INTERNATIONAL DATING VIOLENCE STUDY
Murray A. Straus
and Ignacio Luis Ramirez
Family Research
Laboratory, University of New Hampshire
Durham, NH
03824 603-862-2594 murray.straus@unh.edu
Straus
website: http://pubpages.unh.edu/~mas2/
ABSTRACT
Numerous studies have found that past-year prevalence rates of dating partner violence (DPV) among university students are in the 25-45% range, which is much higher than the rates for married couples. The high rate of DPV is an important health problem because physical injury rates are also high (about 11% of the victims), as are psychological injuries. The proposed research will provide data on rates of perpetration of physical assault, physical injury, sexual coercion, and psychological aggression. This is a more comprehensive range of aspects of DPV than is usually investigated. The study will also differentiate between “minor” violence such as slapping and shoving, and “severe” violence such as punching, choking, and attacks with objects.
Societal-level, individual-level, and contextual theories of risk factors for DPV will be tested. The results of testing these theories could contribute to designing primary prevention programs and treatment programs. For example, one objective of the study is to test what we call the “differential risk factor theory” which argues that the risk factors for minor violence tend to reflect social characteristics such as male dominance and couple communication problems, whereas severe violence tends to reflect psychological and psychopathological characteristics such as anti-social personality and criminal history. If this theory is supported it suggests that prevention and treatment programs should develop differentiated approaches to minor and severe violence.
Members of a multidisciplinary research consortium in 30
countries will obtain the data.
Cross-national data is needed to test the societal-level theories and
the social context theories. The
consortium members will use a standard and tested instrument, which also
includes a section to permit each to add measures to test culture-specific
theories. The N for each site will vary
from 80 to 500, with a combined N of over 7,000. A variety of statistical approaches will be used because the
theories to be tested require different statistical approaches. For example, multinomial logistic regression
will be used to test the differential risk factor theory, and Hierarchical
Linear Modeling will be used to test contextual effect theories such as the
effect of stress in the context of a male-dominant society as compared to a
more egalitarian context.
Table of Contents
Aim 1. To Provide Information On Prevalence And
Chronicity Of Dating Partner Violence (DPV)
Aim 3. To Test Theories
About The Effect Of Socio-Cultural Context On DPV
2. What
societal-level variables explain (in the statistical sense) site-to-site
variation in DPV?
3. How do
socio-cultural context variables affect relationships at the individual-level?
B.
BACKGROUND AND SIGNIFICANCE
B1. Limitations And
Importance of College Student Dating Violence
B2. Definition And
Cross-National Applicability Of Dating..
B3. Prevalence And
Chronicity Of Four Aspects Of Dating Violence
B4 Risk Factors For Dating Partner Violence (DPV)
B5. Psychological and Social Risk Factors
1. The
Differential Risk Factor Controversy.
2.
Limitations of Research On Social and Psychological Risk factors.
B8 Summary Of Strengths of the Proposed Study
C1. Work of The Principle Investigator
C2. Development of Instruments For This Study
Conflict
Tactics Scales (CTS).
Personal And
Relationships Profile (PRP).
Research
With This Instrument.
Identification
of Severe Violence.
C3 The International Dating Violence Research Consortium
D. RESEARCH
DESIGN AND METHODS
Consortium
List Server And Virtual Conferences.
Definition
And Terminology For Dating.
Demographics
And Structure of The Dating Relationship.
Revised
Conflict Tactics Scales (CTS2).
The Personal
and Relationships Profile (PRP).
D4. Data Analysis – General
Approaches
Descriptive
Analyses of Prevalence and Chronicity.
Gay and
Lesbian Relationships.
Confounding
with Social Desirability.
D6. Differential Risk Factor Hypothesis Tests
Interaction
of Psychological and Social Risk Factors.
Differential
Risk Factor Hypothesis.
D8. Other Aspects of DPV As
Dependent Variables
D9. Societal-Level and Socio-Cultural Context
Analyses
Culturally
Relevant Measures For Site-Specific Analyses.
Hierarchical/multilevel
modeling methods and the HLM3 program..
Control for
Characteristics Of The Student Body.
D10. Publications and Time Schedule
Articles
Describing The Extent Of DPV In Specific Sites.
Articles Comparing
The Extent Of DPV In Different Nations.
Theory-Testing
Comparative Articles.
The aims listed below will be
pursued with data from 30 countries by a group of investigators referred to as
the International Dating Violence Research Consortium and shortened in this
proposal to the Consortium (see C. Preliminary Studies).
It is now widely recognized that relationships between partners in marital, cohabiting, and dating relationships are often violent (Barnett, Miller-Perrin, & Perrin, 1997; Gelles & Straus, 1988; Watson, Cascardi, Avery-Leaf, & Daniel, 2001). Less widely recognized is the fact that relationships of dating partners are, on average, more likely to be physically violent than those of married partners (Stets & Straus, 1989; Sugarman & Hotaling, 1989). The past-year prevalence rates found in many studies of dating couples are in the 25 to 45% rate, compared to 10-15% for married couples (Cano, Avery-Leaf, Cascardi, & O'Leary, 1995; Stets & Straus, 1989; Sugarman & Hotaling, 1989). Most of this difference is because college data couples are much younger, but some of the difference remains, even when compared to married couples of the same age (Stets & Straus, 1989).
This international study will provide rates of perpetrating dating partner violence (DPV) derived from a standardized and well-validated instrument that has been used in many countries – the Conflict Tactics Scales (Straus, 1979, 1990a; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). These data will permit investigating the extent to which DPV is a world-wide phenomenon, at least among this educated sector of society. The study will provide data on nation-to-nation differences in the past-year prevalence and also chronicity (mean number of times among the subgroup who engaged in the behavior) of the following four aspects of DPV.
There is controversy over the extent to which violence between intimate partners has its origins in the psychological characteristics such as depression and borderline personality, as compared to social characteristics of the setting or the relationship such as cultural norms permitting violence, or characteristics of the relationship such as dominance by one partner; or as argued by Dutton (1994) the interaction of psychological and social risk factors. This study aims to provide data on that issue by including both psychological and social risk factors in the same study so that they can be more readily compared.
Psychological and psychopathology theories are exemplified in work such as Dutton (1993), Holtzworth-Munroe (1999; 1994), Flournoy (1991), O’Leary (1993 ), and Saunders (1992). The social theories are exemplified in the work of Dobash and Dobash (1979), Rouse (1990), Coleman and Straus (1990; 1994), and Yllo (1984), and by the PI’s book The Social Causes of Husband-Wife Violence (Straus & Hotaling, 1980). Part of the reason for the controversy probably lies in the disciplinary training of the investigators. Psychologists naturally, and appropriately, tend to focus on psychological causes, and sociologists naturally and appropriately tend to focus on social causes.
Another important reason for the controversy is that the two types of risk factors are usually investigated in different studies, making the results difficult to compare. Finally, we suggest that some of the inconsistencies arise because of failure to take into account the severity of the violence. Psychological factors, and especially psychopathology, may be most important for understanding offenders who engaged in chronic and severe assaults and often inflict injuries that require medical attention. On the other hand, social factors may be most important for understanding what Johnson (1995; 2000) calls “common couple violence” and Straus calls the “ordinary violence” of American married life (1990b).
The possibility that minor and severe violence against a dating partner have a different etiology, led to what will be called the “differential risk factor” hypotheses, i.e., that psychological characteristics such as antisocial personality features are more strongly related to severe violence than are social characteristics; whereas social characteristics (e.g., dominance by one partner, norms permitting violence) are more strongly related to minor violence than are psychological characteristics.
In addition to the broad categorization of psychological and social risk factors, the study will also provide data to investigate 22 specific risk factors. For example the psychological risk factors include poor anger management, antisocial personality, and alcohol problems; and the social risk factors include dominance in the relationship, communication problems, and weak integration into society.
The study will investigate the ways in which the risk factors for DPV are similar and different in the diverse socio-cultural contexts provided using data from 45 sites in 30 nations. The data will be obtained and analyzed by local investigators at each site. The study is designed to provide a more culturally informed understanding of DPV and to test theories about socio-cultural context effects. The following types of questions will be investigated:
Although every site will use the same core questionnaire consisting of demographic questions, the revised Conflict Tactics Scales (CTS2) and the Personal And Relationships Profile (PRP), every site will also add to the questionnaire measures to test hypotheses concerning the issues that the Consortium member judges to be locally salient issues. For example, the preliminary study in El Paso, Texas, where the student body is about 80% Mexican American, included a scale to measure Acculturation to the dominant society in order to investigate the extent to which acculturation is related to DPV. A measure of acculturation might be included in other sites with large immigrant or minority population such as Tel Aviv, Israel.
One of the most important social theories of intimate partner violence is that the more patriarchic the society, the higher the level of violence against women (Dobash & Dobash, 1979; Straus, 1973, 1974a; Yllo & Bograd, 1988). This is a widely accepted theory and is the basis for most programs to prevent and treat intimate partner violence. Consequently, it will be the primary societal level theory to be investigated. It will also have priority in this research because of questions that have been raised about this theory. For example, Dutton’s review of the empirical research (Dutton, 1994) concludes that it explains little of the variance. This may be because, although women are still disadvantaged in the US and Western Europe, the level of male dominance does not approach the patriarchic system found, for example, in Iran and Pakistan. Dutton also argues that the lack of strong empirical support for the patriarchic society theory may be due to inappropriate use of individual-level data for an issue that require societal-level data. Consistent with this explanation, when Straus (1994) used macro-level data on US states he found that the lower the status of women relative to men, the higher the rate of physical assaults on women by male partners. The proposed research will use a similar macro-level scale to measure gender equality in each nation. The theory predicts that the rate of DPV against women will be greatest in the most patriarchic societies, and the 30 nations and 45 sites planned for this study can provide the variance in patriarchic social organization that is needed to test this theory.
The same acts may have different meanings and consequences in different societies. For example, in the US, spanking and other legal corporal punishment has been found to be related to an increased probability of the child, later in life, hitting a partner (Carroll, 1980; Simons, Lin, & Gordon, 1998; Straus, 2001; Straus & Yodanis, 1996; Swinford, DeMaris, Cernkovich, & Giordano, 2000). But for African Americans, this relationship, although present, is attenuated. Some scholars argue that this is because corporal punishment is widely accepted in the African American community and children do not perceive it as parental hostility. This leads to the hypothesis that the more normative the use of corporal punishment in a society, the lower the correlation between the amount of corporal punishment and assaulting a dating partner. This hypothesis and other contextual effect hypotheses will be tested using multilevel linear and non-linear modeling (Bryk & Raudenbush, 1992; Goldstein, 1988).
Aggression is an extremely important aspect of psychological functioning and social life, and one that has profound implications for physical and mental health of both the aggressor and the targets of aggression. Important strides have been made toward understanding all types of aggression, including aggression against intimate partners. However, two recent reviews have pointed out there is relatively little cross-national data on aggressive behavior (Segall, Ember, & Ember, 1997; Tedeschi & Bond, 2001). The importance of a cross-national approach to family violence was indicated by the January 1999 special topic issue of the American Psychologist on Psychology and Domestic Violence Around and The World, and by the World Health Organization conference on comparative risk factor assessment in Auckland, NZ in December 2000. Intimate partner violence was identified as an avertable risk factors for mental disorder (Andrews, Ustun, & Kessler, 2000). The proposed research, in addition to its focus on understanding aggression in intimate partner relationships, is intended to also contribute to understanding of the socio-cultural aspects of aggression. The research is posited on the assumption that identifying social and cultural context variables that enhance or inhibit aggression will provide a more adequate understanding of both physical aggression between dating partners and of human aggression in general.
Student samples have important
limitations to which readers of the articles and the monograph reporting the
results of this study will be alerted.
In all the countries they tend to be from a relatively privileged sector
of society and therefore are not representative of their age group. In some countries females will be
underrepresented. (As indicated in
sections D4 and D9, to a certain extent these two potential confounds can
controlled statistically.) Student
culture also tends to go to extremes.
In some situations students are much more secular and liberal than the
general population, and in other situations such as the student take-over of
the American Embassy in Iran, they represent more fundamentalist religious
ideology. Each member of the consortium
will take this into account when interpreting the results for their site. They will also provide the PI with a
statement on this aspect of the cultural context so that it can be considered
when interpreting the results of the cross-national comparative analyses.
Despite these limitations, college
students are an important population for research on partner violence because,
as was indicated in the Specific Aims section, they are a population at an
extremely high risk of physical and sexual aggression against a partner. In addition, in many countries they
constitute a sizeable population. In
the USA, for example, there are about 15 million currently enrolled in post-secondary
institutions. Moreover, college
students are at a formative period in their lives, especially in relationship
to the development of appropriate patterns of behavior with an intimate
partner. The patterns manifested at
this age are often enduring features of their relationship (Murphy & O'Leary, 1989; O'Leary et
al., 1989; O'Leary, Malone, & Tyree, 1994; H. S. Pan, P. H. Neidig, &
K. D. O'Leary, 1994). The proposed
cross-national study will permit societal level and social context analyses
that can provide a more adequate understanding of this phenomenon.
For purposes of this study dating is
defined as a dyadic relationship involving meeting for social interaction and
joint activities and an explicit or implicit intention to continue the
relationship until one or the other party terminates or until some other more
committed relationship is established such as cohabiting, engagement, or
marriage. The names used for this
relationship have varied over time, and in the US some regard it as
anachronistic. However, whatever the
current terminology, this type of dyadic relationship involves most young
people in the US. One of the objectives
of the study will be to estimate the proportion of students in a current or
recent dating relationship at each of the sites.
The social norms for dating, and
actual dating behavior, differ according to many dimensions, including
individual differences, racial/ethnic and socioeconomic group differences,
historical era, and cultural context.
In the American cultural context, dating implies individual choice of
partners, but that is not an essential part of the phenomenon being address in
this study. Despite these differences,
there are also some inherent structural similarities; for example, it is a
dyadic relationship and the parties usually invest time and energy. Therefore social interactional process
typical of dyads are likely to apply regardless of whether the relationship was
arranged by parents or friends, by newspaper or by internet, or by one party
initiating the development of a relationship.
Further, if the relationship continues beyond the initial date, it
implies the possibility of mutual attraction, ego involvement, and the
possibility of becoming a more committed type of pair relationship. Thus, the principles of theories such as
exchange theory and conflict theory could to apply to these types of
relationships (whatever, they are named) in the national contexts to be studied. This research will investigate the extent to
which theories concerning intimate partner violence apply to student couples in
diverse societies. Evidence indicating
that at least some characteristics apply cross-nationally comes from a study of
undergraduates in North Carolina and at the National Taiwan University (Lin Yuan-Huei & Rusbult, 1995). They found
that in both countries, students were more committed to their partners when
they were satisfied with the relationship and if they perceived a high degree
of investment in the relationship. They
concluded that the same basic processes were involved in changing to a more
committed type of relationship.
To investigate the proportion of
students are in a dating relationship in each the nations in this study
requires culturally appropriate terminology and descriptions at the start of
the questionnaire for each site. Our
approach to developing culturally appropriate terms and descriptions is
described in the Research Design And Methods sections D1 (paragraph 2) and D3
(paragraph 2). We expect to find large
differences in the proportion of students in dating relationships. If so, we plan to use this as one of the
social context variables to investigate whether dating relationships are
different when dating is relatively infrequent.
The Conflict Tactics Scales (CTS),
which is the instrument to be used to obtain the data on partner violence,
provides rates of occurrence and chronicity (how many times it occurred among
those who engaged in each behavior) for the following aspects of partner
violence:
Physical assault
` Physical
injury
Psychological aggression
Sexual Coercion
The prevalence rate and the chronicity of these four types of aggression against a partner across the many nations is the most basic type of information that will result from this study. Assuming that differences across nations will be found, the challenge is to understand why those differences occur. A number of explanations for cross-national differences will be tested. These will use a variety of approaches starting with relatively simple controls for demographic data to examine confounding with such things as the age of respondents. If students in the sites with high rates of partner assault tend to be younger, age could explain the difference between sites in assault because the younger the person, the greater the probability of assaulting a partner (Stets & Straus, 1989; Suitor, Pillemer, & Straus, 1990). Theoretically focused approaches are described below, especially, the patriarchic society theory. The PI’s previous research has shown that male dominance is associated with partner assault, using individual-level data (Coleman & Straus, 1990) and also using societal-level data on state-to-state differences in gender equality in the US (Straus, 1994, see Appendix 7). Parallel procedures will be used for this cross-national study, i.e., analyses will be done using both individual-level and societal-level data on gender equality.
Despite the large number of studies of DPV, few studies provide data on the severity of the physical attacks, and even fewer on chronicity (how often it occurs among those who experience physical violence), or the level of physical injury. Our preliminary analyses using the Injury scale of the CTS2 indicate a surprisingly high rate of attacks resulting in injury, for example in one sample of 370 students, 15% of the male respondents and 9% of the female respondents reported having inflicted physical injury sufficient to cause visible marks or bruises or to feel pain continuing for a day or more, or more serious injury of a partner (Straus et al., 1996). Data on two other samples (as yet unpublished) found similar rates. A study by Rosenbaum and Brown (private communication, September 12, 2001) also found similar rates and, in addition, found substantial percentages of students who missed classes, could not study properly, and had lower grades because of violence by a partner. These results indicate that injury from DPV is an important public health problem that is rarely perceived, and which needs further investigation.
There are many studies showing high rates of sexual coercion by university students, including the PI’s own research which found that 37% of male students and 18% of female students reported having coerced a partner into a sexual act in the past year (Straus et al., 1996). The rates vary widely from study to study, probably because of variation in the way sexual coercion is defined and measured in these studies. Some studies focus on use of physical force and others include non-physical coercion. The research described in this application will standardize the measurement by using the Sexual Coercion scale of the CTS2 in all the societies in the study. This instrument provides data on both an overall score that includes all types of coercion and thus permits rough comparability with previous studies that measured sexual coercion broadly. The CTS2 also provides scores on subscales that focus on verbal pressure, threat, and use of physical force, thus permitting rough comparability with studies that have focused on each of these modes of sexual coercion. Table 1 (from Straus, 2000) gives rates based on the CTS Sexual Coercion scale for a sample of 651 US students, and shows a very high rate of using physical force, that is parallel to the legal category of rape (see Appendix 4 for this and other items in the CTS).
Table 1. Prevalence of Sexual Coercion by Male and Female Dating Partners
_______________________________________________________________________
Both Males Females
Severity (N=651) (N=211) (N=440)
X2
Sexual Coercion Overall 20.1 26.1 17.3 6.86**
Threat
2.5 3.3 2.0 0.96
Physical
2.0 4.7 0.7 12.00**
_______________________________________________________________________
* p <.05 ** p <.01
The second type of issue to be
investigated will use the Personal and Relationships Profile or PRP (Straus, Hamby, Boney-McCoy, & Sugarman,
1999; Straus & Mouradian, 1999) and other data to test hypotheses concerning risk
factors for DPV. The PRP measures 22
risk factors for partner violence. Some
of the PRP scales measure psychological and psychopathological variables such
as depressive symptoms, antisocial personality symptoms or substance abuse, and
some measure characteristics of the relationship (as reported by the partner
who completed the PRP) such as Communication Problems and Dominance. There is also a scale to measure social
desirability response set. This
research will permit examining the extent to which these 22 PRP variables are
risk factors for partner violence in US and non-US cultural contexts. The analyses will control for the gender,
age, family background, and other demographic characteristics, and will also
control for spurious relationships that could be produced by variation in
tendencies to respond in ways that present the respondent in a socially
desirable light.
The data to be analyzed, like most
previous research on intimate partner violence, are cross sectional. Strict use of the concept of risk factor
requires that it represents an attribute or characteristic that meets three
general criteria (Kleinbaum, Kupper,
& Morgenstern, 1982; MacMohon & Pugh, 1970; Nagel & Neff, 1979). First, the
characteristic must covary with partner violence. Second, the covariation with DPV must not be due to such sources
of error as sampling error or measurement error. Third, changes in levels of the characteristic must temporally
precede changes in level of the violence measure. However, in this proposal, we use “risk factor” to refer to
cross-sectional data that do not necessarily meet the temporal priority criterion. For example, inadequate anger management
skills could be a cause of violence in relationships for some violent
partners. But for some others it might
be an effect of relying on physical force and therefore never having to
learn to effectively control anger.
Although that is an important limitation, if this research shows that
inadequate ability to manage anger is associated with partner violence, that is
important information, regardless of whether it is cause or effect. It will take on added importance if, as
could well be the case, this relationship applies to some of the societies and
not others. That would lead us to
examine characteristics of the two groups of societies to try to determine what
might account for the difference. An
answer to that question would likely provide further understanding of DPV, and
might suggest approaches to primary prevention as well as treatment.
Early psychiatric analyses of battering emphasized psychopathology, although principally on the part of the victims (Gayford, 1975; Snell, Rosenwald, & Robey, 1964). Such analyses suffered considerable criticism (Hilberman, 1980; Pagelow, 1981) and, in the 1980s, were largely replaced by sociological analyses of domestic violence that emphasized social risk factors such as male dominance and conflict (Dobash & Dobash, 1979; Gelles, 1983; Pagelow, 1981; Straus, Gelles, & Steinmetz, 1980; Yllo & Bograd, 1988). Many of these authors explicitly rejected psychopathology as a major aspect of the etiology of intimate partner violence, arguing that only a minute proportion of offenders manifested clinical level psychopathology. The belief that sociological factors were more important than psychological factors had wide acceptance among family violence researchers and was believed to apply to all kinds of violence and formed the basis of most primary prevention and treatment programs.
More recently, however, the primacy of social risk factors has been challenged. Numerous researchers are now asserting that individual-level psychological risk factors are at least as important, if not more important, than social factors (e.g., Dutton, 1994; Gottman et al., 1995; O'Leary, 1993). Others, however, continue to focus on sociological factors, and the dispute between those focusing on psychological mechanisms and those focusing on sociological ones has received considerable attention in the professional literature (Bowker, 1993; Renzetti, 1994; Stets, 1992; Walker, 1993; Yllo, 1993). We will refer to these different perspectives as the differential risk factor controversy, and refer to the explanation for the controversy that we will investigate as the differential risk factor theory.
Difficulties with currently available research on the relative importance of psychological and social risk factors occur at the design, operationalization, and analysis levels. At the design level, most studies are not intended to address both psychological and social domains and hence within-study comparisons are rarely possible. Even fewer studies attempt to sample these two general domains equally well (by including a similar number of indicators in each domain). At the measurement level, measures vary considerably with regard to reliability and validity. Thus, it can be questionable to compare measures directly, even when they are used in the same study, because differences in effects may be due to differences in the adequacy of the instruments (e.g., the well-validated MMPI and the Blood and Wolfe decision-power scale, whose validity has been widely questioned). Still another problem occurs because social constructs have not been operationalized as thoroughly as psychological constructs and because psychological measures are often used as proxies for social variables (Bagarozzi, 1989; Dutton, 1994). Perhaps the most important limitation is explained in the following section: not differentiating minor and severe violence.
We suggest that much of the inconsistency in research on
intimate partner violence occurs because it draws on populations that differ
drastically in severity of violence.
One population consists of community samples in which minor violence
predominates and in which what Johnson calls “terroristic” violence (Johnson, 1995; Johnson & Ferraro,
2000) is rare. The
other population consists of clinical samples, such as men in batterer
treatment programs and the partners of women who have sought refuge in
shelters. For this population there is
a high rate of severe violence, sometimes at the terroristic level (Straus, 1990b, 1999). There is
evidence suggesting that the etiology of minor and severe violence may be
different. For example, a survey of
15,023 men in the U.S. Army (H.S.
Pan, P.H. Neidig, & K.D. O'Leary, 1994) found that severely physically aggressive men
reported more depressive symptoms and more substance abuse problems than less
severely aggressive men did. Our own research team has found a number of
differences between minor and severe relationship violence. A longitudinal study by Feld and Straus (1989) found that perpetrators of minor violence were much
more likely to cease than perpetrators of severe violence. Using the 1985 National Family Violence
Survey, Surgarman, Aldarondo, and Boney-McCoy (1996) found that severely violent husbands were more likely
to also be violent outside that marriage.
As
indicated previously, the research just cited, as well as the research by
Ramirez in the Preliminary Studies section C2, and other research (Campbell, 1995; Downs, Miller, &
Panek, 1993; Gottman et al., 1995; Holtzworth-Munroe et al., 1999;
Holtzworth-Munroe & Stuart, 1994), lead to what
we call the “differential risk factor” theory.
This theory predicts that extreme
forms of violence are more likely to reflect individual deviance or
psychopathology, while less extreme forms of violence are more likely to
reflect social characteristics. The
procedures described in Research Design section D6 will permit empirical
examination of this theory. In focusing
on the distinction between minor and severe violence, however, we do not mean
to imply that “minor” violence is at all acceptable or of less concern than
severe violence. In fact, we have
argued that the results of research on minor violence is the most appropriate
empirical basis for programs of primary prevention (Straus, 1990b, 1999).
In this project, we will also
differentiate between minor and severe for each of the other three aspects of
partner violence for which data will be available (physical injury, sexual
coercion, and psychological aggression).
A primary
reason for conducting this cross-national study is to test theories to explain
cross-national differences in DPV. As indicated in Specific Aim 3, we will give
priority to feminist theory, which predicts that patriarchal societies
have the highest levels of violence against women (Dobash & Dobash, 1979; Straus, 1973, 1974a; Yllo &
Bograd, 1988). However,
other theories can be and are likely to be investigated, by the PI, by other members of the
consortium, and by social scientists generally because the data set will become
generally available after it is deposited with Interuniversity Consortium For Political
And Social Research (ICPSR). Data sets
from two of the PI’s previous NIMH funded projects are in the ICPSR archive and
many papers have been published by researchers who obtained the data. Examples of the theories that might be
tested with the data from this study include:
Cultural
spillover theory (Baron &
Straus, 1989; Baron, Straus, & Jaffee, 1988) which predicts that the more violent the society (as
measured by indicators such as the percent of per capita income devoted to
military expenditures and the homicide rate), the higher the rate of DPV.
Humanization
of society theory which suggests the hypothesis that greater the societal
commitment to human rights (using the Human Rights Index for each nation and
other data) the lower the rate of DPV.
Social
stress theory which leads to the hypothesis that the higher the level of
stressful events in a society, the higher the rate of DPV (Linsky,
Bachman, & Straus, 1995; Linsky & Straus, 1986; McWilliams, 1998).
The feminist movement brought the issue of intimate partner violence into public consciousness and created two new social institutions to deal with it – shelters for battered women and batter treatment program for offenders. In part because of this, feminist theory has provided the predominant basis for interventions to prevent or treat intimate partner violence. However, there have been only a few empirical studies that used societal-level data to test the patriarchal society theory (Browne & Williams, 1993; Levinson, 1989; Straus, 1994; Yllo & Straus, 1990) and the empirical status of this theory has been questioned by Dutton (1994) and others. Most studies have used individual-level data and Dutton points out that this is not necessarily the same as male dominance at the societal-level. His review of the empirical research concludes that there little support for the patriarchy theory.
Since the
theory that violence against a partner is carried out to maintain male
dominance in both the family and the society is the basis for most primary
prevention and treatment programs, such an important theory needs further
empirical research. The proposed study
will provide a unique opportunity to contribute to that need because it will be
able to measure the level of male dominance at both the individual-level and
the societal-level, and the interaction of the two levels. There will be two types of societal-level
data on male dominance: (1) Archival
data such as the percent of women holding prestigious jobs, the ratio of income
earned by women to that earned by men, and the percent of women holding
political offices, as in the Gender Equality Index created by the author and
colleagues for each of the 50 US states (Straus,
1994; Sugarman & Straus, 1988). (2)
Aggregated individual-level data for each site using the Dominance scale of the
PRP (Hamby, 1996). This will be
created using the Aggregate procedure of SPSS to assign the mean or median
Dominance scale score for each site to the records of all cases for the
site. These data will permit testing
the hypothesis that DPV is highest in sites with the least equality for women
and highest levels of male dominance.
Violence between intimate partners,
including dating partners, can be considered a manifestation of important
mental health problems of the perpetrators.
For the victims it is widely recognized as a risk factor for mental
health problems such as depression and relationship distress. As indicated previously, violence in
university student dating relationships is also associated with missing
classes, difficulty studying, and lower grades. This is not surprising in view of the increasing
evidence that family violence is implicated in a range of chronic physical
health problems (Hastings &
Kaufman Kantor, In press; Koss, Koss, & Woodruff, 1991; Leserman et al.,
1996; McCauley, Kern, Kolodner, Derogatis, & Bass, 1998), including the PI’s preliminary study which found an 11% rate of physical
injury. The results of cross-national
tests of etiological theories can have important health implications because we
anticipate that they will provide clues to the development of more culturally
relevant prevention and treatment approaches for intimate partner violence.
The high prevalence of DPV and the tendency for DPV to carry over into relationships later in life, suggests that, for purposes of primary-prevention (Cowen, 1978; O'Leary & Sweet Jemmott, 1995), it is vital to increase understanding of the etiology of dating partner violence (DPV) in order to provide part of the empirical basis for primary prevention programs. To what extent, for example, are cultural norms that define hitting a partner as sometimes acceptable, part of the etiology of DPV (as suggested by some research e.g., Paschall & Flewellng, 1997, Kaufman Kantor, 1987; Straus et al., 1980); and to what extent is it a function of conditions such as antisocial personality, or both? An answer to this question can suggest quite different foci for efforts at primary prevention of DPV.
If the hypothesized differences in
etiology for minor and severe violence are supported, primary prevention
programs might need to be differentiated to focus on minor violence as well as
severe violence. At present, public
service announcements on television and in other media focus almost entirely on
severe wife beating. Straus &
Kaufman Kantor (1994; 1997) have argued that this may be one reason their study
of trends in intimate partner violence from their 1975, 1985, and 1992 national
surveys found that there has been a decrease in the rate of severe violence by
men against their partners, but virtually no change in the rate of minor
violence against female partners. They
speculate that men who occasionally slap or throw something are not reached by
these announcements because such men can distance themselves from the extreme
violence behavior displayed.
If the hypothesis that the risk factors are different for minor and severe partner violence risk is supported, programs for assaultive partners may also need to be modified to deal differently with perpetrators of minor and severe violence than is now typical. The results of this study can also suggest diagnostic criteria for matching the mode of intervention to the characteristics of the client. In addition, as indicated above, DPV, like violence between married and cohabiting couples, is a risk factor for many other mental health problems such as depression (Larimer, Lydum, Anderson, & Turner, 1999; Morrow & Sorell, 1989; Sato & Heiby, 1991; Stets & Straus, 1990). Another aspect of mental health relevance is based on the previously cited studies showing that violence in dating relationships tends to carry over to marital relationships. Furthermore, growing up in a household with violent parents is associated with numerous mental health problems for the children (Holden, Geffner, & Jouriles, 1998; Straus, 1992). Thus, ending DPV can contribute to reducing mental health problems of both the current generation and the next generation.
·
It will
provide prevalence rates and assessments of risk factors for DPV that can
contribute to developing primary prevention and treatment programs.
·
The
cross-national approach permits testing theories concerning socio-cultural
context differences on DPV by examining the issue in diverse cultural settings
covering all the major world regions, and in some countries such as the USA,
Canada, and Israel, among different groups within each society such as
Francophone and Anglophone Canadians; and in the US, African American, Euro
Americans, and Mexican Americans.
·
It draws on
the experience and knowledge of a multidisciplinary team of investigators who
are familiar with the socio-cultural settings of each site.
·
It makes
available to the members of the Consortium the experience and assistance of a
Principle Investigator who has conducted many studies of family violence, and
who has also done previous cross-national research on the family; and who has
had 25 years of experience mentoring post-doctoral fellows.
·
The procedures
have been field tested with more than a thousand students at three different
universities in the USA, including minority populations. The Conflict Tactics Scales or CTS (see
Appendix 4) part of the questionnaire has been used successfully in more than
30 countries.
·
, The data set
will be archived and will permit other researchers to investigate many
important issues that cannot be accomplished in the three years of support
requested (for example, risk factors for psychological aggression), and an even
larger number of issues that only other researchers taking a fresh look at the
data perceive. The richness of the data set and the fact that many papers have
been published that report results from analyses of the archived data from the
PI’s two previous NIMH funded projects, suggests that this is will be an
important contribution.
This section supplements the
description of the preliminary work in the Background and Significance section
and the Research Design sections.
The Principle Investigator, Murray
A. Straus, is one of the pioneer researchers on intimate partner violence. The book he co-edited Violence In The
Family (Steinmetz & Straus,
1974) and his other publications in the 1970’s (Gelles
& Straus, 1979; Steinmetz & Straus, 1973; Straus, 1971b, 1973, 1974a,
1974b, 1976, 1977a, 1977b, 1977c) helped define “family violence” as a field of
scientific research. He was a member of
the National Academy of Sciences panel on child abuse research, and the
American Psychological Association Presidential Task Force On Violence And The
Family.
Straus has also done pioneering
research on cross-national comparative studies of the family. His study of working class and middle class
families in Minneapolis USA, San Juan Puerto Rico, and Bombay India resulted in
both substantive contributions (e.g.,
Straus & Straus, 1968; Straus, 1968a, 1971a) and methodological contributions (e.g., Straus, 1968b; Straus, 1969; Straus
& Tallman, 1971). The
combination of his interest in family violence and cross-national comparative
research led to a theoretical article on “Societal Morph genesis And
Intrafamily Violence In Cross-Cultural Perspective” (Straus, 1977a). The
International Dating Violence study brings these two interests together again.
In collaboration with a number of
colleagues, Straus has investigated many issues that are central to this
proposal, including the first national epidemiological survey to estimate the
prevalence of intimate partner violence and studies of risk factors for partner
violence, especially social risk factors (Straus
et al., 1980), and also research using societal-level variables to
test the patriarchic society theory of intimate violence against women (Straus, 1994, see Appendix 7).
The questionnaire for this study has been developed, pre-tested, revised, and used in research on dating violence with more than a thousand students at four universities in the USA, including Mexican and Mexican American students at the University of Texas, El Paso (Ramirez, 2001), and is currently being administered to a sample of about 300 students at the Universidad Autonoma de Ciudad Juarez, Mexico. The questionnaire consists of four parts: a demographic section, the revised Conflict Tactics Scales, the Personal and Relationships Profile, and a section for questions to measure variables of specific relevance for the society and culture of each site. The two core components (the CTS and the PRP) are briefly described below and additional information is given in the Research Design section. Some of the preliminary results from the use of this questionnaire were presented in the preceding Background and Significance section.
The development of the CTS (Straus, 1979, 1990a) (Straus et al., 1996) is one of the PI’s most important contributions to the development of family violence as a field of scientific research, and to the research proposed in this application. The CTS was instrumental in enabling a phenomenon that was previously restricted to clinical samples to be investigated by epidemiological survey methods, and it makes the proposed research feasible. It is the predominant instrument in family violence research and has been used in many countries. About ten journal articles based on data obtained using the CTS are published every month.
A major preliminary work for this investigation is the development of the Personal And Relationships Profile (Straus et al., 1999; Straus & Mouradian, 1999). The PRP was designed with one of the theoretical foci of the proposed research in mind: the Differential Risk Factor controversy. It measures both psychological and social risk factors and thus permits the comparison described in Specific Aim 2, as well as providing the data to test other key psychological and social theoretical issues such as the patriarchic society theory.
There have been three uses of the questionnaire containing these four parts (demographics and dating information, the CTS, the PRP, and locally salient measures) for comparative research on DPV. The first by Ralf Lindman of the Abo University in Finland used a preliminary version of the PRP to investigate the relation of “gender hostility” (as measured by the Hostility to Women and the Hostility to Men scales of the PRP. Lindman found that hostility to women is higher in the US than Finland, and that it is a risk factor for American dating couples, but not for Finnish couples (Aromaki, Lindman, & Straus, 2001).
The second study is by co-investigator for this project, I. Luis Ramirez compares Mexican American students with non-Mexican students (Ramirez, 2001). Ramirez used multinomial logistic regression to investigate the relation of PRP scales for Criminal History and Social Integration to violence against a partner. The analysis controlled for social desirability and nine other variables. He found that (1) Mexican American ethnicity was associated with an increased probability of minor violence, but was not related to severe violence, and (2) Criminal History was associated with an increased probability of severely assaulting a dating partner, but not associated when the measure of violence was restricted to minor assaults. Because the Criminal History scale of the PRP is one of the pathology risk factors and the Social Integration scale is one of the social risk factors measured by the PRP, these results are consistent with the differential risk factor theory, and also illustrate the potential of the PRP as a tool for testing hypothesizes derived from that theory.
As mentioned previously, a third comparative use of the instrument for this research is now in progress. Ramirez translated the questionnaire into Spanish and is administering it to students at the Universidad Autonoma de Ciudad Juarez, Mexico.
The preliminary
studies, as well as many other studies of DPV, indicate that although minor
violence predominates (30%) in university student samples (as in general
population samples), there is a sufficiently high prevalence of severe violence
(9%) and injury from both minor and severe violence (11%) to permit
investigating differences in risk factors for minor and severe violence.
Part of the preliminary work that makes the proposed study possible was the development of a multidisciplinary international consortium who will gather and analyze the data for their location. At the time this application was written, the consortium already included investigators at 35 sites, in 21 nations. The members of the consortium have agreed to translate and back translate the questionnaire, and to administer it under the rules established by the University of New Hampshire IRB, and in compliance with specifications for protection of human subjects of their own institutions. They have further agreed not to modify the instrument, except where an issue of “conceptual equivalence” (Straus, 1969) makes that essential, and only after consultation with the PI.
The research will be conducted in
each site by members of the International Dating Violence Research Consortium
who are familiar with local conditions and the culture of the site (see the
biographical sketches). They will use a
core questionnaire that will be as consistent as possible across all sites, but
modifications needed to maintain “conceptual equivalence” (Straus, 1969) will be made after careful review and consultation
between the PI and the consortium member.
In addition, space is provided in the questionnaire for each member of
the Consortium to include questions needed to measure issues that are uniquely
important for a site. For example, the
study of Mexican American students by Ramirez (2001), described in the Preliminary Studies section C2,
included a measure of acculturation to Anglo American culture. These procedures allow the benefits of both
standardized measures for all the sites, and also the benefits of culturally
informed investigations of unique issues in each site.
The proposed budget
includes funds for the PI and Co-Investigator to attend two international
conferences per year at which we will convene meetings of members of the
consortium to discuss and develop procedures.
However, as valuable as they are likely to be, at best, only about a
quarter of the members will be able to attend any one of these meetings. Consequently, an important step to achieve
the sharing of ideas and to achieve a balance between cultural appropriateness
and consistency across the sites will be the creation of a consortium list
server. Consortium members will be able
to raise issues and discuss problems on the list serve so that the entire group
can gain from participation in the problem solving process. One of the first issues to be dealt with in
this way is the definition and terminology for “dating” to use in the
questionnaire for each site. See the
definition in section B2 and the procedures in section D3.
The research will
be conducted under the terms of an explicit agreement (see Appendix 1) that specifies the rights and
obligations of the consortium members and Principle Investigator. This agreement is intended to protect the
intellectual property rights of all the investigators, while also facilitating
cross-national analyses and publication of the results.
The participants will
be university students who are age 18 years or older attending universities in
30 countries. At the time this
application was submitted, investigators at 35 sites in 21 countries had joined
the consortium. Additional recruitment
steps to add nine nations for a total of 30 nations will include contacting the
members of the International Division and the Women and Crime Division of the
American Society of Criminology, the International Division of the National
Council On Family Relations, and other similar groups. These steps were temporarily put aside to
prepare this application. The
probability of reaching the target of 30 nations seems high because the recruitment
of the current group of members from 35 sites in 21 nations was accomplished in
only two months (July and August 2001).
A further indication of the high probability of adding to the consortium
occurred at an international conference the PI attended the week before
finalizing this application. In the
course of telling some others at the conference about the project, four
expressed an interest and have been sent materials. Of course, there is also likely to be attrition from the 35 sites
already arranged. But even if the
additional members were only sufficient to make up for the attrition, a total
of 21 nations and 35 sites will still provide the basis for accomplishing
almost all of the Specific Aims.
The
sites are the university locations of the investigators who have agreed to be
part of the Dating Violence Research Consortium. The number of cases to be tested at the 35 sites so far arranged
totals to an N of approximately 7,000.
With the planned 30 sites the N will be much larger. Although these sites will not be a random
sample of the world’s nations, all continents and many ethnic groups are
included and they will provide data on differences in national socio-cultural
characteristics that are needed to test key social context theories, such as
those identified in section B6. The
sample also includes African American and Mexican American students, permitting
multicultural approach to the US data.
When there is more than one site in a country (as in Canada, Israel,
Switzerland and the USA) the consortium members will form a country-consortium
to carry out within-nation comparative analyses. Where there is only one site in a culturally diverse nation such
as Mexico, reporting of results will remind readers that the data refer to a
specific site and may not describe students at other sites in that country.
Because the
questionnaire is administered during a regularly scheduled class period, one of
the most serious threats to the validity of the sample – self-selection bias –
is minimal. This is indicated by the
preliminary studies in the US and Finland where over 90% of the questionnaires
distributed were completed. That some
blank questionnaires were turned in indicates that the voluntary nature of
participation was effectively communicated.
INVESTIGATOR |
SITE |
EMAIL ADDRESS |
TARGET N |
|
Hawkins, Russell Flinders
University, |
Australia |
|
450 |
|
Aldrighi, Tania Sao Paulo
Catholic Univ. |
Brazil |
|
150 |
|
Bennett, Lorna Memorial
University of Newfoundland |
Canada |
|
100 |
|
Brownridge,
Douglas University of
Manitoba |
Canada |
200 |
|
|
Cantin, Solange Université de
Montréal |
Canada |
|
300 |
|
Gagne,
Marie-Helene Universite Laval, |
Canada |
100 |
|
|
Hébert, Martine McGill University |
Canada |
150 |
|
|
Laporte, Lise McGill University Malo, Claire Research
Institute, Soc. Dev of Youth Chamberland, Claire University de
Montreal |
Canada |
llaporte@mtl.centresjeunesse.qc.ca cmalo@mtl.centresjeunesse.qc.ca |
100 |
|
Rinfret Raynor,
Maryse University de
Montreal |
Canada |
100 |
|
|
Sirkia, Diane T. Univ. of British
Columbia, |
Canada |
500 |
|
|
Walsh, Christine McMaster
University Trocme, Nico University of
Toronto MacMillan,
Harriet Queen’s
University |
Canada |
200 |
|
|
Lindman, Ralf Abo University, |
Finland |
|
200 |
|
Chan, K. L.
Edward University of
Hong Kong |
Hong Kong &
Shanghai |
|
150 150 |
|
Panchanadeswaran, Subadra University of Maryland |
India |
|
200 |
|
Enosh, Guy Haifa University Goldblatt, Hadass Barllan
University |
Israel |
280 |
|
|
Finzi-Dottan,
Ricky Haifa University |
Israel |
60 |
|
|
Shapira-Berman,
Ofrit Hebrew University |
Israel |
|
180 |
|
Yassour-Borochowitz, Dalit Emek Yezreel College |
Israel |
|
380 |
|
Ramirez, Ignacio
Luis Univ. of New
Hampshire |
Mexico |
|
C. Juarez 300 El Paso * 400 |
|
Rose-Junius,
Hetty. University of
Nambia |
Namibia |
|
50 |
|
Lamers-Winkelman, Francien Vrije University |
Netherlands |
|
100 |
|
Wurtzburg, Susan
J. Univ. of
Canterbury |
New Zealand |
|
100 |
|
Fawad, Rana University of the
Punjab |
Pakistan |
|
150 |
|
Vilchez Roman,
Carlos Vara Horna,
Aristides Minorities
Defense Assoc. |
Peru |
|
200 |
|
Figueiredo,
Bárbara University of
Porto Paiva, Carla University of
Porto |
Portugal |
|
100 |
|
Maw, A. Anastasia University of
Cape Town |
South Africa |
|
100 |
|
De Paul, Joaquin Univ. of Paris,
Vasco |
Spain |
|
300 |
|
Yodanis, Carrie University of Switzerland Sieber, Corinne Boeckmann, Irene University of Switzerland |
Switzerland Site French Site German Site |
|
300 |
|
Koski, Douglas Univ. of New
Hampshire |
Taiwan |
|
? |
|
Stanley, Nicky University of
Hull. London |
UK |
N.E.Stanley@comhealth.hull.ac.uk |
200 |
|
Ramirez, Ignacio
Luis Univ. of New
Hampshire |
USA Mexican-American
. |
|
C. Juarez 300 El Paso * 400 |
|
Smithey, Martha Univ. Texas, El
Paso |
USA |
African-American 300 |
|
|
Straus, Murray A. Univ. of New
Hampshire |
USA |
|
* 500 |
*Data already available
The data for this research will be
obtained by a questionnaire (see Appendix 3) that takes one class period to
complete. As indicated in the Preliminary
Studies section, this questionnaire has been used in previous research in four
US universities, in Finland, and is currently being administered in
Mexico. . The questionnaire consists of
four parts: (1) Demographics and
structure of the dating relationship, (2) the Conflict Tactics Scales (to
obtain data on partner violence), (3)
the Personal And Relationships Profile (to obtain data on risk factors for
partner violence), and (4) questions to measure locally salient variables.
An important step
in the development of the questionnaire for use at each site is to determine
locally appropriate terminology to use in the questionnaire to refer to and
describe the dyadic relationship that in the USA is called a “dating.” As explained in section B2, where dating was
defined, there are likely to be important site-to-site differences in the
terminology and the way dating relationships are identified and in the nature
of dating relationships. Consequently,
one of the first tasks for each member of the consortium will be to use
qualitative methods to investigate the terminology and definitions used by
students at their site. They will then
post this information on the list serve described in section D1 so that other
members of the consortium and the PI can read and comment on it (Instant messages might be even better
approach, but that technology is not available to all members of the
consortium.), after which the consortium member in consultation with the PI
will post the terminology and descriptive phrases to be used in the
questionnaire for that site.
These questions are on the demographic background characteristics of the respondent (age, sex, parent’s education, etc.) and the structural characteristics of the dating relationship, such as the length of the relationship at the time of completing the questionnaire, the sex of the partner, whether the relationship includes sex, and whether the respondent is cohabiting with the partner.
The CTS2 (Straus et al., 1996) is a revision of the Conflict Tactics Scales (Straus, 1979, 1990a). The CTS2 is a behavioral self-report instrument that includes scales to measure the four types of maltreatment of a partner in a dating, cohabiting, or marital relationship that are the focus of this study: Physical Assault, Injury, Sexual Coercion, and Psychological Aggression (see the paper on the CTS in Appendix 4). For each of these aspects of partner violence, the CTS provides annual prevalence rates (the percent who experienced the behavior in the previous 12 months), severity (minor and severe in the sense of high risk of injury), and chronicity (the frequency of the behavior among those who experienced it). Although the proposed research will give priority to perpetration pf physical assaults on dating partners, data from the Injury, Sexual Coercion, and Psychological Aggression scales will also be analyzed, and time permitting, we will explore creating typologies and other methods of taking the occurrence of two or more of these forms of DPV into account (see section D6 for examples of typologies).
As indicated under Preliminary
Studies, the CTS is the most widely used instrument for obtaining data on
partner violence, and has sometimes been described as the standard
instrument. It is the only instrument
designed to differentiate between Minor and Severe aggression between partners
and to have explicit procedures for doing so for each of the four aspects of
partner violence. It is also the only
standardized instrument that provides data on the chronicity (how many times it
occurred among those who engaged in each behavior) for each aspect of partner
violence. A unique feature of the CTS
is that it obtains separate data on both victimization and perpetration. This is psychometrically appropriate because
both are behaviors the respondent has personally experienced. Thus, the CTS permits investigation of both
victimization and perpetration of acts of violence against a partner. The analyses described in this application
refer to perpetration because we want the results to inform programs of primary
preemption and treatment of offenders.
We may also subsequently apply for funds to analyze victimization. In addition, the data will be archived and
available to other investigators for whatever purpose they choose, including
analyses of victimization.
The CTS2 differs from the CTS1 mainly in the addition of scales to measure Injury and Sexual Coercion. The two core scales (Physical Assault and Psychological Aggression) differ from the CTS1 mainly by having a few additional items. Thus, the psychometric data reported below, which is mostly based on the CTS1, is likely to also apply to these scales in the CTS2.
Both the original and the revised
CTS have strong psychometric characteristics, as shown in Appendix 5 which
provides “Tabular Summaries Of Methodological Characteristics Of Research Using
The Conflict Tactics Scales” (Yodanis, Hill, & Straus,
1997) and a
recent meta analysis (Archer, 1999). The tables in the appended article summarize
methodological data from more than a hundred papers and books published up to
1997 by many investigators. Since 1997,
many studies reporting data obtained by use of the CTS have appeared and some
include additional psychometric information.
Table 3 of Appendix 5 provides
reliability coefficients from 80 studies.
The mean alpha married and cohabiting heterosexual couples is .88. The reliability coefficients for ethnic
minority groups and for gay and lesbian relationships are lower but still
adequate (mean of seven studies = .79).
Table 4 in Appendix 5 summarizes 31
factor analyses and shows that almost all either confirm the three-factor model
used to design the CTS (Negotiation, Psychological Aggression, Physical
Assault), differentiate the physical assault items into minor and severe
factors. This is a remarkable degree of
consistency for the results from a mode of analysis that is noted for its
fragility.
Table 2 in Appendix 5 summarizes a
variety of validity evidence from 72 different studies; for example, there are
31 studies reporting the degree of agreement between partners and almost all
show a high degree of concordance such as correlations of .42, .88, .51, and as
high as 80% agreement. However, these
figures can be viewed as inflated because the main area of agreement occurs because
most couples report no violence. The
agreement on the number of times or on specific assaultive acts is much
lower. However, when the specific acts
are combined into a scale, a study by Moffitt et al (Moffitt et al., 1997) led them to conclude that “reports of abuse can be
aggregated to form internally consistent scales that show strong interpartner
agreement.” Appendix 5 does not include
construct validity because the evidence is too extensive to make it practical
to include in a tabular summary (but see (Straus,
1990a) for an earlier summary of construct validity
evidence).
Particularly relevant for this application is the fact that the CTS part of the questionnaire has been used in more than 30 countries, including male-dominant societies such as Palestinian Arabs (Haj-Yahia & Dawud-Noursi, 1998; Haj-Yahia & Edleson, 1994). Table 7 of Appendix 5 lists 74 studies which used the CTS with racial/ethnic minorities in the US and in other nations. Moreover, since that table was compiled, many other studies have come to our attention. For example, the CTS was used in studies of maternal health sponsored by the World Health Organization in more than 20 countries. (Unfortunately, the investigators in many of these WHO studies modified the instrument in ways that were not necessary and which undermined the integrity of the data and also made cross-national comparisons inappropriate). An example of a recent major prospective study that used the CTS is the Dunedin {New Zealand} Multidisciplinary Health And Development Study (Moffitt & Caspi, 1999). It is also currently being used in many other studies, such as ADHEALTH, which is a large scale prospective study of American Adolescents.
A widely
used instrument such as the CTS is likely to also be closely scrutinized, and
there have been a number critical reviews (Dobash,
Dobash, Wilson, & Daly, 1992; Frieze & Browne, 1989; Pagelow, 1984;
Yllo & Bograd, 1988) as well as favorable reviews (Archer, 1999; Barling, O'Leary, Jouriles, Vivian, &
MacEwen, 1987; Grotevant & Carlson, 1989; Herzberger, 1991; Messer &
Reiss, 2000). See the
section on Comparison Of Original and Revised CTS in the Appendix 4 paper on
the CTS (Straus et al., 1996) for a response to some of the main criticisms.
The PRP is a 23 scale instrument designed explicitly for
research on partner violence. The
instrument is described in the paper in Appendix 6 (Straus & Mouradian, 1999). One of the
23 scales is adapted from the Reynolds short form of the Marlowe-Crowne Social
Desirability Scale (Reynolds, 1982). This permits
controlling for an important potential source of spurious associations (see
below). The 22 risk factor scales were
selected on the basis of a review of research on the correlates of couple
violence and theories concerning the etiology of couple violence, with
attention to including scales that measure variables to test psychological
theories and sociological theories of partner violence (see Specific Aim 2 and
Background section B5). The paper in
Appendix 6 (Straus & Mouradian,
1999) provides information on the
theoretical basis and measurement strategy used to create the PRP, and preliminary reliability and validity
coefficients. The scales in the PRP
are:
Personal or Intrapsychic Scales
ASP Antisocial Personality Symptoms
BOR Borderline Personality Symptoms
CH Criminal
History
DEP Depression Symptoms
GHM Gender Hostility to Men
GHW Gender Hostility to Women
NH Neglect
History
PTS Post-Traumatic Stress Disorder
SD Social
Desirability
SI Social
Integration
SUB Substance Abuse
STR Stressful Conditions
SAH Sexual Abuse History
VA Violence Approval
Relationship Scales (scales which include items that refer to behavior towards
or beliefs about the partner.)
AM Anger
Management
CP Communication
Problems
CON Conflict
DOM Dominance
JEL Jealousy
NA Negative
Attribution
RC Relationship
Commitment
RD Relationship
Distress
There is only limited data on the cross-cultural and cross-national utility of the PRP. It has been used in a study of Mexican Americans and a study in Finland, and data for a study in Mexico was being gathered at the time this application was being completed. Consequently, results are available for only two comparative analyses and there is no assurance that the PRP will be valid cross-nationally. However, the results for these two studies (summarized in Preliminary Studies section C2) are promising. The use of the Conflict Tactics Scales in more than 30 countries also suggests that instruments developed in the USA can have cross-national applicability. This is also suggested by many examples in the Journal of Cross Cultural Psychology of productive and presumably valid use of standard instruments across nations, such as the Holmes and Rahe stressful live events scale (McAndrew, Kande, Turner, & Sharma, 1998) and the Schwartz Value Survey or SVS (Inglehart, Basanez, & Moreno, 1998; Schwartz, 1992). Analyses of SVS data from 63 nations confirmed the 10 basic values structure of the instrument, and correlations of the value scores with other variables provided extensive evidence of construct validity. However, deviations from the 10 values structure have been found in sub-Saharan Africa, India, and Malaysia.
Schwartz’s argued that the problem in those societies was that the SVS “demands a high level of abstract thought and presents the value concepts outside of any specific context. The samples in which the {ten basic values} theory has failed to obtain support are almost exclusively from populations that have not been educated in Western school that emphasize abstract, context-free thinking” (Schwartz et al., 2001). He therefore developed an alternative instrument that uses more concrete items to measure the same 10 value constructs. When this instrument was used in South Africa, Uganda, Italy and Israel, the results were consistent with the 10 basic values theory. The differences between the SVS and the new instrument include (1) The items are brief descriptions of people such as “He wants people to do as he says” rather than abstract concepts ‘Dominance over other people” (2) Respondents are asked to indicate how similar the person described is to them, rather than to rate the importance for them of the abstract principle. (3) The number of response categories was reduced from 9 to 6. (4) Vocabulary and sentence structure were simplified. The design of the PRP follows these same four principles.
The fact that the PRP items at the 5th to 6th grade reading level facilitated translation into Spanish, and is likely to also facilitate translation into other languages needed for this research. Each translation will be checked for accuracy by two independent back-translations, and the “conceptual equivalence” of the items (Straus, 1969) will be carefully reviewed.
The success we have had using the PRP in two comparative studies, the many examples of successful use of American designed instruments in cross-national research (including the CTS), the fact that the PRP items correspond to the item and response structure that Schwartz has found effective in non-Western societies, and the careful translation and back-translation and review of conceptual equivalence, all suggest that the PRP is likely to be suitable for use in the 30 nations of this study.
The final part of the questionnaire consists of questions to measure locally salient variables and will provide data for one of the ways of meeting Specific Aim 3. These measures will be added to the questionnaire for their sites by the members of the Consortium (as in the example of the Acculturation scale in the study of Mexican Americans). The locally salient measure questions will be the final part of the questionnaire because these question will differ from site to site and placing them last avoids the possibility that they could influence responses to subsequent questions in ways that differ from site to site.
Some previous attempts at cross-national consortium based research have been handicapped by data management problems (Erbacher, Hyman, & DuCette, 2000), such as errors in data entry, and incorrect scoring of scales, and some who gathered the data did not provide copies of the data for cross-national analysis. The procedures to be used for this study are designed to ease the data management tasks for the Consortium members and to avoid these problems. If pre-testing indicates that this mode of response can be used at the site, respondents will answer the questionnaire using optically scanned answer sheets. The members of the consortium will send answer sheets to the PI for scoring and “data cleaning.” The PI will provide each consortium member with an SPSS, SAS, Stata, Excel, or ASCII data file containing responses to all questionnaire items, and also the scores for each respondent on the many scales computed from these items. In accordance with the agreement between the PI and each consortium member (see Appendix 1), each consortium member has the sole right to use the data for articles giving the results for his or her site, but gives the PI the right to retain a copy for purposes of cross-national analysis.
It is always important to begin data analyses with a careful consideration of the distribution of all variables to uncover distribution problems and to consider the implications of the central tendencies, range, and standard deviation of the variables. For the present study, the descriptive statistics take on added importance because they provide the data on the prevalence and chronicity of each of the four aspects of partner violence being studied (physical assault, injury, sexual coercion, and psychological aggression). In fact, these descriptive statistics are so important that the members of the consortium have agreed that the first papers written from the data for each site will present the rates for each of the four aspects of DPV measured in this study (see the suggested outline in section D10 below). We believe there will be wide public and professional interest in those rates and that they will be another stimulus to create prevention and treatment programs in a number of countries.
It is critical that research on intimate partner violence take into account the gender of the participant. Consequently, all analyses will be replicated for males and female respondents. When differences are found, the multivariate analyses will be designed to include tests for interactions with gender of the respondent.
Given the large sample size, there may be enough respondents
who are in gay and lesbian relationships to permit investigating the ways in
which those relationships are different and similar to heterosexual couples in
the prevalence and chronicity of physical assault, injury, sexual coercion, and
psychological aggression. We also plan
to include the gender composition of the relationship as an independent
variable in the regression analyses.
The results from the planned analyses might be spurious if both the PRP risk factor variables and the CTS variables are associated with social desirability (SD) response set. Although a number of studies have shown that the CTS is only minimally correlated (r’s of about .20) with SD (Straus, 1990a; Sugarman & Hotaling, 1996), the PRP scales have higher correlations with SD (r’s of about .30). However, the relationship between risk factors and violence as measured by the CTS remains after controlling for SD (Hunsley, Vito, Pinsent, James, & Lefebvre, 1996). Our own research using the PRP also found that when SD was controlled, in all but a few cases, after controlling for SD, the relation of risk factors to DPV remained significant (Straus & Mouradian, 1999; Straus & Ramirez, 1999).
The internal consistency reliability of each of the CTS scales and each of the PRP scales will be estimated for each site using Cronbach’s alpha. This is particularly important for a cross-national study because the reliabilities already established for these instruments in the USA do not necessarily apply in other social contexts.
We will conduct a confirmatory factor analysis (with the AMOS structural equation modeling program) as a means of establishing the extent to which the risk factors in the PRP fall into psychological and social domains. If this analysis confirms a two domain model, we will follow the pattern used for the Achenbach Child Behavior Checklist and create two “broad band” factor scores in addition to the “narrow band” scores for each of the specific risk factors.
The hypothesis
concerning differential risk factors (see Specific Aim A2 and section B5) will
be tested by comparing the strength of the association between social risk
factors and psychological risk factors with minor violence (measured as
occurrence of minor violence, but no instance of severe violence) with the
strength of the association between these two types of risk factors with severe
violence. Before testing these
hypotheses, the distribution of each of the risk factor variables will be
examined. Departures from normality
will be corrected by an appropriate transformation. In addition to using these two “broad band” factors, we will also
run analyses for the “narrow band” scales in the PRP to see if the
psychological risk factor scales are more closely related to severe assaults,
and whether the social risk factors scales are more closely related to minor
assaults on a partner.
We will use multinomial logistic
regression (Hamilton, 1998) to test the Differential Risk Factor hypothesis. The dependent variables for these analyses
will be constructed by classifying the respondents into mutually exclusive
categories based on the severity or severity and chronicity of violence. One version will be a three category
variable:
1 = No violence
2 = Minor violence only
3 = Severe violence.
We will also conduct
analyses using categories that take into account whether the assault resulted
in injury and the chronicity of the violence.
For example, a typology that considers the chronicity of assaults could
have the following five categories:
1 = No violence
2
= Low frequency minor violence, defined as one or two instances of minor
violence in the past 12 months but no severe assault
3 = Chronic minor violence defined
as three or more instances of minor violence in the last 12 months and no
instances of severe violence
4 = One Instance Severe assaults
5 = Repeated Severe Assault defined as 2 or more
instances in the previous 12 months.
These methods of
classifying respondents into mutually exclusive types are necessary because
separate measures of minor and severe violence are correlated due to the fact
that almost all men who severely assault their partner also engage in minor
violence. The five-category typology
also deals with the fact that some respondents who do not severely assault
their partner engage in almost constant minor violence.
Similar categorizations
have been successfully used in previous research (Hotaling, Straus, & Lincoln, 1990; H. S. Pan et al., 1994;
Sugarman et al., 1996; Sugarman & Hotaling, 1996). The study by
the Co-Investigator described in the Preliminary Studies section C2 provides an
example of using the three-category typology as the dependent variable in a
multinomial logistic regression analysis to test the differential etiology of
minor and severe violence (Ramirez,
2001). Ramirez’s
study also illustrates the use of “conditional effect plots to provide a more
detailed understanding of the significant relationships, and as a tool for
communicating complex statistical information to professionals who are not at
home with regression coefficients (see
examples in Straus, 2001; and Straus
& Yodanis, 1996).
Much of the previous discussion focuses on the differences between psychological and social risk factors. psychological and social theories may be complementary rather than competing theories, and there may be interactions of psychological and social factors. For example, the most severe levels of violence against a partner may tend to occur when social risk factors and psychopathology are both present. Consequently, similar to the planned tests for interactions between societal-level and individual-level variables, we will also explore interactions between the psychological and the social risk factors.
In addition to the ”broad band” scales to measure psychological and social risk factors, the data will also allow examining the relation of the specific “narrow band” PRP scales to partner violence. For example, as sociologists, the PI and the Co-Investigator are particularly interested in the relation of male dominance and social integration to DPV. Consequently, analyses of those two risk factors will be the focus of much of our work. The PI and others have found that corporal punishment by parents is associated with an increased probability that the child will later as an adult, use the equivalent of “physical discipline” on a partner (Simons et al., 1998; Straus & Yodanis, 1996; Swinford et al., 2000). The data from the 30 nations in this study will permit research to test the view of cultural relativists who argue that in societies where corporal punishment is culturally defined as legitimate and necessary, it will not result in an increased tendency to violence as an adult. The other members of the Consortium will have data on 22 specific risk factors, as well as the variables of local interest that each built into the questionnaire for their site, and will therefore be able to examine still risk factors that are of particular interest to them.
Power analyses
needed to compare social and psychological risk factors (Specific Aim 2, the
differential risk factor hypotheses) were conducted using procedures outlined
by Cohen (1969). For the
hypothesis that social risk factors are more closely related to Minor Violence
than to severe violence, we assumed the correlation between social risk factors
and violence would be .40, and that the correlation between social risk factors
and Severe Violence would be .20.
Although these correlations are moderate, the difference represents a
fourfold difference in the size of these effects (16% versus 4% of the
variance). For the hypothesis
psychological risk factors are more closely related to severe violence, we
expect to find the reverse pattern. We
believe these estimates for the effect sizes correspond to those commonly found
in the partner violence literature (Hotaling
& Sugarman, 1986; Sugarman & Hotaling, 1991).
The difference between these two
effect sizes, or q, is .22, because the difference between the z
transformations of the above r’s is used to correct for the scaling of r. To achieve an alpha of .05 and power of .80,
this effect size produces a point estimate of 327 participants for both the
Minor Violence and Severe Violence groups.
For some of the sites, the N will be this larger or larger (see Sample,
above) and the total number of participants for the study (approximately 7,000)
far exceeds 327.
The societal-level
analyses (described below in section D9) will use 45 sites or 30 nations. The statistical power of 30 is low, but N’s
of this size are frequent in macro-level cross-national research. Consequently, to reduce the risk of Type II
error, the .10 level is often used, as in the recent study of helping behavior
observed in a sample of cities in 23 nations (Levine,
Norenzayan, & Philbrick, 2001). This study
used the .10 level for analysis involving the correlations between the measures
of helping because they expected the correlations to be relatively large; and
the .15 level to evaluate the statistical significance of tests of the theory.
The results from the PI’s previous
macro-level research (using the 50 states of the US as the units) suggest that
correlations of societal-level data are often larger than correlations of
equivalent data at the individual level (Baron
& Straus, 1989). Those
results, and the results of societal level research by others. lead us to
expect correlations of at least .20.
When the units are nations, an N of 30, using a one-tailed test (which
will apply to the analyses described in this application) provides sufficient
power because correlations of .175 or higher will be significant. When the units are sites and the N is 45,
correlations of .144 and higher will be significant. Therefore the planned N’s of 30 and 45 and alpha set at .05,
provide sufficient power for bivariate analyses. However, for multivariate analyses, the situation may be
different. For example, multiple or
partial correlations using five variables, require a correlation of .257 when
the N is 30, and .216 when the N is 45.
If the actual results show a consistent and theoretically meaningful
pattern of multivariate results that do not quite reach the .05 level of
significance, we plan to use the .10 level to keep the risk of Type II error in
multivariate analyses to an acceptable level.
While the principal focus of this
study is on physical assaults, use of the CTS2 will also provide data on
Injury, Sexual Coercion, and Psychological Aggression. The Consortium members are committed to papers
presenting the rates for these aspects of DPV at their site. We would also like to conduct analyses to
investigate the relationship of the risk factors in the PRP to each of these
aspects of DPV. Given the range of data
to be analyzed for just the aspect of main interest (physical assault), we are
not sure how much of that will occur within the three year period of
funding. However, some consortium
members may do this and, as pointed out elsewhere, the data will be archived
and available to researchers generally.
Three approaches will be used to
obtain data on culturally relevant social context variables.
Because of the
increasing availability of international social indicators, there is data to
measure hundreds of socio-cultural characteristics of the nations that are
relevant for this research. What once
took many months of library and fieldwork to compile is now readily available in
compendia listing the data by nation. The data is derived from many sources such as the census, public
health, national income, and crime reporting agencies. Some examples are:
Adult literacy rate and percent of
university graduates (UNDP, 1999)
Female literacy as a percent of male
literacy (UNICEF, 1998)
Income Inequality (UNDP, 1999)
Homicide death rates (Newman, 1999)
Human Development Index (UNDP, 1999)
Human Rights Index
Stressful Social Environment Index. This will be created using the same type of
indicators that the PI and colleagues used to create a social stress index for
each of the 50 US states (Linsky et al.,
1995; Linsky & Straus, 1986).
Percent of Gross Domestic Product spent on
the military and on education and human services, and the ratio military to
education and human services expenditures (UNDP,
1999)
Many variables from the Cross Cultural
Sourcebook On Human Values and Beliefs (Inglehart
et al., 1998) such as religious orientation as measured by “How
important is God in your life” and the
frequency of religious participation.
There are also composite scales on many beliefs and values
As indicated previously, the standard questionnaire includes provision for a block of questions to permit each consortium member to include data to measure variables that are of specific relevance for their socio-cultural setting or population, as in the previously described example of an acculturation scale used in the study of Mexican Americans in El Paso, Texas (Ramirez, 2001).
The individual-level data for each site will be used to create additional social context variables for each site. For example, the Violence Approval scale of the PRP measures the extent to which the participant subscribes to beliefs and attitudes approving violence. The SPSS procedure AGGREGATE will be used to attach the mean or median Violence Approval score for the site to the record of each participant to provide an indicator of the extent to which Violence Approval is characteristic of that social context. The same procedure can be used to aggregate the data from the Communication Problems scale of the PRP to create an indicator of the extent to which communication between partners is a problem at each site. These variables can be used for macro-level analyses in the same way as census data on median education level of each country is used to investigate whether the education level of a nation is associated with intimate partner violence. The PI, for example, used this approach to create a measure of social stress for each of the 50 states and to examine the extent to which social stress is associated with psychological problems, health problems, and criminal behavior (Linsky et al., 1995).
An important objective of this research is to investigate the effect a macro-level variables, such as the degree to which the national or site context is male-dominant, on an individual-level relationship, such as the relation of alcohol abuse to physical or psychological aggression against a dating partner. The hypothesis would be that there is a stronger relation between alcohol abuse and DPV in a male-dominant context. Many national-level variables mentioned above that are available in compilations of national characteristics, or which can be created by aggregating the individual-level data, can be used to test other contextual effect theories.
A problem with the types of analysis described above is that cases within a given context tend to be more similar to each other than cases randomly sampled from the entire population. For example, students in a university in a particular nation are more similar to each other than to students randomly sampled from universities in the nation, and obviously even more similar to each other than to university students from all nations. Students within a particular university tend to come from a community or community segment that is more homogeneous in terms of values, family background, socio-economic status, race or ethnicity, religion, and educational preparation than the population as a whole. Further, students within a particular university share the experience of being in the same environment, which may lead to increased homogeneity over time. Consequently, observations based on these individuals are not fully independent. However, most analytic techniques require independence of observations as a primary assumption for the analysis and this assumption is violated in the presence of hierarchical data. In turn, this inflates the probability of rejecting a null hypothesis.
(Bryk & Raudenbush, 1992; Goldstein, 1988) were designed to deal with these problems in research on contextual issues and to permit disentangling individual and group context effects on DPV. The HLM3 program can analyze multiple levels. For this research, both two-level models (nations and individuals) and perhaps also three-level models (Nation, site, individual) will be used. The program can perform non-linear analysis appropriate for counts and binary, multinomial, or ordinal data. Using the HLM3 program, each of the levels in the data structure (persons within, sites, within nations) is represented by its own sub-model. In a two-level analysis, the level-1 model will represent the relationships among the individual-level variables and the level-2 model will capture the influence of national level factors. For example, in the proposed research, regression equations will be computed for the relation of the Dominance Scale score of the PRP to violence against a dating partner for each nation surveyed. These regressions indicate the micro-level effects of dominance of one partner on violence against the other partner. At the macro-level of analysis, saved intercept and slope values form outcome variables of statistical models of higher-level (national) effects. Separating the individual effects from either contextual or compositional effects can be done by subtracting the national mean from the individual score on an independent variable.
Multilevel linear modeling can also be used to control for
potential confounds that might lead to spurious results. For example, the participants in this
research will all be university students, but this does not assure that they
come from equivalent sectors of the population because there are large
differences between nations in the selectivity of universities. Consequently, one of the macro-level
variables to be analyzed will be the percent of the eligible age population who
attend a university in each of the nations.
Similarly, we will also treat the percent of women students and the
percent of students in dating relationships as contextual variables.
There will be at least four types of
publications. First priority will be
given to publication of journal articles in refereed journals.
These will cover physical assault, injury, sexual coercion, and psychological aggression. The articles will, to the extent possible, follow a suggested standard format (see below), and will emphasize interpreting the results in the light of the socio-cultural characteristics of the site. The PI will be a consultant to the consortium members doing the data analyses and writing of these articles.
PREVALENCE
AND CHRONICITY
OF DATING PARTNER VIOLENCE AMONG
A SAMPLE OF {insert site reference such as
Mexican} UNIVERSITY STUDENTS
INTRODUCTION
Importance of the
issue, including local relevance
Gender differences
Prevalence and chronicity in the USA (I will
provide the consortium member one or two draft paragraphs to use or not use as
seems appropriate)
What is known about DPV
in the Consortium members country or site
METHODS
SAMPLE
How the sample was selected and how the questionnaire was
administered. Give the number of
cases. Include a table such as Table 1
in paper PR15 (available on my website.
Note this is not the same as CTS15 mentioned below) and describe the
implications
MEASURES
I will provide consortium members with the draft of a section describing the CTS.
DATA ANALYSIS
I will
provide consortium members with a draft of this section to work from.
RESULTS
PHYSICAL ASSAULT
INJURY
SEXUAL COERCION
PSYCHOLOGICAL AGGRESSION
I suggest you present the data in a table like Table 4 of paper CTS15 (this paper is in the meeting handout, or you can print it out from my website http://pubpapges.unh.edu/~mas2). However, note that this is a methodological paper and only presents the data in a table. It does not discuss each of the statistics, as you will be doing when reporting the results for your site. Also, it presents only the results for the overall scale scores. For IDV consortium papers, it will be important to also present results for minor and severe levels of each scale. Finally, I suggest adding a frequency distributions for some or all of the scales.
Another example that may be useful is a paper on the prevalence and Chronicity of corporal punishment by parents. This is paper CP36 in the Corporal Punishment section of my website
DISCUSSION
Some
examples of things that you might discuss are
(1) What do the findings tell us about the nature and extent of violence
in these types of relationships, and perhaps other types of intimate partner
relationships, at your site? (2) What
accounts for the findings at your site?
(3) What are the limitations of the results and cautions that readers
need to keep in mind? (4) What are the
implications of the findings for understanding, preventing, and treating
violence in intimate partner relationships in your social context?
For an example of a paper on prevalence and chronicity paper that is in some ways analogous to the paper in this outline, log into my website http://pubpapges.unh.edu/~mas2. Look in the section on Corporal Punishment for paper CP36.
These articles will report on the experience of administering these instruments in very diverse societies, and report the results of psychometric analyses such as the degree of similarity and difference across nations in the factor structure of each instrument. The PI has previously published articles of this type and will draw on this experience to encourage and assist Consortium members to do such analyses. However, the chances are that the psychometric papers will mostly be the responsibility of the PI and the Co-Investigator.
Articles are planned comparing the prevalence of physical assault, injury, sexual coercion, and psychological aggression in the societies covered by the study. These articles will use ANCOVA and bar charts to compare sites and offer theoretical interpretations of the similarities and differences. The ANCOVA’s will control for demographic and other national characteristics that could result in spurious correlations such as the median education of the nation and the eliteness university attendance. Although four articles are planned (one for each aspect of DPV measured by the CTS), my past experience indicates that once data analysis and writing is underway, unforeseen opportunities for additional important articles emerge.
The paper in appendix 8 on criminal history and partner violence (Straus & Ramirez, 1999) is an example of a specific risk factor that will be examined cross-nationally. This paper shows, as have a number of previous studies, that many university students report having committed criminal acts, and that a criminal history is related to partner violence, after controlling for nine other variables. A number of other such papers, using one or more of the risk factors measured by the PRP will be written in addition to the paper testing the differential risk factor hypotheses. For each such article, email letters will be sent to members of the consortium inviting those interested in research on that particular risk factor to be a co-author of the paper.
Some of the cross-national articles will be macro-level analyses and some will be multi-level contextual analyses. Some papers are planned which will follow the widely used format of testing theories by multiple regression analysis of macro-level data, as in the PI’s research on the relation of state-to-state differences in gender equality to rates of partner violence against women (Straus, 1994). Priority will be given to a paper similar to the one just cited to test the patriarchic society theory by regressing the rate of DPV on the degree of male dominance in the society. The analysis will be replicated using several different measures of male dominance, some of them from official statistics and some from aggregating the individual-level data (as explained above). Finally, there will be papers using multilevel modeling, as described above, to test theories involving contextual effects. One example of the many contextual-effect theories that can be tested concerns the relation of corporal punishment to later physical aggression against a partner (Straus & Yodanis, 1996). Critics of this study believe that this occurs only if corporal punishment is experienced in a cultural context where it is defined as an expression of parental hostility rather than as a normal part of parental behavior (Gunnoe & Mariner, 1997). This cultural context theory will be tested by using HLM to examine the relation of corporal punishment to DPV in societies where corporal punishment is normative and in others such as Sweden that have made all corporal punishment by parents illegal.
A monograph with the tentative title VIOLENCE IN DATING
RELATIONSHIPS: A MULTI-NATION STUDY (table of contents is below) will be
written in the final year of the study.
A monograph length report is necessary because, although the journal articles will document
and interpret the basic results of the research, they will do so on a
site-by-site basis, or on a topic by-topic basis. Consequently, there is also a need for a monograph integrating
the results from all the sites and paying attention to the interrelation of the
four aspects of DPV covered by the study.
This will be a time-demanding and difficult work to produce. However,
that this is an attainable goal is suggested by the publications from two previous
NIMH grants to the PI for large scale studies
The grant applications for both indicated a monograph length report, and
three were published (Gelles &
Straus, 1988; Straus & Gelles, 1990; Straus et al., 1980).
VIOLENCE IN DATING RELATIONSHIPS:
A MULTI-NATION STUDY
Part I. ISSUES AND METHODS
1. Violence Between Dating Partners In World Prospective
(Introductory chapter
to explain the theoretical and practical contributions of the study)
2. Method
(In
addition to the usual description of the sample, measures, and data analysis
techniques, this chapter will give reliability estimates for key measures for
each site)
Part II.
PREVALENCE AND CHRONICITY
(Each chapter will analyze differences
between sites in prevalence and chronicity)
3. Physical Violence
4. Injury
5. Sexual Coercion
6. Psychological Aggression
Part III.
RISK FACTORS FOR VIOLENCE
7. Demographics
8. Social Characteristics
9. Psychological Characteristics
PART IV. THE FUTURE
10. Psychological and Social Risk factors
10. Implications For Prevention And Treatment
APPENDIX
1. Core Questionnaire
2. Additional Measures For Specific Socio-Cultural Contexts
Andrews, G.,
Ustun, T. B., & Kessler, R. C. (2000, December 4-6). Shutting the stable door: Identifying avertable risk factors for mental
disorders. Paper presented at the WHO Burden of Disease Meeting, Auckland,
New Zealand.
Archer, J. (1999). Assessment of the
reliability of the Conflict Tactics Scales: A meta-analytic review. Journal of Interpersonal Violence, 14(12),
1263-1289.
Aromaki, A. S., Lindman, R. E., &
Straus, M. A. (2001). Cross-Cultural
prevalence of gender hostility: an exploratory study. Unpublished
Manuscript, Department of Psychology, Abo Akademi University, Turku, Finland.
Bagarozzi, D. A. (1989). Family diagnostic
testing: A neglected area of expertise for the family psychologist. AThe American Journal of Family Therapy, 17,
257-272.
Barling, J., O'Leary, K. D., Jouriles, E.
N., Vivian, D., & MacEwen, K. E. (1987). Factor similarity of the conflict
tactics scales across samples, spouses, and sites: Issues and implications. Journal of Family Violence, 2(1), 37-54.
Barnett, O. W., Miller-Perrin, C. L., &
Perrin, R. D. (1997). Family violence
across the lifespan: An introduction. Thousand Oaks, CA: Sage Publications.
Baron, L., & Straus, M. A. (1989). Four theories of rape in American society: A
state-level analysis. New Haven: Yale University Press.
Baron, L., Straus, M. A., & Jaffee, D.
(1988). Legitimate violence, violent attitudes, and rape: A test of the cultural
spillover theory. In R. A. Prentky & V. L. Quinsey (Eds.), Human Sexual Aggression Current Perspectives
(Vol. 528): Annals of the New York Academy of Sciences.
Bowker, L. H. (1993). A battered woman's
problems are social, not psychological. In R. J. Gelles & D. R. Loseke
(Eds.), Current controversies on family
violence (pp. 154-165). Newbury Park, CA: Sage.
Browne, A., & Williams, K. R. (1993).
Gender, intimacy, and lethal violence: Trends from 1976 through 1987. Gender and Society, 7(1), 78-98.
Bryk, A. S., & Raudenbush, S. W. (1992).
Hierarchical linear models. Newbury
Park, CA: Sage Publications.
Campbell, J. C. (Ed.). (1995). Assessing Dangerousness: Violence by sexual
offenders, batterers, and child abusers. Thousand Oaks, CA: Sage
Publications.
Cano, A., Avery-Leaf, S., Cascardi, M.,
& O'Leary, K. D. (1995). Correlates
of dating violence in a high school sample. Paper presented at the The
Fourth International Family Violence Conference, Durham, NH.
Carroll, J. C. (1980). A cultural
consistency theory of family violence in Mexican-American and Jewish-Ethnic
groups. In G. T. Hotaling & M. A. Straus (Eds.), The social causes of husband-wife violence (pp. Chapter 5).
Minneapolis: University of Minnesota Press.
Cohen, J. (1969). Statistical powers analysis for the behavioral sciences. New York:
Academic Press.
Coleman, D. H., & Straus, M. A. (1990).
Marital power, conflict, and violence in a nationally representative sample of
American couples. In M. A. Straus & R. J. Gelles (Eds.), Physical violence in American families
(pp. 287-304). New Brunswick, NJ: Transaction.
Cowen, E. (1978). Demystifying primary
prevention. In D. G. Forgays (Ed.), Primary
Prevention of Psychopathology (pp. Chapter 2). Hanover, NH: University of
New England Press.
Dobash, R. E., & Dobash, R. P. (1979). Violence against wives: A case against the
patriarch. New York: Free Press.
Dobash, R. P., Dobash, R. E., Wilson, M.,
& Daly, M. (1992). The myth of sexual symmetry in marital violence. Social Problems, 39(1), 71-91.
Downs, W. R., Miller, B. A., & Panek, D.
D. (1993). Differential patterns of partner-to-woman violence: A comparison of
samples of community, alcohol-abusing, and battered women. Journal of Family Violence, 8(2), 113-135.
Dutton, D. G. (1994). Patriarchy and wife
assault: The ecological fallacy. Violence
and Victims, 9(2), 167-182.
Dutton, D. G., & Starzomski, A. J.
(1993). Borderline personality in perpetrators of psychological and physical
abuse. Violence and Victims, 8(4),
327-337.
Erbacher, T. A., Hyman, I. A., & DuCette,
J. (2000, August 4). Cross cultural
perspectives on corporal punishment: Methodological issues. Paper presented
at the Symposium at the National Convention of the American Psychological
Association, Washington, DC.
Feld, S. L., & Straus, M. A. (1989). Escalation
and desistance of wife assault in marriage. Criminology,
27(1), 141-161.
Flournoy, P. S., & Wilson, G. L. (1991).
Assessment of MMPI profiles of male batterers. Violence and Victims, 6(4), 309-320.
Frieze, I. H., & Browne, A. (1989).
Violence in marriage. In L. Ohlin & M. Tonry (Eds.), Family violence (Vol. 11, pp. 163-218). Chicago: University of
Chicago Press.
Gayford, J. J. (1975). Wife battering: A
preliminary survey of 100 cases. British
Medical Journal, 1, 194-197.
Gelles, R. J. (1983). An exchange: Social
control theory. In D. Finkelhor & R. J. Gelles & G. T. Hotaling &
M. A. Straus (Eds.), The dark side of
families: Current family violence research (pp. 151-165). Newbury Park, CA:
Sage.
Gelles, R. J., & Straus, M. A. (1979).
Determinants of violence in the family: Towards a theoretical integration. In
W. R. Burr & R. Hill & F. I. Nye & I. L. Reiss (Eds.), Contemporary theories about the family
(Vol. 1, pp. Chapter 21). New York: Free Press.
Gelles, R. J., & Straus, M. A. (1988). Intimate violence. New York: Simon &
Schuster.
Goldstein, H. I. (1988). Multilevel models in educational and social
research. London: Oxford University Press.
Gottman, J. M., Jacobson, N. S., Rushe, R.
H., Shortt, J. W., Babcock, J., La Taillade, J. J., & Waltz, J. (1995). The
relationship between heart rate reactivity, emotionally aggressive behavior,
and general violence in batterers. Journal
of Family Psychology, 9(3), 227-248.
Grotevant, H. D., & Carlson, C. I.
(1989). Conflict Tactics Scales (CTS), Family
Assessment: A guide to methods & measures (pp. 293-298). New York: The
Guilford Press.
Gunnoe, M. L., & Mariner, C. L. (1997).
Toward a developmental- contextual model of the effects of parental spanking on
children's aggression. Archives in
Pediatric Adolescent Medicine, 151(August), 768-775.
Haj-Yahia, M. M., & Dawud-Noursi, S.
(1998). Predicting the use of different Conflict Tactics among Arab siblings in
Israel: A study based on social learning theory. Journal of Family Violence, 13(1), 81-103.
Haj-Yahia, M. M., & Edleson, J. L.
(1994). Predicting the use of conflict resolution tactics among engaged
Arab-Palestinian men in Israel. Journal
of Family Violence, 9(1), 47-62.
Hamby, S. L. (1996). The dominance scale. Violence and Victims, 11(3), 199-211.
Hamilton, L. C. (1998). Statistics with stata 5. NY: Brooks/Cole Publishing Company.
Hastings, D. P., & Kaufman Kantor, G.
(In press). The relationship of women's victimization history to surgical
intervention. Association of Operating
Room Nurses Journal.
Herzberger, S. D. (1991). Conflict Tactics
Scales (CTS). In D. J. Keyser & R. C. Sweetland (Eds.), Test Critiques, Volume VII (pp. 98-105).
Austin, TX: Pro-ed.
Hilberman, E. (1980). Overview: The
"wife-beater's wife" reconsidered. American Journal of Psychiatry, 137, 1336-1347.
Holden, G. W., Geffner, R., & Jouriles,
E. N. (Eds.). (1998). Children exposed to
marital violence: Theory, research, and applied issues. Washington, Dc:
American Psychological Association.
Holtzworth-Munroe, A., Meehan, J. C.,
Herron, K., & Stuart, G. L. (1999). A typology of male batterers: An
initial examination. In X. B. Arriaga & S. Oskamp (Eds.), Violence in intimate relationships.
Thousand Oaks: Sage Publications, Inc.
Holtzworth-Munroe, A., & Stuart, G. L.
(1994). Typologies of male batterers: Three subtypes and the differences among
them. Psychological Bulletin, 116(3),
476-497.
Hotaling, G. T., Straus, M. A., &
Lincoln, A. J. (1990). Intrafamily violence and crime and violence outside the
family. In L. Ohlin & M. Tonry (Eds.), Family
Violence (Vol. 11, pp. 315-375). Chicago: The University of Chicago
Press. Also in Straus, Murray A. and
Richard J. Gelles (Eds.) Physical violence in American families: Risk factors
and adaptations to violence in 8,145 families.
Pp. 431-470.
Hotaling, G. T., & Sugarman, D. B.
(1986). An analysis of risk markers in husband to wife violence: The current
state of knowledge. Violence and Victims,
1(2), 101-124.
Hunsley, J., Vito, D., Pinsent, C., James,
S., & Lefebvre, M. (1996). Are self-report measures of dyadic
relationshiips influenced by impression management biases? Journal of Family Psychology, 10(3), 322-330.
Inglehart, R., Basanez, M., & Moreno, A.
(1998). Human values and beliefs: A
cross-cultural sourcebook. Michigan: University of Michigan Press.
Johnson, M. P. (1995). Patriarchal terrorism
and common couple violence: Two forms of violence against women. Journal of Marriage and the Family, 57(May),
283-294.
Johnson, M. P. (2000). Conflict and control:
Images of symmetry and asymmetry in domestic violence. In A. Booth & A. C.
Crouter & M. Clements (Eds.), Couples
in conflict. Hillsdale, NJ: Erlbaum.
Johnson, M. P., & Ferraro, K. J. (2000).
Research on domestic violence in the 1990's: Making distinctions. Journal of Marriage and the Family, 62(4),
948-963.
Kleinbaum, D. G., Kupper, L. L., &
Morgenstern, H. (1982). Epidemiologic
research: principles and quantitative methods. Belmont, CA: Wadsworth.
Koss, M. P., Koss, P. G., & Woodruff, W.
J. (1991). Deleterious effects of criminal victimization on women's health and
medical utilization. Archives of Internal
Medicine, 151, 342-347.
Larimer, M. E., Lydum, A. R., Anderson, B.
K., & Turner, A. P. (1999). Male and female recipients of unwanted sexual
contact in a college student sample: Prevalence rates, alcohol use, and
depression symptoms. Sex Roles, 40(3/4),
295-308.
Leserman, J., Drossman, D. A., Li, Z.,
Tomey, T. C., Nachman, G., & Glogau, L. (1996). Sexual and physical abuse
history in gastroenterology practice: How types of abuse impact health status. Psychosomatic Medicine, 58, 4-15.
Levine, R. V., Norenzayan, A., &
Philbrick, K. (2001). Cross-cultural differences in helping strangers. Journal of Cross-Cultural Psychology, 32(5),
543-560.
Levinson, D. (1989). Family violence in cross-cultural perspective. Newbury Park, CA:
Sage.
Lin Yuan-Huei, W., & Rusbult, C. E.
(1995). Commitment to dating relationships and cross-sex friendships in America
and China. Journal of Social and Personal
Relationships, 12, 7-26.
Linsky, A. S., Bachman, R., & Straus, M.
A. (1995). Stress, culture, and
aggression. New Haven: Yale University Press.
Linsky, A. S., & Straus, M. A. (1986). Social stress in the United States: Links to
regional patterns in crime and illness. Dover, MA: Auburn House Publishing
Company.
MacMohon, B., & Pugh, T. F. (1970). Epidemiology: Principles and methods.
Boston, MA: Little & Brown.
McAndrew, F. T., Kande, A., Turner, S.,
& Sharma, Y. (1998). A cross-cultural ranking of stressful life events in
Germany, India, South Africa, and the United States. Journal of Cross-Cultural Psychology, 29(6), 717-727.
McCauley, J., Kern, D. E., Kolodner, K.,
Derogatis, L. R., & Bass, E. B. (1998). Relation of low-severity violence
to women's health. Journal of General
Internal Medicine, 13, 687-691.
McWilliams, M. (1998). Violence against
women in societies under stress. In R. E. Dobash & R. P. Dobash (Eds.), Rethinking violence against women (pp.
11-140). Thousand Oaks, CA: Sage.
Messer, S. C., & Reiss, D. (2000).
Conflict Tactics Scales (CTS). In American Psychiatric Association (Ed.), Handbook of Psychiatric Measures (pp.
251-254). Washington, DC: American Psychiatric Association.
Moffitt, T. E., & Caspi, A. (1999). Findings about partner violence from the
Dunedin multidisciplinary health and development study (NCJ 170018).
Washington DC: National Institute of Justice: Research in Brief.
Moffitt, T. E., Scaspi, A., Krueger, R. F.,
Magdol, L., Margolin, G., Silva, P. A., & Sydney, R. (1997). Do partners
agree about abuse in their relationship?
A psychometric evaluation of interpartner agreement. Psychological Assessment, 9(1), 47-56.
Morrow, K. B., & Sorell, G. T. (1989).
Factors affecting self-esteem, depression, and negative behaviors in sexually
abused female adolescents. Journal of
Marriage and the Family, 51, 677-686.
Murphy, C. M., & O'Leary, K. D. (1989).
Psychological aggression predicts physical aggression in early marriage. Journal of Consulting and Clinical
Psychology, 57(5), 579-582.
Nagel, S. S., & Neff, M. (1979). Policy analysis in social science research.
Newbury Park, CA: Sage.
Newman, G. (Ed.). (1999). Global report on crime and justice. NY:
Published for the United Nations Office for Drug Control and Crime Prevention,
Oxford University Press.
O'Leary, A., & Sweet Jemmott, L. (Eds.).
(1995). Women at risk: Issues in the
primary prevention of AIDS. New York: Plenum Press.
O'Leary, K. D. (1993). Through a psychological lens: Personality traits, personality disorders, and levels of violence. In R. J. Gelles & D. R. Loseke