UNT Digital Library/67531/metadc279080/... · VanHorn, Barbara, Effects of family of origin...
Transcript of UNT Digital Library/67531/metadc279080/... · VanHorn, Barbara, Effects of family of origin...
3 7 9
//Si
Ao. 15-00
EFFECTS OF FAMILY OF ORIGIN VIOLENCE ON PARTNER VIOLENCE:
A COMPREHENSIVE REVIEW AND META-ANALYSIS
THESIS
Presented to the Graduate Council of the
University of North Texas in Partial
Fulfillment of the Requirements
For the Degree of
MASTER OF SCIENCE
By
Barbara VanHorn, B.A.
Denton, Texas
May, 1998
VanHorn, Barbara, Effects of family of origin violence
on partner violence: A comprehensive review and meta-
analysis . Master of Science (Psychology), May, 1998, 114
pp., 12 tables, 1 figure, references, 128 titles.
Meta-analyses with 144 correlations from 44 studies to
assess the relationship between experienced, father-to-
mother, and mother-to-father violence in the family of
origin and partner violence for males and females in
clinical, community and student samples. The effect size for
all family of origin violence was small. Type of sample was
the most useful moderator. The effect size for clinical
samples approached moderate strength, followed by community
and student samples, respectively. This pattern held for
each type of family of origin violence. There was little
difference in effect size between witnessed and experienced
violence or sustained or expressed partner violence across
gender. However, for males the relationship was strronger
for expressing, rather than sustaining, partner violence.
3 7 9
//Si
Ao. 15-00
EFFECTS OF FAMILY OF ORIGIN VIOLENCE ON PARTNER VIOLENCE:
A COMPREHENSIVE REVIEW AND META-ANALYSIS
THESIS
Presented to the Graduate Council of the
University of North Texas in Partial
Fulfillment of the Requirements
For the Degree of
MASTER OF SCIENCE
By
Barbara VanHorn, B.A.
Denton, Texas
May, 1998
TABLE OF CONTENTS
Page
Chapter
1. INTRODUCTION 1
Terminology Theory-Limitations of the Research Research Results
The National Surveys Studies of Men Studies of Women
Dating Violence Research Male Students Female Students
Conclusions
2. METHOD 37
3. RESULTS OF THE META-ANALYSIS 48
4. DISCUSSION 57
APPENDICES 67
REFERENCES 97
LIST OF TABLES
Table page
1. Samples not Identified by Gender 71
2. 1975 National Survey 72
3. 1985 National Survey 73
4. Male Community Samples 74
5. Male Clinical Samples . 76
6. Female Community Samples 7 9
7. Female Clinical Samples 81
8. Males Student Samples 33
9. Female Student Samples 86
10. Any Family of Origin Violence: Population Effect Size and y? Homogeneity 89
11. Witnessed Family of Origin Violence: Population Effect Size and %2 Homogeneity 91
12. Experienced Family of Origin Violence: Population Effect Size and x2 Homogeneity 93
xv
LIST OF FIGURES
Figure P a g e
1. Stem-and-leaf-display for 144 effect sizes in the family of origin variables data set 96
V
CHAPTER 1
INTRODUCTION
Prior to the early seventies, little notice was taken
of family violence by the social science research community.
Even family researchers failed to perceive the problem
(Straus, 1990). Gelles and Straus (1979) noted that the word
violence did not appear in the title of any article
published in the Journal of Marriage and the Family prior to
1971. Grassroots movements during the 1970s helped raise the
awareness of social and behavioral scientists (Dobash &
Dobash, 1979; Martin, 1976; Pagelow, 1981) and called for
shelter for victims of domestic violence, legislative
change, and research. A national survey (Straus, Gelles, &
Steinmetz, 1980) was conducted in 1975 to establish the
prevalence and incidence of violence within couples. They
reported that 28% of men and women had been victims of
physical aggression at some point in their marriage.
Moreover, 11.6% of the men and 12.2% of the women had been
the target of a partner's physical aggression in the
preceding year.
2
Even before the prevalence of violence evident to the
grassroots movement was confirmed by research, questions
regarding its etiology were raised. Early theorists cited
patriarchal norms of Euro-American societies as instrumental
in supporting and sanctioning the violence (Dobash & Dobash,
1979; Straus, 1976). Gelles (1976) and Owens and Straus
(1975) suggested the cultural norms were transmitted within
families. Thus, exposure to violence in the family of origin
could function as a predictor of partner violence as an
adult.
Research increased exponentially during the past two
decades. Exposure to violence in the family of origin has
been repeatedly implicated in the theory, research, and
treatment literatures with conflicting opinions, results,
and applications. Family of origin violence is most often
examined as one of many possible correlates of partner
violence. Despite this attention, there has been no
comprehensive review and no consensus on the strength of the
relationship. Therefore, a meta-analytic review of the
literature is proposed.
Several issues need to be addressed. Because the terms
used are diverse and have changed across time and
disciplines, terminology must be clarified. At least a dozen
theories have been offered, most of which support the view
of an intergenerational transmission of violence. Finally, a
3
comprehensive review describes relationships found between
effects of witnessed and experienced violence in the family
of origin and expressing and sustaining of violence in
partner relationships.
Terminology
A broad range of terms have been used to describe
various aspects of family violence, some of which appear
similar but may be used in disparate ways. For example,
abuse, assault, aggression, battering, beating, and violence
have been used to describe similar, but not necessarily
interchangable behavior. The early literature used vague,
general terms which included aggression against children,
siblings, spouses, and elders (Gelles & Straus, 1979). When
forms of violence were distinguished, it was usually in
regard to their effects. For example, when the "battered
woman syndrome" (Walker, 1979) was identified, research was
conducted on specific effects of violence by husbands.
Researchers soon recognized that acts of aggression were not
limited to men, nor was physical aggression limited to
spouses. Consequently, terms that were less gender specific
and which could also incorporate violence between cohabiting
or dating partners began to appear. Finally, it became clear
that the word abuse was too broad because it included
psychological abuse, such as threats, and sexual abuse as
well as physical acts.
4
For present purposes, the term violence is most
appropriate for physical acts that may cause physical harm.
According to Gelles and Straus (1979), violence is "an act
carried out with the intention or perceived intention of
physically hurting another person" (p. 554) . Authors have
proposed that legitimacy (Goode, 1971), motivation or
function (Herzberger & Tennen, 1988), and injury (Pagelow,
1992) associated with violence have theoretical importance.
However, none of these distinctions have dominated the
literature and none are directly relevant to clarify the
relationship between violence in the family of origin and in
later relationships.
The changing structure of the family has required
researchers to expand and clarify their definitions of
family members (Kalmuss & Seltzer, 1989) . Early research
limited samples to married couples (Straus et al., 1980),
reinforcing the assumption that spouse abuse was a marital
problem. The importance of not restricting the definition is
evidenced by the U.S. Department of Justice (1994) estimate
that in almost three fourths of reported spouse assaults,
the victim and perpetrator were living in separate
households at the time. Additionally, researchers have found
higher rates of violence in the dating relationships of
college students (Laner, 1985; Marshall & Rose, 1987, 1988;
Sugarman & Hotaling, 1989) than among spouses. It is,
5
therefore, obvious that the terms must be sufficiently broad
to include all these groups. The term partner or
relationship violence includes both males and females who
express and/or sustain acts of violence in their dating,
cohabiting, or marital relationship.
Diverse types of relationships also occur in the family
of origin. Violence may occur within an individual's
generation (e.g., between siblings) and across generations
(e.g., parent to grandparent, grandparent to child). Most of
the research has been on acts by parents or those persons
who were in a position like parents to the child.
Consequently, this review is limited to acts done by a
parental figure to the child or done between parents.
Theory
Theories developed to explain relationship violence
have ranged from intrapsychic to macro-sociological and
cultural, each suggesting the possibility of differing
variables that may predict violence. Intrapsychic theories
cite factors within the perpetrator, such as psychopathology
(Maiuro, Cahn, Vitaliano, Wagner, & Zegree, 1988),
personality (Dutton, 1995) or alcohol abuse (Kaufman-Kantor
& Straus, 1989) or within the victim (e.g., learned
helplessness, Walker, 1979). At the other extreme, feminist
theories have proposed a cultural basis, citing patriarchal
norms accepted by the predominant culture as responsible for
6
violence perpetrated by men against women (Dobash & Dobash,
1979). Regardless of the level of analysis, most theories
implicate violence in the family of origin as a causal
factor. Each theory has problems and they are often accepted
without adequate scientific evidence. Furthermore, few
theories address the range of violent behavior between
intimate partners.
Aspects of Bandura's (1973) social learning theory have
been incorporated in cultural, intrapersonal, and
interpersonal theories of relationship violence. Theorists
postulate that the family is a training ground for violence.
Childhood experiences offer role models and examples for
imitation which may be utilized later in adult life. When
parents are violent with each other or their children, they
provide models of aggression as a coping style and as
punishment. For example, a parent who is frustrated by his
or her child's behavior and hits the child for that
behavior, models coping with frustration through aggression
as well as violence as a method of correcting undesirable
behavior (O'Leary, 1988) .
Violence researchers often note that behavioral norms
are learned in the family. Consequently, exposure to
violence as a child teaches approval of violence (Owens &
Straus, 1975; Ulbrich & Huber, 1981). Specifically, Straus
et al. (1980) suggest that the effects of physical
7
punishment teach a child three lessons: 1) those who love
you most are those who hit you, 2) the moral rightness of
hitting other members of the family, and 3) violence is
permissible when other things do not work.
Specific application of social learning theory would
suggest that behaviors modeled by the same sex parent may be
most influential. Therefore, it is important to
differentiate mother-to-father, father-to-mother and gender
of the child as well as gender of the parent who inflicts
violence on the child. However, research has seldom made
these distinctions.
Feminist and patriarchal theories have drawn on another
aspect of social learning theory that suggests that children
would be more likely to model behavior of those with
perceived social power in the family (Walker, 1979).
Similarly, violence theorists argue the cultural norms of
power are used by men to justify violence in their
relationships (Martin, 1976) .
Several intrapersonal theories have been proposed. For
example, Walker (1979) applied Seligman's (1975) learned
helplessness theory. Seligman hypothesized that when an
individual learns through experience that he or she has no
control over an unpleasant stimulus, that individual loses
the motivation to change the situation. Although Seligman's
theory has only been applied to battered women as a result
8
of their partner's violence, children may also learn they
have no control over their parent's acts. Abramson,
Seligman, and Teasdale's (1978) reformulation of learned
helplessness utilized attributional dimensions. Negative
situations which elicit internal, stable and/or general
attributions are believed to have detrimental effects.
Theorists have suggested that perhaps the experience or
observation of violence in the family of origin contributes
to the perception that a person has no control and is unable
to change the unpleasant situation (i.e., violence) in adult
intimate relationships.
Symbolic interaction theory is concerned with processes
involved in defining the act of violence. Learning to label
behavior as acceptable or not, whether a slap or a severe
beating, is likely to be influenced by what was acceptable
in the family of origin (Gelles, 1974). Exchange and
resource theories, assuming people maximize rewards and
minimize costs, have also been applied. Goode (1971)
hypothesized that violence is used as a resource when all
else is perceived to have failed or when the violence may
bring about desired behavior.
Ney (1992) suggested a transgenerational theory in
which the triangle of perpetrator, victim, and observer
rotates over time with an individual changing roles. Either
from the same or a different position on the triangle,
9
individuals tend to reenact violence in an attempt to
conclude unresolved and energy consuming conflicts.
These and other theories suggest some relationship
between the violence children observe between their parents
or sustain from their parents and their own expression or
sustaining of violence in adult intimate relationships.
However, each theory emphasizes a different factor and
leaves major issues unresolved. It is not clear whether
witnessing parental violence is more or less influential
than personally sustaining violence. The effects of gender
are not resolved. Girls who witness or experience their
mother's violence may be affected differently than boys in
the same situation. Girls and boys are also likely to be
affected differently by their father's acts. Regardless of
the gender of the role model, family of origin violence may
affect females differently than males. Widom (1989) reported
that male children who had sustained violence were more
likely than children who had witnessed parental violence to
exhibit acting out behavior problems, such as conduct
disorders. This may or may not extend into the adult
relationships of those children.
To understand the relationships between family of
origin violence and partner violence as an adult it is
necessary to clarify research. Study populations have varied
from nationally representative samples of more than 6,000
10
people to small samples from specific populations. Reports
from both men and women, married and cohabiting, have been
examined. A large number of studies done with student
populations examined family of origin variables in relation
to dating violence. Some samples have only included persons
identified as victims, usually women {Mitchell & Hodson,
1986; Pagelow, 1981; Ryerson & Fishel, 1993; Snyder &
Fruchtman, 1981; Walker, 1984). Others have only included
perpetrators, usually men (DeMaris & Jackson, 1987;
Schuerger & Reigle, 1988). To date no comprehensive review
has organized the research including all types of samples.
Limitations of the Research
Several points should be made about the state of the
literature. In addition to the specific issues of variations
in gathering and reporting data, the broader issues
resulting from self report data must be considered.
Expressing or sustaining violence is often measured
using modified versions of the Conflict Tactic Scale (CTS;
Straus, 1979). However, in many studies, the data are often
reduced to a discrete classification, the presence or
absence of violence (e.g., Hotaling & Sugarman, 1990).
Sometimes researchers have followed Straus et al.'s (1980)
detailed inquiry into the frequency of acts using the CTS
(mother's acts toward father, father's acts toward mother,
mother's acts toward child and father's acts toward child).
11
Others have used single items. For example, Follingstad,
Rutledge, McNeil1-Harkins and Polek (1992) asked "Did you
ever see adult members in your home use physical force?" (p.
131) and Strube and Barbour (1984) assessed "exposure to
abuse as a child (none vs. witness or victim)" (p. 839).
Time frames also vary widely. Partner violence has been
measured for the preceding year or two years, ever in the
relationship, or ever expressed or sustained by the
participant with a partner which could presumably include
acts across may relationships. Some researchers asked
whether the acts ever happened while the subject was growing
up (Follingstad et al., 1992), others requested reports for
the "worst" year of childhood (Burke, Stets, & Pirog-Good,
1988), and still others asked about when the respondent was
a teenager (Straus, et al, 1980).
The construct for family origin violence has also been
measured several ways. In some studies, separate questions
or sets of questions were asked for witnessing acts between
parents and experiencing acts from the parents. However,
these distinct types of acts are often combined into a
single variable for analysis. This makes it impossible to
tease out differences in effects.
The effects of recall are especially relevant when
asking respondents to report experiences that may have
occurred 20 to 40 years previously, when they were children.
12
Azar (1997) argued that people recall best those events that
are memorable, rather than mundane or unremarkable. Although
every incident may not be remembered, violence that is
witnessed or experienced may be sufficiently traumatic to a
child to be remembered. The more severe acts are probably
remembered better than "mild" acts. For example, Grayson and
Smith (1981) item "Do you remember being excessively beaten
or hit when you were a child?" (p. 192), is more likely to
prompt recall than a more general question of whether or not
the respondent remembers being hit. However, the use of the
word excessive is unnecessarily subjective and may elicit
other response biases. Further, that question would not
elicit affirmative responses from, for example, respondents
who were slapped several times a month although such acts
may have had an effect.
Self-reports of violence may be limited by social
desirability response bias. However, Sugarman and Hotaling's
(1995) meta-analysis found that the overall effect of this
bias was small (mean r = -.16). It was somewhat greater for
reports of expressing (mean r = -.21) rather than sustaining
(mean r = -.12) violence and gender had little impact (Male
mean r = -.19; Female mean r = -.17). Reports of violence
during childhood may be less vulnerable to this bias than
are reports of partner violence because physical punishment
of children is normatively accepted (Straus et al., 1980).
13
Self-reports have also been questioned for their
accuracy and the possibility of over-reporting
retrospectively. Widom and Shepard (1996) compared
retrospective reports of early childhood physical abuse with
official cases documented through court records and found
good discriminant validity. However, they also noted that
there were problems with under-reporting, with almost 40% of
the validated cases failing to report. They too suggest this
problem may be due to the type of questions asked to elicit
the reports. Overall, these results would suggest that
retrospective reports may be conservative in their findings.
Another limitation is that most studies have not
included both expressed and sustained violence. Many studies
continue to focus on women as victims and men as
perpetrators. It is important to remember that the injuries
caused by a 240 pound man punching his 125 pound partner are
likely to be much greater than the reverse (Cantos, Neidig,
& O'Leary, 1994; Koss, 1990), but it is equally important
not to ignore women's use of violence.
The issue of who reports the acts is a problem in this
body of literature. A common occurrence is to ask the
respondent to report his or her own behavior as well as the
partner's behavior. For example, the national surveys by
Straus and his collegues had one informant per household,
randomly choosing a male or female. Research using these
14
data often do not distinguish whether the respondent was
male or female reporting their own or their partner's
behavior. Not infrequently, respondents also report events
in their own and their partner's family of origin (Gondolf,
Fisher, & McFerron, 1990; Pagelow, 1981; Walker, 1984; Tutty
& Rotherty, 1997) . Although it is not unreasonable to accept
an individual's self reports of acts she or he expresses
(i.e., own behavior) and sustains (i.e., partner's
behavior), it is more problematic to accept one's view of
his or her partner's childhood.
It is also important to recognize differences among the
populations sampled. Many studies were done using samples
identified specifically by the presence of violence that are
likely to differ in many ways from samples drawn from more
general populations. Clinical (i.e., identified) samples
have been drawn from women's shelters, batterer's treatment
programs, or mental health clinics. People who have sought
assistance specifically for violence are likely to differ
from people in more representative community samples who
report violence in their relationships. Therefore, studies
using participants from clinical populations should be
considered separately from studies using only community
people.
Unfortunately, statistics are reported inconsistently,
making it difficult to compare results across studies. It is
15
important to understand not only the probability of presence
or absence of an effect, but also the size of the effect,
regardless of the size of the sample (Cohen, 1994; Hunter,
1997). Therefore, zero order correlations are reported when
available. A related problem is the inconsistent use of
statistical tests. Many studies do not test for significant
difference or significant relationships. Other reports may
indicate a significant relationship was found but not give
an indication of its strength or size. Consequently, in this
review the means, percentages, or correlations are included
when available and significance of the findings are reported
when possible.
Despite these problems, family of origin violence
continues to be cited as an important factor to be
considered in treatment and prevention of adult relationship
violence (Brennan, 1985; Fortune, 1993; Gerbi, 1994; Hanks,
1992; Mackey, 1992; North, Thompson, Smith & Kyburz, 1996;
Rounsaville, 1978; Russell, 1988; Rynerson & Fishel, 1993).
Treatments are implemented without regard to the possible
inconsistencies of findings. Therefore, meta-analysis of the
results reported in the literature for each of the family of
origin variables, using data of reported effect size, rather
than relying on a tally of significance, is necessary to
clarify the actual relationship.
16
Research Results
In the majority of studies, family of origin violence
was not the focus of the research. Although a number of
investigators have asked questions regarding violence
experienced or witnessed as a child, many of them reported
relationships for outcome variables other than partner
violence (e.g., coping strategies, self-esteem) that are not
relevant. Only data regarding the relationship of family of
origin violence with either sustained or expressed partner
violence are included here. The national surveys will be
considered separately. Because it is important to understand
gender differences, results from studies with men are
reported separately from results of women. Studies examining
community samples are separated from clinical samples.
Finally, dating violence among students may be distinctly
different. Consequently, male and female student samples are
reported in separate sections. Details of the applicable,
published results of each study are presented in table
format to allow for comparisons. Several researchers
reported results without identifying gender, therefore,
these studies are reported separately in Table 1 and will be
included in the omnibus meta-analysis. However, the samples
represented are too diverse to allow for specific
comparisons within this group of studies.
17
The National Surveys
The first large scale survey was conducted in 1975 by
Straus and Gelles (Straus et al., 1980). Data were collected
via face-to-face interviews with a nationally stratified
random sample of one partner in a married or cohabiting
relationship. These 960 men and 1183 women completed the CTS
(Straus, 1979) assessing violence in the family of origin
and family of procreation. Respondents reported their own
and their partner's acts and their own teenage experience.
More men (35%) and women (26.7%) who had witnessed parental
violence as a teenager had been violent toward their partner
in the preceding year than men (10.7%) and women (8.9%) with
nonviolent parents. The authors stated that witnessed acts
by fathers had more effect than acts by mothers but did not
affect sons more than daughters, nor did the mother's
violence affect daughters more than sons. The same trends
held when the more stringent criteria of including only
severe assaults was applied (Straus et al.). Unfortunately,
they did not report data to support these conclusions.
People who had not experienced parental violence had
the lowest rate of violent marriages (6.3%). As the severity
of the violence experienced increased, so did the likelihood
of violence in the marriage. For example, among men and
women who reported sustaining the most severe violence from
their parents, 20% reported acts of partner violence. Straus
18
et al. (1980) concluded that males and females punished by
fathers rather than mothers were more likely to be violent
in their marriage but did not report the supporting data.
In addition to Straus and his collegues, several
investigators have conducted secondary analysis on data from
the first national survey including family of origin
variables. Table 2 lists these studies. When researchers
combined reports from males and females, witnessed violence
was associated with expressed partner violence (Seltzer &
Kalmuss, 1988). When males were considered separately, a
relationship was found between witnessed parental violence
and expressed partner violence in two studies (Howell &
Pugliesi, 1988; Straus et al., 1980) but not the third
(Cappell & Heiner, 1990). For females, witnessing parental
violence was associated with expressing (Cappell & Heiner;
Straus et al.) but not sustaining (Hotaling & Sugarman,
1990) partner violence. Although the correlation reported by
one study was statistically significant, the variance
accounted for by witnessing parental violence was minimal.
A relationship between experienced parental violence
and partner violence was found by both studies reporting
combined male and female reports (Seltzer & Kalmuss, 1988;
Straus et al., 1980). The one study on experienced parental
violence and females' sustained partner violence did not
find significant differences (Hotaling & Sugarman, 1990).
19
However, two of the six factors they identified as risk
markers were experienced and witnessed violence in the
family of origin, but only two items comprised each factor
and frequency of father hitting mother loaded on
experiencing violence at a level, the authors considered
acceptable for a primary loading (.30). Therefore, the
factors were not as distinct as would be desirable.
In 1985, Straus and Gelles (1990) conducted the
National Family Violence Resurvey via telephone interviews
with 6,002 households. Respondents were married, cohabiting,
divorced or separated within the preceding two years. Single
parents with a child under eighteen living in the home were
included. Whether respondents were in a relationship at the
time of the survey is not clear, Findings for family of
origin variables have been reported from secondary analyses
listed in Table 3.
Two studies (Aldarondo & Sugarman, 1996; Stets, 1990),
examined the effects of witnessed parental violence on
married men's expressed violence, with Stets reporting no
relationship for all married men in the study. In a follow
up study involving a smaller number from the original sample
Aldarondo & Sugarman found a significant relationshipbetween
witnessing and expressing violence. Witnessed parental
violence was associated with married women's expression of
violence. Results were significant for the effects of father
20
to mother violence for married or cohabiting women
sustaining both minor and severe partner violence (Kaufman-
Kantor & Straus, 198 9). Two studies examined effects of
experienced violence (Aldarondo & Sugarman; Stets) finding a
significant relationship for women's, but not men's
expression of partner violence.
It appears that effects of family of origin violence
are inconsistent across the national surveys with differing
effects for men and women. According to the 1975 data, more
men had witnessed parental violence and witnessing and
experiencing father's violence had a greater effect than
mother's on the expression of violence. Across both surveys,
witnessing family of origin violence had somewhat consistent
effect on men's and women's expression of violence and an
inconsistent effect on women sustaining violence. Results
across studies are less clear for experiencing parental
violence. When men's and women's reports are combined,
experiencing violence appears important. However, when
examined separately women's, but not men's, experienced
violence may be important to their expression of violence.
Finally, results are inconclusive for effects of
experiencing on women sustaining violence. However, it is
important to note that all reported effects were small.
21
Studies of Men
In order to determine whether the effects of family of
origin violence reported for men in the national surveys are
consistent with the remainder of the body of literature,
results from researcher's data for men must be assessed.
Although much of the research has focused on men
specifically identified as batterers, some data are
available from community samples, who are more likely to be
similar to men in the national survey samples. These studies
are addressed first. It is important to note that subjects
for the both community and clinical samples were included
for diverse attributes and are not necessarily equally
representative of the population as a whole.
Community samples. Aside from the landmark national
surveys, the few studies that have used community samples
are listed in Table 4. The largest of these was a
longitudinal study supported by a grant to O'Leary in 1982
who studied couples at four points. Malone, Tyree and
O'Leary (1989), reporting zero order correlations, found
that witnessing violence as a child was significantly
related to a man's use of violence toward his partner before
marriage but not 6 or 18 months later. Doumas, Margolin, and
John (1994) studied a community sample of 181 couples also
finding that, for men, witnessed parental violence was
related to expressing violence.. A study of men in treatment
22
for alcohol or drug abuse (Bennett, Tolman, Rogalski &
Srinivararaghavan, 1994) and a study of married males
originally identifed as part of the National Youth Survey in
1976 (Mihalic & Elliott, 1997) found no relationship for
witnessed violence and expressed violence for the men in
their sample. Both Doumas et al. and Mihalic and Elliott,
the only researchers to publish results examining the
relationship of witnessing to the violence men sustain,
reported no signficant relationship.
Malone et al.(198 9) found that the violence men
experience from their parents was related to expressed
violence at premarriage and at 6 months, but not 18 months
of marriage. Three studies, (Bennett et al., 1994; Doumas et
al., 1994; Mihalic & Elliott, 1997) found no relationship
for experienced violence with the violence men expressed or
sustained. A fourth study (Rouse, 1988) consisted of
community men who responded to a questionnaire mailed to
people who paid property taxes. All but 4 men were married
to their current partner. Experiencing parental violence as
a teenager was associated with expressed partner violence.
Rouse's zero order correlations demonstrated a moderate
effect size. Unfortunately, she did not ask about witnessing
parental violence.
Results from these community samples support the
finding of the national surveys that violence witnessed in
23
the family of origin is more likely to be related to men's
expressing violence against their partner. In contrast to
the national surveys, two of the four studies demonstrated
that experienced violence was important to men's expression
of violence. There was no relationship for men sustaining
partner violence, however, only two studies reported data
for this variable.
Clinical samples. Numerous studies have addressed
family of origin violence among batterers. However, outcome
measures have varied (e..g., by typology, completion of
intervention programs, recidivism of violent behavior, or
willingness to respond to research questionnaires). Only
those that assessed family of origin violence in relation to
men's use of violence or sustained violence with their
partners are relevant here. All studies represented in Table
5 as "clinical" included at least one group of men seeking
treatment for partner violence.
Some studies have compared men identified as batterers
to a non-violent control group. Others have merely reported
descriptive statistics. For example, Edleson and Gruzinski
(1987) reported results from three separate studies of
batterers in treatment among whom the percentage who had
experienced childhood violence ranged from 59% to 72.3%.
Hamberger and Hastings (1986) reported that for their sample
of batterers in treatment 26.5% had witnessed and 19% had
24
experienced parental violence. Rynerson and Fishel (1993)
used a sample of 85 men from a mental health center's
batterer program. Many (38.8%) had witnessed father to
mother violence and 29.4% observed mother to father
violence. Almost all (97%) had experienced violence from a
parent. None of the studies examined effects of family of
origin violence on the violence men sustain from their
partner.
One study (Barnett & Hamberger, 1992) combined
variables for witnessing and experiencing family of origin
violence. They compared batterers in treatment, community
men who reported violence, maritally discordant but non-
violent men, and maritally satisfied, non-violent men.
Batterers had witnessed and experienced more parental
violence than maritally discordant or satisfied men.
Ten studies (Barnett, Martinez & Bluestein, 1995;
Beasley & Stoltenberg, 1992; Caesar, 1985, 1988; DeMaris &
Jackson, 1987; Dewhurst, Moore, & Alfano, 1992; Hamberger &
Hastings, 1991; Johnston, 1988; Rosenbaum & O'Leary, 1981;
Schuerger & Reigle, 1988; Stith & Farley, 1993) reported
effects of witnessed violence without distinguishing
father's and mother's acts. All but two (DeMaris & Jackson;
Scheurger & Reigle) included a nonviolent control group, and
three (Dewhurst et al.; Hamberger & Hastings, Stith &
Farley) included comparison groups comprised of sex
25
offenders or alcohol abusing men. A relationship was found
between witnessed and expressed violence in nine of the ten
studies. However, in Schuerger and Reigle's study of
batterers, the reported relationship at the time of follow
up was negative.
Both studies that specifically examined the effects of
father's acts toward mother (Barnett & Fagan & 1993; Murphy,
Meyer & O'Leary, 1993) found a relationship. Barnett & Fagan
also examined effects of witnessed mother's acts and found a
significant difference between groups. It is interesting to
note that when Caesar (1985, 1988) considered father to
mother and mother to father violence separately, there was
no difference between violent and nonviolent groups,
however, when she combined the variables, significance was
found. This may be an artifact of sample size, since effect
sizes were not reported.
Overall, it appears that witnessing parental violence
is related to expressed violence among men. Reported effect
sizes, although greater than those from the national
surveys, varied widely, ranging from .71 to -.19. It is
interesting to note that alcohol use may be a confound. When
violence groups were separated according to alcohol use,
higher levels of witnessed family of origin violence were
found for the alcohol abusing group (Hamberger & Hastings,
1991).
26
Of the ten studies which examined effects of
experienced parental violence only three distinguished which
parent was doing the acts. In the seven remaining studies
(Caesar, 1985, 1988; DeMaris & Jackson, 1987; Hamberger &
Hastings, 1991; Johnston, 1988; Murphy, Meyer & O'Leary,
1993; Rosenbaum & O'Leary, 1981; Schuerger & Reigle, 1988)
expressed violence was associated with partner violence in
all studies except Schuerger & Reigle's follow up data for a
portion of their sample. Of the three studies which
differentiated parental acts (Barnett & Fagan, 1993;
Barnett, Martinez & Bluestein, 1995; Dutton, Starzomski, &
Ryan; 1996) both father's and mother's acts were found to be
important.
It is apparent from these results that, across the
clinical samples, both witnessed and experienced violence
were associated with acts of partner violence. However, it
must be remembered that individuals who sought or were
ordered into treatment may differ a great deal in many
unknown ways from men in the community population who have
expressed violence toward a partner.
Studies of Women
Research with women has reported family of origin
variables less often than studies with samples of men. Very
few researchers used community samples. With the exception
of the community samples, women's acts of violence are
27
rarely reported. Again, when zero-order correlations were
available they are reported in the tables.
Commun11y samp1es. Apart from the national surveys, few
researchers reported findings from community samples of
women (see Table 6). Samples were recruited from a variety
of sources including legal aid offices, marital therapy
clinics, and advertisement in the community. They included
women from divorcing as well as intact couples.
Two of the three studies (Doumas, et al., 1994; Malone,
et al., 1989; Mihalic & Elliott, 1997) that examined effects
of witnessing interparental violence on women's expressed
violence found a relationship. However, that relationship
remained significant at premarriage, 6 and 18 months of
marriage. Although Doumas et al. found no relationship for
expressed violence, witnessing was related to women's
sustained violence. Mihalic and Elliott also found a
relationship between witnessing and minor, but not severe,
sustained violence. Lockhart & White (1989), reporting
effects of father's and mother's acts separately, found that
neither was significantly related to women's expression of
violence. However, witnessed father to mother, but not
mother to father, violence was related to the sustaining
partner violence. In contrast, Sedillo (1997) found that
father to mother violence was related to expressing, but not
sustaining partner violence. Parker and Schumacher (1977)
28
also found father's acts were related to sustaining
violence. In VanHorn, Sedillo & Marshall's (1997) study of
low income women, both father to mother and mother to father
violence were related to expressing and sustaining partner
violence. In this study, the strongest relationship was
found between mother's acts and women's expressed violence.
The seven studies (Cascardi, O'Leary, Lawrence, &
Schee, 1995; Doumas et al., 1994; Grayson & Smith, 1981;
Malone et al., 1989; Mihalic & Elliott, 1997; Sedillo, 1997,
VanHorn, Sedillo & Marshall, 1997) that examined experienced
family of origin violence did not distinguish between
parents' acts. Four of the five studies (Doumas et al.;
Malone et al.; Mihalic & Elliot; VanHorn et. al) that looked
at effects on women's expressed violence found a
relationship and three (Grayson & Smith; Mihalic & Elliott;
VanHorn et al.) of the five studies that examined effects on
sustained violence found significance.
These results suggest that witnessing parental
violence, particularly father's acts, is more consistently
related to women's sustaining rather than expressing
violence. It is possible that the relationship between
mother's acts and expressed violence suggest that women who
witness mother defending herself are more likely to defend
themselves. For these community women, experienced parental
violence is more consistently related to expressing than
29
sustaining violence which is consistent with the results of
the national surveys.
Clinical Ramnl.es. The majority of women's clinical
studies draw their samples from battered women's shelters or
other agencies offering services to women who are victims of
partner violence. None of the studies report women's
expressed violence and only two include a non-violent
control group. Research questions of the studies in Table 7
focus on differences between groups of battered women or
report percentages within the sample. Therefore, it is less
appropriate to draw conclusions from the data.
Only one study (Koslof, 1984) examined effects of
witnessing parental violence on sustained partner violence.
No difference was found between groups, However, the sample
size was extremely small (nine women in each group) and
therefore, the study lacked sufficient power to find
significance.
Percentages of women who report experiencing family of
origin violence were high, varying from 32% to 100%.
However, for the two studies (Koslof, 1984; Lake, 1993)
reporting significance, experienced parental violence was
not significantly correlated with sustaining partner
violence. Downs, Miller, Testa, and Panek (1992) examined
violence experienced from father and mother separately. In
each case they found significant differences between women
30
who had not sustained violence and women who were victims
and/or had alcohol abuse problems. However, as with male
clinical samples, the group that reported both alcohol
problems and sustained violence had the highest means for
sustained violence, suggesting again that alcohol may be a
confound.
Dating Violence Research
There has been a great deal of research looking at
dating violence using college students. This literature
examines effects for a population that is younger, never
married, and dating. Therefore, these studies should be
considered apart from those using married or cohabiting
respondents. Furthermore, researchers in this area more
often report family of origin effects on both sustained and
expressed violence, consequently, there are two sections
(expressed and sustained) for each gender. Studies for male
and female students are reported separately in Table 8 and
Table 9.
Male Students
Expressed Violence. Both witnessed and experienced
family of origin variables were combined by two researchers.
Bernard and Bernard (1983) reported parental violence for
more than twice as many male students who expressed violence
as those who were nonviolent. Ryan (1995) reported effect
31
sizes for expressed violence for two different samples, one
of which was very low, the other moderate.
Seven studies (Alexander, Moore, & Alexander, 1991;
Breslin, Riggs, O'Leary, & Arias, 1990; Marshall & Rose,
1990; O'Keefe, 1997; Sack, Keller, & Howard, 1982; Stets &
Pirog-Good, 1990; Tontodonato & Crew, 1992) examined
witnessed parental violence without distinguishing which
parent did the acts. In only three (O'Keefe; Stets & Pirog-
Good; Tontodonato & Crew) was expressed violence associated
with this variable However, when each parent's acts were
considered separately (Barnes, Greenwood, & Sommer, 1989;
Breslin et al.; Marshall & Rose, 1987), the picture was less
clear. Only one (Barnes et al.) of three studies found
father to mother violence important and mother to father
violence was related to expressed violence in one (Breslin
et al.) of two studies on male students.
Eight studies (Gwartney-Gibbs, Stockard, & Bohmer,
1987; Laner, 1985; Marshall & Rose, 1987, 1990; O'Keefe,
1997; Riggs, O'Leary, & Breslin, 1990; Stets & Pirog-Good,
1990; Tontodonato & Crew, 1992) examined effects of
experienced parental violence on men's expression of
violence with six reporting a significant relationship. For
Laner's sample, experienced violence was important for
nonMormon, but not Mormon, students. The only study to
32
differentiate parents' acts (Alexander et al., 1991) found
that father's, but not mother's, acts were important.
Experiencing parental violence, particularly father's
acts, appears to be more consistently associated with the
male student's expression of violence than witnessing
violence. Inconsistencies may be based on the statistics
used to test significance. Although effect sizes varied
widely, two of those reported for experienced violence (d =
.44 & R = .32) were moderately large.
Sustained Violence. None of the three studies
(Alexander et al., 1991; O'Keefe, 1997: Stets & Pirog-Good,
1990) that examined effects of witnessed parental violence
on sustained violence found significance. Even when father
and mother's acts were considered separately (Marshall &
Rose, 1987) witnessing was not important. Of the six studies
(Gwartney-Gibbs et al., 1987; Laner, 1985; Marshall & Rose,
1987, 1990; O'Keefe, 1997; Stets & Pirog-Good) that examined
experienced parental violence, only three found a
relationship. When considered separately (Alexander et al.,
1991), father's, but not mother's, acts were related to
sustained violence. Thus, for male students, neither
witnessing nor experiencing family of origin violence
emerged as a consistent predictor.
33
Female Students
KyprfisseH vi nl ence. All but two of the researchers who
gathered data from males also reported for female students.
The one study combining variables reported percentages. Of
those students expressing violence, 50% witnessed or
experienced family of origin violence compared to 23% in the
nonviolent group (Bernard & Bernard, 1985). Nine researchers
examined effects of witnessed parental violence on female
students' acts (Alexander et al., 1991; Breslin et al.,
1990; Gwartney-Gibbs et al, 1987; Laner, 1985; Marshall &
Rose, 1990; O'Keefe, 1997; Sack et al. 1982; Stets & Pirog-
Good, 1990; Tontodonato & Crew, 1992), with only two finding
a relationship. When father's and mother's acts were
considered separately (Breslin et al.; Marshall & Rose,
1987) neither was found to be important to expressed
violence.
Experienced parental violence was related to female
students' expressed violence for four of the seven studies
(Follingstad et al., 1992; Laner, 1985; Marshall & Rose,
1990, Riggs et al., 1990; O'Keefe, 1997; Stets & Pirog-Good,
1990; Tontodonato & Crew, 1992). When father's and mother's
acts were reported separately (Alexander et al., 1991;
Marshall & Rose, 1987) neither was important. Again, the
results across studies were inconsistent, with experienced
parental violence emerging as more consistently related to
34
women later expressing violence than was witnessed violence.
Regardless of significance, reported effect size was small.
Sustained Violence. The effects of witnessed parental
violence on the partner violence female students sustain was
reported by eight researchers (Alexander et al., 1991;
Coffey, Leitenberg, Henning, Bennett & Jankowski, 1996;
Follingstad et al., 1992; Gwartney-Gibbs et al., 1987;
Laner, 1985; Marshall & Rose, 1990; O'Keefe, 1997; Stets &
Pirog-Good, 1990) with only four finding a relationship.
Further, Marshall and Rose (1987) found no relationship when
father's and mother's acts were reported separately.
Finally, only four (Coffey et al., Goguen, Stepaniuk & Foy,
1996; Marshall & Rose, 1990; O'Keefe) of eight studies found
a relationship between experienced parental violence and
sustained partner violence. Thus, for female students family
of origin violence is not consistently related to the
violence they sustain in their relationships.
Conclusions
The results of this review suggest that there is some
relationship between family of origin variables and partner
violence. For samples drawn from community populations,
witnessing parental violence is consistently related to
men's expression of partner violence. Witnessing is more
consistently related to women's sustained than expressed
violence. Experienced violence appears to be related to
35
expressed partner violence for both men and women. The
picture that emerges for clinical samples is somewhat
different. Both witnessing and experiencing parental
violence were important for men in treatment for partner
violence. For women identified as victims of partner
violence, their experience of parental violence was more
important than interparental violence.
The relationships found for family of origin violence
lends support to social learning explanations of partner
violence. Since these results are more consistent for men,
it is possible that the reinforcement of patriarchal norms
increases the effects of social learning in the home. In
other words, what boys watch their parents do may have more
effect on their own behavior because such acts are endorsed
by the culture. For women, the results are less clear.
However, the fact that both witnessing and experiencing
violence were somewhat related to women sustaining partner
violence may offer support for the learned helplessness
theory. There is no clear pattern across studies for
father's vs. mother's acts. Although it is possible that
these results negate the modeling theories of social
learning, sufficient studies reporting these variables
separately were lacking.
Studies that examined dating violence found
experienced, rather than witnessed violence emerging as a
36
more consistent predictor. Within this body of research the
results suggest that the violence experienced in the family
of origin was related to males expressing and females
sustaining partner violence.
It is important to note that these conclusions are
based on significance testing rather than effect size.
Statistical significance is impacted by sample size which
varied widely across these studies from 18 to more than
2000. A meta-analysis based on effect size will provide a
more accurate foundation for drawing conclusions.
CHAPTER 2
METHOD
Identification of Studies
Potential studies for the meta-analysis were located
through a computerized literature search of the
Psychological Abstracts database from 1974 to 1997. Key
words representing violence (e.g., domestic violence, spouse
abuse, family violence) were combined with "family of
origin" to direct the initial search. This search yielded
only 44 potential studies. However, data are reported in
many studies which do not necessarily include relevant
information in abstracts or key words. Therefore, an
extensive search was performed using Marshall's personal
database which consists of over 2000 journal articles and
presentations on partner abuse. This search yielded 287
sources. In addition, the reference lists of major research
reviews (e.g., Feldman & Ridley, 1995; Hotaling & Sugarman,
1986; Pagelow, 1992) were manually searched. A total of 317
studies were identified as potentially eligible for a meta-
analysis on the relationship between family of origin
violence and subsequent relationship violence.
37
38
Of the 317 originally identified studies, 40 could be
classified as reviews, 17 opinion articles, 43 focused on
theory, and 214 reported research. Based on initial coding
of potential research studies, estimates are given for the
number of studies excluded for various reasons. First,
researchers looked at groups not included in the meta-
analysis such as children or gay/lesbian couples (n = 23).
Variables other than expressed or sustained violence were
used, including treatment outcome or remaining in
relationship (n = 15), sypmtomology or character traits (n =
30), coping strategies (n = 5), attitudes or attributions (n
= 9), or differing types of violence (n = 3). Many studies
reported family of origin violence differences by
demographic groupings such as gender, ethnicity, economics
or other nonviolence related variables (n = 35). Studies
also combined either family or origin or partner violence
variables with psychological or sexual abuse (n = 13).
Finally, studies measured, but did not report, statistics
for family of origin violence (n = 8) or multiple
publications reported data from the same sample (n = 13).
For multiple publications of data from the same sample,
only results from the initial wave in a longitudinal study
were included. Several researchers reported results from the
national surveys. These studies were included if they
reported results for a different part of the sample. If more
39
than one researcher reported results for the siniilar parts
of the sample, the report than included, the most complete
distinction by gender and violence variables was included.
For example, results were included for men who witnessed
father to mother violence as opposed to a report of parental
violence, not identified by gender. Eliminating these 148
studies left 66 potential studies for the meta-analysis.
Criteria for Inclusion
To be included, studies had to meet four criteria.
First, respondents had to have reported data about himself
or herself. This eliminated studies in which subjects
reported their partner's past experiences. Second, the
report had to include the respondents' acts of expressed
and/or sustained partner violence. Third, experienced and/or
witnessed family of origin violence must be reported.
Fourth, sufficient statistical information had to be
available, either in the published study or through personal
communication with the author.
The last criterion required information on which to
base effect size to indicate the strength of relationships
between two variables. Three statistics commonly used in
meta-analyses are Cohen's d (1977), Glass's (1976) g, or
correlations. Both d and g, computed using t-tests and
standard deviations, reflect the magnitude of differences
between groups. Therefore, they are most appropriate when
40
comparing experimental conditions or categorical groups
(e.g., gender). In contrast correlations (r) describe the
direction and strength of the relationship between two
variables. They are most appropriate when the focus is on
bivariate relationships and when the underlying constructs
are considered to be continuous.
Both violence in the family of origin and the partner
violence are considered continuous in the literature. These
variables are usually measured by such instruments as the
Conflict Tactics Scale (Straus, 1979) and the Severity of
Violence Against Women Scale (Marshall, 1992). Therefore,
correlations are the most appropriate method -to measure
effect size when examining the relationship between two
continuous variables (Rosenthal, 1984). Consequently, effect
size as used here refers to zero-order correlations or
transformations to an estimated r for an individual
analysis.
Studies in which correlations were neither reported nor
provided by the author were included in the meta-analysis if
sufficient information was provided in the form of F or %
statistics. These statistics were used to compute an
estimated r. Because it was important to determine how
accurate these estimates were, raw data from the VanHorn,
Sedillo and Marshall (1997) study were grouped by presence
or absence of each of the family of origin variables and
41
ANOVAs on partner violence were computed. These F statistics
were then used to compute an estimated r for each family of
origin variable with each partner violence variable. The
resulting correlations were then compared to the Pearson
correlations. In every case, the estimated r was
conservative. The original correlations ranged from no
difference between the measures to 2.5 times the estimated
£. For example, the correlation for parent to subject
violence and sustained partner violence was r = .26 and the
transformed estimate was r = .10. Consequently, it was
reasonable to use estimated correlations as effect sizes.
Sixty six studies were found that met the first three
criteria. Only 16 of these studies reported zero-order
correlations. Letters and forms were sent to the other
authors requesting correlations between each family of
origin variable and expressed and/or sustained violence
(Appendix A).
As a result of compiling published data, personal
communication from the authors, and conversion of
appropriate statistics it was possible to include data from
44 studies in the meta-analysis. There were 144 separate
measures because several different correlations were
included in some studies. (Results used in the meta-analysis
are shown in bold type in Table 1 through Table 9.)
42
Although there are relationships among these types of
family of origin violence, it is important to assess the
different effects of each. To avoid combining non-
independent results, multiple correlations from the same
study were only included as effect sizes if they measured
different constructs (e.g., witnessed father to mother
violence and experienced parental violence). For studies
reporting repeated measures over time of the same construct,
only the measurement from the initial wave of a longitudinal
study was included. Because data from the national surveys
were examined by several researchers, the most specific
version of reports (e.g., experienced from father, rather
than experienced parental) was used in the meta-analysis.
Coding
Different factors may affect the strength of
relationships observed in the studies. To determine whether
moderators account for variability in the effects of family
of origin violence, a coding system was employed. The review
of the literature suggested that effects appear to be
dissimilar across the various types of samples. Therefore,
each individual effect size included in the meta-analysis
was coded for type of population from which the sample was
drawn (i.e., clinical, community, and student). Because
efects were also dissimilar for men and women, each effect
size was coded for gender. To determine effects associated
43
with each specific variable, studies were coded to
distinguish type of family of origin violence (witnessed
parental, father to mother, mother to father, parent to
subject, father to subject, and mother to subject).
One decision involved how to address the issue of
expressed and sustained partner violence. Many studies,
particularly with clinical samples, often examined only
expressed or sustained acts. Yet there is a moderately
strong relationship between these two variables (Gray &
Foshee, 1997; Marshall & Rose, 1990; VanHorn, Sedillo &
Marshall, 1997). Therefore, each effect size was coded for
type of partner violence. Separate analysis of expressed or
sustained violence for each gender should be run only if
heterogeneity was found after other moderating variables had
been considered.
Individual effect sizes were further coded for several
study characteristics. Date of publication was coded to
determine whether improved methodology of later studies was
important. Studies were coded based on whether family of
origin violence was the focus of the report. Finally,
individual effect sizes were coded to distinguish reported
correlations from transformed statistics to determine
whether transformation influenced the results.
Finally, the number of participants used in each
correlation in reported results was used to weight effect
44
size. This procedure was important because larger samples
produce more reliable estimates of effects and have less
sampling error (Rosenthal & Rosnow, 1991).
Analysis
The meta-analysis for each of the family of origin
variables was executed in three steps using the 5.3 Revision
of Meta-Analysis Programs (Schwarzer, 1989). First, an
overall population effect size estimate was computed for a
global measure of experienced and/or witnessed family of
origin violence across all studies. Tukey's (1977) stem and
leaf plot of individual effect size estimates was inspected
to examine the distribution of effect sizes and determine
whether outliers were present. When using correlations
Rosenthal and Rosnow,(1991) suggested using the Binomial
Effect Size Display (BESD) to evaluate the practical
significance of an effect size in terms of the change of
rate of an outcome due to variable under consideration.
Binomial refers to the fact that results are cast into
dichotmous outcomes which can be interpreted in this case as
partner violence occurred or did not occur, based on the
reported effect size. BESD was used to evaluate the
practical significance of the effect of global family of
origin violence on partner violence.
Second, homogeneity of the population effect size was
evaluated. This provides strong evidence that differences in
45
the results are negligible across studies. Homogeneity
indicates that the variability is due to sampling error and
that the effect size represents effects of the variables
measured. Heterogeneity suggests that moderators may better
explain the variability in individual effect sizes. Non-
significant (p > .05) %2 tests indicated homogeneity.
Third, if effect size was found to be heterogeneous for
the initial meta-analysis of a type of family of origin
violence, separate meta-analysis were executed based on
likely moderators. Because effects of family of origin
violence appeared dissimilar by type of sample, separate
meta-analyses for each sample type across gender were
executed. Likewise, a meta-analysis of global family of
origin violence for each gender was executed. Xf homogeneity
was not found, then separate meta-analysis for expressed and
sustained violence across gender were executed. If
homogeneity was not found for each gender, separate meta-
analyses were executed for each type of sample within
gender. Next, partner violence was examined within gender.
Finally, if necessary, expressed and sustained violence were
examined within type of sample for each gender.
Differences between population effect sizes are
reported. Traditionally, t tests are used to determine
whether there is a statistically significant difference
between groups. However, due to the large sample sizes
46
included in many of the population effect sizes, sufficient
power exists to find significance when there are negligible
differences in strength of the association. Consequently, no
t tests are reported.
The file drawer problem (Rosenthal, 1984) is often
identified as a shortcoming for meta-analysis, with the
assumption that published articles are more likely to report
significant findings. Consequently, Rosenthal's Fail-safe
was computed. This is a method of estimating the number of
non-significant results that would be required to bring
about a non-significant finding for the overall population
effect size.
After analysis of the effects of global family of
origin violence, separate meta-analysis for each of the
specific types (i.e., any witnessed parental, father to
mother, mother to father, any experienced parental, father
to subject, and mother to subject) were performed.
Homogeneity for each population effect size obtained by
meta-analysis and BESD are reported. If the population
effect size was heterogeneous, further meta-analyses were
performed employing the moderator variables using the order
established by the first meta-analysis of global family of
origin violence. Although we should stop when homogeneity is
reached and variation is explained by a single moderator,
there are several important issues in the literature.
47
Therefore, it was meaningful to continue to examine results
by gender even if homogeniety had been reached.
CHAPTER 3
RESULTS OF THE META-ANALYSIS
Table 10 shows the results of meta-analyses across all
types of family of origin violence. The initial procedure
included 144 correlations from 44 studies. The global (k =
144) weighted population effect size of £. = .14 was
statistically significant, Z. = 27.58, p < .0001. This is a
small relationship. The BESD indicated a change in
likelihood of outcome for absence or presence of partner
violence from 43.1% to 56.8%. This initial analysis did not
meet the test for homogeneity, %2= 354.40, p < .0001.
There was too much variability in the individual
correlations. Therefore, Tukey's (1977) stem and leaf plot
of effect size estimates (Figure 1) was examined. An outlier
was identified that measured the relationship between
witnessed and expressed violence in a clinical sample of
males. Johnston (1988) reported a relationship of .71 which
was much stronger than others have found. (The next
strongest was r = .53.) After removing the outlying
correlation, the meta-analysis was re-executed, but the
effect size and measure of homogeneity were unchanged. The
48
49
lack of change is not surprizing, given the large number of
subjects and effect sizes included in this global analysis.
However, the extreme relationship found by Johnston (1988)
could affect the outcome of other analyses. Therefore this
correlation was dropped from subsequent analyses.
As shown in Table 10, further meta-analyses were
performed for the global family of origin variable to
discern likely moderating factors. The overall population
effect size varied conspicuously by type of sample. The
lowest population effect size was found for students (k =
72, r = .10, p < .0001). In contrast, using clinical samples
resulted in the highest population effect size (k = 29, £
.26, £ < .0001) which also met the test for homogeneity.
Separate analyses were conducted by gender to see
whether homogeneity could be obtained. There was little
difference between the overall population effect size for
males (k = 72, r = .15, p < .0001) and females (k = 64, r =
.13, p < .0001) but neither was homogeneous. For males the
overall population effect size was much greater for
expressed (k = 56, r = .18, p < .0001) than for sustained (Jc
= 16, r = .08, p < .0001) partner violence. The latter was
also homogeneous. In contrast, there was little difference
between expressed (k = 32, r = 12, p < .0001) and sustained
(k = 32, r = .13, p < .0001) violence for females, neither
of which met the test for homogeneity. Therefore, the
50
moderator of type of partner violence (expressed or
sustained) was tested for males, but not for females, in
subsequent meta-analyses of specific family of origin
variables.
When meta-analyses for each type of sample within
gender were executed, the population effect size for males
in clinical (k = 26, r = .27, p < .0001) and community (3< =
10, r = .14, £ < .0001) samples and females in clinical
samples (Js = 3, £ = .23, p < .0001) each met the test for
homogeneity. For these groups, the effect was homogeneous
regardless of whether the family of origin violence was
witnessed or experienced. Because the moderator of type of
sample met tests for homogeneity across and within gender,
it was the first moderator tested for each specific family
of origin variable.
Witnessed Violence
A meta-analysis including all studies reporting any
violence witnessed in the family of origin was executed.
(See Table 11.) The overall population effect size for
witnessed parental violence (k = 76, r = .13, p < .0001)
varied little; from the population effect size across all
types of family of origin violence and was heterogeneous.
Because population effect size was similar to that for
global family of origin violence, an additional BESD was not
necessary. When moderator variables were examined, type of
51
sample seemed most important. Without regard to gender,
population effect sizes for clinical (k = 13, r — .26, e <
.0001), community (k = 23, r = ..15, e < .0001), and student
(k = 38, r = .09, e < .0001) samples each met the test for
homogeneity.
Meta-analysis of witnessed violence for each gender
revealed a slightly higher population effect size for males
(k = 36, £ = .15, E < .0001) than for females (k = 38, £ =
.12, e < .0001), neither of which was homogeneous.
Subsequent analyses distinguishing sample type for males
yielded homogeneity for all three groups. The pattern was
consistent with the measures of overall family of origin
violence, with the strongest relationship found in clinical
samples (k =13, r = .26, e < -0001) and the weakest effect
in student samples (k = 18, r = .10, e < .0001). A
difference was again evident between the overall population
effect size for males expressing (k = 27, r = .17, e <
.0001) and sustaining (k = 8, r = .07, e < -0001) partner
violence. The latter population effect size was homogeneous.
Among females, the population effect size was homogeneous
for student samples (k =20, r = .08, e < .0001) but not for
community samples (k = 18, r = .14, e < .0001) . There were
no individual effect sizes measuring witnessed violence for
females in clinical samples.
52
Meta-analysis of studies that reported father to mother
violence produced an overall population effect size (k - 20,
r = .11, E < -0001) that was only slightly smaller than the
population effect size for any witnessed violence. Moreover,
it was heterogeneous. Only clinical and student studies with
males and community and student studies with females
reported fathers' acts separately. Consequently, meta-
analysis for type of sample across gender was not
appropriate. When each gender was examined separately, the
population effect size for males (lc = 8, r = .12, E < -0001)
was heterogeneous, but analyses by type of sample produced
homogeneity for both clinicial (k = 2, r = .34, e < .0001)
and student (k = 6, £ = .07, p < .0001) samples. Further,
for males the population effect size for sustaining (k = 3,
£ = .07, p < .0001), but not expressing (k = 5, r = .16, E <
.0001) partner violence was homogeneous. For females,
regardless of sample type, the population effect size (k -
12, £ = .11, p < .0001) was homogeneous.
The population effect size for mother to father acts (k
= 15, £ = .14, e < -0001) exhibited only a minimally greater
effect on partner violence than father's acts and was
heterogeneous. Again, differentiation by sample type across
gender was not done because there were no clinical female or
community male samples available. When each gender was
considered separately, the popialation effect size for males'
53
(k = 5, r = .16, p < .0001), but not females' (k = 10, r =
.13, P < .0001) mother to father violence met the test for
homogeneity. When type of sample was examined for females,
population effect sizes for community (k = 6, r = .18, p <
.0001) and student {k = 4, r = .03, p < .0001) samples were
homogeneous.
Experienced Violence
Table 12 shows the results for experienced violence.
The overall population effect size (k = 66, r = .14, p <
.0001) was heterogeneous and similar to that of witnessing.
Type of sample was an important moderator for clinical
samples (k = 17, r = .26, p < .0001) which reached
homogeneity, although that was not found for community or
student samples. Neither the population effect size for
males (k = 35, £ = .16, p < .0001) nor females (k = 26, r =
.14, p < .0001) was homogeneous.
When further partitioned within gender, population
effect sizes for males in clinical (k = 14, r = .26, p <
.0001) and community (k = 5, r = .12, p < .0005), as well as
females in clinical (k = 3, r = .23, p < .0001) and
community (k = 7, r = .19, p < .0001) samples were
homogeneous. Subsequent meta-analyses separating effects on
expressed and sustained partner violence for males did not
produce homogeneous population effect sizes. However, meta-
analyses of male student samples produced a homogeneous
54
population effect size for expressed (k = 9, r = .12, p <
.0001), but not sustained (k = 7, £ = .09, p < .0001)
partner violence.
When violence experienced from father was considered
separately, the overall population effect size {k = 10, r -
.16, p < .0001) was heterogeneous. Again, sufficient studies
were lacking to look at type of sample across gender. Meta-
analyses by gender demonstrated more similarity between
population effect sizes for males (]c = 6, £ = -19, p <
.0001) and females (k = 3, £ = .11/ £ < -002) than for
mothers' acts, but neither met the test for homogeneity. The
variability was greater for population effect sizes for
males' expressed (]c =4, r = .25, p < .0001) than sustained
(k = 2, r = .05, p = .19) partner violence. Again, the
males' population effect size for sustained violence was
homogeneous, but the effect size itself was not
statistically significant. Only three studies reported
fathers' acts separately for females. The population effect
size for the student samples (k = 2, £. = -.01, p = .42) was
considerably lower than the individual effect size for the
clinical sample (r = .29). The population effect size for
students was homogeneous but not statistically significant.
A meta-analysis including all studies that reported
violence experienced from mother separately (k = 9, r = .18,
p < .0001) was heterogeneous. Sufficient studies were
55
lacking to look at differences for sample type across
gender. There were dissimilar population effect sizes for
males (k = 4, x = .23, p < .0001) and females (k = 3, £ =
.12, p < .0005), however, neither was homogeneous. All four
studies that examined effects of: mothers' acts for males
reported only expressed partner violence. Meta-analysis for
males from clinical samples yielded an overall population
effect size (k = 3, r = .28, p < .0001) which was
homogeneous and considerably higher than the single
individual effect size from the student study (r = .07).
Only three studies reported mothers' acts for females. The
population effect size for the female students (k = 2, r =
.05, p = .14) was homogeneous, but not significant. However,
it was considerable lower than the individual effect size
for the single clinical sample (£ = .23).
Study Characteristics
Further analyses were conducted to examine the
possibility of variation due to study characteristics. For
example, studies in which the focus was family of origin
violence (k = 68, r = .13, p < .0001) were compared to those
which tangentially reported these variables (k = 76, r = 14,
p < .0001). There was little practical difference between
the two groups. Furthermore, a Failsafe N = 243, suggests
that more than 243 additional studies would be required to
reduce the global family of origin population effect size
56
below a critical r of .05. This effect size was identified
as critical because population effect sizes of that
magnitude were not found to be statistically significantly
different from zero.
Transformations from other measures to an estimated r
were performed for 46 of the 144 effect sizes. Therefore,
the population effect size of transformed data (k = 46, r -
.17, p < .0001) was compared to that of the non-transformed
relationships (k = 98, £ = .13, p < -0001). However, 24 of
the 46 transformed statistics were from clinical samples
which exhibited consistently higher population effect sizes,
with only 6 of the non-transformed statistics from clinical
populations. This difference likely accounts for the higher
effect size with transformed statistics.
Finally, a meta-analysis was performed to look for
differences in earlier compared to later publications. The
median split for year of publication for the 144 individual
effect sizes was nearest to 1991. Therefore, the population
effect size of all studies published prior to 1991 (k = 69,
£ = .11, p < .0001) was compared to that of studies
published later (k = 75, r = .16, p < .0001).
CHAPTER 4
DISCUSSION
This meta-analytic review examined the relationships
between witnessed and experienced violence in the family of
origin and subsequent partner violence. Overall, only small
population effect sizes for the relationship of family of
origin variables to partner violence were found. The largest
population effect size obtained approached only moderate
strength. Thus, family of origin violence does not appear to
account for a great deal of the variance in partner
violence. Theories positing a strong causal relationship
should be reconsidered. It may be more effective to focus on
understanding moderators that increase the likelihood that
early experiences in the family will be associated with
later partner violence. This focus may be more helpful in
directing future theory as well as prevention and treatment
models.
Conclusions about the overall effects of witnessed or
experienced violence must be tempered by the heterogeneity
of the results. It is likely that moderator variables may
better explain the various effects. Although nearly all
57
58
comparisons of moderator groups would have been
statistically significant, due to the large number of
subjects included, only meaningful differences are noted
here.
Gender as a Moderator
There appeared to be little difference in the effects
of either witnessed or experienced family of origin violence
for males and females until data were further partitioned.
When expressed and sustained partner violence was considered
separately, there was little difference for females. In
contrast, the magnitude of the relationship was consistently
twice as strong for expressed as sustained partner violence
among males. Further, witnessing and experiencing parental
violence are equally likely to affect males expressing,
rather than sustaining, partner violence.
Gender differences remained when parents' acts were
considered separately, but these results are tentative.
Although the sample size for studies that distinguished
parents' acts were reasonably large, there were few effect
sizes in these meta-analyses. An unanticipated finding was
that, for males, witnessing mothers' acts had a somewhat
greater effect on partner violence than witnessing fathers'
acts. This contradicts social learning theory that would
predict the same sex parent would have the greatest
influence. On the other hand, the stronger relationship for
59
community females between mothers' acts and partner violence
would tend to support the same sex theory. However, both
contradict the argument from feminist and patriarchal
theories that children would be likely to model behavior of
fathers, who have perceived social power in the family. It
is possible that, within the family, mothers, rather than
fathers, have the greater perceived social power.
Type of Sample
Before considering the effects of witnessing or
experiencing violence separately, type of population was
clearly a moderator. Regardless of gender, the effect size
for clinical samples was homogeneous and approached moderate
strength. Consistently across analyses, the pattern was for
weaker relationships among community samples for the
smallest effect to be found for student samples.
When witnessing parental violence was considered
separately, the pattern was even more apparent. For males,
all three sample types had homogeneous effect sizes. There
were no reports for females in clinical samples, but the
effect size for community samples was similar to that of
males and the effect for female* students was homogeneous and
small. In each instance, homogeneity was achieved without
regard to whether the partner violence was expressed or
sustained. Similar magnitudes of effect size and homogeneity
were found for experienced parental violence. Thus, type of
60
sample may be the best predictor of the effects of family of
origin violence on partner violence.
There are several possible reasons for the relative
strength of the effects among clinical samples. It must be
remembered that these studies only included persons who had
sought or been referred specifically for treatment for
partner violence. Much research suggests that people try to
make sense of their current experience, especially negative
experiences, in terms of their own self-perceptions and
beliefs. For ejxample, in the search for explanations,
cognitive activity increases when one observes or
experiences a negative or unexpected event (Anderson, Krull
& Weiner, 1996).
Women entering shelters and men entering batterer
treatment programs are likely to engage in ruminative
thinking regarding their current predicament. This could
cause heightened awareness of not only current but also past
experiences because they are focused on violence. Therefore,
any and all physically aggressive acts would be more salient
for them than for samples drawn for a general population of
community people or students. Salient acts are, by nature,
over-emphasized and may be over-reported (Nisbett & Ross,
1980).
Further, lay people are not unlike researchers. If they
thought about violent acts (as they would if they were
61
entering treatment), they would be likely to perceive a
relationship between family of origin and later violence.
When people believe a relationship exists they are much more
likely to notice and recall confirming instances rather than
disconfirming instances (Trolier & Hamilton, 1986) .
Consequently, men and women entering treatment may perceive
such an illusory correlation exists and over-report acts in
the family of origin. For example, women entering shelters
may attempt to make sense of their situation by reporting
more parental violence for their partner than for themselves
(Pagelow, 1981) or for their battering partners than non-
battering partners (Walker, 1983).
It is also possible that persons who enter treatment
are somehow different from the general population. Many
findings from shelter and treatment samples are not
replicated in normal samples (Gondolf, Fisher, & McFerron,
1990). Although the effects of their childhood experiences
may be more salient, it is possible that other factors
account for the relationship between family of origin and
partner violence found in clinical samples.
Reasons for the differences in effect between community
and student samples are less clear. Presumably, both are
drawn from a larger, more comprehensive population. However,
both type of relationship (married versus dating) and age
are usually different for these groups. Childhood
62
experiences likely would be both recent and salient for
students who are may feel more connected to their family of
origin than older adults now living in their own family of
procreation. However, these experiences appear to have less
relationship with violence in dating rather than married
relationships. This difference is less likely to be a
function of partner violence because is rates and means are
at least as high for student as community samples (Gray &
Foshee, 1997, Sack, Keller & Howard, 1982) . Consequently,
there must be other factors which mediate the effect of
childhood experiences. There is some evidence that peer
influence is greater than family influence for student
samples (Gwartney-Gibbs, Stockard, & Bohmer, 1987).
Differences in age or stage of development may also be
factors. Students are not only younger, but most have not
yet begun their own families of procreation. Thus, they may
recall or experience the effects of their family of origin
violence differently than older people or those at a
different stage of life.
Limitations
The greatest limitation of this meta-analysis is the
variability that exists in the methodology of the studies
that were included. These differences existed at three
different stages. First, data collection methods varied from
the use of scales to determine each parent's acts separately
63
to a single question regarding the presence or absence of
witnessed or experienced violence. Second, even when
detailed, specific data were collected using scales, the
results were often collapsed into a dicotomous variable.
Finally, the statistics employed varied with some
researchers mesrely reporting percentages. Unfortunately,
many studies lacked sufficient information to adequately
weight individual effect sizes according to quality of the
methodology or to code all studies so that methodology could
be examined as a possible moderator. Effects and homogeneity
were found despite inconsistent and poor methodology which
suggests that meta-analysis may be robust to methodological
integrity. The somewhat larger effect size for later
studies, revealed that improved methods may have made
effects more likely to be identified. Effect sizes remained
small, however, suggesting that the weak relationship found
between family of origin and partner violence was accurate.
It is unlikely that the obtained population effect
sizes were inflated by the bias of publishing significant
results. A large Fail-safe N and finding little difference
between studies focusing on family of origin violence and
those that tangentially reported these variables suggest
that these results were not subject to publishing bias.
It is important to note that of the 70 meta-analyses
executed, 17 included five or fewer individual effect sizes.
64
Further, only three of the studies employing female clinical
samples reported sufficient statistics to be included in the
meta-analysis, and none of these included reported effects
for witnessing.Clearly, some results should be viewed with
more caution than others.
There were also few studies reporting parents' acts
separately. For example, for males, only 8 studies reported
effects of witnessing father's acts and 5 reported mothers
acts. Even fewer individual effect sizes were available for
effects of experiencing each parents' acts. Therefore, these
results provide less evidence to support a relationship
between individual parents' acts and partner violence.
Consequently, those results should be considered tentative.
Future Research
Despite the limitations, this meta-analytic review has
produced a clearer representation of the differences in the
effects of violence experienced or witnessed in the family
of origin on adult partner violence. The overall weak
relationship suggests that investigators must reconsider the
application of current theories as an explanation for
partner violence. For example, social learning theory may be
applicable, but in a much more complex form than currently
used. Although a relationship exists, it is obvious that
other factors must account for most of the variance in
expression of partner violence.
65
Furthermore, although some theories have implied a
causal relationship, caution should be employed in making
any causal inferences from this meta-analytic investigation
(Knight, Fabes, & Higgins, 1996). This analysis was based on
correlations as measures of effects size. Consequently,
although associations were found, this investigation did not
employ statistics that are explanatory.
It is not possible to discern whether these findings
are the result of variation in methodology rather than a
true measure of effect. Therefore, future investigators
should use scales rather than single items. Future
researchers should also differentiate witnessed and
experienced parental violence and report findings without
collapsing data into dicotomous variables.
Research is needed to clarify the moderate relationship
between witnessed or experienced violence and men's
expression of violence toward their partner. Differences in
effect based on type of sample should also be examined to
better understand how type of relationship, age, or other
undetermined factors may contribute to differential effects.
Given the weak relationship, mental health professionals
should recognize that treatment; models should look to more
factors than merely witnessed or experienced violence, for
both prevention and treatment.
66
Conclusion
Regardless of the widespread impression held by the
research community that the relationship between family of
origin and partner violence is well documented, this is not
the case. Although 214 relevant studies were found, only 66
reported statistics that directly assessed that
relationship. Further, of the 66 studies that met the first
three criteria, only 44 included sufficient statistics to be
included in this meta-analysis. Most important, however,
this meta-analytic review which included over 40,000
subjects from the 44 studies, contradicts the impression
held by the research community. There is only a weak
relationship between the violence men and women witnessed or
experienced in their family of origin and the violence in
their adult partner relationships.
APPENDIX A
LETTER
68
3 University of North Texas Colle^e of Arts and Sciences Department of Psychology
June 15/ 1997
Ola W. Barnett, Ph.D. Social Science Division Pepperdine University Malibu, CA
Dear Dr. Barnett:
I am working on a program of research with Dr. Linda L. Marshall. My area of interest is the effects of f a m^ 1^ violence on partner violence. I am in the process of completing comprehensive literature review and plan to do ® 'h* would like very much to include data from your 1992 article, Assessment of Maritally Violent Ment on the California Psychological Inventory". In order to include your data in tms, I will need zero order correlations for the relationship or family of origin violence variables with sustained and/or expressed partner violence. These correlations were not published in your article.
To this end, I would appreciate you taking the time to complete the enclosed form, reporting the relevant statistics. To make the meta-analysis as complete as possible, please report witnessed violence separately from experienced family of origin violence, if possible. I would also appreciate information on sample size included in each correlation.
I enclosed a self-addressed, stamped envelope for your convenience. If you have any questions regarding the:infomation requested you can reach me by phone at (940) 484-8649 or e " ^ X
(vanhorn0jove.acs.unt.edu) or Linda Marshall at (817) 565-4145 o marshall0terrill.unt.edu). I appreciate your taking the required time to gather the statistics I am requesting. I intend to r u n
the meta-analysis before the end of October. We hope that a meta-analysis involving these variables will clarify the effects of family of origin violence. t
Thank you very much for your assistance in this researcn
effort.
Sincerely,
Barbara VanHorn
P.O. Box 13587 • Denton. Texas 76203-6587 (817)565-2671 » FAX (817)565-4682 • TDD (800)735 2989
69
V a v a
a w 55 M H CO D tO
Q M 55 M H — co c D — to
to w •J
u o z w *9 O
cc w z h s Cm
V a
a w h u t-4 fcu Z
CO w 1-3 § w
i W u 2 w o
a; w z
v a
a w H u M ta z
OJ ON 0%
u 0) cn M 0) <0 32
(1) G f0 OQ
CO w iJ § M
I »-•
M a; o fcu o >4 a
"D a)
—» •H *0 0) •H •H u *W 0) •H a U (0 <u c a 3 w e 4J 4J 3 M •H u o
a) <D w <D <v x: x: •n *n
•H JJ 4J TJ iJ X) w o <0 rH 3 3 ' S 1*4 •H to to
(0 x: JU 0 o u o o c 4J u JJ -p a> to u M u m M fO 0) ai a> <D a x: XI T3 x: XI
4-> JLi 0) 4-> 4J c fO o C fO o a> t* E £u 32 0) <TJ 3 4J W a) 3 OQ CO
CO W
3
> M 13 H-4 OS o [*4 o >* •J
•o a>
•H u a> a to c 3 M M
<D <U x; x: -H u W> o <d s W -U> o o c J-J 4J a) u u M m 0) a) a x: x:
4J £ m O QJ iu s 0) 3 4J 0) CQ
•a a)
u a) a w c: 3
X3 u (0 <0
T3 a> c
to a to
4-> u 0) •r-i JQ 3 <0
M 0) x: 4J (0 [*4
U 0) T-i ja 3 (0 o JJ M <U je 4J O s
APPENDIX B
TABLES
0* 9 SDD M *< 01 H-03
K O rt m
f t
0) M
0
CD h "«*
K hrj
O r t £T CD hi
r t O
H> $> r t
CD H
CO r t C Q, H-0) 00
H-»
8*
ft
H-»
0 H £
a a>
a
H-£
1 r t a> i
a
i
m o
H P 2 f t P »
& o
8 r #
s <*
IN
H3 M ft
H co a tf* Zf
£h
lr* H-" 0 a iO £3-H* 05 ft
n o £ t* H
tft
(J1 o
w 3 X K
*0 ft 0 & H ® M" 03 ® tfl £ f» a pt (D a
m m
H K ® m. w (ft oi to # <i> a a
T* f t o H ft
Pu
o o o o U) M
a 5? 9 8
f t ft
M F* H
' J~* a f t f t
H **J Mb H m
H
f"
•6 J» f t Hv
8 f t tn
01
f t *0 H Ift H -k j 0 ~
p I
\0 (XI
5
i 0 m to o 6
I f ! I
H ,
CI*
O (JT
n ,
H
*4
i** m
a o 0 sr * a> 2 H
H n> 0
-̂N» Hi m M KO "*>
CO <T\ 03 <x> w H <1
0 jfc* 0 i*r o 2 •a
til IS ^ rt o H* £ {J* iQ-a* o sr f& D t* M ft m
to in cr%
5
£ to fl)
a.
Ca M £ . X m *xs f t H $ ® M - CQ £ m <D fl> Gkf CL
o H
hi © •d Q hi rt d> a
£ o f t
o in
a o >§
0» m
CO
o
3 3$ G P-rt rt
&•
53 f& H h Cft f t Eft EJ 0 a m m f& a
m m hji KA
tJ *tJ-h H 0 fl> w m w m a> (P a. a
U) (Jl o o oP oP
A
r r i*
f . "
p
^ M H
H m H'
H n a
! s m m
4
6P
m
N H 02 ct
II It Pi rt
o U) p, * * m H *a: rt
H* o
!3-. Id
!3- o. Id
to Cn
CO P) B
!0 f~4 CD 0)
P O
H a a> p r t H-H) H-a>. a
cr s<-
o fD y a. CD h>
Kl P> t r
<D
7 1
72
0 > 3H 3 CO
rH d C o
"H 4~>
CM <d S3
0) rH t o ,jQ r > aj (Ti EH rH
p i
u -•ri -P tO *H .p cd 4J CO
- »
M O
•M
& -p
*d m
IH o
> i £ H H
00 O
CO O &
•Hi £ <8 M fo o
>1 73
-P CO
73 4) to w <t> M & K m
1
<n
d)
0
4J
Ik
rH t^ O CD
4) o & 2J & s O —
yi W c a
tH O o
<r>
vo
ei to o 0
a> ffl r j rH 73 • a> i g
»H §£ S <u ftS S o) o S &* lw
to > i <U rH a <i> -H -p -H
H m
• tfl cm nj
73 73 © <D i
a -H -H W
ftS 4J 44 w tQ to P4
w CO CO m
73 <D
73 0 t j d) 15 # to <9
•H w •iU M # a 0) c: -p & -H « ss •w
A0
0 #
*rf 0
•3 H £
t8
W
-r-i »rl CO CO
cn 0 <3V cn 0 CO <3V
!> » *4 «ri O '* r * CsJ tH « * *
H 11 It 11 11 tt 11
C k N N Ml Mi
M& m
H
** ID
« w
<#>
o t>
IA o 00 rH
I> CTi o CO
XO 0D CM
O CM
M o X I 4J 3 cd
CP
X I 4J
•H
fl
O rH - P <tS D H d
73 73 0) 0) W to to to # d) M M .o, a X X C£J |x?
73 fl} -01 4-» t0 "ri m &
73 73 m a) to to to w # © u M Qi Qt $*5 *<
m m
73
to to 0* a
73 73 0 0 CO CQ to to ® ® M M a , o<
W W
*d 4) O a 4) a* •H ^ M W m
•a 4J -P -P PU 4J •H O -H G * 0 & % S6 fij
<Y> 00 O a> V0 H tH o> tH
W w (S> <D
CO TH to rH <y fd fd
H S rH flS <u cd X hi CM
V o to C D
a) cr> r-t tH r—j *s"'* <u 0 H
•P * U % S
0 *»** tft 0CI M 00 a CO CO td -H cri HI m M <D
0 H -p 4J m w CO CO
O U
rH nj CJ O «
to CO M -H <D CO Pu >1
rH >1 (d fit fl
fd X3 t CD <d
73 4^ "tH 0) > & O M 0) ft XI
4J CO D
-rH • H •p to 7^
-H <1> - P 7J as ^5
•P rH CO D
a rH "H <d a T5 0 H
-rH Q -M
TS 7^ •H < * %
i w
«H 0 7^ 4-i d O 4J
a W
73
CD 1> M 0 m
r H
a a
- H - p
0 0 fd 5 ?
CD r-H t n
0 0 05 CTs
£ h t H
tH O W
0 1 • «
u •+>
W -H -P (d 4J to
00 tH <NI rH
n ii *
mi ^ *V«
M G
0) d
M rti
04
O
£ l
M 0) •&
<y
> i t r
t o
7$ 13 0 fl) 93 m s i to <y < r M M ft ft X tA m w
m CM
cm
OJ m
00 tH
11
0) *D 4-i CO -P
fd <y dp D H O
r l Gn
0) £ O
o -H a a «j <u
P H t s 81 o
, -H 0 -H fa 13 M >
1$
I •8 «* *> w
I ! i
Tf 0 £ *r"t €) «s w -P a) o) > 3 <U cn w
t S I 0
IS V 0 & w 0 «
C5 10 -H w -H (U M i a <D H 4J ft 43 M "H 2 i
O a s W
m o CM
0) H
to !
^ tH
D m -P a i fa fd
if} tH m (fl O O £l £» * #
4J -H ^ O j ^ O H O H H M O O O O
CM * * #
*X$ TS xt *& a) <d 0 a) w m m to m m m m 0 0 a) a> u u u M ft ft ft ft X M ft. M mm w m
"6 0 m TO <y 0 4J •H
TJ 0 O 13 & 0 0) W
«H W U 0 0 d ft - P X *H m &
T3
rH o> vo o
a\ rH rH
O tr» cr>
to 4J 4) P m
(d d (d i <tl
4-> CD
cd
£ H
CD * d P H u a
t s H
£ H
"d
0 ) •Hi
P 4~J £/)
M CD
X I 4~> O S
O 4-1
U CD
- P (ti
MH
a ) -p
o
74
CD I—I & FO
EH
QL
O •H -P CO
-H -P (U 4J cn
M o
0 O M 3 O
. cn
>1 T3 3 -P CO
CO CO G G
OV OI o o
II II
Ml Ml
TJ id U a) a 0 10 0} G to (0 -P a a) U M M m a a Cu X X
W W
TJ 0)
4-1 TJ O o A) G
to a) >1 G CO >H
r--| -H a> M •H CJt G A) S -H a rd M •H x UA O S3: w
CO IO
4J O •H TJ T> <#
A (U S rH O EH
G TD
£ •§> FLJ M RJ
^ H 01 -P ^ AS -P <0 > ^ (D H rl «s* a id G en G Gn-H O* <D O M rH GQ PS CO W
O to CO CO • A C C
TJ 0} 10 TO a) G 4J
£
00
TJ a) O G <D
•H
a! &
W
§• O 0
A •H rH
M
I
O Q
LO LD LO O (0 CO O O CO
• G G * • G CO CO G G
CO CO G G
rH o o •s* <M CN rH rH o CD rH rH -p
* • • rH • • to . 0 II II II 1! II II II -p
Ml Ml Ml Ml Ml Ml M -Pi
t r 00 CD <X> <SO M CM R- • <£>
m * cn CNJ •
rH * \ rH G \ \ rH \ aj o o R- o cu CTI <N • <y\ S rH
'
(U a)
TJ TJ TJ TJ TJ (d CO (d CO "O X5 <D 0) a) 0) 0) •H co x : •H CO x: a) 0 (0 G (0 G to M x: -p M x: -P CO G (0 •H CO •H to M -P G M -P G CO 0 -H 0 a) td a) (0 <D m G O <d G o <D M U M M M M 4J M 4J M i 0 B n 0 e M O 0 •p O 0 P< (0 a to P» 0) fi a) g a G > CO G > X 3 X X M OD M 00 X •H 0 3 •H 0 w CO w CO W & <X> <—1 a KO RH w 6 to W fi CO
CO 0 A G
-H •P G O O
XI M EH
T» A) to to A)
5
£
TJ 0) 0 G 0)
•H
a! &
W
TJ 0 CO CO 0 G -P
to a) >>
\ o 2
M O> M
CO A)
H &
O a
o
CM
-P T3 4J rd 0 *H Z >.
-H X5 x; 0 SH S -P > M x: O a M fU U o M o P E Q u 4H 01
4J -P o
•H 0) O\ rH 0) 00 i—1
cr> M CTi w cn rH £ rH w ' >t o G 0) M -H XJ G fd i—1 O O O (l) O0 hi rH x: <Y\
iS •» •H *—i S o 2
75
o i
a • H 4J CO
- H P 03 - P C/3
S I
u o
T3 TJ <D d)
(D CO C C CO <D - H CD P CU jq n 03 M M M M O CD P O CD 03 a C > CO C >
O i X - H 0 3 *H CD W g « CO g CO
M-l O
> 1 c i—i - H • H Cn £ m u
U4 o
CD o u 3 o co
> 1 T5 3 4J CO
to O to CO CO
• c c c
• p CO <D p
-Pi
<D 00
d) (M
T3 CD a c CD
*r4 u CD a x
w
o <Ti CM
P " 0 P 03 <D • H S
- H xa X <D M .03 6 p > U X I O 3 u fO O M 0 3 6 O U m >-• CO
O "0 - H (D r~i P <—I C W - H
P ^ £
o o o
• H — rH r -fd
JC c t i •H «—I s —
CTk O O
O CO
Ml
* 0 a) 0) w a) M &
w
TJ a) a a 0)
•H M a)
w
os
CO - H CO ! > i • H m c (U i f d 4 J CD E 0
x ; 4 J
c - H
T S CD
T J D
i—I a a
• H
CD M CD £
T J H 0
A
CO 0
• H T 5 3
- P 0Q
CD - P O
s
76
CO Q)
i—1 Q » E m
CO
<—i ro a
- H c
• H LO i—1
O CD
i—1 CD X3 •—1 no 03
EH S
0 1
o -H 4-> CO
-H 4J
m
ui
S I
u o
M CD £ -P
rci 0u
MH T* O " 0>
w > i G CO
i—1 -H 0} -H tj* a £ •'-* -P 03 U •r j t i O *
vo G|
<d o u 0 0
CO PQ
> i TS 3 4J CD
o o o o
i n
O rH o o
GO CO
<T> II V0
PmI VD
CM O CTi
CN CM rH \ W ID (Ti O
|| O O rH
i n
\ &
H CO 00 ^ kO <T\ 00 ^
CM CM tH rH \ W
rH CM rH 00 ^ O (Ts CO
2 1 r o n h H r o oo r—f r-H
V 0 CO to CD u * w
q cn S 2
cn Oi cn
a (d
r * fa
-P -P 0) a u id
n
TJ a) (0 (0 Q) M &
m
u a
X! •p
2 \ M a)
frS w fa
co cm o i r j
GO
CM
fal
TJ 0)
d) M
8" w
T3 a) CO w <D a
TJ a) o a a
•rl u a)
m >
t—1 r -ID O ;—1 o O o o o
** cn r * c*1 i n <j\ rH i n
*> O rH VD i n r -
a II II II II II
§ i N fal M M
•p t i a* •H a x ^ i d H
i n CN CM 00 r f <3 TJi
CO 2
TJ *3 T5 a) a) a) W 0) w co w w <D a) a M M M
fr 0 - B* w w w
fa 2
T5 • d a> a)
T3 a a a) G c 0) a) a) to *H •ri 0) M M a 0 0 4J P« & •H X X £ w w
vo vo *3
m > S
cm r -
CM rH
CO rH
li II
S I S I
*G <d
u & W
T* a) CO (0 0)
5
i s
cn i n co
PQ
uT) LO O O
2 s x : cn
<N
II
T3 T3 (D CD (Q CO (0 CO 0) CD M M 04 a 5T X w u
T3 CU
TJ O a) c w 0) •H 0) M c 0 "p a •H X 3 : CJ
V0 GO CM r-4
m
& M 0 n
N • 9 d) ^ a) a m «p
•H rH •P cn 0 M rH 4J i n td — CO 00 S ch
a ui rH -H
+) a j >i 4J 4J a j ^ M
CM (U CO H CM (d CTi a a CO Oi cn Oi M 3 cd cn a) TH nj h a) rH <d w pq m CQ ~ a
CO (D 3 C
G o o
X) fO EH
77
a
-p co
-H 4J rd -p CO
SI
u o
0 a
D O CO
LD O CO
G
CM LO O r—I
IT) O
«N (S
CQ
lD O
r -
CM
II
o o\0 LO
-o 13 T3 u 0 0 0} 0 co CO V) C en CO CO •P 0 0 0) M U M rd a a , &* Pu X X
W W W
•P 0 c T?
/ 0 4-1 T3 o T» 0 0 C '• a)
CO 0 (0 >i C CO -H 0)
i—I *H 0 U 0) •H tJ> C 0 CJ £ *H •P a 4J rd U •H x rl t4 O 3: w *
00 d LO H CM
a a
a> <w co o
CM CM
ffl >
CM r -O rH o o
CM
II CM
•PI co
<£> .. . CT\ U) VQ • 0 N CM V-i ^ \ 0 • oo 0 r-f CO H 0 0
T> <D CO 0} a) M &
w
M 0)
M
a)
W Cm o in <3 ^
PO
a 0 nJ a %• tw d UJ
•rl a u> 0 0) to a)
M M 0 0 0 N
0 U b a «J
4J c» CO
(0 0) •H ^ M — a —
>i M r - 0 CM 0 VD T3 M oo cn 4J Ok
5̂ 2 cn ^ o\ 4-J 4-> Q) H <D r-4 ^ rH CO Q — P — Q ^
CM i—t O O O O
r-0
^ ro 01 S
o\° o\° o\° • • 0 0 o\o
O CO • Ch LO CM CO • w \ r -o\o o\o o\° \
r - lo o\° . . . LO
r-1 O r—i • Lf) CM H CTi
T3 <l> W 0) (9 M fr
W
TJ a) a a
to a) n a)
•H M a)
4J a •h * !k u
00 O vo 00 cn MD CM VD
9 03 >
0)
a •H -p
01
s
r - r - r - T—1 w o o rH o 0 o o o o 0
TH
a •H -P c o
CM O 0
r-t r - 00 CM CM 0
rH CO 00 1
2
0 rH
II XI
11 II II 11 II II fd 11 II
N H
Ml M Ml Ml N x'—
*0 a) 0) w <d M &
w
Tl <D w 0) a)
B
£ in o *-4
•a 0
0 W m a> 0 " > a)
M 0 a rl _ . . g co g
fr,fl
-a a) o c a>
•rl M 0) fr
W
>
« 5s
35 w
00 00 a*
a o 4J (0 a
.C 0
i—I CTt *—I
CM CM
dP o\o dP II II II
H H LO CM CM S I S I S I
T3 a>
a> M
w
•a
a) 5
is
Tt 0)
d) M fr
w
15 <D o a a) •H
a! *
W
^ ^ ^ CM CM CM
nSS
& fl) a)
0
ui
M a >i
1
£ CO fk o\ U OS a -
78
pi
s i
u Q
CI
<D O U 2 O CO
m o o * i
O CTk md m
^ on
II II C* M
TJ TJ (I) (1) CO 0] (0 (0 <D d) M M Oi Q. X X W «
O O O C\| CN CM
«§§
flj fl)
o uj
LO LT) LT) o O O CO
- . - c
oo ^ h r -r-i tH * O
• • I
II II It 11
Ml Ml M| M|
V TS TJ T3 (D <|) (L) CD W W to CO (0 CO CO CO <D 0 CD CD M M M M ( i i a a * * x x W W W M
TJ TJ TJ <D <u a)
4-4 t j a TJ 0 T3 o O CD C <u a a) G
w a) oi a) co CD >i C CO -r| CO H CO -H i—I -H 0) M <D M CD U •H Cn C <D G 0 G 0 & -H 4J a •P &4-> a 03 M •H X •rj X H x t i o £ H £ W £ w
0 in 01
o <̂ r
m
00 00 <7*
•HI <D
5 O
O — m-4 a
OQ P
CO C
CTk O
II
Ml
> IS
P £ 0 g
-P fd CD P
TJ 0) CO W a) M fl-pa
TJ a) w 0} a) c +j
CNI Cft m on
< PQ
m o\ cr>
>i a) H M
fa
£ M <4
>̂i iH 0) T3 a) 00 P 4J
CO o> •rH -P 0 rH a 4J co tf CO CO
PQ
T3 CD
•H
CO •H P (0 0)
4-1 O
•H 6 m tl4
P CD
T3 G D
TS CD
T3 3
rH O G
•H
CD P CD £
0 •Q
G 1—1 • -H m CD G
p CO -H -P -H 3 O p XI CO G rd fd CD
6 •H P rH X5 CD 11 0 G P JG -P CD w o cn -P 2 a -P «—i • fd •V < U X! •p CD
G p x: II <d o 4->
TJ o > P s
O p • s o G o -p CO CD -p G *H 6 CD TS p p g CO CD P >i a) x: m rH p 4J CD rH +j fd P as MH p -P G
•H •H il G P -H fd G 2
6 CD P 6 CD II • V P II p CD Q CD -P 2 < JG P +J rd •V • »k O XI G CO g
0 M II g CD II
P CQ >< CD •
P -P CO v -H P •V •H CD G <d p CO O 3 JQ 0 P g x: i—1 D S- Cn 4J fd O o G fd G co o -H M-l fd
CO i P -p 3 II fd CD G X* -P T5 CD fd CD G rH g D O i—i
-H o G CD . > JG -H x: CD 1 o p P G o •H O O 1—1 P G ss G fd o -H
7 9
CD i—I XI ra
EH
Ql
U •H P co -H -P <T3 p CO
SI
u o
u Q) a P M H3 a•
t-n o >I G rH -H •H CP g *H 03 U iu O
a> o M 3 O CO
-a a) G •H fu p co 3 CO
TS a) Q G 0
•H
0 a x C4
r -cr\
«H t>1 m a
03 x: g p
LO CTI Ch
>< a) m a) oj x: cu o ^ 00
o <-»
ld in co o o co G * • G
in en vo in O rH HI O
U II il II
Ml Mi Ml Ml
TJ <D W 0) 0) M &
» 3 H CO W W
TJ <D 0} <0 a) 5
is
TJ a) a a <D *H M 0 &
W
(0 a) H §• O o
c £
£ UJ
c *H H O tn M
•H 0 T3 o 5-1 G 03
>1 03 D <d "O o U ai 3 CO 2 Ok P 03 03 O T—1 CQ A Q W
TJ 0 P 5H O a <D M
P — 0 oo G CT>
m U tn T3 O 0) rH m 0 rH
CO tl O CO 0\P 3 3 11 r - X> GOP JQ G m a3 — co rd ^
T3 0 G *H 03 P CO 3 cn
T3 0 o G 0 •H M 0 a X w
r -CM 00
>1 p •H G 3
0 0
rH 00 cn rH
x: P •H 6 Ul
t 3
G O CO >1 03 M O
CM LO M LO LO LO CO O (0 CO o o o o O o G
*
G G
GO rH <N TH <?* O CM o GO CO CM CM in
• . CM H rH rH rH iH rH VO H • • * • * •
II II 1! II II II il II II II
ixil N pcil (zil
S S I
TJ 13 0) Q)
a) a)
5 5
£ is
in in
a a id o o H
•H M M a) s i
c\ 00 Oi
a) •P •H
4J M a) X3
o o
co 0 3 G •rH
P G O a
M{ Ml Ml Ml Ml Ml
XI m EH
(9 0 t»
TJ TJ TJ TJ TJ (d CQ (d CO a) <0 a) 0) a) •H CO XI *H CO x; CO G W a CO M x: P M x: -p 0) •H 0) •H (0 M P G M P G 0) flJ 0) (d a) (d G 0 rd G O M 9*
4J (0
M P<
4J W
M P«
flog a) g
3 o 0) g
6
W 0 5? a 5< M oo M 00 m to w CO w Ot̂ Q rH &KO rH
TJ a)
a>
5
£
TJ a) a a ai •H M a) fr
w
ro cn m
0) d) H
8-0 a
«d a)
o
a) a) M
ai ~ c c* o CD H cr» s =
80
Q l
a -H - p co
rd - p GO
-p CO a ) • p
H I
tJ* r - ( O cn 0 C\J VD CM a) LO 0 r H 0 0 0 0 0 0
M . . • • • • • •
S I MH CM H CM *—i 0 0 *—1 \ cn \ *—1
M C LO 0 r - r - <T\ 0 0
0 rc co l o O r H cn 0 0 co a> • * • • • • • •
o\o 2 r H r H T-H
- a T3 - a TS TJ u CD CD 0 0 a) CD CO G CO C a C CO CD *H a) CO CD •rH a> •H -P CD U U (T3 M SH CD M M 03 M M ai M 5H O CD -P 0 CD U O CD 4J O Q) 4J (tJ a C > co a > a a > CO C > <n Ot X -H CD 3 •H CD X -H CD D •H CD
w g CO cn g CO W g CO CO g CO CO
h h
o
>* a i—I -H -H tJ> g *H m JH Cx< O
CI
>1 T3 D - P CO
LO o
LO o
LO LO o o
<y\ CO CM r - 0 r H r H r H r H
LO LO LO O r H *—1 r H cn O O O O
0 O O O rH r H CO \—1 r H O O O O
T5 CD CO CO to CD <D £ -P \ •H O 3 : 2
T5 0) U C CD -H U CD £X X £xj
TJ + j as CD <D -H Z > 1 O • h x i x ; cd
U M 03 g - P > 3 SH X I O 3 M O fd M O SH O cn g O O 4H >h
W
o cn
Ml
T J 0) (0 to (1) a +>
•H *
O tn
tJ* G
•H a M o >
•H T J
M <0
00 CTl CO If) rH CTk CM O *H O H O
II tl II II II i! Ml Ml Ml Ml Ml Ml
Tl TJ TJ Tf TJ (1) (D a) <D (!) CO CJ w C 0) 0) •H W H W <u <d a a) dJ M +> M 4J M 9* CO & w & w X 3 X W w w w W
6 I TJ 0)
TJ T3 0 a) 0 a (0 w a) (0 (0 •H 0) <D M a CJ a)
4-> a« •H •H X * * w
CO o\
I ? •rl
O 0
0 -K at
P Oi
0 Cfi CO rH ta
0 M H 0) r - rH
r - •H M o\ TJ <d rH 0 04 w CO
vo vo i n r * o vo H rH CM H CM CM
II II II: II II
Ml Ml Ml Ml Ml Ml
TJ TJ TJ TJ TJ TJ a) a) a) 0) 0 0} w a w a W G W rl w •H CO H a) «j a) 0 a) M 4J M M 4J a « & CO 04 to X 3 X p X P w w w CO M W
W (1) a 4J
§1
LO CO 00
0) 3 0
* a 0 G
H 'H
T J a> 0) to <D
£
u c a)
•rl
&
w
rH •rl TJ 01 03
a M o EC g
>
c H
1 3 CD
T3 3 H O G
CD
P 0 £
-o
0 .Q
C H
CO
CD
H T5 3
P
W
M CD
P O
JC P
3 m
CD
jC P
x : p
a s O
O
0 1—I
- p 03
C
M O
CD CO
J C U p CD
0 a
6 > 1
II
tl4 n S Q)
X 5
• V • H p > CD 0
x : M p a
0
E CO
0
0 p P
CO p • -rH <D CO P
X I * H r d
p CO P
f d > 1 CO
r H Cd r H
11 C 03
03 C
s 1 O CX4 f0 - H
P P
CD * H • 6 T 5
<D T3 P CD <
O x :
2 4J
81
CO CD i—1 0 e f0 cn 1—1 m o
-H c
•rH i—1 o
r -a) <D <—i
i—i m Xi e 03 a>
EH 1*4
Qi
O •rH -P CO -H -P rd -p cn
5H o
a> a 4J u rd a.
MH o >1 G i—I -H •H Cn g •' rd M ki O
CI
CD O >H P O CO
>i TJ 3 4J cn
0 • TJ g •H O O •H g G O O SC G
CO (Ti
0 c £ o u CQ
o o
r* o in oo
in oo
II II
M N
•<& CM CO CO C\J VjD *—I
O O O tH
in o r- cm oo cri oo h
co co G G
co co
0 0 & jC CO CO
* H * H
Cxj ttj
CO g
in o
II
Ml
O O o t—1 o\o o\o o\o o\o o\o o\o T—1 r- on o\o o in oo o\o
o\o o\o II II II II , . * o\o • • • • • o
rH rH r- oo CM CM oo LO HOOO r- <X> M
M
2 S rH r- oo CM 00 kO 00 CM rH
X) TJ TJ T3 TJ T3 •d TJ TJ TJ TJ TJ TJ -a 0 a> 0 0 0 0 (D 0 0 0 0 0 0 0 G c G G G G a G G G G G G G
-H •H •rH •H -H •H •H •H •rH •H •rH •rH -rH -H m n} tX3 aj m rd (0 rd m rd rd rd rd rd -P -P •P -p -P -P •P -p -p -P •p -P -P -P CO W CO co co CO (0 CO CO CO CO CO CO CO P 3 3 P P 3 P 3 0 p p p p cn 0} cn cn cn cn CO cn cn cn cn cn cn cn
M S fa 0 "0 TJ V TJ V TJ TJ Cx4 S TJ
0 ai 0 0 <D 0 0 0 T3 O a TJ 0 TJ o a TJ o TJ o TJ TJ a 0 G a 0 G 0 G a 0 G 0 G 0 0 G CO 0 o CO 0 CO 0 0 CO 0 CO 0 CO CO 0 CO *H •H CO *H CO •rH •H CO •H CO •rH CO CO -H 0 u M 0 M 0 u U 0 U 0 u 0 0 M G 0 <D Pm G 0 G 0 (I) G 0 G 0 G G 0 -P a P< -P a -P a P* "p a -p a 4J jj a •H x ST - •H X -H X w •H XI •H XJ •H -H X S: w W S IS U 3: w m S w IS w 3: S M
LO o CM o o cn n rH o LO
C\3 oo in VO 00 cn cn 00 <sQ 00 VD
a! 5 •H
I
ui
a a
u 0 -P I—I 0 & cn
I>1 •p •H c P
o o
TJ 0) a o CQ •H M
a*
5-1 <D •p rH 0 x; C/3
5h 0) •p I—I 0 x: cn
o u CD
CD -p
M CD g 4-1 U -P rd m O-i CQ
G O •—1 CO 0 TJ x:
CM O rH CO CTi nc 00 •H cn 00 CTi fr4 rH a\ cn US rH
r-H <J\ —' o\ 1—1
44 rH rH G a) 4H w 0 — o O — a O -G ^ i—i CO 00 nJ i—1 <D a oo 0 U cn 04 CO 44 P fT\ Cn 0 cn o cd -H i—1 J>1 rH
p? s — 04 cr: —
CO CD p c
-H -p g o o
X! (d Eh
82
Ql
O •H
CO •H 4J
-P CO
SI o\°
u ID 0 o\o • o\o o\o o\o o\o
1> O ^ CTi 1—1 o\o on r- ^ CM "sT
TJ T3 TS "0 TS T3 U 0 0 0 0 0 0 0 g G a g G G G •H *H •H -H •H -H 4J 03 fT3 fO rti n3 as U -P -P -P -P -P -P rfl cn CO CO CO CO CO Cu 0 13 3 D 3
co CO CO CO CO CO
2 UA [14 s T3 -o k* S 0) 0 0 0
<4-1 o o XJ T3 u u O G •g 0 0 G G
0 0 CO CO 0 0 >i c •H •H CO CO «r-4 *H rH "H M U 0 0 U U -H CP CD 0 G G 0 0 j~i -H a, a -p -P a a 05 M X X •H *H X X k* o w w 3: !S W W
en <N in CI rH o CN
1—1
"C U 0
<D 0 0 M U -p -P 0 U rH 4J 0 0 a) -P o sz £3 CO CO • CQ
g fis
"•s >i -p •p jG u o 0 .
3 JC 4-3 00 o <J\ PC r—i <-#
M H, u >1 0 rH >1 r- 0 T3 T3 co -P CTi D >1^ -p CTi T—i -P C t—1 2 tH rfl co CO EH —' S
CO •H CO !>i H m c 03 1 03 -P (D e
CD x: -p
c *H
TS CD T5 2 i—I O C •H
0 U CD £
0 A
c -H
CO CD -H T5 3 -P C/3
CD -P O IS
83
CO <D r-H Q e m CJQ +j c CD T3 D
00 4J CO
<d i—i CD XI rH no EH S
01
o •P CO •H -P (T3 4J co
SI
o
H-4 o >1 G rH -H 6 --rd M kj o
CI
CJ> tji t/1 to -H -H to CO G G co co c G
cn i-i VO O CM *H O O O • H H O • • I . . . |
II II II II II II M| Ml Ml Ml Ml Ml
Tl Q) CO CO (I) 5 is
CM in
fc
<D a c © •rl a) & w
T) a) o fj a) •H a! * w
ta
fl) ̂ M rH 0 Cft 0 cn s i
M M a) a T) CJ G >i rt nj
73 X X a <D 0) -P H H CO
o CM
II Ml
-a a) 0) (0 ai 5 is
CM O CM
<4
T3 0 0 S c <D at 0) oo M o* 0 rH
(0 M a) a) a g m 3 flj o PQ CO
-P G (D
-P <—1 4H G mh O 0 CD O -H
i—1 > dP O o\° G on -H CM O r- > 00 G
T5 T3 T3 Tl TJ Tl T3 U 0 (1) 0) 0) (1) a) 0 0 Q) co G (0 G CO a (0 CO G CO H CO H (0 •H CO CO -P a) (d <D rd a> (d a) d> M M 4J M -P M 4J M U rd & <0 0* W a w a a &4 X EJ X 3 X 0 * X
H CO W CO W CO w w
M O TJ
(D T5 O CD G CO <D (0 *H 0 ^ G CL) -m a •H X 3E U
r- ro o o co • * G
rH ID a>
tn ^ II II
<N fN
cn i 2 T3 TJ X) a) ai a> CO CO co CO CO CO ci ai cd q q g 4J 4J 4J •H H -H £ £ £
r- CD CM CO rH rH II II
w cs
T3 0) O G 0) •H a! & w
0* CO CO -H CO G G CO G
£> 03 Eh
t3 Tl T3 V Tl Tl TS TJ "O a) a) a) a ai <D CD CD CD W CO CO to G co G CO G (0 <0 CO CO H CO *H CO -H d) d) 0) d) (4 d> fd a; m M M M M -P u 4-> M -P 04 04 0-4 a to a co a <n 5? x x X 0 x a x a w w w W CO w cn w co
TJ (D O G <y
M <d a pa
00 in 00 rH CM VD CM 00
1—1 rH
s, ^ - T? 00 >1 M GO u (d CTk cd rH a) a •W 0 -P T5 o CO M cd "" *<-«»
G CO co r-M 43 CD CD 3̂ CTk m •H O •rl rH LO & cn 0 — CO CTi | a\
rH >t M rH T5 G — (D 0) ''
M •H G g rd H CO •P 5 u G co at M O <D M a) H rt a G (D M M 3: fd PQ m < O
T3 a) o G a> -H n CD a x w
V£> o
84
a
s i
u o
u CD C -P U ftJ 04
VD O O O O C\! H O O H O
i n o n cm c o ^ O rH rH CM O CM
II II 11 II II II
U M M M M M
T3 TJ 0) 0) CO G (0 rl (U rd M -U
S 3
TS <u a w G CO H 0) a* M 4J a « x a
TJ T) a) <» « a 01 H a) «j M 4-»
& 3 w CO w . CO W CO
TJ a) o a OJ
•r|
a! & w
4H
S 1
T* TJ O a) 0}
CO CO > i C W CO
i—1 -H 0) <1) •H CP a q g -H 4J 4J n3 M >H -H k-4 O *
^ r* h cm r o cm on rH
CO
O rH O CO
o ^ cn co ct> co • O O O rH O
I
II ii II II II II
u M M M M M
*d a) 0} CO a> M
a*
TJ <1> C
•H nJ 4J
& 3
T3 0) (0 (0 <D
B
S
TJ <0 (0 CO <D a
•M
£
"d a) o a a}
M a) & w
O LD CO O O C • *
H CT* U3 O CM O rH rH
II II II II
Ml Ml Ml Ml
B" 3 w CO w CO w CO
ftto X 3 X 3
W CO H W
TJ 0)
0) a -M
3c
T5 0 a a a)
a> &
w
•H
jh o 4J
ID O _ O
no
CO ^ 0
3 G
-P c
T3I
CO
o T3I
CO o\ co a
•H rH U CO <D
a • O > i <—i (1) O +-> rH X) A •
. II O rd m 0 C H
U H II Ml MH rd ^
t J T3 T5 "0 0) <» a) <U (0 CO (0 to CO CO CO CO <U (1) (1) <D M M M M a & Pu Q. X X X X
W W w W
M M 0 TJ o *d TJ
a) a) a) T) 0 - o 0 a TJ (D c d) a c a) CO 0) CO 0) a) CO CO 'H CO H •H CO 0) M <D M M (D a a a a) <1) c -P 04 -P CL, p< 4J •H X H X X •H | 5 W ^ W w IS
CT» <n i n 0 i n CM i n 00 CM 0 CM <Ti rH CM CO CM rH rH
r* 0 a> T3 CD M M o\ o\ m m rH rH
a) a)
t* 0 a :
CO CO fH <-a 0 0 r - H P£J pcj o\ cd M P£J pcj
o\ a) (D C8 rH i n rH
*«-«» cn rH rH H o\ 0 0) rH rH a rH RJ 0} tN w *» — .
t>i X a) CO 0 OJ V CO (0 0) a tn cr» M 00 0 M M K td tP <n O cn 4J , 3 S •H rH n3 i—1 CO s s O pci P£j w CO —
85
QJ
O
-P
co
21
u o
o
>1 c rH -H •H CP g -rH 03 M
1X4 O
O CO CO CO • c c c
in cn cm ** r-f O O O
II it II II
Ml Ml Ml Ml
TJ 0) w 0} (1)
5 *H 15
T) (!) 0 a a) •r|
a! &
w
LD o
TH O rH H
II II
Ml Ml
XI Tl t3 *d V *d M ai a) a a) <D a) a> 0) a w a W CO c <0 •H 0) H W CO -p 0) td <D flJ 0 0) u M -P M -P M M as a CO a. w P* ft a, X 3 X X
w CQ w. w W w
<D
a
& w
in r -CI CO
CO m
o oT ON
ON ON rH H
w
T> !* 0 a) 0 M O u
tn td 0 M 0 •H -P 04 (d a it* 0
V >. 0} 0 13 -p -p 3 0 a 4J •P 0 c/3 CO Eh
CO CD H
•P cn
M CD O
XJ x : •p -p rd
m fd
ll CD x ;
Cxj 4->
•v x ; M -p 0 -H
x : £ -p o c e o
-rH ii -P
fd 2 o
•H c
U 3 CD §
XJ S -P o O o e
1—1 o fd •P c
o CO
(D M X! CD -P a fd hA >1
• JQ 11 01
•H 13 2 CO CD Cu 73
1—j •H rd > a O
a) fd M x : i a •p fd rd -P CO
<+H CD o e •H
o •p -P c CO
-H *H u -P <D TS fd
x : <D •P •P T5 CO O P e •—I I—1
O fd ii a C
-H o Zu "H
•d -P H -H
• 0 T5 CD A TS -P < O c ss -H *
C O e M O S
C O 6
0 e 1 c o 52
86
CO a )
i—i
0 ,
6 m
c n
• P c a )
T 5 D
- p
w CTk
CP a ) i—i
»—i 03 J Q S ca CD
E h Cl4
Ql
S I
u Q
t-w O
>1 C rH -H *H c n & *H (T3
O
C|
T5 <1) U U O Qu 0 M
CM i n ^ o o o o CN o o o • o o . . * I
II II II II II II
Mi Ml Ml Ml Ml Ml
"D D)
(0 0) G
•P
£
00 CM CM
Cm
TJ 0) 0 a D) •H M 0) &
W
TJ 0) 0 A
•H M (1)
W
•» TH 0) CTK M CTK O RH 0 ^ A T
M - <D M TJ O A TJ (D G X
>I (D A) T3 X 3 AI 4; P H CO <* U0
- p
a Q
-i—> i—i c o <D - H
rH > O C
-H O > G
oV> o\o O 0 0 LO cm
T* TJ TF TJ TJ TJ TJ TJ M a) 0) a) a) <D a) 0 0 0) (0 G w a co a CO CO A CO •HI (0 H (0 H CO CO -P 0) <TJ (!) (d a) <d 0 0 M M 4 J M -P M -P U U A3 CO O i co a co c u A DM W 3 X 0 X 3 >C X
W CO m c o W CO w u
CM
LO O
o\
i n
II
CN X
CO G
\ CO a
T* <D
0) M
W
u O "0
0 T> TJ O TJ 0 0 £ 0) CO CO 0 CO CO CO -H 0} 0 0 M a) C Ua G 0 q 4J S P A -P •H \ •H X •H 3 : S 3 : W Cw
r o 0 e n 0 0
CM
O CM O O
CM 1 0 • i n
M * H i n
II II
N (N
• d T3 a> a) G a
•H H (d at
• P - P (0 0)
c o co
TJ - ® TJ 0 a) a (0 0) (0 H AI M G a) + j pu H X
£ pa
r** O\
O CO • g
R-o
V0
TJ TJ 0 ) C D a g
•rl *H (d m •P - P CO CO 3 3
co cn
T5 0
TJ O 01 G 01 0 CO -rH a) m G A) -P a •H X I k w
o rH CM
CO M 0 0 3 . G
•H -P C O O
0 0 HI 0
• 1—I JQ
II td EH
Ml - —
TJ 0 A
•rl (d - p
0)
CO
TJ <D O G 0)
•r l M A> &
W
o o
> 1 0 0 >I A) O OO U ** H CTi (D
0 D) 0
rH (D 0 CFI OI UJ
>—- Hi • d us A) "«•
T3 O M rH M u a) - P •P -03 n 4 J KO p CO •rl G CO —> A A) CTk OTI G G M t n 0 0) G o\ r l CD 0 CN cr> -P A rH - ^ gu
m •H •rl 0) 1' TJ M 0) rH A) M (D (D •P
Hi •H •P A CO M CO 1
n A w - ^ CO CRT H 0. u •H M >i G z G H ^ G — cd H -H <D -RL 0 •rl H CM 0 ^0 G CO M M-L G H 0 Cft p O»
0 < «W G G H % Cft & cr> 0 M 0 a FD O O rH O rH PQ P3 0 a : b A ~ O —
87
Ql
o -H -P CO
-H -P rd -P en
SI
to o
5-1 CD £ -P to fO Oj
co to G C
. CM
W cri
« II
Tl Tl o a> co a (0 rl <D aJ to -P & §
W CO
CP c o c o c o c o c o - h c o c o c c c d c w d d
CO X! m -p co CO o to o
T3 TJ T3 TJ Q) <D
CO C CO -H d) (13 to P a CO X 3
a . co X 3
-a T5
a co X 3
Qu co x ^
w c n u w u c n u w
T3 T3 <D <D
<4-4 TJ TJ O T3 0 O 0 CD C 0 C
(0 CO <D CO (1) >1 C 10 CO CO
r—1 *H <D a) to Q> to •H CT a c <D £ 0 £ -H P a P a ftJ to •H -H x •H X [n O is w IS w
CM (M o rH
CI 00 00 CTi CI tH i—1
M T ) H CM CO ^
CM ^ r-4 CM CO CM
on cm \o in n o vo o o o o • o •
I
tl II II
M U M
I
II II I!
to to to
T3 T3 13 TJ T3 T3 d) (0
a 0 d) 0) a> d) (0 a w a CO c
0} •H 0) rl (0 •rl a) aJ d) (tf a to 4J to 4i to -M
CO a* o) a W 5? a K 3 X 0
w CO M w m CO
TJ a> 0) (0 <u
5
£
T5 0) 0)
*15
TJ a) o c
<D •H
a5 &
w
CD o\
to TH <D d "W
* J 0 (0 0 w £ CQ 0
rt d ta LO 0 00 ta i (T\ >i U rH H d) to — ' rH C (fl - Rj
>1 P ^ r - to 43 TJ to o oo (U 0) 3 aJ o cn a to P M h m S CQ a co ^ h3 S
CO O H O rH o
• • o
KD H ro ^ o
o o o o o o o o
H CO H h o o o o
o CM
h CO CM O H H CM CM
CO 0) 3 G
-H •P C
O
o
II II II II II II
to to to to to to
T> xf a) <d co a w H
2 3
& 3
T) d) 0) to (0 to
& 3
TJ a) c
•H aJ 4J
T) (D o> a) 0)
5
is
in rH CO
o o> a t
o
(d
CQ to
T3 (I) (0 w 0)
£
i s
13 d) a a ai •rl to <d
w
it II II tl
to) tol tol tol
TJ 0) (0 co d) to
Tl (1) a
•H rt 4J
W C O W W M C O W 01 W CO
T3 — ® TJ 0 a) a 0) CD W H 0 to c a)
3 * !s w
m in
r -o> o>
0) <w a> $
88
Ql
-P CO •H -P 03 -P
CO
M O
o >1 c
i—I -H •H t n 6 -H 03 M DM O
Cft rH
tl Ml
T5 (U a c a>
•H
a!
a -w
o GO CM
i*
Cn U C O -H -P TJ O 03 O (13 O
CO •H CO !>i
M O P '—I o m 03 C
£n cd
- a 0) CO CO 0) e p •H 3 :
<j\ i—I
M O flj ov CM a) at <L> CO rH rH CTi ^ "W» i—1 t—1 o 0 a s*s -H T3 !>i CO H M T3 d> CO 03 & <D o 3: -P H M 03 o CO Pi PQ CO tc
CO CO to CO C G C C
h in in r-o o o o
II il II II
Ml Ml Ml M|
*0 <l)
0) a -P
00
*d a) a a a) •H M (1)
8 -W
o CT> H
TJ O O
Ul 0
to fcl -P O
LO CO o c •
as o O rH
II
Ml Ml
"d TJ TJ Tl Tl T3 TJ Tl u a> a) a) a) a) 0 a) a) o 01 CO to g to a to to c to CO to -H to -H (0 0) -p a) a> 0) 03 0 03 a) tu M M u M 4J M P M M 03 P* a Pi to 0| CO fr ^ 04 5? 5? a « 3̂ X X w w w w W CO w w
w
o o i n
0) -P co at
UJ
o -P CM rt ct> a o
TJ ^ o •P * c a) o M E-i U
CO CD H
T5 3 P cn
M 0)
, G P rx5
4H
U4
M CD x: P o
II s
M CD JC P o 6 o p
u CD
j c p rc3
MH
S
M CD
JC P ra
*w
o •p
<D , G P o S
CX4 S
CD P O
52
CO •H CO > .
i—I A3 G CTJ I rti P CD e CD
X P
G *H
T3 0
- a
rH O g
•H
o A g
•rH
CO u CD
T5 G CD
MH MH O
CD i—I -rH G CD >
3
G O 6 u o
s
g
o e M 0 £ 1 G O G
89
a) i—I X5 m Eh
t>i -P •H CD
. £ CI) ts> O 6 o
*0 £ 03
CD N -H cn
4J u 0)
w
c o •H •P as i—i 3 Q, o cu
CD o £ <D I—I o •H
•H
f0 Pu
Ql
SI
CD M O 0 G G 0) -P rH U O rd -H Du >
co 0)
r—r XI 03 -H M rd > G •H Cn •H
• M O
4-1 O
>i 1—1 i—1 ro -H X e o rd 1—[ Cx4 o
in o
o o
CM O O
O O O
O o o
o o o
o o o
o CD <X> 00 CTi r- CO 00 00 IT) o CM CTi
rl r- CO CTi LO ID *—1 KD 00 O (J\ CO 1—1 %—1 *—1 CM
in CTi
o
!>1 g <
<dD CM
CTi CM
<J\ oo
CO
VD CO
r-CM o
CM r-
o CM
CNJ r-
LO
r-
oo
! T3 -D 0 0 CO G
>1 >i >i >i CO -H G G G G 0 rd < < < < U -P
a, CO X P w CO
(0 w CD a> i—I CO H 03 CD rd 6 rH n CD m 0) U, g CO CO 1X4 CD 0 0
[tl I—1 (—1 Uf 03 03
01 6 £ (0 CD 0 0 dJ i—1 co tl-4 CX4 H rd CD
1 2 •—i
rd CO 0
S CO rH CO CO rH 4J CD 0} 0 0
-H -P i—1 S 1—t i—1 0 G G 03 0 03 03 *r| P a) 2 2 2 c § T3 •H c p i—( <—i i—i rH H O 4-> ! 1 (—i )—1 i—1 O U CO < < <C <
H CO CM o in as \o in t—1 o r- CM O *o O o o rH
LO CM ro o r- GO LO CM rH
U) VO r> CO CO LO rH CM CO
rH
CM oo o
00 CO KO o VO r- \o CM rH CO LO rH
r- CO VD %o in o O CTi o 00 CO 00 CM ro CO rH o ro a\ 00 in
*- **
00 ro CM o ro CM CM 1—1 rH
! ! T3 Tt 0 CD CO a
>i CO •H G 0 03 < U 4->
a (0
w CO
co 0 i—i 03 s
(0 0) <D CJ rH CO rH JS 0 J 2 i—1 s rO
2 CO rH -P 0
•H •P rH 0 6 G 03 •H P 0 2 a T3 •H S P •—i rH o -P i—i a o CO <
0) 0)
H
3
CO 0 P c -H -P G o o
XI 03 &H
ro CM
CO
CO V£> VD
S
co 0) i—i ro e 0) [14
to (V H
m
a En
flj 0 •H CJ *H rH u
90
v
Q l
(D G CD Cn O e O
I D CM O o O \—! O o O O O o
VD !—1 t—l 00 KD O
LD VO oo O O
*^r LO vo r - o CO o
6 (tJ En
° l j l LO o CM on HI rH rH rH *—i
LO CM CM CM 00 CO CO
2 1 KO CO rH VD 2 1 LO r - LO
<—1 CM 00 m (13 «w •*
-P o CM o CM O T—J rH rH rH EH
* o T3 a) CD CP
n o CO G CD G >1 > 1 CO *H
G cd a G CD Cfl
4J r-t < < iH 4J 5-1 O a CO rd *H X 3 cu > w 00
CO 0
rH
X3 <T3 CO • H a> U rH CO
m m cd > S rH CU rfl
G • lu £ CO CO - H 0 a) cd CJt [l4 rH i—I •H -M fO CO M -H • -P £ g O G G CD CD
3 <D fxj Exj <4-4 i T3 O fE; £3 i—1 '—1
0 4-> >—1 1—1 > i a w rt!
i—I
- H CO
•—i (TJ G m i <TS - P CD
B a> x: P
T 5 CD T> 3
V
Q l
CO • H
M
CD N
- H cn 4->
o CD
M-4 m
w
- p
0)
6 CO
- H
I D o
A
Q l
• V o G TJ G O >i 0) -H •H P G P •H
CO m CD (G (D i—i G 4J is] 3 CD CO •rH a tr> 3 03 o 0 CO PL4 6
P o M O "0 JG O 0 <D m 4J u "0 <4H M 0 G CD 0 UH fO a
rH CD p "0 fO u CO CD 3 CD CO T3 JC p CO -rH o CD > m UH
-H 0 •H a T3 X G P TJ W -H o H
<4H 0 II <4H pQ
O CD CO O G
U G -H CD r0
O JQ O CO \ e •H •H us MH
G -H CD W G M
II Cn 3 • •H Cn
CD cn •H P Cl4 O • . .
*—1 CM r o
91
0) i—| JQ
n3 Eh
>t -P
CD C CD D' O e o
•• as
n
"0 C rd
CD N *H CO
U CD 4-4 M-l bP
4J 03 i—| 3 Q,
O CU
CD O c CD <—I o -H
c •H
M O
o
s 03 pL4
"O (D CO U) CD £ -P •H 3:
v >, 01
P •H CD c CD tT> O g( O ^ 33
M
CD H O <D C C CD P r-H ^ o fTJ -H a. >
CD CO a CD c i—1 CD X) rH rO o -H -H U > ro > >—i;
03 a P •H C Cn CD *H U M H3 O
4H 73 O CD
co >1 CO i—1 (D •H C & P (U -H Dt, 3:
o o o
o CT\ as
no oo r-
CM
>i C <
CM CM CM
m
tn
£
w 0)
•a
2 & t*
to 0)
rd o •H C •r| H U
r-o
CM
CM on
GO (31 o
vo •̂r
oo
rH 00 CM CM CM VO o o CM CTk rH O Ok o o CM r- rH O
rH CM 00 00 00 00 LO <£> 00 VO CM CO
00 "sP in rH CM VO r- rH CM LO
VO CM
cn
m rH VO
in rH
00 CM
O* O
CD cn
o*
!
(0 &|
0} 0 H
I
£ •H a 3
o u
<d
i <4
0) a)
-M a a) T3 a -u CO
<£> 00
O
CM
00 00
ID 00 00
! c < !>i C <
CO a) i—i rd 2
CO (D i—I f0 S CD Ext
VO CM
00
ro TH
o
IT)
CM in
00
m r-
I I ! "0 <D CO CO CD U a X u
TJ d) a *H (d 4J 0)
co
CO CD I—I 03 s
ca ai H
o •H a •H rH U
<D H
£ * •H a 3
o a
-M £ rS P +J CO
CO CD i—I 03 S
cn 0)
CO
oo CD
CTi oo CM 3 CM o o C o CM o -H
• • • P C O o
<D 00 CM CM i—| 00 00 <Xi XI
• • • <TJ (S\ oo CM EH CM CM
00 o
r- 00 00 o o CM rH CM CM
o\ <S\ VO r-a\ VO cn 00 rH 00 r-
LO rH r- VO kD
>1 >f c
>1
s s
0 o c CD i—1 o -H > U CD X! -P
co O CD s <—1 0} m 0 o S H p CD rt Cxj u
d) CD >1 PM x: •p p •H 4J 0J c c CX4 D a)
CO g Tl T3 Q) s a) '—1 o 4J CO A3 o CO CO e CD CD c tM p
•H s:
92
V 1—1 rH rH
01 o to CM O
01 1 o ID CM
o * • O •
-H <y c a; cn o s O .. vc x
Ml
*1 0 0 CM vo in
SI o on r* oo CM i n to VJD 1—1 IT) CM CM 0 0 ra H -P rH
<D U O Q) C C Q)
4J rH M O fd *H cu >
CO cu i—i J3 ftj -H M 03 >
Cn •H U o u-i o
-H e nj Cl4
CM 0 0
Ot CM o o
o rH co vo O M3 o o r -
.54
.34
.42
.78
.39
m m
O O 6
8 6
8 4
8
23
26
o vo on CNJ
o VO rH 3
6.
r -
27
.
10.
H
CN r* on o r -o
>i c < I ! T3 <D CO cn a) u a x w
c
0} a)
at a •H a •H r-4 U
0} ai
+) c a> tj 5 CO
CO <D i—I rtf S
I> i n io
TJ 0) a •H (d 4J (0 p CO
I
03 a)
(0 & a) b
>i <
a> a c a) •—i o -H > U 0) x -P 03 Ua
M <D X -P O
T3 0 CO CO dJ c •p
-H 3:
VD 00
CM LO in o KD rH t—1 T—i
r* r~ on KP <N rH o r- 00 KO CM o 00 rH rH
** CM in 00
CM
! >i c •a:
w 0) H I
H c
CO dJ <—i fd e CD
>1 £
(1)
«!
I &
•r| e a
o u
0) rH ft d) En •U a Tj 3 4J CO
TS CD C H CO •P CO P CO 5H o
TJ C 03 TS CD CO CO CD M a K W
en
o
o o o
CO o . -H •
CO >i V i—1 f0 Ql c Ql 03 CO 1 -H as •
- P s. - P
CD M CD 6 6 CD CD CO x: N -rH - P -rH
CO £ LO
*H - P o 0 •
73 CD CD 4H A T5 3 w Ql i—1 U c C 0 -H •PH - P
- P •H CO CO CD (D »—1 C N D CD *H a Cn CO o 0 e 4J O O TU x: <D CD 4-H 4-> M
M o CD O 4-1
a i—1 CD -p flj M CO 3 CD TS jC • P
O > (0 CD -H
T5 £ •o
O H 0
<U A O CD
U c M C •H
CD F0 XI O CO e •H -H a MH c •H CD c M II CN 3 -*1
-H Cr> -*1 CO -H PD
• • •
I—i CNJ 00
93
-P
<D G CD D i O e o
rc
CN
X3 G m
CD N •H CO
O CD
W
*0
D i O
04
» •
CD U C (U i—1 o -H > c
•H 0 •H M O
0 > rH
. e 03 PM T3 CD O
CNJ C i—1 CD
•H a) M i—i CD £) Q, «5 X EH w
>» -P -H (D G Q) CJi
i r
o a:
Q) U O OJ G £ <U 4J r-H u o <T3 -H ^ >
CO 0 —I J3 ru
H M (13 >
G H tn -H u
O
hh o
O CN CN o to i—I o o o
I—I o o
o o o
o o
CTfc lO rH rH r -oo H KD o cr\
ch <X> IT) rH 00 KD CM CN VD CO KO
i—i
<s£> vo
>1 c <
VD CM
00 CM 00
LO 00 CM
rH 00 "3* if) CM CO o CN r - LO rH o <JS rH 00 O
•* r - CM <J\ VD (Ti rH
(Ti
I
>1 c >1 c <
>1 c <
>1 G
CO 0 H (If
01 <D H
<tf a •H G •H H a
CO <D
CO 03 a> s rH cu 03
[x4 e CD Cn
CO CD
>—1 CO ra 0) s 1—1
03 >1 2 CO 4-> aj •H 4-J •—i G G no
a) S £ T3 s D "—I o 4J i—I
o CO <
V) a> i—i 03 a <D Cxj
r H 00 rH 00 r H CN rH <X> <J\ rH O rH m CN rH
O O O o r H O
O O rH LO CM 0 0 cn r * r - i—1
r H cn cy> o\ KD CM KD CM CM LO : rH r H rH
* * *
*0 CM CM H
rH 00 CTi CM rH rH O H o
O kj* O CM
!
un
m oo oo
KO CM
CM 00 *31
! G
r - 1—1 (N o rH ID 00 LO kO o\
• V rH rH rH
T3 T3 TJ TJ a) d) a) <P CO G (0 G CO •H CO -H CD (T3 0 03 n 4J M +-> a CO P< CO x P 3
C£J cn CO
co CD
<—I (T3 S
3
rt a
•H G
•rl H U
& •H G g o a
CO CD <—I m S
4J G 0) T3 3 4-) 00
CO 0 i—1 (13 S
CO 0 i—i cd S
CO 0)
JJ G
'O 5 CO
CO a)
rH nj S
G a)
T3 3 4-> co
w a> G
X! 03
94
v > , 0 4
-P •H CD G CD
o e o «. ac ^
Ml
23
<D U O d) G G CD -P <—I u o 03 -H CU >
o\ o r -o i vo o iH O O
o rH
co CM
CO
co VD vo
! !
CO 0
r-4 JQ 03 CO
•H w a) 0 H
(T3 rH flj > flj 3
i a) G 0) fa -H EM
fa D> >1
*H . H M fll •H O O G
•H <4-1 CO G O <D •H ft
rH rH 0 I>i 03 a o
rH e o
•H a; g tl4 03 Cu
rH 00 CM i n i—1 to 1—I CO a\ CM rH O O C\ o o rH io CM O O rH CM o o o o o O m
TH CO r - cn i n c* *—1 i n CO i—I TH 00 cn <vO o o CO 00
CM rH m LO i n rH o KD o rH CO on rH rH CN] rH
CT\ i n CM
!>i G
>1 G
U CD
JC P 03 1X4
& 0 M
CO (D 0) 0
<—1 G m a) £ <—1 CD o Exj •H CO
> a) -P •—i G T3 0J CD CD S T3 O D G "—i P Q) i—I CO -H <
M CP a X w
* i n o
CN
*0 -a ai (!) CQ a w •H
G 0) (d < M -M
9* w X 3 w CO
0) 0) H H
I I H H
3 3
r - CM a\ CM <30 nj« r -o "NT *—t rH O o VO r - 00 o r - m r -
*»
CM LO 1—1 rH
CM
ID i n
£>i G
CO 0>
I—I 03 fi CD £u
LO r -Ul
!
co CM
s>i G <
co on
i n
LO co
rn
00 o r -
TJ CD co co CD U a x M
!
M a) x: p 0 s
a o U
0) MH (!) 0)
h a) a) (U o rH 0 G <g <I) CD s fa rH
o HI 4J -H to (d G > CD 0 a) rH H
x) -a 03 G p a> S H -P a rH CQ G rH CJ
<D rH •H < M CD Q X W
co CU I—I m S CD Cu
* * * *
CD eg m CN rH o
CM
KO r» KO i n i n m r - *j<
> i G f=C !
(0 a)
H f t
-P C 0)
TJ
3 CO
TJ 0)
fU P CO 3
en
M O \ TS C 03
" 0 CU CO CO CD M a X
pa
CO
u o \
w
CO ftf
-P a CU a X <1)
%k rH O O O
CO O •rH •
CO !>t V
i—I a
a a
cd i M
^ — «
- p P CU CO CU e -H g
CD CU CO x : N •rH -P •H
CO a LO
•H -P o a •
- a CU CD MH A
T5 mh D M Of
i—1 O a c o > i
•rH -H P P • -H
CO m l o CU a) rH O C N 3 O CU
•H a . o Cn CO o • 0
0-i 6 -P V o O T3 X CD cu a
MH P * M m SH * O CD O * MH
a i—i CU P aJ U CM CO 3 O CU
TJ JU o P •H o • > *0 V CM
•rH CU •H T3 a C 5H -K "0
- H O -K rH ^H 0
rQ o cu m
o o c M c • *H 0) (XS
Xi O A CO S •H -H p MH a c -H -)< Q)
c U II •• D
^ 1 -H TJ Cn
^ 1 cn CU -H P
• • O •
rH CM C 00
APPENDIX C
FIGURE
95
96
-.9 I - . 8 I -.7 1 -.6 I -.5 1 -.4 I -.3 I -.2 1 -.1 I -.0 I 0122333 + . 0 I 011122333334444555555555555666777888888999999999 + . 1 I 0000000011112222233344456666677777888899999 + . 2 I 00000001112222344555667778899 + . 3 I 0012222378999 +.4 I 16 +.5 I 13 + .6 I + .7 I 1 + .8 I + .91
Fiqure—l_j_ Stem-and-leaf-display for 144 effect sizes in the
family of origin variables data set. The values in the left
column represent the first digit of the effect size (r), and
the values within the graph represent the second digit of
the effect size.
REFERENCES
References marked with an asterisk indicate studies included
in the meta-analysis.
Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D.
(1978). Learned helplessness in humans: Critique and
reformulation. Journal of Abnormal Psychology. 87. 49-74.
Aldarondo, E. & Sugarman, D. B. (1996). Risk marker
analysis of the cessation and persistence of wife assault.
Journal of Consulting and Clinical Psychology. 64. 1010-
1019 .
^Alexander, P. C., Moore, S., & Alexander E. R. (1991).
What is transmitted in the intergenerational transmission of
violence? Journal of Marriage and the Family. 53. 657-667.
Anderson, C. R., Krull, D. S., Weiner, B. (1996).
Explanations: Processes and Consequences. In E. T. Higgins &
A. W. Kruglanski (Eds.) Social Psychology Handbook of Basic
Principles (pp. 271-296). NY: The Guilford Press.
Azar, B. (1997, January). Poor recall mars research and
treatment. APA Monitor- p. i, 29.
Bandura, A. (1973). Aggression: A social learning
analysis. Englewood Cliffs, NJ: Prentice-Hall.
*Barnes, G. E., Greenwood, L., & Sommer, R. (1989).
Courtship violence in a Canadian sample of male college
97
98
students. Paper presented at the NCFR 51st Annual
Conference, New Orleans.
*Barnett, 0. W. & Fagan, R. W. (1993). Alcohol use in
male spouse abusers and their female partners. Journal nf
Family Violence. 8. 1-25.
*Barnett, 0. W. & Hamberger, L. (1992). The assessment
of martitally violent men on the California Psychological
Inventory Violence and Victims. 7. 15-28.
*Barnett, 0. W. , Martinez, T. E., & Bluestein, B. W.
(1995). Jealousy and romantic attachment in maritally
violent and nonviolent men. Journal of Interpersnnal
Violence. 10. 473-486.
Barnett, 0. W. , Showalter, S. M. , & Reynolds, T. P.
(1997, June) . Relationships between internersona 1. V I Q I C T P P
and interpersonal—control. Paper presented at the 5 th
International Family Violence Research Conference, Durham,
NH.
Beasley, R., & Stoltenberg, C. D. (1992) Personality
characteristics of male spouse abusers. Professional
Psychology:—Research and. Practice. 23. 310-317.
Bennett, L. W., Tolman, R. M., Rogalski, C. J., &
Srinivasaraghavan, J. (1994) . Domestic abuse by male alcohol
and drug addicts. Violence and. Victims. 9. 359-368.
Bernard, M. L., & Bernard J. L. (1983). Violent
intimacy: The family as a model for love relationships.
Family Relat-.-i one-, ^ 283-286.
99
Brennan, A. F. (1985). Political and psychosocial
issues in psychotherapy for spouse abusers: Implications for
treatment. Psychotherapy, 22, 643-654.
*Breslin, F. C., Riggs, D. S., O'Leary, K. D., & Arias,
I. (1990). Family precursors: Expected and actual
consequences of dating aggression. Journal—of—Interpersonal
Violence. 5, 247-258.
Browne, A. (1986). Asssault and homicide at home: When
battered women kill. In M. J. Saks & L. Saxes (Eds.),
Advances in applied social psychology (vol. 3, pp. 57-79).
Hillsdale, NJ: Erlbaum.
Burke, P. J., Stets, J.E., & Pirog-Good, M.A. (1988).
Gender identity, self-esteem, and physical and sexual abuse
in dating relationships. Social Psychology Quarterly,—5JL_
272-285.
*Caesar, P. L. (1985). The wife beater: Personality and
psychosocial characteristics. Presented at American
Psychological Association National Convention, Los Angeles.
Caesar, P. L. (1988). Exposure to violence in the
families-of-origin among wife-abusers and maritally
nonviolent men. Violence and Victims, 3. 49-63.
Cantos, Al L., Neidig, P. H., & O'Leary, K. D. (1994).
Injuries of women and men in a treatment program for
domestic violence. Journal of Family Violence, 9. 113-125.
100
*Cappell, C. & Heiner, R. B. (1990). The
intergenerational transmission of family aggression. Journal
of Family Violence. 5, 135-152.
Cascardi, M. , O'Leary, K. D., Lawrence, E. E. & Schlee,
K. A. (1995) . Characteristics of women physically abused by
their spouse and who seek treatment regarding marital
conflict. Journal of Consul ting and Clincial Psychology, 63,
616-23.
*Coffey, P., Leitenberg, H., Henning, K. Bennett, R.
T., & Jankowski, M. K. (1996). Dating violence: The
association between methods of coping and women's
psychological adjustment. Violence & Victims,—IL. 227-238.
Cohen, J. (1977). Statistical power analysis for the
behav:oral sciences. New York: Academic Press.
Cohen, J. (1994). The earth is round (p<.05). American
Psychologist. 49. 997-1003.
*DeMaris, A., & Jackson, J. K. (1987, October).
Batterers' reports of recidivism after counseling. The
Journal of Contemporary Social Work. 458-465.
*Dewhurst, A. M., Moore, R. J., & Alfano, D. P. (1992).
Aggression against women by men: Sexual and spousal assault.
Journal of Offender Rehabilitation. 18, 39-47.
Dobash, R. E., & Dobash, R. P. (1979). Violence against
wives: A case against patriarchy. New York: Free Press.
*Doumas, D., Margolin, G., and John, R. S. (1994). The
Intergenerational Transmission of Aggression Across Three
Generations. Journal of Family Violence. 9, 157-175.
101
*Downs, W. , Miller, B., Testa, M., & Panek, D. (1992).
Long-term effects of parent-to-child violence for women.
Journal of Interpersonal Violence,—7_<_ 365-382.
Dutton, D. G. (1995) . Intimate abusiveness. Clinical
Psychology: Science and Practice,—2j_ 207-224.
*Dutton, D. G., Starzomski, A., & Ryan, L. (1996).
Antecedents of abusive personality and abusive behavior in
wife assaulters. Journal of Family Violence,—11> 113-132.
Edleson, J. L., & Grusznski, R. J. (1987). Treatxng
men who batter: Four years of outcome data from the domestic
abuse project. Journal of Social Service Research,—12, 3-22.
Feldman, C. M., & Ridley, C. A. (1995). The etiology
and treatment of domestic violence between adult partners.
Pl ini rial Psvcholnav: Science and Practice, 2, 317-348.
*Folligstad, D. R., Rutledge, L. L., McNeill-Harkins,
K., & Polek, D. S. (1992). Factors related to physical
violence in dating relationships. In E. C. Viano (Ed.),
Intimate violence: Interdisciplinary perspectives (pp.
121-135). New York: Hemisphere.
Fortune, M. M. (1993). The nature of abuse. Pastoral
Psychology. 41. 275-288.
Gelles, R. J. (1974) . The violent home. Beverly Hills,
CA: Sage.
Gelles, R. J. (1976). Abused wives: Why do they stay?
Journal of Marriage and the Family, 38, 659-668.
Gelles, R. J., & Straus, M. A. (1979). Determinants of
violence in the family: Toward a theoretical integration. In
102
W. R. Burr, R. Hill, F. Ivan, & I. Reiss (Eds.) Contemporary
theories about the family: Research.—based—theories, (Vo1. 1,
pp. 59-81). New York: The Free Press.
Gerbi, L. (1994) . Spousal violence: Understanding and
intervention techniques. Journal of Family Psychotherapy,—5_
(4̂ . 19-30.
Glass, G. V. (1976). Primary, secondary and mata-
analysis of research. Educational Researcher,—19, 3-8.
*Goguen, C. A., Stepaniuk, H. A., Foy, D. W. (1996,
September). The relationship between dating violence and
other traumas among female adolescent—offenders. Paper
presented at the American Psychological Association Women's
Health confernece, Washington, DC.
Gondolf, E. W., Fisher, E., & McFerron, J. R. (1990).
The helpseeking behavior of battererd women: An analysis of
6,000 shelter interviews. In E. G. Viano (Ed.), The
victimology handbook: Research treatment and public policy,
(pp. 113-127). New York: Garland Publishing.
Goode, W. J. (1971) Force and violence in the family.
Journal of Marriage and the Family. 33, 624-636.
Grayson, J. & Smith, G. (1981). Marital violence and
helpseeking patterns in a micropolitan community.
Victimology: An International Journal, 6 , 188-197.
Gray, H. M., & Foshee, V. (1997). Adolescent dating
violence: Differences between one-sided and mutually violent
profiles. Journal of Interpersonal Violence. 12. 126-141.
103
*Gwartney-Gibbs, P. A., Stockard, J., & Bohmer, S.
(1987). Learning courtship aggression: The influence of
parents, peers, and personal experiences. Family Relations,
36. 276-282.
Hamberger, L. K., & Hastings, J. E. (1986). Personality
correlates of men who abuse their partners: A cross-
validation study. Journal of Family Violence,—l_i_ 323-341.
*Hamberger, L. K., & Hastings, J. E. (1991).
Personality correlates of men who batter and nonviolent men:
Some continuities and discontinuities. Journal—of—Family
Violence. 6. 131-147.
Hanks, S. E. (1992). Translating theory into practice:
A conceptual framework for clinical assessment, differential
diagnosis, and multi-modal treatment of maritally violent
individuals, couples and families. In E.C. Viano (Ed.),
Intimate violence: Interdisciplinary Perspectives, (pp.
157-176). New York: Hemisphere.
Herzberger, S. D. & Tennen, H. (1988). Applying the
label of physical abuse. In G. T. Hotaling, D. Finkelhor, J.
Fitzpatrick, M. A. Straus (Eds.), Research and policy
perspectives. Newbury Park, CA: Sage.
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. 101-123.
Hotaling, G, T., & Sugarman, D. B. (1990). A risk
marker analysis of assaulted wives. Journal of Family
Violence. 5. 1-13.
104
*Howell, M. J., & Pugliesi, K. L. (1988). Husbands who
harm: Predicting spousal violence by men. Journal of Family
Violence. 3. 15-27.
Hunter, J. E. (1997). Needed: A ban on the significance
test. Psychological Science, JL. 3-7.
*Johnston, M. E. (1988). Correlates of early violence
experience among men who are abusive toward female mates.
In G. T. Hotaling, A. Finkelhor, J. T. Kirkpatrick, & M. A.
Straus (Eds.), Family abuse and its consequences, (pp 192-
202). Beverly Hills: Sage.
*Kalmuss, D., & Seltzer, J. A. (1989). A framework for
studying family socialization over the life cycle: The case
of family violence. Journal of Family Issues,—10i 339-358.
Kaufman-Kantor, G., & Straus, M. A. (1989). Substance
abuse as a precipitant of wife abuse victimizations.
American Journal of Drug and Alcohol Abuse.—15(2) , 173-189.
Knight, G. P., Fabes, R. A., & Higgins, D. A. (1996).
Concerns about drawing causal inferences from meta-analyses:
An example in the study of gender differences in aggression.
Psychological Bulletin. 119. 410-421.
Koslof, K. E. (1984) . The battered woman: A
developmental perspective. Smith College Studies—in Social
Work. 54(31. 181-203.
Koss, M. P. (1990). The women's mental health research
agenda: Violence against women. American Psychologist.—45,
374-380.
105
*Lake, E. S. (1993). An exploration of the violent
victim experiences of female offenders. Violence—and
Victims. 8, 41-51.
Laner, M. R. (1985) . Unpleasant, aggressive, and
abusive activities in courtship: A comparison of Mormon and
non-Mormon college students. Deviant Behavior,—6_<_ 145-168.
*Lockhart, L., & White, B. W. (1989). Understanding
marital violence in the black community. Journal of
Interpersonal Violence. 4. 421-436.
Mackey, T. F. (1992) . A psychological model for
analysis of outcome related to trauma. In C. M. Sampselle
(Ed.), Violence against women: Nursing research,—education,
and practice issues (pp. 45-68). New York: Hemisphere.
Maiuro, R. D., Cahn, T. S., Vitaliano, P. P., Wagner,
B. C., & Zegree, J. B. (1988). Anger, hostility, and
depression in domestically violent versus generally
assaultive men and nonviolent control subjects. Journal—of
Consulting and Clinical Psychology. 56, 17-23.
Malone, J., Tyree, A., & O'Leary, K. D. (1989).
Generalization and containment: Different effects of past
aggression for wives and husbands. Journal of Marriage—and
the Family. 51. 687-697.
Marshall, L. L. (1992) . Development of the severity of
violence against women scale. Journal of Family Violence,—l_j_
103-121.
•Marshall, L. L., & Rose. P. (1987). Gender, stress and
violence in the adult relationships of a sample of college
106
students . Journal of Social and. Personal Relationships,—4 ,
299-316.
Marshall, L. L., & Rose, P. (1988). Family of origin
violence and courtship abuse. Journal of Counseling and
Development. 66. 414-418.
•Marshall, L. L., & Rose, P. (1990). Premarital
violence: The impact of family of origin violence, stress,
and reciprocity. Violence and Victims. 5_i_ 51-64.
Martin, D. (1976) . Battered wives. New York: Pocket
Books.
Mihalic, S. W. & Elliott, D. (1997). A social learning
theory model of marital violence. Journal of Family
Violence. 12. 21-47.
Mitchell, R. E. & Hodson, C. A. (1986). Coping and
social support among battered women: An ecological
perspective. In S. E. Hobfoll, (Ed.), Stress,—social
support and women (pp. 154-169) Washington: Hemisphere.
*Murphy, C. M., Meyer, S., & 0' Leary, D. (1993).
Family of origin violence and MCMI-II psychopathology among
partner assaultive men. Violence and Victims. JL. 165-176.
Ney, P. G. (1992). Transgenerational triangles of
abuse: A model of family violence. In E. C. Viano (Ed.),
Intimate violence: Interdisciplinary perspectives (pp.
15-25). New York: Hemisphere.
107
Nisbett, R. E. & Ross, L. (1980). Human inference:
Strategies and shortcomings of social judgment. Englewood
Cliffs, NJ: Prentice-Hall.
North, C. S., Thompson, S.J., Smith, E. M., & Kyburz,
L. M. (1996). Violence in the lives of homeless mothers in a
substance abuse treatment program: A descriptive study.
Journal of Interpersonal Violence. 11, 234-249.
*0'Keefe, M. (1997). Predictors of dating violence
among high school students. Journal of Interpergonal
Violence. 12. 546-568.
O'Keefe, N. K., Brockopp, K., & Chew, E. (1986). Teen
dating violence. Social Work. 31. 465-468.
O'Leary, K. D. (1988). Physical aggression between
spouses: A social learning theory perspective. In V. B. Van
Hasselt, R. L. Morrison, A. S. Bellack, & M. Hersen (Eds.),
Handbook of family violence (pp.31-55). New York: Plenum.
Owens, D. J., & Straus M. A. (1975). The social
structure of violence in childhood and approval of violence
as an adult. Aggressive Behavior. 1. 193-211.
Pagelow, M. D. (1981). Factors affecting women's
decisions to leave violent relationships. Journal of Family
Issues. 2. 391-414.
Pagelow, M. D. (1992). Adult victims of domestic
violence. Journal of Interpersonal violence. 7. 87-120.
*Parker, B., & Schumacher, D. N. (1977). The battered
wife syndrome and violence in the nuclear family of origin:
108
A controlled pilot study, amprican Journal of Public Health,
67. 760-761.
* Post, R. D., Willett, A. B., Franks, R. D., House, R.
M., & Back, S. M. (1981). Childhood exposure to violence
among victims and perpetrators of spouse battering.
Victimology, 6. 156-166.
*Riggs, D. S., O'Leary, K. D., & Breslin, F. C. (1990).
Multiple correlates of physical aggression in dating
rniiplps. Journal of Interpersonal Violence,—5_i_ 61-73.
*Rosenbaum, A., & O'Leary, K. D. (1981). Marital
violence: characteristics of abusive couples. Journal of
Consulting and Clinical Psychology,—49x- 63-71.
Rosenthal, R. (1984). Msl-.a-ana]yH c procedures for
social research. Beverly Hills, CA: Sage.
Rosenthal, R. & Rosnow, R. L. (1991). Essentials of
behavioral research: Methods and data analysis. New York:
McGraw-Hill.
Rounsaville, B. J. (1978). Theories in marital
violence: Evidence from a study of battered women.
Victimology: An International Journal, 2L_ 11-31.
*Rouse, L. P. (1988) . Conflict tactics used by men in
marital disputes. In G. T. Hotaling, A. Finkelhor, J. T.
Kirkpatrick, & M. A. Straus (Eds.), Family abuse and its
consecruences (PP- 176-191). Beverly Hills: Sage.
109
Russell, M. (1988). Wife assault theory, research, and
treatment: A literature review. Journal of Family Violence,
193-208.
*Ryan, K. M. (1995) . Do courtship-violent men have
characteristics associated with a "Battering Personality"?
Journal of Family Violence. 10. 99-120.
Rynerson, B. C., & Fishel, A.H. (1993). Domestic
violence prevention training: Participant characteristics
and treatment outcomes. Journal of Family Violence,—8j_
253-266.
Sack, A. R., Keller, J. F., & Howard, R. D. (1982).
Conflict tactics in dating situations. International—Journal
of Sociology of the Family. 12. 89-100.
*Schuerger, J. M., & Reigle, N. (1988). Personality and
biographic data that characterize men who abuse their wives.
Journal of Clinical Psychology. 44, 75-81.
Schwarzer, R. (1989). Meta-analvsis programs. Berlin:
User supported software.
*Sedillo, D. M. (1997). Effects of family of origin
violence on severely psychologically abused women.
Unpublished master's thesis, University of North Texas,
Denton, Texas.
Seligman, M. (1975). Helplessness: On depression.
development and death. San Francisco: Freeman.
*Seltzer, J. A. & Kalmuss, D. (1988) Socialization and
stress explanations for spouse abuse. Social Forces.—67,,
473-491.
110
Snyder, D. K. & Fruchtman, L. A. (1981). Differential
Patterns of Wife Abuse: A data-based typology. Journal—of.
Consulting and Clinical Psychology. 49. 878-885.
Stets, J. E. (1990). Verbal and physical aggression in
marriage. Journal of Marriage and the Family,—52, 501-514.
*Stets, J. E., & Pirog-Good, M. A. (1990).
Interpersonal control and courtship aggression. Journal of
Social and Personal Relationships. 371-394.
*Stith, S. M., & Farley, S. C. (1993). A predictive
model of male spousal violence. Journal of Family Violence,
iL_ 183-201.
Straus, M. A. (1976). Sexual inequality, cultural
norms, and wife-beating. Victimology: An International
Journal. 1. 54-70.
Straus, M. A. (1979). Measuring intrafamily conflict
and violence: The Conflict Tactics (CT) Scales. Journal of
Marriage and the family. 41. 75-88.
Straus, M. A. (1990). Social stress and marital
violence in a national sample of American families. In M. A.
Straus & R. J. Gelles (Eds.), Physical violence in American
families: Risk factors and adaptations to violence in 8,145
families (pp. 181-201). New Brunswick, NJ: Transaction
Publishers.
Straus, M. A., & Gelles, R. J.(1990). Physical
violence in American families: Risk factors and adaptations
to violence in 8.145 families. New Brunswick, NJ:
Transaction Publishers.
Ill
Straus, M. A., Gelles, R. J., & Steinmetz, S. K.
(1980). Behind closed doors: Violence in the American
family. Garden City, NY: Doubleday.
Strube, M. J., & Barbour, L. S. (1984). Factors related
to the decision to leave an abusive relationship. Journal—of
Marriage and the Family. 46. & 837-844.
Sugarman, D. B., & Hotaling, G. T. (1989). Violent men
in intimate relationships: An analysis of risk markers.
Journal of Applied Social Psychology. 19. 1034-1048.
Sugarman, D. B., & Hotaling, G. T. (1995, July).
Intimate violence and social desirability: A meta-analytic
review. Paper presented at the meeting of the Fourth
International Family Violence Research Conference, Durham,
NH.
Telch, C. F. & Lindquist, C. U. (1984). Violent versus
nonviolent couples: A comparison of patterns. Psychotherapy,
21. 242-248.
*Theodore, R. M. (1992). The relationship between locus
of control and level of violence in married couples. In E.
C. Viano (Ed.), Intimate violence: Interdisciplinary
perspectives (pp. 37-48) . New York: Hemisphere.
*Tontodonato, P., & Crew, B. (1992). Dating violence,
social learning theory, and gender: A multivariate analysis.
Violence and Victims. 7. 3-14.
112
Tukey, J. W. (1977) . Exploratory data analysis.
Reading, MA: Addison-Wesley.
Trolier, T. K., & Hamilton, D. L. (1986). Variables
influencing judgments of correlational relations. Journal of
Personality and Social Psychology. 50, 879-888.
Tutty, L. M. & Rothery, M. A. (1997, June). Women who
seek shelter form wife assault: Risk assessment—and service
needs. Paper presented at the 5th International Family
Violence Research conference. Durham, NH.
Ulbrich, P., & Huber, J. (1981). Observing parental
violence: Distribution and effects. Journal of Marriage and
the Family. 43. 623-631.
U.S. Department of Justice (1994). Violence against
women: A national crime victimization survey report.
Washington, DC: Bureau of Justice Statistics.
*VanHorn, B., Sedillo, D. M. & Marshall, L. L. (1997,
May). Childhood violence: Effects on women sustaining
partner violence among three ethnic groups. Paper presented
at the ninth annual meeting of the American Psychological
Society, Washington, DC.
Walker, L. E. (1979). The battered woman. New York:
Harper and Row.
Walker, L. E. (1983) . Victimology and the psychological
perspectives of battered women. Victimology: An
International Journal. 8. 82-104.
113
Walker, L. E. (1984) . b a t t e l woman syndrome. New
York: Springer.
Widom, C. S. (1989). Does violence beget violence? A
critical examination of the literature, psychological
Bulletin. 106. 3-28.
Widom, C. S. & Shepard, R. L. (1996). Accuracy of adult
recollections of childhood victimization: Part 1. Childhood
physical abuse. Psvcholocrical Assessment?—£L,_ 412—421.