However, the difficulty women had in answering the questions might have been due to their inability to remember the number and nature of violent incidents which occurred between the age of 16 and the present time. For a woman who has to recall events that occurred over twenty years ago, especially those incidents which were not perceived as important, there is bound to be unreliability in her reporting of facts. One study reported that older respondents exhibit greater response errors on questions requiring recall of factual material (Groves and Magilavy 1989:300). There have also been suggestions that telephone surveys suffer greater nonresponse among older persons; even among those who do respond, there is a greater tendency to fatigue even during a 30-minute interview (this survey took over 30 minutes). This suggests that there are reporting biases present  163 in the survey data. If Statistics Canada experienced this problem in pretesting, there should have been studies to determine the extent of under-reporting, the presence of reporting biases, and ways to reduce or control for these problems (Smith 1990:268). Researchers must accept the challenge to ask sensitive questions in such a way that encourages respondents to give responses which are close to the truth, even though the truth may not be socially desirable or may be painful to remember (Mangione 1995:30-31). This will help counter reporting biases but it will also be affected by the wording of those questions.  5.5.3 The Problematic Wording of Questions One of the advantages of conducting a survey today is the ability to use the mistakes of past surveys to focus the research and minimize problems (de Vaus 1986:29; Sheatsley 1983:201). Although prior consultation and pretesting allowed for the improvement of many questions, this survey still fell prey to problems that were also associated with past surveys. A criticism of previous surveys is the use of questions, characteristic of the Conflict Tactics Scale, which include poorly conceived categories of violence, combining threatened, attempted and acts of violence that are not mutually exclusive (Dobash and Dobash 1981:450). For example, Question #J14 asks, "Has he ever threatened to or used a gun or knife on you?" Even after extensive pre-testing, the wording of some of the questions was still problematic for those critiquing the survey. Another example is the question asking a woman whether her husband had ever forced her into any sexual activity by threatening her, holding her down, or hurting her in some way. The wording of this question made the categorization of severe abuse more  164 problematic than it needed to be. It would be more useful to be able to distinguish between a husband who threatened his wife with denying her sex in the future, for example, and a husband who beat his wife into submission (Lenton 1995:322). Poorly conceived categories do not increase our understanding of domestic violence.  5.5.4  The Problem with Equating Researchers' with Respondents' Perceptions Another problem in determining the quality of measurement is the equating of the  researcher's perceptions with those of the respondents. After consultation with the many groups in the development of the survey, a number of new items were included in the questionnaire which would be causes of fear for a woman. However, the questions on the survey include items, such as unnecessarily leaning close, blowing a kiss, and inappropriate comments, which are not always interpreted as a precursor to violence, nor do they always cause fear. There were no questions which asked the respondent if these actions caused her to fear a violent attack, if she enjoyed them, or if she found them, at most, irritating; what one woman feels is a violation is not a violation for all women. There were also no questions about the situational and individual context of these actions. As Gartner reports, "even though the researchers acknowledge that the questions they ask about incidents or events ignore the meaning of the acts to the persons themselves and ignore their place in the particular relationships, the researchers themselves ignore the implications of this limitation" (Gartner 1993:315). For another example, attention should have been given to the level of fear generated by such experiences in women who have been victimized in the past and in women who have not been victimized. Past violent victimization will surely allow for a  165  woman to feel threatened by acts which are insignificant for other women. This changes the nature of the presentation of the findings on the levels of fear experienced by the respondents. Even without taking this into consideration, the researchers should have been aware that their perspective may be different from the perspective of the respondents; the perceptions of some respondents will also be different from those of other respondents. Especially in the case of women who have never experienced violence by men, questions should have been asked regarding their perspectives on violence against women by men for possible insights into prevention strategies. The implication of this limitation is the possibility that a different reality is presented than the reality experienced by the respondents. Researchers need to be more aware of the perceptions of the respondents. This can be accomplished by asking the respondents what their perceptions are, in order to compare with the perceptions and interpretations of the researchers. For example, in the sexual harassment section, the respondent could be asked to provide her definition of sexual harassment and then to provide contexts for each item experienced.  5.5.5 The Use of the Conflict Tactics Scale The quality of measurement is also determined by the use of reliable and valid tools. One measurement tool used in this survey is a version of the Conflict Tactics Scale (CTS), used to measure psychological and physical abuse.  6  This survey made numerous changes to the CTS in order to improve on its many weaknesses. One change made was the rejection of the typical introduction, stating instead, "We are particularly interested in learning more about women's experiences of  166 violence in their homes. I'd like to ask you to tell me if your husband/partner has ever done any of the following to you" (Johnson and Sacco 1995:292). This revised CTS also did not include the verbal reasoning or verbal aggression scales, which are used to ease respondents into questions about violence. Johnson and her research team believe that these scales were not necessary because the survey was concerned with violence against women; thus, that context would already have been established by the point these questions were presented (Johnson and Sacco 1995:292). Although the verbal reasoning and aggression scales are necessary when using the original CTS, the introduction does not allude to these scales, thus warranting its exclusions. Other modifications included altering some of the question wording, such as changing "threatened to hit or throw something at you" to "threatened to hit you with his fist or anything else that could hurt you" (p. 293). The scale also included an item concerning sexual attacks (p. 293). Although the CTS was drastically modified, it now reflected the specific measurement goal to obtain incidence and prevalence rates of violence against women, rather than conflict tactics, giving the scale content validity (DeVellis 1991:54). It will now be important to repeat the use of this version of the scale in order to determine its reliability (Carmines and Zeller 1979:50). The use of the CTS requires the asking of questions in a variety of ways and at different times throughout the investigation. Researchers have discovered that many reports of violence are made by respondents while answering seemingly unrelated questions or towards the end of an interview. In order to determine the reliability of these data, surveys must use carefully designed measures which create more than one opportunity for respondents to disclose abusive experiences (Carmines and Zeller  167 1979:40; Hippensteele, Chesney-Lind and Veniegas 1993:69; Mangione 1995:28). These measures must be valid, measuring what they are intended to measure (Carmines and Zeller (1979:17). Although the CTS was improved by including other questions about the respondents' experience with sexual violence, there is concern that the CTS lacks construct validity. For example, there were no other measures used to reveal other experiences with physical violence, indicating a lack of construct validity. There is also concern that the CTS lacks content validity, by not specifying the full domain of content relevant to physical violence and psychological abuse.  5.6 The Theoretical Assumptions Before any research project begins, one or more theories lay the groundwork for the study. These theories set out the problems for the research project, staking out new objects for examination and directing empirical inquiry (Blumer 1954:3). The research project provides the fundamental test of those theories and their verification. "Through the use of research methods elements of the causal proposition are brought together, and new observations are brought forth to modify, verify, and change the theory under examination" (Denzin 1989:64-5). In this study, there were five theories or perspectives which provided the framework. These theories are implicitly acknowledged by Johnson through specific reference to hypotheses associated with these theories. The first theoretical perspective was a feminist perspective. Feminist theories explain through studies, such as Dobash and Dobash (1979), that men who physically assault their spouses do so because their partners have violated, or are perceived as violating, the ideals of familial patriarchy (DeKeseredy and Kelly 1993a: 138).  168 DeKeseredy's study (1988b) also reported similar results. A feminist perspective highlights relevant themes such as an insistence upon women's obedience, respect, loyalty, dependency, sexual access, and sexual fidelity (DeKeseredy and Kelly 1993a: 138). For example, respondents were asked about examples of power, control and emotional abuse demonstrated by current and previous husbands/partners in Sections H and K, such as "he is jealous and doesn't want you to talk to other men," "he tries to limit your contact with family or friends," "he insists on knowing who you are with and where you are at all times," "he calls you names to put you down or make you feel bad" and "he prevents you from knowing about or having access to the family income, even if you ask." The findings reveal a higher percentage of women reporting controlling behaviour by a spouse than the percentage of women reporting violence (Johnson and Sacco 1995:297), providing nominal support for feminist theories in their explanation of violence against women. Another theory which underlie the assumptions of this study was the social learning theory. Social learning theory explains that human behaviour is learned through modeling (O'Leary 1988:33), and that behaviour is supported by other's attitudes reinforcing that behaviour. In relation to violence against women, findings have been presented to suggest that when an individual's attitudes are shaped that support the use of physical force against a woman, the likelihood of physical aggression increases (O'Leary 1993:24). This study provided data to support this theory in its finding that women with violent fathers-in-law were three times as likely as women with non-violent fathers-inlaw to be assaulted by their partners (The Daily 18 November 1993, p.5). These data  169 were the results of questions in Section M about the history of abuse against the mother, both for the respondent and her husbands/partners. Questions about the respondent's and her current husband's/partner's age, employment, health status, income and marital status are guided by sociological theories which attempt to identify risk factors of wife abuse. These factors constitute social structure and within this structure, there is inequality, subordination and superordination, and thus social stratification (McCall and Shields 1986:100). The problems within this social structure affect the family, one of its sub-structures and thus helps identify risk factors of violence in intimate relationships, such as young age and those married or living common-law for less than two years (Rodgers 1994:20), as revealed by the findings. Theories of alcohol and substance abuse were also considered. Although these theories have not proved that alcohol and drugs cause violence, the theories have pointed to the correlation between alcohol and drugs and violence (O'Farrell and Murphy 1995; Frankel-Howard 1989). In this study, each respondent was asked, "In the past month, how often did your husband/partner drink alcoholic beverages?" and "How many times in the past month has your husband/partner had five or more drinks on one occasion?" Alcohol was found to be a prominent factor in women's experiences of violence; perpetrators had been drinking in more than 40% of violence incidents (The Daily 18 November 1993:5). The rate of wife assault for women currently living with men who drank at least four times a week was triple the rate of those whose partner did not drink at all. The rate increased to six times as high for women living with men who drank 5 or  170 more drinks at one time (p. 6), thus providing support for the correlation of alcohol abuse and domestic violence. This study also contributes to victimology research, which has helped make victims of violence more visible (Rock 1994). The contribution helps to identify the impact of crime on women and their fear of crime (Viano 1990:3). For example, this survey included items such as sexual harassment and women's perception of fear and safety in Sections A and B. The findings revealed that 87% of the respondents experienced sexual harassment, either by a stranger or by men they know (Johnson and Sacco 1995:299). Other findings revealed that 60% of respondents who walk alone in their own areas after dark report feeling "very" or "somewhat" worried doing so (The Daily 18 November 1993:1). This victimization survey allows for "an elaboration of our understanding of the broad range of violence women experience and the impact of these experiences on women's lives" (Johnson and Sacco 1995:293).  5.7 Conclusion Statistics Canada initiated a new trend in researching violence against women. Statistics Canada was the first to ask questions of the experiences of women and their victimization in Canada. Its numerous improvements over past studies are to be commended. These data from this survey have also allowed numerous researchers to test further hypotheses, such as Lenton (1995) who analyzes the etiology of wife abuse through power and feminist theories, Wilson, Johnson and Daly (1995) who compared demographic patterns of risk of lethal and nonlethal violence against wives between data from Statistics Canada's Homicide Survey (1974-1992) and this survey, and Keane  171  (1995) who addressed the relationship between victimization and fear of crime. However, there are a number of problems with this survey, which have given researchers the knowledge for improvements in future studies of domestic violence. For example, there is concern with the assumption that the questions asked on traditional victimization surveys fully incorporate the experiences of men. Without men's responses to the questions in this survey, there is no room for comparison of both men's and women's experiences with violence. If the victimization surveys in the past are not relevant for women, there should be testing of their relevance for men. In the same way that Statistics Canada was asked by the federal government to design a victimization survey for women, researchers must consult with relevant interest groups and conduct pretests to design a survey for men. Personal interviews with men in clinical situations and telephone interviews for community samples will best tap into men's experiences. The questions should be asked in a context of understanding and sensitivity, exploring physical, sexual and psychological abuse and men's perceptions of the effects of this abuse. Other examples include the many areas which have not, or rarely have been, included in domestic violence research, which would add to the context of the violent incidents and add to the understanding of violence against women in future research. These areas include women's fear in the home while with others, especially if there is a violent family member in the home, or a man known to her, apart from her marital partner; women's fear outside of the home with men known to her; the respondents' drinking habits (see Sommer, Barnes and Murray 1992); women's involvement in the violent incidents, considering "issues such as self-defense, intentions, motivating factors,  172 and precipitants" (Lenton 1995:321); threats of physical attacks, physical attacks and psychological attacks by other women; unwanted sexual touching in dating relationships; whether the violence was conflict-related or whether it was unprecipitated; and the respondent's perceptions of the actions reported. Johnson and her research team have provided the field of domestic violence research with more accurate estimates of the incidence and prevalence of woman abuse 8  in Canada, as well as the testing of a number of hypotheses based on a number of different theories. Although there are criticisms and concerns raised, it is important to note that "...it is not always possible to provide sufficient details about the actual planning and conduct of research to address adequately all of the methodological questions or criticisms that might arise" (Hay 1993:54). This survey must be commended for its numerous improvements and strengths, as well as for its weaknesses which will provide future research with direction and guidance. In the next chapter I will discuss how the strengths and weaknesses of this survey, together with the previous two surveys assessed, will provide direction and guidance to future research in domestic violence. Notes: For further discussion of telephone interviews and interviewers, see Chapter Two. The final estimates will be biased to the extent that the subpopulation without telephones differs from those with telephone access (Johnson, Norris and Paton 1994:10). " For further discussions of open- and close-ended questions, see Chapter Two. Eichler (1988) also refers to the problem of gender insensitivity — the failing to ask whether a situation has different implications for men and women (Eichler 1988:78); her work reveals the dangers of accepting men's opinions about women to be facts. However, the same danger exists when accepting women's opinions about men to be facts. For further discussion of reporting biases, see Chapter Two. See Chapter Two for further discussion of the CTS. See Chapter One for a discussion of the theories relevant to domestic violence. 1  2  4  5 6 7  Q  See Chapter Four for definitions of incidence and prevalence.  173  CHAPTER SIX:  PROPOSED REVISIONS TO DOMESTIC VIOLENCE SURVEYS  6.0 Introduction In this chapter the strengths and weaknesses of the three surveys previously assessed will be used as the catalyst to propose revisions to domestic violence research. Revisions are necessary in order to not only refine the strengths and improve the weaknesses, but also to propose alternative theoretical perspectives and enhanced methodological tools, as well as to extend existing theories. The goal of such revisions is to widen our knowledge and understanding of domestic violence for the purpose of prevention, protection and treatment. These revisions must be accepted as proposals, similar to hypotheses which cannot be refuted until tested. It is only through this process that research is generated. The revisions came from a number of sources: analyses of domestic violence research conducted in other countries, as well as retrospective critiques of Canadian research published by the researchers themselves. Revisions also came from other researchers or by other individuals within academia who have different political agendas. Other suggestions were retrieved from the media, especially through newspaper editorials and magazine articles. Discussions with individuals provided ideas for revisions, whether from a clinical background or from personal experience. Care was taken to consult as many sources as possible. However, no claim is made to the exhaustiveness of the sources consulted as the field of domestic violence, being linked to studies of family  174 violence, dating violence, child and elder abuse, in sociology, psychology, nursing and medicine, is vast and comes with an ever increasing bibliography. The following section will focus on each survey assessed, highlighting revisions to be made and repetitions to be considered specific to each study. A discussion of the similarities between the three studies will follow, noting similar weaknesses and how future research can be improved by making revisions based on these weaknesses.  6.1 The 1987 All Alberta Study  Criteria 1. Who paid for the results? 2. When did data collection begin and end? 3. How were the data collected? 4. Of what group is the data representative? 5. How was a representative sample acquired? 6. Exactly how many people participated? 7. How many of the people contacted refused to participate? 8. What was the exact wording of the questions?  All Alberta Study (1987) The University of Alberta Between January and March 1987 Telephone and face-to-face interviews by men and women (often unsupervised) All persons 18 years of age or older living in a dwelling unit in Alberta Face-to-face interviews: random selection from a sample frame; Telephone interviews: random digit dialing 1343 298 (22.2%)  Both open- and close-ended questions including a shortened version of the CTS; questions about domestic violence included the qctions and victimization of both men and women 9. How accurate are the data? Concern with short version of CTS; questionable response categories; lack of definitions of terms 10. What ethical issues were No discussion of ethical issues considered?  175 The 1987 All Alberta Study, the first study in Canada to investigate domestic violence, provided the basis for numerous revisions to be made in future research. At the same time, it also encouraged the repetition of parts of the survey designs. One area that should be repeated, in particular, is the sample population consisted of roughly equal numbers of men and women. Each respondent was questioned about their experience with violence as both victim and abuser. It is important to encourage offenders to identify both experiences. Questions assuming male violence only place men on the defensive, bringing the reliability of those data into question (Sudman and Bradburn 1982:5; Dillman 1978:101). Another part of this survey to be repeated is the question regarding the initiation of most serious physical violence. It was an important question to help identify the process of violence and its initiators in violent situations. The process of violence has been known to begin with psychological abuse, sometimes years prior to the violence. However, this knowledge has demonstrated the need for including questions about psychological abuse in domestic violence research and also the importance of identifying the actions and reactions of the victim and aggressor immediately prior to the violence. Ideally, the question would ask what happened following the initiation of violence; the interviewer would then be required to record the events that followed in order. Besides the encouragement of similar studies in other provinces in Canada, a major contribution to the field of domestic violence research is the use of longitudinal studies, similar to the All Alberta or Edmonton Study series. These studies allow university researchers and local community agencies to investigate and explore social issues, of importance to the specific populations in an on-going research framework  176 (Kinzel 1987:16). However, in the future, researchers undertaking study series need to repeat questions specific to domestic violence in subsequent surveys. Repetition allows for comparisons and increases the reliability of the results. The areas to be revised in future research include the supervision of interviewers, validity and reliability of the questions as measurement tools, the use of an omnibus victimization survey, improved consultation or consideration of ethical issues, and the consideration of gender effects of the interviewers. Each of these concerns provides direction for future research by encouraging the repetition of the measurement tools to increase the reliability of these data and by determining the extent to which the measurement tools measure what they are intended to measure to increase the validity of these tools! One important revision involves the use of CATI for telephone surveys in order to better supervise the interviewers. As was already noted in Chapter Three, the telephone interviewers in Calgary were allowed to work unsupervised from their homes. The necessary monitoring of interviews was not given careful consideration by the researchers, and could have influenced the quality of data collected. Future research should consider the use of the CATI system, as used by Statistics Canada in 1993 after considering the disadvantages of past surveys and the advantages of using such a system, as discussed in Chapter Two. Another area of revision for future research involves the validity of the questions as measurement tools. For example, the relation of introductions to the questions is problematic. Introductions must be true to the corresponding questions by specifying the full domain of these questions. In the All Alberta study, the revised CTS was introduced  177 in a similar way to the introduction used for the original CTS, which indicates that there are many different ways of settling disagreements. However, the questions were not comprehensive, asking only about physical violence; the questions omitted verbal reasoning and verbal abuse. If the introduction does not represent the questions asked, there is concern that the introduction will not have its intended effect. The questions should also reflect the reality of the respondents by creating exhaustive response categories (Mangione 1995:21). Caution must be taken in the use of close-ended questions. One part of a four part question which asks, "What did the police do?" (Population Research Laboratory 1987:29), did not allow for such a reflection. The researchers anticipated some of the possible responses but they may not be sufficient for this question. Researchers need to anticipate all possible responses. A comprehensive list of responses is essential for standardized coding (Sudman and Bradburn 1982:21), but it is also essential to prevent placing respondents into categories which do not fit. If researchers do not anticipate all possible responses, they should then allow the respondents to provide their own responses, in order to improve future research questions. Another revision involves the use of questions containing 11 digit response categories. Respondents need enough categories to represent their feelings or experiences, but not so many that the respondents make inconsequential distinctions (Mangione 1995:12). An 11 digit response scale is too arbitrary. The respondents are more likely to answer truthfully to a scale question when the possible responses are given (Dillman 1978:206). Researchers should either provide the possible responses such as a continuum beginning with "never" to "always" or the respondent should be allowed to provide a response in their own words.  178 This study was also subject to the criticisms often given to omnibus victimization surveys. This survey did not directly address the issue of domestic violence, but allowed for the introduction of such questions into Canadian research. These sensitive questions about domestic violence were presented in the middle of the questionnaire in order to cushion their threatening characteristics with other issues which were non-threatening (Fowler 1988:102). Surveys such as this one encouraged the creation of surveys specific to domestic violence, and especially the creation of a victimization survey for women, as created by Statistics Canada. There appears to be a lack of prior consultation with groups or persons involved in this area. The survey questions are based on earlier U.S. surveys of domestic violence. However, as the first Canadian survey, Kennedy and the Population Research Laboratory should have consulted with relevant interest groups in Alberta to identify issues pertaining to men and women in Alberta. For example, prior consultation was used in future research, such as DeKeseredy and Kelly's student survey and Statistic Canada's Violence Against Women survey, assisting in the wording and content of questions. The decision by the Population Research Laboratory to include questions about domestic violence also did not appear to reflect any ethical issues, although such sensitive questions can have serious implications for both the respondents and the interviewers. Precautions must be taken to ensure the participation, and well-being, of the respondents. This consideration would be reflected in information provided at the conclusion of the survey, such as lists of services for victims of domestic violence provided by both Statistics Canada and DeKeseredy and Kelly in their surveys.  179 This study also reveals the need for same-sex interviewers. From what is known about the interviewers in this study, both male and female interviewers conducted the interviews for both male and female respondents, which will have had an effect on these data. Interviewer effects, for men interviewing women and women interviewing men, have been reported in surveys, especially ones with sensitive questions about domestic violence and one's participation in that violence. If both men and women conduct the interviews, then the respondents are not being presented with the same set of stimuli; this result therefore affects the quality of data (Denzin 1989:116). Lastly, it is important to note that findings of surveys provide direction for future research. For example, the findings of this study should help guide the questions in future studies, such as why high rates of violence exist in common-law relationships involving young men and women with low incomes in urban areas. However, this direction can most effectively be given when these data are extensively analyzed. In this study, the findings on the violent experiences of the male respondents were not published, although these data were available through the interviews conducted with both men and women. Whatever the reasons for non-reports, it is important for future research to have all data analyzed. Subsequent researchers, such as DeKeseredy and Kelly, made the effort to have all of their survey data analyzed by including other researchers, but are still waiting to accomplish this task. This survey will be discussed in more detail in the following section.  180 6.2 The Study of Male-Female Dating Relationships in Canadian Universities and Colleges  Criteria 1. Who paid for the results? 2. When did data collection begin and end? 3. How were the data collected? 4. Of what group is the data representative? 5. How was a representative sample acquired? 6. Exactly how many people participated? 7. How many of the people contacted refused to participate? 8. What was the exact wording of the questions?  DeKeseredy and Kelly's dating violence survey (1992) Health and Welfare Canada's Family Violence Prevention Division Between September 1992 and January 1993 Self-administered questionnaires in classroom settings, conducted by the researchers (a man and a woman) All men and women enroled in a Canadian college or university in the fall of 1992 Multi-stage, systematic sampling strategy, using both nonprobability and probability sampling methods 3,142 fewer than 1 %  Both open- and close-ended questions, including a revised version of the CTS; separate questionnaires for men and women to identify women's victimization and their use of help groups and to identify men's offenses and the role of peer pressure 9. How accurate are the data? Criterion-validity in question; lack of comparable data from the two different questionnaires; lack of definitions and vague wording used; concern with use of revised CTS 10. What ethical issues were Participants were provided with a list of local support considered? services for survivors and abusers, and were encouraged to talk to the researchers about the dating violence or the survey  The study of male-female dating relationships by DeKeseredy and Kelly was the first national survey about dating violence in college and university populations. This study was an improvement over past surveys in that DeKeseredy and Kelly consulted with a number of groups before the administration in order to ensure proper wording and  181 content of the questions. DeKeseredy and Kelly wanted to balance the need to replicate previous studies with the necessity of avoiding their methodological problems. However, I have raised a number of concerns that should encourage revisions to future research. These areas include concerns with the sampling design, the sample composition, gender effects of the research team, and the validity of these questions and the reliability of these data. For example, in this sampling design, a properly calculated response rate was not provided. In this study, there was the possibility of nonresponse in all stages of the sample selection. Institutions could refuse to participate, as well as departments, instructors and students, each affecting the overall sample size and representativeness. The precise number of students who did not participate is not known (DeKeseredy and Kelly 1993a: 144). However, the rate of nonresponse could have been more precisely determined by obtaining class lists for each of the selected classes (Hay 1993:58). Another example of an area for revision in future research is the need for representation of the population. In this study, there is concern with the size of the sample in the overrepresentation of women in the student population, and the lack of justification for this overrepresentation. Women do not make up the majority of student populations, so survey populations should not reflect such overrepresentation. Overrepresentation brings the generalizability of these data into question, because these data will not be representative of the actual population. DeKeseredy and Kelly should have determined the percentage of women in the student population and then adjusted the estimates to reflect that percentage when making gender comparisons.  182  Another concern to be carefully considered in future research is the gender of the research team which administered the questionnaires, and whether or not there were any biases due to their gender. The research team consisted of a man and a woman. According to Denzin (1989), interviews are "gendered, interactional productions" affecting these data constrained by the "gendered identities that are enacted in the interview encounter. Gender filters knowledge" (p. 116). Thus, the gender of the research team may have affected the responses given by the men and women who participated in the survey. Their presence was probably beneficial in increasing the response rate (DeKeseredy and Kelly 1992:9); however, attention must be given to the possible effects of gender biases. Researchers might consider separating the men and women respondents and conducting the survey in two different rooms with the appropriate gender. Scales should also be created to identify possible gender effects, exploring, for example, the respondent's perceptions of the opposite sex. Similar to the All Alberta study, there are concerns with the validity of the questions as measurement tools, that the questions measure what they are intended to measure. Again, the introductions need to correspond to the questions. As discussed in Chapter Four, this survey was introduced to the respondents in the context of problems in heterosexual dating relationships. However, male respondents are asked about their own abusive actions and their friends' support of such behaviour, not about their own experiences or perceptions of the relationships. This must be viewed as a problem because such questions can place men on the defensive, affecting the reliability of these data. If questions are demeaning, embarrassing, or upsetting, respondents may terminate the interview or falsify their answers (Sudman and Bradburn 1982:5). Researchers need  183  to maintain a level of consistency between the introductions and their corresponding questions. The choice of measurement tools should also be carefully considered in future research. In this study, following each question was a line with 0% and 100% at either end and percentage points spaced equally by tens. Such a scale requires a lot of thinking on the part of the respondent; the responses are many in number, difficult to choose and too time consuming to answer accurately. Other valid measurement scales must be designed for increased reliability of these data. Personal or telephone interviews may allow researchers to explore the issue of self defence to better create valid measurement tools. Group-administered questionnaires should allow respondents to use their own words to reveal their perceptions about self-defence. This study, in general, has encouraged numerous revisions to future research projects, in terms of question topics. For example, as previously discussed in Chapter Four, future research should produce comparable studies to cohorts in the work force. Comparisons should also be made between students who live in residences on campus, students who live alone or with friends off campus, and students who live with family. The associations with different groups could also be revealing in the types and frequency of victimization and perpetration by women and men. Future research should also include reports of the psychological abuse of men by their dating partners or girlfriends, which has been identified as a problem by other research. This study also encourages researchers to identify risk markers of dating violence, such as ethnicity, educational status, and the level of intimacy (DeKeseredy and Kelly 1993a: 155). Other risk markers to be discerned include the amount of time spent together  184 and level of commitment. Past history of violence, parental familial patriarchal values and norms, the structural fact of male dominance within the family, absence of deterrence and a closer look at memberships in social groups such as fraternities are also possible factors for researchers to consider in future research (DeKeseredy and Kelly 1993b:4748). Such information will help identify those men and women who are at the greatest risk of being abused or being abusive, while also assisting in theory development (DeKeseredy and Kelly 1993a: 155). One last contribution of this study to future research is in demonstrating the need for couple data. One problem with data from this study involves large gender differences in reporting the incidence and prevalence of abuse. As noted in Chapter Four, men were more likely to report having used a weapon than women were to state having been victimized by this form of abuse. Also, more men reported threatening a date with a weapon than women reported being threatened (DeKeseredy and Kelly 1993a: 153). Social desirability and miscommunication can play a role in differences in reporting. However, the differences in reporting may also be attributed to the lack of dating relationships between the male and female respondents. Until couples or ex-couples are studied, there can be no guarantee of the validity of the measurement tools or the reliability of these data. Couples and ex-couples must be sampled in both clinical and community populations using both personal and telephone interviews. Questions in such studies should allow for the comparison of the perceptions of each member of the couple about their relationship and their interactions. Researchers also need to use couple data for another reason. These data from this study cannot provide gender comparison because the men and women were asked  185 different questions, eliciting men's reports of abusive behaviours and women's reports of victimization. The identical questions that were asked on both questionnaires were not asked in the same order, further complicating the reliability of these data. Research on order effects reveal how people's responses to one item in a survey are consciously or unconsciously influenced by their responses to a previous item (Sigelman 1989:144). These data must be gathered using the same questions, in the same order, for both men and women, to ensure reliability of these data and validity of the measurement tools used.  6.3 The Statistics Canada Violence Against Women Survey  The Statistics Canada survey on Violence Against Women, as the first national survey of its kind, has encouraged a number of revisions, but also encouraged the repetition of many of its questions and design. For example, it improved on past victimization surveys by tailoring the entire survey to women's experiences with violence and harassment. Statistics Canada also treated the selection and training of interviewers as critical factors in enabling a relationship of trust to develop between interviewers and respondents; this relationship of trust would allow respondents to feel safe and comfortable discussing their experiences. Only women were selected for the job of interviewing because of the concern that many female respondents would not feel comfortable discussing these personal experiences with male interviewers. This would have a negative effect on participation rates (Johnson and Sacco 1995:288).  186  Criteria 1. Who paid for the results? 2. When did data collection begin and end? 3. How were the data collected? 4. Of what group is the data representative? 5. How was a representative sample acquired? 6. Exactly how many people participated? 7. How many of the people contacted refused to participate? 8. What was the exact wording of the questions?  Statistics Canada's Violence Against Women Survey (1993) Health and Welfare Canada's Family Violence Prevention Division Between February and June 1993 Computer-assisted telephone interviews by women only All English- and French-speaking women aged 18 to 64 years old in households in the ten provinces of Canada Random digit dialing 20,000 7700 (36%)  Both open- and close-ended questions, including another revised version of the CTS; topics included personal safety and fear; sexual harassment; physical, sexual and psychological victimization; use and satisfaction with support services and the justice system 9. How accurate are the data? Concern with single sex survey; lack of controls for reporting biases; vague wording of questions; the equating of researchers' perceptions with respondents'; and the use of another revised CTS 10. What ethical issues were provision of a list of local support services to respondents considered? and a toll free telephone number to complete or talk about the survey; special training of female only interviewers and an in-staff psychologist to help interviewers deal with stressful calls  This survey also encouraged further use of the CATI system. Although the CATI system allows interviewers to make their calls from the same room under the watchful eye of supervisors, it cannot prevent interviewer effects. However, it does offer the potential for greatly reducing them (Dillman 1978:251). The CATI also allowed interviewers to simply activate a special computer key to access services in the respondent's geographic area in order to respond to participants' distress. With the CATI  187 system, this can be done without compromising the role of the interviewer or the objective of collecting statistical data (Johnson and Sacco 1995:288). Another important aspect of the survey was the opportunity for the respondents to choose when and where they wished to participate. A toll-free telephone number was given to each respondent in case she was unable to complete the survey for reasons of safety or time constraints (Johnson 1995b: 128). Both the CATI system and the toll-free number improved the survey and should be continued in future research. Similar to DeKeseredy and Kelly's survey, Johnson sought advice about the methodology from many groups during the development of the survey, ensuring that the wording and content of the questions were acceptable. Also, similar to DeKeseredy and Kelly's survey, this study demonstrated the need for gender comparison within studies. If there is no comparison between the experiences of men and women as victims, researchers can never claim that one is different from the other. Attention should be made to both the characteristics of men and women involved in violent activities, and also to the characteristics of men and women not involved in violence. To include both characteristics, it is necessary to conduct a survey with both men and women responding to questions about both men's and women's experiences. Only then will such violence be placed within a broader social context. The Statistics Canada survey was also criticized for problematic wording of questions, use of the CTS, equating researchers' perceptions with those of the respondents and reporting biases. These weaknesses are shared by all three studies assessed and will be discussed generally in the next section.  188 6.4 Similarities Between The Three Surveys  There are a number of similarities between the three studies assessed, both in terms of encouraging repetition, and revisions, to future research in both methodological and theoretical areas. For example, all three studies used surveys to obtain these data, although different types of surveys were used. The social survey is the most commonly used method in domestic violence research; however, a disadvantage of these crosssectional surveys is that the temporal order of the independent and dependent variables cannot be established or measured (Gray and Guppy 1994:50), thus only describing associations and correlations. Other survey methods should be considered, such as panel and longitudinal studies, which would help establish the temporal order of variables. Another problem encountered in surveys involves the potential for memory bias; memory bias can result in underreporting, overreporting or changes in perceptions of past events. The information obtained in these surveys consists of reports about the respondents' behaviours and experiences or, more specifically, the respondents' beliefs about those behaviours and experiences (Dillman 1978:83). One way to avoid such memory bias is to not ask respondents to summarize past events, but rather to ask them to recall actual events or episodes (Gray and Guppy 1994:90). Other suggestions to prevent or reduce reporting biases include using only necessary and relevant questions, requiring the least amount of detail as possible. The use of introductions and context also encourage respondents to answer questions more honestly (Mangione 1995:31-32). Although researchers must make every effort to decrease these ill effects, questions about past events are important for survey projects such as this one.  189 Future research should be encouraged to continue the use of community samples, as used in all three surveys, in order to provide estimates for rates of abused men and women outside of clinical populations. Most information about domestic violence has been provided by those in shelters and transition homes, and to those who have sought help from the police or hospitals. Because few people use shelters, other crisis services or hospitals, or report violent events to the police, these statistics undercount the incidence of domestic violence. However, researchers should also continue to study clinical samples in order to further compare between community and clinical samples. Other similar revisions encouraged by all three surveys include the need for similar tools and definitions in order to make Canadian data comparable. To date, there are no such data because researchers and clinicians have yet to agree on the terms, labels, or measures to be used in discussing domestic violence (McHugh 1993:58). For example, the All Alberta study referred only to physical violence but did not define or clarify the word "hitting." Such a word will vary in interpretation for both men and women as gender groups and as individuals, as well as for the researchers. Statistics Canada, on the other hand, included physical, psychological and sexual abuse of women, as defined by criminal law. DeKeseredy and Kelly's dating violence survey focused on the physical, psychological and sexual abuse of women, but also briefly addressed the physical and sexual abuse of men, excluding the psychological abuse of men. These studies reveal the need for the study of men's and women's definitions of domestic violence in order to improve the validity of its measures and to determine the level of miscommunication between men and women. The importance of definitions and similar tools can be measured in terms of the quality of measurement of these data  190 gathered. In order to ensure the reliability of the reports, it is essential that all respondents have the same understanding of the questions. This allows the researcher to be confident that the same definitions have been used by all respondents (Fowler 1995:13). For example, one question in DeKeseredy and Kelly's female questionnaire (#24) asks if the woman has ever been sexually harassed. No explanation of sexual harassment is given, nor do the researchers attempt to determine what sexual harassment means to the respondents. This lack of definition decreases the effectiveness of this question and the reliability of these data obtained from this question. Researchers must provide definitions for ambiguous terms and identify the respondents' perceptions of those terms. Particularly in the case of dating and domestic violence, there is no clear agreement on what constitutes such violence and how to measure it. It is, therefore, important to measure it using several different tools and methods and to also ensure that researchers measure all of its different dimensions (Babbie 1989:126). For example, researchers require different tools to determine the context of abuse, and to determine the respondents' views about relationships, sex, and boundaries. Another example of different measurement tools is the use of a different version of the CTS in the three respective surveys. The modifications to the CTS were often in response to criticisms regarding a lack of context or motives. However, as each survey used a different version of the CTS, there is no test of the reliability of each version. Each version has not been repeated and there are no similar results which help determine the reliability of the scale (Carmines and Zeller 1979:11). Researchers must repeat past  191 studies, while improving methodological weaknesses, in order to increase the reliability of these data obtained. Although these studies add to the small amount of Canadian data on the incidence and prevalence of domestic violence, they still suffer from many faults that preclude a comprehensive understanding of domestic violence (DeKeseredy and MacLean 1990:24). A problem still exists with a lack of checks on the accuracy of reporting when using the CTS, as well as with poorly conceived categories of violence, which combine threatened, attempted and actual categories of violence that are not mutually exclusive (Dobash and Dobash 1981:450). There is still a need for more context when using the CTS. Although DeKeseredy and Kelly and Statistics Canada argue that context was achieved by using questions about self-defense, sexual harassment and assault, there is still much more to be learned about the context and dynamics of domestic violence. More can be learned about domestic violence via questions eliciting the perceptions of those involved. All three surveys fell prey to equating the researchers' perceptions with those of the respondents. For example, a number of new items were included in the Statistics Canada survey questionnaire which would be causes of fear for a woman. However, the questions on the survey included items such as unnecessarily leaning close, blowing a kiss, and inappropriate comments, which are not always interpreted as a precursor to violence; nor do these actions always cause fear. The researchers did not ask the respondents if these actions caused them to fear a violent attack, if they enjoyed these actions, or if they found them, at most, irritating; what one woman feels is a violation is not a violation for all women. There were also no questions about the context of these actions.  192 Such questions reflect the need for a combination of qualitative and quantitative methods in order to produce the most thorough understanding of domestic violence. Qualitative methods uncover subjective perspectives and the social context in the creation of theories, while quantitative methods evaluate the accuracy of claims through deductive logic (Murphy and O'Leary 1994:219). Quantitative measurements were the primary method in survey research for a number of years until alternative techniques, such as open-ended questions began to be used. Open-ended questions to build interviewerrespondent rapport, allow respondents to qualify their responses, encourage interaction and collaboration between researcher and subject, and elicit versions of violent events that reflect the experiences of victims (Smith 1994:115). Not only does future research require both open- and close-ended questions, it also requires expanding the content of the questions into areas such as the complexities of intimate relationships in which there is intensity in emotion and physical sensations. Researchers need to explore the desires of men and women in relationships, women's roles in violence and also more about the nature of relationships, such as the strengths and weaknesses which reflect people's decisions to leave or stay in an abusive relationship. Another example, from both DeKeseredy and Kelly's and Statistics Canada's survey, includes questions about a woman's perception of safety during the day. Questions also need to address issues such as the relationship between violence and dual-income families versus single-income families, having children versus no children, and employment versus unemployment/welfare. Future research also needs to include questions for men about their own experiences, especially their victimization, as well as questions of both sexes in terms of  193 self-reports of violence. Researchers need to ask questions which encourage men to report their victimization, just as there are special survey questions for women to report their own victimization. The questions must also encourage men and women to discuss their perceptions about their violent relationships. Investigating both men's and women's perceptions allows for a better understanding of domestic violence and also allows for the treatment and prevention of violence in these relationships. Researchers who investigate the perceptions of both men and women will exhibit a sensitivity demonstrating that not all women are victims, nor are all men aggressors (Crowley 1994:10-11). This revision will respond to criticisms of the single sex surveys, such as DeKeseredy and Kelly's survey and Statistics Canada's survey, who gathered information on violent acts from only one member of a couple, making it difficult to assess the reliability of those reports (McHugh 1993:57). Researchers must also continue to use numerous theories to allow for the testing of many more hypotheses, as mentioned above. Although the three surveys based their questions on feminist, sociological, social learning, alcohol abuse, victimology and interactionist theories and perspectives, these theories require strengthening with increased testing of hypotheses. As demonstrated in the earlier chapters, many of these theories demonstrate weaknesses which must be addressed in future research. Revisions must be made to these theories, through changes and expansion, to include a more encompassing focus. These existing theories will be best advanced through syntheses to create new theories which will provide a more comprehensive understanding of domestic violence.  194  Each theory must be acknowledged to be necessary, then combined to allow the weaknesses of each theory to be strengthened by the others. For example, feminist, sociological and social learning perspectives all provide necessary information but neither is sufficient on its own to provide an effective explanation of domestic violence. Considered together, these theories can explore domestic violence multidimensionally. However, these three theories are still not enough. Evidence has been provided to support the use of psychological and interactional theories which, I believe, provide the most potential to provide a more effective explanation. The importance of personal factors and how men and women communicate in intimate relationships cannot be stressed enough in the study of domestic violence. The role of alcohol and substance abuse in violent situations should also be studied further. Domestic violence is complex and requires many perspectives to be fully understood.  6.5 Conclusion The importance of assessing these three surveys is revealed in the suggestions for theoretical and methodological revisions which will provide direction to future research in domestic violence. The next critical stage in domestic violence research involves the syntheses of existing theories to create new, more effective theories. Other perspectives must also be considered, such as psychological and interactional theories, as well as the role of alcohol in domestic violence. Following these theoretical revisions, researchers should explore changes in methods of data collection and in the direction of the questions. For example, longitudinal and series studies would be effective to identify the variables which interact  195  to encourage and discourage abuse and violence. Cross-sectional studies should be continued but improved in future research by reducing reporting biases. The emphasis of these studies needs to shift to a more comprehensive outlook on domestic violence by exploring the role of both men and women in the violence, thus including men and women in samples and using similar questions. The similarities need to be acknowledged and studied to determine how surveys should be designed to elicit appropriate responses from both men and women. More specifically, researchers need to identify the perceptions of men and women about domestic violence, incorporating these perceptions in the surveys and our understanding of domestic violence. The assessment of past surveys allows for the identification of strengths and weaknesses, both methodologically and theoretically. However, these assessments and suggestions reveal the need for even further research and revisions. Methodological and theoretical issues have been identified which are not directly related to the three surveys discussed, but would greatly enhance future studies in domestic violence, as will be discussed in Chapter Seven.  196 CHAPTER SEVEN: CONCLUSION  For the past twenty-five years, researchers have been faced with the problem of studying, both theoretically and methodologically, domestic violence. This thesis has provided an analysis of the strengths and weaknesses of Canadian research on domestic violence. The following is a summary of my recommendations: •  Researchers need to create new or revise old hypotheses, (for example, multifactor causal models), in order to compensate for the existing, necessary but not sufficient, theoretical perspectives. New theories are necessary in order to eliminate all possible lines of inquiry and advance our understanding of domestic violence.  •  The use of longitudinal studies should be encouraged, as well as the continued use of clinical and community samples, in order to determine rates of violence in both samples.  •  Researchers need to combine quantitative and qualitative measures in order to produce the most thorough understanding of domestic violence.  •  In order to increase the reliability of data, memory bias should be reduced by asking respondents to recall specific events or episodes, instead of asking for a summary of past events.  •  The quality of data can be increased by encouraging researchers to use same-sex interviewers, ensuring that respondents are presented with the same stimuli.  Researchers need to use couple data for gender comparisons, not only to identify gender gaps in reporting, but also to identify the experiences of the men and women as both victim and offender. It is important to identify the process of violence and its initiators in violent situations, especially the role of psychological abuse. As well, the actions and reactions of the victim and aggressor immediately prior to the violence must be identified. Questions must include areas such as: (1) the complexities and nature of intimate relationships, including the desires of men and women; (2) the further identification of risk markers of dating and domestic violence, such as level of commitment and intimacy and time spent alone together; and (3) the psychological abuse of both men and women. Dating violence research would be advanced by producing comparable studies to cohorts in the workplace to identify if rates are higher in college or university settings. Studies should also compare rates of violence between students who live in residences on campus, those who live alone or with friends off campus, and those who live with family. Researchers need to agree on the definitions used in studies of domestic violence as well as the research tools used. However, researchers also need to study men's and women's definitions and perceptions of domestic violence in order to improve the validity of measures used, and to determine the level of miscommunication between men and women.  198 The revisions I have suggested provide further directives for the study of domestic violence in Canada and internationally. Although the three surveys assessed reveal methodological and theoretical weaknesses and suggest revisions, a number of other methodological and theoretical revisions should be considered in future research of domestic violence. Researchers must continue to challenge one another in the search for more effective methods and improved theoretical understanding. Methodologically, researchers need to consider the use of multiple measurement tools and varied questions. Theoretically, researchers must continue to challenge the assumptions underlying studies of domestic violence. Theories need to wholly reflect what is known about domestic violence and continue to test the hypotheses generated by these theories. In order to meet the methodological challenge, it is important to consider the move from a deductive model, in which research is used to test theories, to the inductive model where theories are developed from the analysis of research data (Babbie 1989:53). There have been numerous findings from the deductive model that do not support domestic violence theories. Now is the time to apply those findings to create and revise new theories using the inductive model. By doing so, theories that can be supported are strengthened, and those which cannot be supported are discarded. The developers of research projects also need to consider other methods, such as the use of third party reports. In interviewing or treating men and women in violent relationships, researchers and therapists should seek third-party corroboration and/or seriously question the accuracy of the reports of both partners (McHugh 1993:58). Informants may provide further information and improve reliability, as long as these  199 informants are questioned about behavior that the informant might know, either from observation or through conversations (Sudman and Bradburn 1982:74). The informant should not be a participant in the violence and abuse. Other methodological improvements include the exploration of different contexts and the consideration of alternative explanations and study limitations (Murphy and O'Leary 1994:214). For example, the Center for the Prevention of Domestic Violence in Cleveland, Ohio uses a unique approach in its intervention with male perpetrators of domestic violence. The information gathered and corroborated by other sources as needed reflects the exploration of context and motive including: (a) the history, duration, frequency, severity, and type of violence perpetrated, and the batterer's attributions about his violent behaviors, (b) his current violence potential, (c) the roles and rules in the intimate dyad in which the violence occurred and his perceptions of the victim and of her role in the violence perpetrated, (d) his alcohol/drug history, criminal/legal history, work history, and current career success/satisfaction status, (e) a physical and mental health screening, (f) a psychosocial developmental history (including separation and loss events and other childhood stressors, and patterns of significant childhood and adult relationships), (g) his expectations of and motivations for treatment, and (h) current sources of stress and support (Carden 1994:567). As thorough as this list appears, however, there are still limitations to such a program. There are no questions regarding the actual incidents of violence or the immediate precursors to the violence, nor are there any considerations of the women's role in the violence. An example of the consideration of alternative explanations is the fact that research has not established whether violent relationships should be identified as wife abuse, husband abuse or mutual combat (Saunders 1988:91). Researchers should be  200 willing to concede to alternative explanations, especially when their own research has not been conclusive. Another methodological revision to future research on domestic violence involves the use of questions that include further variety in content. For example, numerous studies have failed to identify consistent risk markers of determinants of violence. It is thus reasonable to assume that there remain gaps in the research; researchers are missing important questions, such as a woman's personality traits prior to her experiences with violence, descriptions of the interactions between men and women, or a woman's or man's reasons for involvement in the violence. Other considerations include the modes of managing fear by men and women. As Stanko (1990) relates, men and women accumulate similar experiences about fear and safety throughout their lifetimes. However, men are taught to keep their distance from fights, to stand their ground when threatened, or to initiate aggression as a way of protecting or establishing a hierarchy. Women, on the other hand, are taught to regularly restrict their activities and to actively monitor their surroundings for danger (Stanko 1990:110-22). Researchers must account for the actions of men and women in relation to fear and safety, in order to gain a fuller understanding of the context of domestic violence. It is also important for questions to incorporate the fact that not all men are violent, nor are all women victims of violence. Stanko also relates that while some young men learn to confront aggression, others discover how to avoid violence by removing themselves from violent situations, by acquiring skills to diffuse tension or by developing the ability to identify trouble in order to avoid it (Stanko 1990:118). Researchers must  201  provide explanations for nonviolent persons, not just for violent ones. Questions must ascertain whether such persons are able to control their anger and can effectively solve conflicts, or whether they blindly avoid conflict and violence. Researchers need to focus on such questions in order to compare the two groups for a better understanding of domestic violence. In order to face the theoretical challenges in studying domestic violence, it is necessary to acknowledge the existence of different types of relationships. I have outlined an example of the different types of relationships in the following chart, in which both men and women can be victims and perpetrators of violence or not involved in violence at all.  A's level of violence Often violent Sometimes violent Nonviolent  B's involvement Yes No mutual A violent only violence A violent mutual only violence B violent "normal" only  To support my breakdown, Riggs (1993) reported in his study of men and women in violent dating relationships that much of the aggression reported by both men and women was mutual. Only 10% of the men and 15% of the women who reported any dating aggression indicated that the man was exclusively the violent member of the relationship. In comparison, 26% of the men and 30% of the women who reported any aggression indicated that only the female member of the couple was physically aggressive (Riggs 1993:26-7).  202 It is necessary to make a separation between men who engage in violence at varying levels, or not at all, in order for the field of domestic violence research to progress. Most importantly, this separation is necessary because several studies of male violence indicate that different variables predict mild versus severe aggression (O'Leary 1993:27). However, it is also important to separate women into groups of varying levels of violence and nonviolence. For example, there is a presumption of homogeneity among women which is problematic for issues related to physical challenge and sexual orientation (Creese and Strong-Boag 1992:6). Researchers must also separate women into categories because theorists and researchers reflect upon the home as the central site of women's oppression. However, for women living in racist environments, the family is a source of support and sanctuary (Creese and Strong-Boag 1992:5). Researchers and theorists must acknowledge and address the heterogeneous nature of men and women, in order to expand our knowledge about domestic violence. It is also important to acknowledge men's and women's use of violence, without dismissing its importance. Although men's violence can have more injurious outcomes for women, women's violence cannot be ignored if prevention is desired. Women cannot expect their husbands or partners to use nonviolent means to resolve conflicts if they themselves engage in violence, especially in contexts where women are the initiators or main aggressors of violence. Violence has been demonstrated to beget violence (Cahn 1996:19) and, therefore, must be viewed as a problem for both men and women. Men and women must work together in order to further the treatment and prevention of violence in relationships. Another important consideration for theorists of domestic violence is the  203 existence of violence in same-sex relationships. A number of studies, such as Bologna, Waterman and Dawson (1987), Kelly and Warshafsky (1987), Myers (1989), and Renzetti (1993), have documented the abuse of lesbian women by other women, raising new questions for researchers (McHugh 1993). Women's violence in these situations cannot be dismissed in terms of self-defense or retaliation, and must be addressed accordingly. Other studies have documented violence in gay and bisexual male relationships, such as Letellier (1994) and Island and Letellier (1991). More research is needed on violence in homosexual relationships in order to make connections, if any, to violent heterosexual relationships. These studies will further define the boundaries of theories of domestic violence. Perhaps the most important revision to future research in domestic violence involves the consideration of multiple factors in discussions of violence in interpersonal relationships. As suggested by DeKeseredy and MacLean (1990), a more intensive examination of the entire context of domestic violence is necessary (p. 25). As revealed by the theoretical review in Chapter One, each theory and approach is necessary, but not sufficient, to address the complexity of domestic violence. This complexity can only be addressed by creating a synthesis among various approaches. For example, the heterogeneity of the batterer population has been empirically established. Therefore, it appears more productive to acknowledge the contributions of multiple psychological, interpersonal, and social variables which, over time, compose the human adult and influence his or her interpersonal behaviors (Carden 1994:573). One of the purposes of sociology is to make the "invisible more visible;" this can only be accomplished through further research. The theories need to contribute to a better  204 understanding of domestic violence, basing interpretations on lived experience. But the purpose of sociology is not just to understand the world, but to change it for the better (Denzin 1989:33). Research needs to formulate treatment and prevention programs with the information to encourage men and women to reject violence and embrace a more open line of communication and conflict resolution. Assertiveness should be encouraged over aggression and passivity. Other recommendations include being aware of the interactional context of relationships characterized by verbal and nonverbal abuse, realizing that certain relationship structures are conducive to violence, and appreciating the fact that due to reciprocal interactions, violence may serve a functional, but harmful, role in the maintenance of some relationships (Cahn 1996:18-9). More research needs to accommodate the interactionist perspective, to further determine the role of communication in violent and abusive relationships. In order for the research to be more valid and reliable, the theoretical assumptions underlying the research must be assessed further. Theories should constantly be tested, refuted or revised. Personal biases of researchers and the influence of government or university funding have prevented many researchers from challenging the current theories of domestic violence, leaving many hypotheses untested. Hypotheses left untested and theories left unrevised or unrefuted mean that the field of domestic violence research does not advance our understanding of domestic violence. Researchers become guilty of reinterpreting counter-evidence in order to make it fit their personal or otherwise chosen paradigms (Karapin 1986:240), rather than re-assessing their personal biases or challenging the purpose of their funding source. The complexity of domestic violence is left unexplored. The desire for truth and a full understanding of domestic violence must  205 be rekindled and reflected in future research. Until then, violence and abuse will be left unhindered to continue its damning cycle which rules and threatens so many lives.  206 BIBLIOGRAPHY  Alani, A. A. 1976. "The Battered Husband." British Journal of Psychiatry 129:96. Ansberry, C. 1988. "Calling Sexes Equal in Domestic Violence, Article Stirs Clash Among Rights Group." Wall Street Journal 5 May:27. Arias, I. and S. R. H. Beach. 1987. "Validity of Self-Reports of Marital Violence." Journal of Family Violence 2:139-49. Arias, I. and P. Johnson. 1989. "Evaluations of Physical Aggression Among Intimate Dyads." Journal of Interpersonal Violence 4:298-307. Arias, I., M. Samios and K. D. O'Leary. 1987. "Prevalence and Correlates of Physical Aggression During Courtship." Journal of Interpersonal Violence 2:82-90. Avakame, E. F. 1993. Explaining Domestic Violence. Unpublished doctoral dissertation. Department of Sociology. University of Alberta. Babbie, E. 1973. Survey Research Methods. Belmont, CA: Wadsworth. . 1989. The Practice of Social Research. 5th ed. Belmont, CA: Wadsworth. Bachman, J. G. and P. M. O'Malley. 1989. "When Four Months Equal a Year: Inconsistencies in Student Reports of Drug Use." Pp. 173-85 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Bandura, A. 1969. Principles of Behavior Modification. New York: Holt, Rinehart and Winston. . 1977. Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall. Bard, M. and J. Zacker. 1974. "Assaultiveness and Alcohol Use in Family Disputes." Criminology 12:281-92. Barling, J., K. D. O'Leary, E. N. Jouriles, D. Vivian and K. E. MacEwen. 1987. "Factor Similarity of the Conflict Tactics Scales Across Samples, Spouses, and Sites: Issues and Implications." Journal of Family Violence 2:37-54. Barnes, G. E., L. Greenwood and R. Sommer. 1991. "Courtship Violence in a Canadian Sample of Male College Students." Family Relations 40:37-44.  207 Barnett, O. W. and L. K. Hamberger. 1992. " The Assessment of Maritally Violent Men on the California Personality Inventory." Violence and Victims 7:15-28. Barnett, O. W., M. Kyson and R. E. Thelen. 1992. "Women's Violence as a Response to Male Abuse". Paper presented at the meeting of the American Psychological Association, Washington, DC, August. Beasley, R. and C. D. Stoltenberg. 1992. "Personality Characteristics of Male Spouse Abusers." Professional Psychology 23:310-71. Ben-David, S. 1993. "The Two Facets of Female Violence: The Public and the Domestic Domains." Journal of Family Violence 8:345-59. Berk, R. A., S. F. Berk, D. R. Loseke and D. Rauma. 1983. "Mutual Combat and Other Family Violence Myths." Pp. 197'-212 in The Dark Side of Families: Current Family Violence Research, edited by D. Finkelhor, R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills: Sage. Berliner, L. 1990. "Domestic Violence: A Humanist or Feminist Issue?" Journal of Interpersonal Violence 5:128-29. Bern, E. H. 1985. "Alcohol Use and Spousal Violence: Implications for Social Service Intervention." Response 8:12-14. Bernard, M. L. and J. L. Bernard. 1983. "Violent Intimacy: The Family as a Model for Love Relationships." Family Relations 32:283-86. Bernard, J. L., S. L. Bernard and M. L. Bernard. 1985. "Courtship Violence and SexTyping." Family Relations 34:573-76. Bersani, C. A. and H. T. Chen. 1986. "Sociological Perspectives in Family Violence." Pp. 57-86 in Handbook of Family Violence, edited by V. B. Van Hasselt, R. L. Morrison, A. S. Bellack and M. Hersen. New York: Plenum. Bersani, C. A., H. T. Chen, B. F. Pendleton and R. Denton. 1992. "Personality Traits of Convicted Male Batterers." Journal of Family Violence 7:123-34. Bethke, T. M. and D. M. DeJoy. 1993. "An Experimental Study of Factors Influencing the Acceptability of Dating Violence." Journal of Interpersonal Violence 8:3651. Bickart, B. and E. M. Felcher. 1996. "Expanding and Enhancing the Use of Verbal Protocols in Survey Research." Pp. 115-42 in Answering Questions: Methodology for Determining Cognitive and Communicative Processes in Survey Research, edited by N. Schwarz and S. Sudman. San Francisco: Jossey-Bass.  208  Billingham, R. E. and A. R. Sack. 1987. "Courtship Violence and the Interactive Status of the Relationship." Journal of Adolescent Research 3:315-25. Bland, R. and H. Orn. 1986. "Family Violence and Psychiatric Disorder." Canadian Journal of Psychiatry 31:129-37. Blum, R. H. 1981. "Violence, Alcohol, and Setting: An Unexplored Nexus." Pp. 11042 in Drinking and Crime: Perspectives on the Relationships Between Alcohol Consumption and Criminal Behavior, edited by J. J. Collins, Jr. New York: Guilford. Blumer, H. 1954. "What is Wrong with Social Theory?" American Sociological Review 19:3-10. Bograd, M. 1990. "Why We Need Gender to Understand Human Violence." Journal of Interpersonal Violence 5:132-35. Bologna, H. J., C. K. Waterman and C. J. Dawson. 1987. "Violence in Gay Male and Lesbian Relationships: Implications for Practitioners and Policy Makers." Paper presented at the Third National Conference for Family Violence Researchers, Durham, NH. Bordin, R. 1981. Woman and Temperance: A Quest for Power and Liberty, 1873-1900. Philadelphia: Temple University. Bowker, L. H. 1983. "Marital Rape: A Distinct Syndrome?" Social Casework: The Journal of Contemporary Social Work 64:347-52. Bradburn, N. M. 1983. "Response Effects." Pp. 289-328 in Handbook of Survey Research, edited by P. H. Rossi, J. D. Wright and A. Anderson. New York: Academic. Bradbury, T. N. and F. D. Fincham. 1990. "Dimensions of Marital and Family Interaction." Pp. 37-60 in Handbook of Family Measurement Techniques, edited by J. Touliatos, B. F. Perlmutter and M. A. Straus. Newbury Park: Sage. Breines, W. and L. Gordon. 1983. "The New Scholarship on Family Violence." Signs 8:490-531. Brinkerhoff, M. and E. Lupri. 1988. "Interspousal Violence." The Canadian Journal of Sociology 13:407-34. Brisson, N. J. 1981. "Battering Husbands: A Survey of Abusive Men." Victimology 6:338-44.  209  Browne, A. 1987. When Battered Women Kill. London: Collier Macmillan. . 1989. "Are Women As Violent As Men?" Commentary presented at the Annual Meeting of the American Society of Criminology, Reno, NV. Browne, A. and D. G. Dutton. 1990. "Escape from Violence: Risks and Alternatives for Abused Women." Pp. 67-91 in Family Violence: Perspectives in Research and Practice, edited by R. Roesch, D. G. Dutton and V. F. Sacco. Burnaby, BC: Simon Fraser University. Brown, B. W. 1980. "Wife-Employment, Marital Equality and Husband-Wife Violence." Pp. 176-87 in The Social Causes of Husband-Wife Violence, edited by M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota. Browning, J. and D. Dutton. 1986. "Assessment of Wife Assault with the Conflict Tactics Scale: Using Couple Data to Quantify the Differential Reporting Effect." Journal of Marriage and the Family 48:375-79. Brush, L. D. 1990. "Violent Acts and Injurious Outcomes in Married Couples: Methodological Issues in the National Survey of Families and Households." Gender and Society 4:56-67. Burke, P. J., J. E. Stets and M. A. Pirog-Good. 1989. "Gender Identity, Self-Esteem and Physical and Sexual Abuse in Dating Relationships." Pp. 72-93 in Violence in Dating Relationships, edited by M. A. Pirog-Good and J. E. Stets. New York: Praeger. Burton, S. and E. Blair. 1991. "Task Conditions, Response Formulation Processes, and Response Accuracy for Behavioral Frequency Questions in Surveys." Public Opinion Quarterly 55:50-79. Caesar, P. L. 1988. "Exposure to Violence in the Families-of-Origin Among WifeAbusers and Maritally Non-Violent Men." Violence and Victims 3:49-63. Cahn, D. D. 1996. "Family Violence From A Communication Perspective." Pp. 1-19 in Family Violence From a Communication Perspective, edited by D. D. Cahn and S.A.Lloyd. Thousand Oaks, CA: Sage. Calzavara, L. 1988. "Trends and Policy in Employment Opportunities for Women." Pp. 287-300 in Social Inequality in Canada: Patterns, Problems, Policies, edited.by J. Curtis, E. Grabb, N. Guppy and S. Gilbert. Scarborough, ON: Prentice-Hall Canada.  210 Cannell, C. F., K. H. Marquis and A. Laurent. 1977. "A Summary of Studies of Interviewing Methodology." Vital and Health Statistics, US Public Health Service, Series 2, No. 69. Cantos, A. L., P. H. Neidig and K. D. O'Leary. 1993. "Men and Women's Attributions of Blame for Domestic Violence." Journal of Family Violence 8:289-302. . 1994. "Injuries of Women and Men in a Treatment Program for Domestic Violence." Journal of Family Violence 9:113-24. Cappel, C. and R. B. Heiner. 1990. "The Intergenerational Transmission of Family Aggression." Journal of Family Violence 5:135-52. Carden, A. D. 1994. "Wife Abuse and the Wife Abuser: Review and Recommendations." The Counseling Psychologist 22:539-82. Carmines, E. A. and R. A. Zeller. 1979. Reliability and Validity Assessment. Beverly Hills, CA: Sage. Carter, J., W. A. Stacey and A. W. Shupe. 1988. "Male Violence Against Women: Assessment of the Generational Transfer Hypothesis." Deviant Behavior 9:25973. Cascardi, M. and D. Vivian. 1995. "Context for Specific Episodes of Marital Violence: Gender and Severity of Violence Differences." Journal of Family Violence 10:265-93. Cascardi, M., J. Langhinrichsen and D. Vivian. 1992. "Marital Aggression: Impact, Injuries and Health Correlates for Husbands and Wives." Archives of Internal Medicine 152:1178-84. Cascardi, M., D. Vivian and S. L. Meyer. 1991. "Context and Attributions for Marital Violence in Discordant Couples." Paper presented at the meeting of the Association for the Advancement of Behavior Therapy, New York, NY, November. Cascardi, M. D., K. D. O'Leary, E. E. Lawrence and K. A. Schlee. 1995. "Characteristics of Women Physically Abused by their Spouses and Who Seek Treatment Regarding Marital Conflict." Journal of Consulting and Clinical Psychology 63:616-23. Cate, R., J. Henton, J. Koval, F. S. Christopher and S. Lloyd. 1982. "Premarital Abuse: A Social Psychological Perspective." Journal of Family Issues 3:79-91.  211  Caulfield, M. B. and D. Riggs. 1992. "The Assessment of Dating Aggression: Empirical Evaluation of the Conflict Tactics Scale." Journal of Interpersonal Violence 7:549-58. Chatterton, M. R. 1976. "The Social Contexts of Violence." Pp. 26-49 in Violence in the Family, edited by M. Borland. Atlantic Heights, NJ: Humanities. Christensen, A. and D. C. Nies. 1980. "The Spouse Observation Checklist: Empirical Analysis and Critique." The American Journal of Family Therapy 8:69-79. Clifton, J. 1982. "Factors Predisposing Family Members to Violence." Social Work Services Group, Violence in the Family and Practice in Social Work, Edinburgh, Her Majesty's Stationery Office. Cohen, B. P. 1989. Developing Sociological Knowledge: Theory and Method. 2nd ed. Chicago: Nelson-Hall. Coleman, D. H. and M. A. Straus. 1983. "Alcohol Abuse and Family Violence." Pp. 104-24 in Alcohol, Drug Abuse and Aggression, edited by E. Gottheil, K. A. Druley, T. E. Skoloda and H. M. Waxman. Springfield, IL: Charles C. Thomas. Coleman, K. H. 1980. "Conjugal Violence: What 33 Men Report." Journal of Marital and Family Therapy 6:207-13. Corvo, K. N. 1992. Attachment and Violence in the Families-of-Origin of Domestically Violent Men. Unpublished doctoral dissertation. Department of Social Work. Case Western Reserve University. Crawford, M. and R. Gartner. 1992. Women Killing: Intimate Femicide in Ontario, 1974-1990. Toronto: Women We Honour Action Committee. Creese, G. and V. Strong-Boag. 1992. "Introduction: Taking Gender into Account in British Columbia." Pp. 1-20 in British Columbia Reconsidered: Essays on Women, edited by G. Creese and V. Strong-Boag. Vancouver: Press Gang. Crenshaw, K. W. 1995. "Panel Presentation on Cultural Battery." The University of Toledo Law Review 25:891-902. Crowley, B. L. 1994. The Road to Equity: Gender, Ethnicity and Language. Toronto: Stoddart. Davidson, T. 1978. Conjugal Crime: Understanding and Changing the Wife-Beating Pattern. New York: Hawthorne.  212 Deal, J. E. and K. S. Wampler. 1986. "Dating Violence: The Primacy of Previous Experience." Journal of Social and Personal Relationships 3:457-71. DeKeseredy, W. 1988a. Woman Abuse in Dating Relationships: The Role of Male Peer Support. Toronto: Canadian Scholars. . 1988b. "Woman Abuse in Dating Relationships: A Critical Evaluation of Research and Theory." International Journal of Sociology of the Family 18:7996. . 1989. "In Defense of Self-Defense: Demystifying Female Violence Against Male Intimates." Paper presented at the Annual Meeting of the American Society of Criminology, Reno, Nevada. . 1993. Four Variations of Family Violence: A Review of Sociological Research. Ottawa: National Clearinghouse on Family Violence. . 1994. "Addressing the Complexities of Woman Abuse in Dating: A Response to Gartner and Fox." Canadian Journal of Sociology 19:75-80. DeKeseredy, W. S. and R. Hinch. 1991. Woman Abuse: Sociological Perspectives. Toronto: Thompson Educational. DeKeseredy, W. S. and K. D. Kelly. 1992. The Incidence and Prevalence of Woman Abuse in Canadian University and College Dating Relationships: Preliminary Survey. Ottawa: Report prepared for Health and Welfare Canada's Family Violence Prevention Division. Department of Sociology and Anthropology, Carleton University. . 1993a. "The Incidence and Prevalence of Woman Abuse in Canadian University and College Dating Relationships." Canadian Journal of Sociology 18:137-59. . 1993b. "Woman Abuse in University and College Dating Relationships: The Contribution of the Ideology of Familial Patriarchy." The Journal of Human Justice 4:25-51. DeKeseredy, W. and B. MacLean. 1990. "Researching Women Abuse in Canada: A Realist Critique of the Conflict Tactics Scale." Canadian Review of Social Policy 25:19-27. DeKeseredy, W., K. Kelly and B. Baklid. 1992. "The Physical, Sexual and Psychological Abuse of Women in Dating Relationships: Results from a Pretest for a National Study." Paper presented at the annual meeting of the American Society of Criminology, New Orleans.  213  Denzin, N. K. 1989. The Research Act: A Theoretical Introduction to Sociological Methods. Englewood Cliffs, NJ: Prentice Hall. Deschner, J. P. 1984. The Hitting Habit. New York: Free Press. de Vaus, D. A. 1986. Surveys in Social Research. 2nd ed. London: Unwin Hyman. DeVellis, R. F. 1991. Scale Development: Theory and Applications. Newbury Park, CA: Sage. Dillman, D. A. 1978. Mail and Telephone Surveys: The Total Design Method. New York: John Wiley and Sons. Dines, G. 1992. "Pornography and the Media: Cultural Representations of Violence Against Women." Family Violence and Sexual Assault Bulletin 8:17-20. Dobash, R. and R. Dobash. 1988. "Research as Social Action: The Struggle for Battered Women." Pp. 51-74 in Feminist Perspectives on Wife Abuse, edited by K. Y116 and M. Bograd. Newbury Park: Sage. Dobash, R., R. E. Dobash, M. Wilson and M. Daly. 1992. "The Myth of Sexual Symmetry in Marital Violence." Social Problems 39:71-91. Dobash, R. E. and R. P. Dobash. 1979. Violence Against Wives: A Case Against Patriarchy. New York: Free Press. . 1981. "Social Science and Social Action: The Case of Wife Beating." Journal of Family Issues 2.4:439-70. . 1983. "Unmasking the Provocation Excuse." Aegis 37:67-8. Dobash, R. E., R. P. Dobash and K. Cavanagh. 1985. "The Contact Between Battered Women and Social and Medical Agencies." Pp. 142-65 in Private Violence and Public Policy, edited by J. Pahl. London: Routledge and Kegan Paul. Dobash, R. P. and R. E. Dobash. 1983. "The Context-Specific Approach." Pp. 261-76 in The Dark Side of Families: Current Family Violence Research, edited by R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills: Sage. . 1992. Women, Violence and Social Change. New York: Routledge. Dollard, J., L. W. Doob, N. E. Miller, O. H. Mowrer and R. R. Sears. 1939. Frustration and Aggression. New Haven, CT: Yale University.  214 Doob, A. N. 1995. "Understanding the Attacks on Statistics Canada's Violence Against Women Survey." Pp. 157-65 in Wife Assault and the Canadian Criminal Justice System, edited by M. Valverde, L. MacLeod and K. Johnson. Toronto: Centre of Criminology. Dougherty, J. 1993. "Women's Violence Against Their Children: A Feminist Perspective." Women and Criminal Justice 4:91-114. Douglas, M. A. and M. Der Ovanesian. 1983. "A Comparative Study of Battered and NonBattered Women in Crisis." Paper presented at the 17th Annual Convention of the Association for the Advancement of Behavior Therapy, Washington, DC. Durst, M. 1987. "Perceived Peer Abuse Among College Students: A Research Note." National Association of Student Personnel Administrations Journal 24:42-47. Dutton, D. G. 1985. "An Ecologically Nested Theory of Male Violence Toward Intimates." International Journal of Women's Studies 8:404-13. . 1988. The Domestic Assault of Women. Boston: Allyn and Bacon. . 1994a. "Patriarchy and Wife Assault: The Ecological Fallacy." Violence and Victims 9:167-82. . 1994b. "Behavioral and Affective Correlates of Borderline Personality Organization in Wife Assaulters." International Journal of Law and Psychiatry 17:265-79. . 1995. "A Scale for Measuring Propensity for Abusiveness." Journal of Family Violence 10:203-21. Dutton D. G. and J. J. Browning. 1983. "Violence in Intimate Relationships." Paper presented at the International Society for Research on Aggression Conference, Victoria, BC, July. Dutton, D. G. and S. D. Hart. 1992a. "Risk Markers for Family Violence in a Federally Incarcerated Population." International Journal of Law and Psychiatry 15:10112. . 1992b. "Evidence for Long-Term, Specific Effects of Childhood Abuse and Neglect on Criminal Behavior in Men." International Journal of Offender Therapy and Comparative Criminology 36:129-38. Dutton, D. G. and K. Hemphill. 1992. "Patterns of Socially Desirable Responding Among Perpetrators and Victims of Wife Assault." Violence and Victims 7:2940.  215  Dutton, D. G. and A. Starzomski. 1994. "Psychological Differences Between CourtReferred and Self-Referred Wife Assaulters." Criminal Justice and Behavior 21:203-22. Dutton, D. G., K. Saunders, A. Starzomski and K. Bartholomew. 1994. "IntimacyAnger and Insecure Attachment as Precursors of Abuse in Intimate Relationships." Journal of Applied Social Psychology 24:1367-1386. Edleson, J. L. and M. P. Brygger. 1986. "Gender Differences in Reporting of Battering Incidences." Family Relations 35:377-82. Eichler, M. 1988. Nonsexist Research Methods: A Practical Guide. Boston: Allen and Unwin. Eichler, M. and J. Lapointe. 1985. On the Treatment of the Sexes in Research. Ottawa: Minister of Supply and Services Canada. Elliot, S., D. Odynak and H. Krahn. 1992. A Survey of Unwanted Sexual Experiences Among University of Alberta Students. Research report prepared for the Council on Student Life, University of Alberta. University of Alberta: Population Research Laboratory. Everett, H. C. 1980. "Battered Husbands." American Journal of Psychiatry 137:5. Faulk, M. 1974. "Men Who Assault Their Wives." Medicine, Science and the Law 14:180-83. Felcher, E. M. and B. Calder. 1991. "Answering Survey Questions: The Case of Behavioral Frequency Data." Paper presented at the Conference of the American Association for Public Opinion Research, Phoenix, AZ, May. Ferraro, K. J. 1984. "An Existential Approach to Battering." Paper presented at the Second Conference for Family Violence Researchers, Durham, NH. Ferraro, K. J. and J. M. Johnson. 1984. "The Meaning of Courtship Violence." Paper presented at the 2nd National Conference for Family Violence Researchers, Durham, NH, August. Fesbach, S. 1964. "The Function of Aggression and the Regulation of Aggressive Drive." Psychological Review 103:119-31. Finkelhor, D. 1979. Sexually Victimized Children. New York: Free Press.  216 . 1984. Child Sexual Abuse: New Theory and Research. New York: Free Press. Finkelman, L. 1992. Report of the Survey of Unwanted Sexual Experiences Among Students of UNBF and STU. University of New Brunswick: Counselling Services. Flanzer, J. 1987. "Double Trouble: Alcoholism and Family Violence." Pp. 33-38 in Growing in the Shadow: Children of Alcoholics, edited by R. J. Ackerman. Pompano Beach, FL: Health Communications. Foddy, W. 1992. Constructing Questions for Interviews and Questionnaires; Theory and Practice in Social Reserach. Cambridge: Cambridge University. Fowler, F. J., Jr. 1988. Survey Research Methods. Newbury Park, CA: Sage. . 1993. Survey Research Methods, 2nd. ed. Newbury Park, CA: Sage. . 1995. Improving Survey Questions: Design and Evaluation. Thousand Oaks, CA: Sage. Fowler, F. J., Jr. and C. F. Cannell. 1996. "Using Behavioral Coding to Identify Cognitive Problems with Survey Questions." Pp. 15-36 in Answering Questions: Methodology for Determining Cognitive and Communicative Processes in Survey Research, edited by N. Schwarz and S. Sudman. San Francisco: Jossey-Bass. Fox, B. J. 1993. "On Violent Men and Female Victims: A Comment on DeKeseredy and Kelly." Canadian Journal of Sociology 18:320-25. Fox, S. B. 1980. Response to Spousal Violence as a Function of Sex-Role Identity and Family Role Salience. Unpublished doctoral dissertation. Department of Sociology. St. Louis University. Frankel, M. 1983. "Sampling Theory." Pp. 21-67 in Handbook of Survey Research, edited by P. H. Rossi, J. D. Wright and A. B. Anderson. San Diego: Academic. Frankel-Howard, D. 1989. Family Violence: A Review of Theoretical and Clinical Literature. Ottawa: Minister of Supply and 'Services. Freeman, M. D. A. 1979. Violence in the Home. Westhead, ENG: Saxon House. Freud, S. 1933. New Introductory Lectures on Psycho-Analysis. New York: Morton.  217 Frieze, I. H. 1980. "Causes and Consequences of Marital Rape." Paper presented at the American Psychological Association meeting. Montreal, Quebec. Garofalo, J. 1979. "Victimization and the Fear of Crime." Journal of Research in Crime and Delinquency 16:80-97. Gartner, R. 1993. "Studying Woman Abuse: A Comment on DeKeseredy and Kelly." Canadian Journal of Sociology 18:313-20. Gayford, J. J. 1975. "Wife Battering: A Preliminary Survey of 100 Cases." British Medical Journal 25:194-97. Gelles, R. J. 1973. "Child Abuse as Psychopathology: A Sociological Critique and Reformulation." American Journal of Orthopsychiatry 43:611-21. . 1974. The Violent Home. Beverly Hills, CA: Sage. . 1976. "Abuse Wives: Why Do They Stay?" Journal of Marriage and The Family 38:659-668. . 1978. "Violence in the American Family." Pp. 169-82 in Violence and the Family, edited J. P. Martin. New York: John Wiley and Sons. . 1983. "An Exchange/Social Control Theory." Pp. 151-65 in The Dark Side of Families: Current Family Violence Research, edited D. Finkelhor, R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills, CA: Sage. . 1990. "Methodological Issues in the Study of Family Violence." Pp. 17-28 in Physical Violence in American Families: Risk Factors and Adaptations to Violence in 8,145 Families, edited by M. A. Straus and R. J. Gelles. New Brunswick, NJ: Transaction. . 1993. "Through a Sociological Lens: Social Structure and Family Violence." Pp. 31-46 in Current Controversies in Family Violence, edited by R. J. Gelles and D. R. Loseke. Newbury Park: Sage. Gelles, R. J. and M. A. Straus. 1988. Intimate Violence. New York: Simon and Schuster. Gil, D. G. 1970. Violence Against Children: Physical Child Abuse in the United States. Cambridge, MA: Harvard University. Giles-Sims, J. 1983. Wife-Beating: A Systems Theory Approach. New York: Guildord.  218 Gillespie, D. L. 1971. "Who Has the Power? The Marital Struggle." Journal of Marriage and the Family 33:445-58. Gleason, W. J. 1993. "Mental Disorders in Battered Women: An Empirical Study." Violence and Victims 8:53-68. Gondolf, E. W. 1985. Men Who Batter: An Integrated Approach for Stopping Wife Abuse. Florida: Learning Publications. . 1988. "Who Are These Guys? Toward Behavioral Typology of Batterers." Violence and Victims 3:187-203. Goode, W. 1971. "Force and Violence in the Family." Journal of Marriage and the Family 33:624-36. Gottheil, E., K. A. Druley, T. E. Skoloda and H. M. Waxman. 1983. "Aggression and Addiction: Summary and Overview." Pp. 333-56 in Alcohol, Drug Abuse and Aggression, edited by E. Gottheil, K. A. Druley, T. E. Skoloda and H. M. Waxman. Springfield, IL: Charles C. Thomas. Government of Canada. 1990. Family Violence in Canada: A Call to Action. Ottawa: Ministry of Supply and Services. Grant, J. 1993. Fundamental Feminism: Contesting the Core Concepts of Feminist Theory. New York: Routledge. Gray, G. and N. Guppy. 1992. Successful Surveys. Vancouver: University of British Columbia. . 1994. Successful Surveys: Research Methods and Practice. Toronto: Harcourt Brace. Greenblat, C. S. 1983. "A Hit is a Hit... Or is it? Approval and Tolerance of the Use of Physical Force by Spouses." Pp. 235-60 in The Dark Side of Families: Current Family Violence Research, edited D. Finkelhor, R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills, CA: Sage. Groves, R. M. and R. L. Kahn. 1979. Surveys by Telephone: A National Comparison With Personal Interviews. New York: Academic. Groves, R. M. and L. J. Magilavy. 1989. "Measuring and Explaining Interviewer Effects in Centralized Telephone Surveys." Pp. 288-303 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: University of Chicago.  219  Hamberger, L. K. 1991 a. "Characteristics and Context of Women Arrested for Domestic Violence: Context and Implications." Marital Violence: Theoretical and Empirical Perspectives, Indiana University Conference on Research and Clinical Problems, Bloomington, IN, May. . 1991b. "Research Concerning Wife Abuse: Implications for Training Police and Physicians." In R. Geffner (Chair), State-of-the-Art Research in Family Violence: Practical Applications. Symposium presented at the meeting of the American Psychological Association, San Francisco, CA, August. . 1994. "Domestic Partner Abuse: Expanding Paradigms for Understanding and Intervention." Violence and Victims 9.2:91-94. Hamberger, L. K. and J. E. Hastings. 1986a. "Personality Correlates of Men Who Abuse Their Partners: A Cross-Validation Study." Journal of Family Violence 1:32342. . 1986b. "Characteristics of Male Spouse Abusers: Is Psychopathology Part of the Picture?" Paper presented at American Society of Criminology, Atlanta, GA. . 1991. "Personality Correlates of Men Who Batter and Nonviolent Men: Some Continuities and Discontinuities." Journal of Family Violence 6:131-47. Hamberger, L. K. and S. D. Hastings. 1989. "Counseling Male Spouse Abusers: Characteristics of Treatment Completers and Dropouts." Violence and Victims 4:275-286. Hamberger, L. K. and T. Potente. 1994. "Counseling Heterosexual Women Arrested for Domestic Violence: Implications for Theory and Practice." Violence and Victims 9:125-37. Hamberger, L. K., J. M. Lohr and D. Bonge. 1993. "The Intended Function of Domestic Violence is Different for Male and Female Perpetrators." Paper presented at the meeting of the Association for the Advancement of Behavior Therapy, Atlanta, GA, November. Harding, S. G. 1987. Feminism and Methodology: Social Science Issues. Bloomington: Indiana University. Hanks, S E. and C. P. Rosenbaum. 1977. "Battered Women: A Study of Women Who Live With Violent Alcohol-Abusing Men." American Journal of Orthopsychiatry 47:291-306.  220 Hart, S. D., D. G. Dutton and T. Newlove. 1993. "The Prevalence of Personality Disorder Among Wife Assaulters." Journal of Personality Disorders 7:1721-40. Hastings, J. E. and L. K. Hamberger. 1988. "Personality Characteristics of Spouse Abusers: A Controlled Comparison." Violence and Victims 3:31-48. Hay, D. A. 1993. "Methodological Review of 'The Incidence and Prevalence of Woman Abuse in Canadian University and College Dating Relationships: Preliminary Results from a National Survey: Preliminary Results from a National Survey,' by Walter S. DeKeseredy and Katharine Kelly." Journal of Human Justice 1993:5366. Hekman, S. J. 1990. Gender and Knowledge: Elements of a Postmodern Feminism. Boston: Northeastern University. Henry, G. T. 1990. Practical Sampling. Newbury Park, CA: Sage. Henton, J., R. Cate, J. Koval, S. Lloyd and S. Christopher. 1983. "Romance and Violence in Dating Relationships." Journal of Family Issues 4:467-82. Hepburn, J. R. 1973. "Violence Behavior in Interpersonal Relationships." Sociological Quarterly 14:419-29. Hergenhahn, B. R. 1982. An Introduction to Theories of Learning. Englewood Cliffs, NJ: Prentice-Hall. Herrenkohl, E. C , R. C. Herrenkohl and L. J. Toedter. 1983. "Perspectives on the Intergenerational Transmission of Abuse." Pp. 305-16 in The Dark Side of Families: Current Family Violence Research, edited D. Finkelhor, R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills, CA: Sage. Hindelang, M. J., M. R. Gottfredson and J. Garofalo. 1987. Victims of Personal Crime: An Empirical Foundation for a Theory of Personal Victimization. Cambridge, MA: Ballinger. Hippensteele, S. K., M. Chesney-Lind and R. C. Veniegas. 1993. "Some Comments on the National Survey on Woman Abuse in Canadian University and College Dating Relationships." Journal of Human Justice 4:67-71. Holtzworth-Munroe, A. and K. Anglin. 1991. "The Competency of Responses Given by Maritally Violent Versus Nonviolent Men to Problematic Marital Solutions." Violence and Victims 6:257-69.  221 Hornung, C. A., B. C. McCullough and T. Sugimoto. 1981. "Status Relationships in Marriage: Risk Factors in Spouse Abuse." Journal of Marriage and the Family 43:675-92. Hotaling, G. T. 1980. "Attribution Process in Husband-Wife Violence." Pp. 136-56 in The Social Causes of Husband-Wife Violence, edited M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota. Hotaling, G. T. and D. B. Sugarman. 1986. "An Analysis of Risk Markers in Husband to Wife Violence: The Current State of Knowledge." Violence and Victims 1:10124. . 1990. "A Risk Marker Analysis of Assaulted Wives." Journal of Family Violence 5:1-13. Hotaling, G. T. and M. A. Straus. 1980. "Culture, Social Organization and Irony in the Study of Family Violence." Pp. 3-22 in The Social Causes of Husband-Wife Violence, edited M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota. Huggins, M. D. and M. A. Straus. 1980. "Violence and the Social Structure as Reflected in Children's Books From 1850 to 1970." Pp. 51-67 in The Social Causes of Husband-Wife Violence, edited M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota. Hyden, M. 1995. "Verbal Aggression as Prehistory of Woman Battering." Journal of Family Violence 10:55-71. Island, D. and P. Letellier. 1991. Men Who Beat The Men Who Love Them: Battered Gay Men and Domestic Violence. Binghamton, NY: Haworth. Jacobs, P. A., M. Brunton and M. M. Melville. 1965. "Aggressive Behavior, Mental Sub-normality and the XYY Male." Nature 208:1351-1352. Jacobson, N. S. and D. Moore. 1981. "Spouses as Observers of the Events in their Relationship." Journal of Consulting and Clinical Psychology 49:269-77. Jarvie, I. 1991. "Pornography and/as Degradation." International Journal of Law and Psychiatry 14:13-27. Joas, H. 1987. "Symbolic Interaction." Pp. 82-115 in Social Theory Today, edited by A. Giddens and J. T. Turner. Stanford: Stanford University.  222 Johnson, H. 1995a. "Seriousness, Type and Frequency of Violence Against Wives." Pp. 125-47 in Wife Assault and the Canadian Criminal Justice System, edited by M. Valverde, L. MacLeod and K. Johnson. Toronto: Centre of Criminology. __. 1995b. "Response to Allegations Made About the Violence Against Women Survey." Pp. 148-56 in Wife Assault and the Canadian Criminal Justice System, edited by M. Valverde, L. MacLeod and K. Johnson. Toronto: Centre of Criminology. Johnson, H. and V. Sacco. 1995. "Researching Violence Against Women: Statistics Canada's National Survey." Canadian Journal of Criminology 37: 281-304. Johnson, H., D. Norris and D. Paton. 1994. Violence Against Women Survey: Planning Document. Ottawa: Ministry of Supply and Services. Johnson, M. P. 1995. "Patriarchal Terrorism and Common Couple Violence: Two Forms of Violence Against Women." Journal of Marriage and the Family 57:283-94. Jones, R. K. 1993. "Female Victim Perceptions of the Causes of Male Spouse Abuse." Sociological Inquiry 63:351-61. Jouriles, E. N. and K. D. O'Leary. 1985. "Interspousal Reliability of Reports of Marital Violence." Journal of Consulting and Clinical Psychology 53:419-21. Julian, T. W. and P. C. McKenry. 1993. "Mediators of Male Violence Toward Female Intimates." Journal of Family Violence 8:39-56. Jurik, N. C. and R. Winn. 1990. "Gender and Homicide: A Comparison of Men and Women Who Kill." Violence and Victims 5:227-41. Kalmuss, D. 1984. "The Intergenerational Transmission of Marital Aggression." Journal of Marriage and the Family 46:11 -19. Kalmuss, D. and J. A. Seltzer. 1986. "Continuity of Marital Behavior in Remarriage: The Case of Spouse Abuse." Journal of Marriage and the Family 48:113-20. Kanin, E. 1957. "Male Aggression in Dating-Courtship Relations." American Journal of Sociology 6:179-204. Kantor, G. K. and M. A. Straus. 1990. "The "Drunken Bum" Theory of Wife Beating." Pp. 203-24 in Physical Violence in American Families: Risk Factors and Adaptations in 8145 Families, edited by M. A. Straus and R. J. Gelles. New Brunswick, NJ: Transaction.  223 Karapin, R. S. 1986. "What's The Use of Social Science? A Review of the Literature." Pp. 236-65 in The Use and Abuse of Social Science, edited by F. Heller. London: Sage. Keane, C. 1995. "Victimization and Fear: Assessing the Role of Offender and Offence." Canadian Journal of Criminology 37:431-56.. Kelly, E. E. and L. Warshafsky. 1987. "Partner Abuse in Gay Male and Lesbian Couples." Paper presented at the Third National Conference for Family Violence Researchers, Durham, NH, July. Kelly, K. D. 1994. "The Politics of Data." Canadian Journal of Sociology 19:81-85. Kelly, K. and W. S. DeKeseredy. 1994. "Women's Fear of Crime and Abuse in College and University Dating Relationships." Violence and Victims 9:17-30. Kelly, L. 1988. "How Women Define Their Experiences of Violence." Pp. 114-32 in Feminist Perspectives on Family Abuse, edited by K. Y116 and M. Bograd. Newbury Park: Sage. Kennedy, L. W. and D. G. Dutton. 1989. "The Incidence of Wife Assault in Alberta." Canadian Journal of Behavioural Science 21:40-54. Kincaid, P. J. 1982. The Omitted Reality: Husband-Wife Violence in Ontario and Policy Implications for Education. Concord: Belsten. Kinzel, C. 1987. 1987 All Alberta Study: Sampling Report. Edmonton, AB: Population Research Laboratory. Kirkpartick, C. and E. Kanin. 1959. "Male Aggression on a University Campus." American Sociological Review 22:52-58. Kolb, B. and I. Q. Whishaw. 1989. Fundamentals of Human Neuropsychology. 3rd ed. New York: W. H. Freeman. Koss, M. P. and C. A. Gidycz. 1985. "Sexual Experiences Survey: Reliability and Validity." Journal of Consulting and Clinical Psychology 50:455-57. Koss, M. P., C. A. Gidycz and N. Wisniewski. 1987. "The Scope of Rape: Incidence and Prevalence of Sexual Aggression and Victimization in a National Sample of Higher Education Students." Journal of Consulting and Clinical Psychology 55:162-70. Kuhn, A. 1985. The Power of Image. London: Routledge.  224 Labell.L. S. 1979. "Wife Abuse: A Sociological Study of Battered Women and Their Mates." Victimology 4:258-67. Lalu, N. M. 1988. Telephone and Face to Face Interviewing: A Comparison. Edmonton: University of Alberta Population Research Laboratory. Lamanna, M. A. and A. C. Reidman. 1985. Marriages and Families. Belmont, CA: Wadsworth. Lane, K. A. and P. A. Gwartney-Gibbs. 1985. "Violence in the Context of Dating Violence." Journal of Family Issues 6:45-59. Laner, M. R. 1983. "Courtship Abuse and Aggression: Contextual Aspects." Sociological Spectrum 3:69-83. Laner, M. R. and J. Thompson. 1982. "Abuse and Aggression in Courting Couples." Deviant Behavior 3:229-44. Langan, P. A. and C. A. Innes. 1986. Preventing Domestic Violence Against Women. Washington, DC: US Department of Justice, Bureau of Justice Statistics. Lengermann, P. M. and J. Niebrugge-Brantley. 1990. "Feminist Sociological Theory: The Near-Future Prospects." Pp. 316-44 in Frontiers of Social Theory: The New Syntheses, edited by G. Ritzer. New York: Columbia University. Lenton, R. L. 1995. "Power versus Feminist Theories of Wife Abuse." Canadian Journal of Criminology 37:305-30. Leonard, K. E. 1984. "Alcohol Consumption and Escalatory Aggression in Intoxicated and Sober Dyads." Journal of Studies on Alcohol 45:75-80. Letellier, P. 1994. "Gay and Bisexual Male Domestic Violence Victimization: Challenges to Feminist Theory and Response to Violence." Violence and Victims 9:95-106. Levinger, G. 1976. "A Social Psychological Perspective on Marital Dissolution." Journal of Social Issues 32:21-47. Lezak, M. D. 1983. Neuropsychological Assessment. 2nd. ed. New York: Oxford University. Lorenz, K. 1966. On Aggression. New York: Harcourt Brace Jovanovich. Loring, M. T. 1994. Emotional Abuse. MA: Lexington Books.  225 Lupri, E. 1990. "Male Violence in the Home." Pp. 170-72 in Canadian Social Trends, edited by C. McKie and K. Thompson. Toronto: Thompson Education. MacAndrew, C. and R. B. Edgerton. 1969. Drunken Comportment. Chicago: Aldine. MacFarlane, K. 1978. "Sexual Abuse of Children." Pp. 81-110 in The Victimization of Women, edited by J. R. Chapman and M. Gates. Beverly Hills, CA: Sage. MacLeod, L. 1980. Wife Battering in Canada: The Vicious Circle. Ottawa: Advisory Council on the Status of Women. . 1987. Battered But Not Beaten: Preventing Wife Battering in Canada. Ottawa: Advisory Council on the Status of Women. MacLeod, L. and A. Cadieux. 1980. Wife Battering in Canada: The Vicious Circle. Ottawa: Canadian Advisory Council on the Status of Women. Mahlstedt, D., D. J. Falcone and L. Rice-Spring. 1993. "Dating Violence Education: What Do Students Learn?" The Journal of Human Justice 4:101-17. Maiuro, R., T. Cahn and P. Vitaliano. 1986. "Assertiveness Deficits and Hostility in Domestically Violent Men." Violence and Victims 1:279-89. . 1988. "Anger, Hostility, and Depression in Domestically Violent Versus Generally Assaultive Men and Nonviolent Control Subjects." Journal of Consulting and Clinical Psychology 56:17-23. Makepeace, J. M. 1981. "Courtship Violence Among College Students." Family Relations 30:97-102. . 1983. "Life Events, Stress and Courtship Violence." Family Relations 32:101-9. . 1986. "Gender Differences in Courtship Violence Victimization." Family Relations 33:383-88. . 1987. "Social Factor and Victim-Offender Differences in Courtship Violence." Family Relations 36:87-91. Mangione, T. W. 1995. Mail Surveys: Improving the Quality. Thousand Oaks, CA: Sage. Mann, C.R. 1988. "Getting Even? Women Who Kill in Domestic Encounters." Justice Quarterly 5:34-51.  226 Margolin, G. 1988. "Interpersonal and Intrapersonal Factors Associated with Marital Violence." Pp. 203-17 in Family Abuse and its Consequences: New Directions in Research, edited by G. T. Hotaling, D. Finkelhor, J. T. Kirkpatrick and M. A. Straus. Newbury Park: Sage. Margolin, G., L. G. Sibner and L. Gleberman. 1988. "Wife Battering." Pp. 89-117 in Handbook of Family Violence, edited by V. B. Van Hasselt, R. L. Morrison, A. S. Bellack and M. Hersen. New York: Plenum. Marlatt, G. A. and D. J. Rohsenow. 1980. "Cognitive Processes in Alcohol Use: Expectancy and the Balanced Placebo Design." Pp. 159-99 in Advances in Substance Abuse: Behavioral and Biological Research, vol. 1, edited by N. K. Mello. Greenwich, CT: JAI. Marshall, L. L. and P. Rose. 1990. "Premarital Violence: The Impact of Family of Origin Violence, Stress, and Reciprocity." Violence and Victims 5:51-64. Matthews, W. J. 1984. "Violence in College Couples." College Student Journal 18:150-58. Maxfield, M. 1984. Fear of Crime in England and Wales. Home Office Research Study No. 78. London: HMSO. McCall, G. J. and S. M. Shields. 1986. "Social and Structural Factors in Family Violence." Pp. 98-123 in Violence in the Home, edited by M. Lystad. New York: Brunner/Mazel. McKenry, P. C , T. W. Julian and S. M. Gavazzi. 1995. "Toward a Biopsychosocial Model of Domestic Violence." Journal of Marriage and the Family 57:307-20. McHugh, M. C. 1993. "Studying Battered Women and Batterers: Feminist Perspectives on Methodology." Pp. 54-68 in Battering and Family Therapy: A Feminist Perspective, edited by M. Hansen and M. Harway. Newbury Park: Sage. McNeely, R. L. and C. R. Mann. 1990. "Domestic Violence is a Human Issue." Journal of Interpersonal Violence 5:129-32. McNeely, R. L. and G. Robinson-Simpson. 1987. "The Truth About Domestic Violence: A Falsely-Framed Issue." Social Work 32:485-90. Means, B. and E. F. Loftus. 1991. "When Personal History Repeats Itself: Decomposing Memories for Recurring Events." Applied Cognitive Psychology 5:297-318.  227 Miller, P. V. and C. F. Cannell. 1989. "A Study of Experimental Techniques for Telephone Interviewing." Pp. 304-23 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Morley, R. 1994. "Wife Beating and Modernization: The Case of Papua New Guinea." Journal of Comparative Family Studies 25:25-52. Murphy, C. M. and M. Cascardi. 1993. "Psychological Aggression and Abuse in Marriage." Pp. 86-111 in Family Violence: Prevention and Treatment, edited by R. L. Hampton, T. P. Gullotta, G. R. Adams, E. H. Potter HI and R. P. Weissberg. Newbury Park: Sage. Murphy, C. M. and K. D. O'Leary. 1994. "Research Paradigms, Values, and Spouse Abuse." Journal of Interpersonal Violence 9:207-23. Myers, B. A. 1989. Lesbian Battering: An Analysis of Power, Inequality and Conflict in Lesbian Relationships. Unpublished doctoral dissertation. Department of Psychology. Indian University of Pennsylvania. Neidig, P. and D. H. Friedman. 1984. Spouse Abuse: A Treatment Program for Couples. Champaign, JL: Research. O'Brien, J.E. 1974. "Violence in Divorce-Prone Families." Pp. 75-81 in Violence in the Family, edited by S. K. Steinmetz and M. A. Straus. New York: Harper and Row. O'Farrell, T. J. and C. M. Murphy. 1995. "Marital Violence Before and After Alcoholism Treatment." Journal of Consulting and Clinical Psychology 63:25662. Okun, L. 1986. Women Abuse: Facts Replacing Myths. Albany: State University of New York. O'Leary, K. D. 1988a. "Physical Aggression between Spouses: A Social Learning Theory Perspective." Pp. 31-55 in Handbook of Family Violence, edited by V. B. Van Hasselt, R. L. Morrison, A. S. Bellack and M. Hersen. New York: Plenum. . 1988b. "Assessing Agreement of Reports of Spouse Abuse." Pp. 218-27 in Family Abuse and Its Consequences: New Directions in Research, edited by G. T. Hotaling, D. Finkelhor, J. T. Kirkpatrick and M. A. Straus. Newbury Park: Sage.  228 /  . 1993. "Through a Psychological Lens: Personality Traits, Personality Disorders, and Levels of Violence." Pp. 7-30 in Current Controversies on Family Violence, edited by R. J. Gelles and D. R. Loseke. Newbury Park: Sage. O'Leary, K. D. and I. Arias. 1988. "Prevalence, Correlates and Development of Spouse Abuse." Pp. 104-27 in Social Learning and Systems Approaches to Marriage and the Family, edited by R. D. V. Peters and R. J. McMahon. New York: Brunner/Mazel. ^ O'Leary, K. D. and A. D. Curley. 1986. "Assertion and Family Violence: Correlates of Spouse Abuse." Journal of Marital and Family Therapy 12:281-89. O'Leary, K. D. and D. Vivian. 1990. "Physical Aggression in Marriage." Pp. 323-48 in The Psychology of Marriage: Basic Issues and Applications, edited by F. Fincham and T. N. Bradbury. New York: Guilford. O'Leary, K. D., J. Barling, I. Arias, A. Rosenbaum, J. Malone and A. Tyree. 1989. "Prevalence and Stability of Physical Aggression Between Spouses: A Longitudinal Analysis." Journal of Consulting and Clinical Psychology 57:26368. Pagelow, M. D. 1981. Woman-Battering: Victims and their Experiences. Beverly Hills: Sage. Pernanen, K. 1976. "Alcohol and Crimes of Violence." Pp. 351-444 in The Biology of Alcoholism, vol.4: Social Aspects of Alcoholism, edited by B. Kissin and H. Begleiter. New York: Plenum. Pleck, E. H. 1987. Domestic Tyranny: The Making of Social Policy Against Family Violence from Colonial Times to the Present. New York: Oxford University. Pollard, J. 1993. Male-Female Dating Relationships in Canadian Universities and Colleges: Sample Design, Arrangements for Data Collection and Data Reduction. Toronto: Institute for Social Research. Population Research Laboratory. 1987. 1987 All Alberta Study Codebook. Edmonton: University of Alberta. Pressman, B. M. 1983. Family Violence: Origins and Treatment. Guelph: University of Guelph.  229 Reid, J. B., P. S. Taplin and R. Lorber. 1981. "A Social Interactional Approach to the Treatment of Abusive Families." Pp. 68-82 in Violent Behavior: Social Learning Approaches to Prediction, Management and Treatment, edited by R. B. Stuart. New York: Brunner/Mazel. Renzetti, C. M. 1993. "Violence in Lesbian Relationships." Pp. 188-99 in Battering and Family Therapy: A Feminist Perspective, edited by M. Hansen and M. Harway. Newbury Park: Sage. . 1994. "On Dancing With A Bear: Reflections on Some of the Current Debates Among Domestic Violence Theorists." Violence and Victims 9:195-200. Research and Statistics Group. 1984. Canadian Urban Victimization Survey, Summary Technical Report. Ottawa: Ministry of the Solicitor General. Richardson, S. A., B. S. Dohrenwend and D. Klein. 1965. Interviewing: Its Forms and Functions. New York: Basic Books. Riggs, D. S. 1993. "Relationship Problems and Dating Aggression: A Potential Treatment Target." Journal of Interpersonal Violence 8:18-35. Riggs, D. S., C. M. Murphy and K. D. O'Leary. 1989. "Intentional Falsification in Reports of Interpartner Aggression." Journal of Interpersonal Violence 4:220-32. Riggs, D. S., K. D. O'Leary and F. C. Breslin. 1990. "Multiple Correlates of Physical Aggression in Dating Couples." Journal of Interpersonal Violence 5:61-73. Roberts, A. 1993. Self-Defense and Assertiveness Training Program. Vancouver, BC. September - November. Rock, P. 1994. "Introduction." Pp. xi-xix in Victimology, edited by P. Rock. Brookfield, VT: Dartmouth. Rodgers, K. 1994. "Wife Assault: The Findings of a National Survey." Juristat 14:121Roscoe, B. and N. Benaske. 1985. "Courtship Violence Experience by Abused Wives: Similarities in Patterns of Abuse." Family Relations 34:419-24. Rosenbaum, A. and R. Maiuro. 1990. "Perpetrators of Spouse Abuse." Pp. 280-309 in Treatment of Family Violence, edited by R. T. Ammerman and M. Hersen. New York: John Wiley.  230 Rosenbaum, A. and K. D. O'Leary. 1981. "Marital Violence: Characteristics of Abusive Couples." Journal of Consulting and Clinical Psychology 49:63-71. Rounsaville, B. 1978. "Theories in Marital Violence: Evidence from a Study of Battered Women." Victimology 3:11-31. Rouse, L. P. 1984. Conflict Tactics Used by Men in Marital Disputes. Paper presented at the Second Conference for Family Violence Researchers, Durham, NH. Roy, M. 1982. The Abusive Partner: An Analysis of Domestic Battering. New York: Van Nostrand. Russell, D. E. H. 1982. Rape in Marriage. New York: Macmillan. . 1986. The Secret Trauma: Incest in the Lives of Girls and Women. New York: Basic Books. Sabourin, T. C. 1996. "The Role of Communication in Verbal Abuse Between Spouses." Pp. 199-217 in Family Violence From a Communication Perspective, edited by D. D. Cahn and S. A. Lloyd. Thousand Oaks, CA: Sage. Sacco, V. F. 1990. "Gender, Fear, and Victimization: A Preliminary Application of Power-Control Theory." Sociological Spectrum 40:485-506. Sack, A. R., J. F. Keller and R. D. Howard. 1982. "Conflict Tactics and Violence in Dating Situations." International Journal of Sociology of the Family 12:89-100. Saunders, D. G. 1986. "When Battered Women Use Violence: Husband Abuse or Self Defense?" Violence and Victims 1:47-60. . 1988. "Wife Abuse, Husband Abuse, or Mutual Combat? A Feminist Perspective on the Empirical Findings." Pp. 90-113 in Feminist Perspectives on Wife Abuse, edited by K. Yllo and M. Bograd. Newbury Park: Sage. . 1989. "Who Hits First and Who Hurts Most? Evidence for the Greater Victimization of Women in Intimate Relationships." Paper presented at the annual meeting of the American Society of Criminology, Reno, Nevada. . 1992. "A Typology of Men Who Batter: Three Types Derived From Cluster Analysis." American Journal of Orthopsychiatry 62:264-75. Schechter, S. 1982. Women and Male Violence. Boston: South End.  231 Schuerger, J. M. and N. Reigle. 1989. "Personality and Biographic Data that Characterize Men Who Abuse Their Wives." Journal of Clinical Psychology 44:75-81. Schulman, M. 1979. A Survey of Spousal Violence Against Women in Kentucky. Washington, DC: US Department of Justice, Law Enforcement Assistance Administration. Schuman, H. and S. Presser. 1981. Questions and Answers in Attitude Surveys. New York: Academic. Schumm, W. R. 1990. "Evolution of the Family Field: Measurement Principles and Techniques." Pp. 23-36 in Handbook of Family Measurement Techniques, edited by J. Touliatos, B. F. Perlmutter and M. A. Straus. Newbury Park: Sage. Schwarz, N., H. J. Hippler, B. Deutsch and F. Strack. 1989. "Response Scales: Effects of Category Range on Reported Behavior and Comparative Judgments." Pp. 16572 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Scott, M. E. 1980. "The Battered Spouse Syndrome." Virginia Medical Journal 107:4143. Sears, R., E. E. Maccoby and H. Levin. 1957. Patterns of Child Rearing. New York: Harper and Row. Sedlak, A. J., G. L. Bowen, G. T. Hotaling and D. B. Sugarman. 1985. "Domestic Violence Risk Factors and Consequences: Current State of Knowledge." Paper presented at the 93rd Annual Convention of the American Psychological Association, Los Angeles, CA. Sheatsley, P. B. 1983. "Questionnaire Construction and Item Writing." Pp. 195-230 in Handbook of Survey Research, edited by P. H. Rossi, J. D. Wright and A. B. Anderson. San Diego: Academic. Shehan, C. L. and G. R. Lee. 1990. "Roles and Power." Pp. 420-41 in Handbook of Family Measurement Techniques, edited by J. Touliatos, B. F. Perlmutter and M. A. Straus. Newbury Park: Sage. Sher, K. J. 1985. "Subjective Effects of Alcohol: The Influence of Setting and Individual Differences in Alcohol Expectancies." Journal of Studies on Alcohol 46:137-46.  232 Shields, N. M. and C. R. Hanneke. 1984. Multiple Sexual Victimization: The Case of Incest and Marital Rape. Paper presented at the Second Conference for Family Violence Researchers, Durham, NH. Shields, N., G. McCall and C. Hanneke. 1988. "Patterns of Family and Nonfamily Violence: Violent Husbands and Violent Men." Violence and Victims 3:83-97. Shupe, A., W. A. Stacey and L. R. Hazlewood. 1987. The Dynamics of Domestic Violence. Lexington, MA: Lexington Books. Sigelman, L. 1989. "Question-Order Effects on Presidential Popularity." Pp. 142-50 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Sigelman, C. K., C. J. Berry and K. A. Wiles. 1984. "Violence in College Students' Dating Relationships." Journal of Applied Social Psychology 5:530-48. Singleton, R., Jr., B. C. Straits, M. M. Straits and R. J. McAllister. 1988. Approaches to Social Research. New York: Oxford University. Skogan, W. and M. Maxfield. 1981. Coping with Crime. Beverly Hills: Sage. Smith, M. D. 1985. Woman Abuse: The Case For Surveys by Telephone. La Marsh Research Programme on Violence and Conflict Resolution. Report No. 12. Toronto: York University. . 1986. "Effects of Question Format on the Reporting of Woman Abuse: A Telephone Survey Experiment." Victimology 11:430-38. . 1987. "The Incidence and Prevalence of Woman Abuse in Toronto." Violence and Victims 2:173-87. . 1988. "Women's Fear of Violent Crime: An Exploratory Test of a Feminist Hypothesis." Journal of Family Violence 3:29-38. . 1990a. "Sociodemographic Risk Factors in Wife Abuse: Results from a Survey of Toronto Women." Canadian Journal of Sociology 15:39-58. . 1990b. "Patriarchal Ideology and Wife Beating: A Test of a Feminist Hypothesis." Violence and Victims 5:257-73. . 1994. "Enhancing the Quality of Survey Data on Violence Against Women: A Feminist Approach." Gender and Society 8:109-27.  233 Smith, T. W. 1989. "That Which We Call Welfare by Any Other Name Would Smell Sweeter: An Analysis of the Impact of Questions Wording on Response Patterns." Pp. 99-107 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Snell, J. E., R. J. Rosenwald and A. Robey. 1964. "The Wife-Beater's Wife: A Study of Family Interaction." Archives of General Psychiatry 11:107-113. Solomon, T. 1980. "History and Demography of Child Abuse." Pp. 63-68 in Child Abuse: Commission and Omission, edited by J. V. Cook and R. B. Bowles. Toronto: ButterWorths. Sommer, R., G. E. Barnes and R. P. Murray. 1992. "Alcohol Consumption, Alcohol Abuse, Personality and Female Perpetrated Spouse Abuse." Personality and Individual Differences 13:1315-23. Sonkin, D. J. and M. Durphy. 1985. Learning to Live Without Violence: A Handbook for Men. 2nd ed. San Francisco: Volcano. Sorenson, S. B. and C. A. Telles. 1991. "Self-Reports of Spousal Violence in a Mexican-American and Non-Hispanic White Population." Violence and Victims 6:3-15. Southam Business Information and Communications Groups. 1991. Corpus Almanac and Canadian Source Book. Toronto: Southam Business Information and Communications Group. Stanko, E. 1990. Everyday Violence: How Women and Men Experience Sexual and Physical Danger. London: Pandora. Statistics Canada. 1992a. Universities: Enrolment and Degrees 1990. Catalogue 81204 Annual. Ottawa: Statistics Canada. . 1992b. Community Colleges and Related Institutions: Postsecondary Enrolment and Graduates 1989. Catalogue 81-222 Annual. Ottawa: Statistics Canada. . 1993a. General Social Survey. Violence Against Women Questionnaire. Ottawa: Statistics Canada. . 1993b. Violence Against Women Survey, 1993: Shelf Tables 1-25. Ottawa: Minister of Supply and Services. . 1994. Violence Against Women Survey: Public Use Microdata File Documentation and User's Guide. Ottawa: Minister of Supply and Services.  234  Steinmetz, S. K. 1977. The Cycle of Violence: Assertive, Aggressive and Abusive Family Interaction. New York: Praeger. Steinmetz, S. K. and J. S. Lucca. 1988. "Husband Battering." Pp. 233-46 in Handbook of Family Violence, edited by V. B. Van Hasselt, R. L. Morrison, A. S. Bellack and M. Hersen. New York: Plenum. Stets, J. E. and D. Henderson. 1991. "Contextual Factors Surrounding Conflict Resolution While Dating: Results From a National Study." Family Relations 40:29-36. Stets, J. and M. Pirog-Good. 1987. "Violence in Dating Relationships." Social Psychology Quarterly 50:237-46. . 1989. "Patterns of Physical and Sexual Relationships." Journal of Family Violence 4:63-76. Stets, J. E. and M. A. Straus. 1989. "The Marriage License as a Hitting License: A Comparison of Assaults in Dating, Cohabiting, and Married Couples." Pp. 33-52 in Violence in Dating Relationships, edited by M. A. Pirog-Good and J. E. Stets. New York: Praeger. . 1990. "Gender Differences in Reporting Marital Violence and Its Medical and Psychological Consequences." Pp. 151-65 in Physical Violence in American Families: Risk Factors and Adaptations in 8,145 Families, edited by M. A. Straus and R. J. Gelles. New Brunswick, NJ: Transaction. Straus, M. A. 1973. "A General Systems Approach to a Theory of Violence Between Family Members." Social Science Information 12:105-25. . 1977a. "A Sociological Perspective on the Prevention and Treatment of Wife Beating." Pp. 194-238 in Battered Women: A Psychosociological Study of Domestic Violence, edited by M. Roy. New York: Van Norstrand Reinhold. . 1977b. "Sexual Inequality, Cultural Norms and Wife Beating." Pp. 59-77 in Women Into Wives: The Legal and Economic Impact of Marriage, edited by J. R. Chapman and M. Gates. Beverly Hills, CA: Sage. . 1979. "Measuring Intra-Family Conflict and Violence: The Conflict Tactics (CT) Scale." Journal of Marriage and the Family 41:75-88. . 1980a. "Sexual Inequality and Wife Beating." Pp. 86-93 in The Social Causes of Husband-Wife Violence, edited M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota.  235  . 1980b. "Social Stress and Marital Violence in a National Sample of American Families." Annals of the New York Academy of Sciences 347:229-50. . 1980c. "Sociological Perspective on the Causes of Family Violence." Pp. 211-34 in The Social Causes of Husband-Wife Violence, edited M. A. Straus and G. T. Hotaling. Minneapolis: University of Minnesota. . 1980d. "Victims and Aggressors in Marital Violence." American Behavioral Scientist 5:681-704. . 1980e. "The Marriage License as a Hitting License: Evidence from Popular Culture, Law, and Social Science." Pp. 39-50 in The Social Causes of Husband-Wife Violence, edited by M. A. Straus and G. T. Hotaling. Minneapolis, MN: University of Minnesota. . 1990. "The Conflict Tactics Scales and its Critics: An Evaluation and New Data on Validity and Reliability." Pp. 49-73 in Physical Violence in American Families, edited by M. A. Straus and R. J. Gelles. New Brunswick, NJ: Transaction. . 1991. "Physical Violence in American Families: Incidence Rates, Causes, and Trends." Pp. 17-34 in Abused and Battered: Social and Legal Responses to Family Violence, edited by D. D. Knudsen and J. L. Miller. New York: Aldine de Gruyter. . 1994. "State-To-State Differences in Social Inequality and Social Bonds in Relation to Assaults on Wives in the United States." Journal of Comparative Family Studies 25:7-24. Straus, M. A. and R. J. Gelles. 1986. "Societal Change and Change in Family Violence From 1975 to 1985 as Revealed by Two National Surveys." Journal of Marriage and the Family 48:465-79. . 1990. Physical Violence in American Families: Risk Factors and Adaptations to Violence in 8, 145 Families. New York: Transaction. Straus, M. A., R. J. Gelles, and S. K. Steinmetz. 1980. Behind Closed Doors: Violence in the American Family. New York: Anchor Books. Sudman, S. 1983. "Applied Sampling." Pp. 145-94 in Handbook of Survey Research, edited by P. H. Rossi, J. D. Wright and A. B. Anderson. San Diego: Academic. Sudman, S. and N. M. Bradburn. 1974. Response Effects in Surveys: A Review and Synthesis. Chicago: Aldine.  236  . 1982. Asking Questions: A Practical Guide to Questionnaire Construction. San Francisco: Jossey-Bass. Sugarman, D. B. and G. T. Hotaling. 1989. "Dating Violence: Prevalence, Context, and Risk Markers." Pp. 3-32 in Violence in Dating Relationships: Emerging Social Issues, edited by M. A. Pirog-Good and J. E. Stets. New York: Praeger. Szinovacz, M. E. 1983. "Using Couple Data as a Methodological Tool: The Case of Marital Violence." Journal of Marriage and the Family 45:633-44. TanVergert, E., M. Gillespie and J. Kingma. 1988. The Measurement of Interspousal Physical Aggression: Using the Conflict Tactics Scale in Survey Research. Edmonton Area Series Report 62. Edmonton, AB: Population Research Laboratory. Taylor, S. P. and K. E. Leonard. 1983. "Alcohol and Human Physical Aggression." Pp. 77-102 in Aggression: Theoretical and Empirical Reviews, vol. 2. edited by R. G. Green and E. I. Donnerstein. New York: Academy. Telch, C. F. and C. V. Lindquist. 1984. "Violent Versus Nonviolent Couples: A Comparison of Patterns." Psychotherapy 21:242-48. Thompson, C. and D. J. Mingay. 1991. "Estimating the Frequency of Everyday Events." Applied Cognitive Psychology 5:497-510. Tournageau, R. and K. A. Rasinski. 1988. "Cognitive Processes Underlying Context Effects in Attitude Measurement." Psychological Bulletin 103:299-314. Traugott, M. W., R. M. Groves and J. M. Lepkowski. 1989. "Using Dual Frame Designs to Reduce Nonresponse in Telephone Surveys." Pp. 79-96 in Survey Research Methods: A Reader, edited by E. Singer and S. Presser. Chicago: The University of Chicago. Turner, J. H. 1991. The Structure of Sociological Theory. 5th ed. Belmont, CA: Wadsworth. Tyree, A., J. Malone and K. D. O'Leary. 1987. Physical Aggression with Others as a Predictor of Spousal Violence. Unpublished manuscript. Stony Brook, NY: State University of New York. US Department of Justice, Bureau of Justice Statistics. 1992. Report to the Nation on Crime and Justice: The Data. Washington, DC: Government Printing Office.  237 Vesterdal, J. 1980. "Psychological Mechanisms in Child Abusing Parents." Pp. 165-69 in Child Abuse: Commission and Omission, edited by J. V. Cook and R. B. Bowles. Toronto: ButterWorths. Viano, E. C , ed. 1990. The Victimology Handbook: Research Findings, Treatment, and Public Policy. New York: Garland. , ed. 1992. Critical Issues in Victimology: International Perspectives. New York: Springer. Vivian, D. and J. Langhinrichsen-Rohling. 1994. "Are Bi-Directionally Violent Couples Mutually Victimized? A Gender-Sensitive Comparison." Violence and Victims 9:107-24. Vogt, W. P. 1993. Dictionary of Statistics and Methodology: A Nontechnical Guide for the Social Sciences. Newbury Park: Sage. Walker, L. E. 1978. "Battered Women and Learned Helplessness." Victimology 2:525534. . 1979. The Battered Woman. New York: Harper and Row. . 1981. "A Feminist Perspective on Domestic Violence." Pp. 102-115 in Violent Behavior: Social Learning Approaches to Prediction, Management and Treatment, edited by R. B. Stuart. New York: Brunner/Mazel. . 1984. The Battered Woman Syndrome. New York: Springer. . 1989. "Psychology and Violence Against Women." American Psychologist 44:695-702. . 1993. "The Battered Woman Sundrome is a Psychological Consequence of Abuse." Pp. 133-53 in Current Controversies on Family Violence, edited by R. J. Gelles and DeKeseredy and Kelly. R. Loseke. Newbury Park: Sage. . 1995. "Understanding Battered Woman Syndrome." Trial 31:30-7. Washburne, C. K. 1983. "A Feminist Analysis of Child Abuse and Neglect." Pp. 28992 in The Dark Side of Families: Current Family Violence Research, edited D. Finkelhor, R. J. Gelles, G. T. Hotaling and M. A. Straus. Beverly Hills, CA: Sage. White, J. W. and J. A. Humphrey. 1994. "Women's Aggression in Heterosexual Conflicts." Aggressive Behavior 20:195-202.  238 White, J. W. and M. P. Koss. 1991. "Courtship Violence: Incidence in a National Sample of Higher Education Students." Violence and Victims 6:247-56. Whitehurst, R. N. 1974. "Violence in Husband-Wife Interaction." Pp. 692-98 in Violence in the Family, edited by S. K. Steinmetz and M. A. Straus. New York: Harper and Row. Wilson, M. and M. Daly. 1994. "Spousal Homicide." Juristat U:\-\3. Wilson, M., H. Johnson and M. Daly. 1995. "Lethal and Nonlethal Violence Against Wives." Canadian Journal of Criminology 37:331-362. Wolfgang, M. D. 1958. Patterns in Criminal Homicide. Philadelphia: University of Pennsylvania. Yllo, K. 1988. "Political and Methodological Debates in Wife Abuse Research." Pp. 28-50 in Feminist Perspectives on Family Abuse, edited by K. Y116 and M. Bograd. Newbury Park: Sage. . 1993. "Through a Feminist Lens: Gender, Power, and Violence." Pp. 4762 in Current Controversies on Family Violence, edited by R. J. Gelles and D. R. Loseke. Newbury Park: Sage. Yllo, K. and M. Bograd. 1988. Feminist Perspectives on Wife Abuse. Newbury Park, CA: Sage. Yllo, K. and M. A. Straus. 1981. "Interpersonal Violence Among Married and Cohabiting Couples." Family Relations 30:339-47.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0087794/manifest

Comment

Related Items