Domestic violence, DOVE, and divorce mediation

Abstract

The primary objective of this article is to describe DOVE, a 19-item instrument designed to assess and manage the risk of domestic violence between partners during and following their participation in divorce mediation. Assessing risk, more specifically how DOVE can be used to assess risk, is described first. The resulting risk scores (TOTDOVE) are used to assign individuals to risk categories. Problems associated with using categorical, frequency, and probability risk assessment formats in interpreting and communicating risk are discussed in the second segment of the article. A dual, categorical/probability format is advocated. Managing risk using Safety Plan interventions that are linked with risk category and predictor subscores on control, substance abuse, anger, relationship problems, mental health problems, and conflict is covered in the final segment.

Full text

FAMILY COURT REVIEW, Vol. 44 No. 4, October 2006 658– 671
© 2006 Association of Family and Conciliation Courts
Blackwell Publishing, Ltd.Oxford, UKFCREFamily Court Review1531-2445© Association of Family and Conciliation Courts, 2006October 2006444Original Article
Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATIONFAMILY COURT REVIEW
DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION
Desmond Ellis
Noreen Stuckless
The primary objective of this article is to describe DOVE, a 19-item instrument designed to assess and manage
the risk of domestic violence between partners during and following their participation in divorce mediation.
Assessing risk, more specifically how DOVE can be used to assess risk, is described first. The resulting risk
scores (TOTDOVE) are used to assign individuals to risk categories. Problems associated with using categorical,
frequency, and probability risk assessment formats in interpreting and communicating risk are discussed in the
second segment of the article. A dual, categorical/probability format is advocated. Managing risk using Safety
Plan interventions that are linked with risk category and predictor subscores on control, substance abuse, anger,
relationship problems, mental health problems, and conflict is covered in the final segment.
Keywords:
divorce mediation
;
domestic violence
;
violence risk assessment
;
violence risk management
;
violence
;
risk assessment
;
risk management
Some critics of divorce mediation claim that it is “unsafe for women” (American Bar
Association, 2000; Fund for Dispute Resolution, 1993; Fischer, Vidmar, & Ellis, 1993;
Ontario Association of Interval and Transition Houses, 1989; Transition House Association
of Nova Scotia, 2000). An investigation of this claim yields the following two conclusions.
First, it is not supported by research findings. Specifically, we found no evidence indicating
that: (a) abused women are more likely than nonabused women to be the victims of physical
violence and/or emotional abuse during and following their participation in divorce medi-
ation or (b) abused women participating in divorce mediation are more likely to be victims
of male ex-partner violence than abused women participating in adjudication or lawyer-
negotiated separations/divorces (Ellis & Stuckless, 1992; Ellis & Stuckless, 1996; Ellis,
2000; Ellis & Anderson, 2005; Ellis, Stuckless, & Wight, 2006).
Second, critics fall into two groups. Group one includes those who flatly reject divorce
mediation and implicitly or explicitly recommend adjudication or lawyer negotiations as
alternatives (e.g., Fischer, Vidmar, & Ellis, 1993; Ontario Association of Interval and Tran-
sition Houses, 1989; Transition House Association of Nova Scotia, 2000).
Group two includes those whose criticisms include suggestions for more effectively
assessing and managing the risk of domestic violence during and following mediation (e.g.,
American Bar Association, 2000; Fund for Dispute Resolution, 1993). DOVE (Domestic
Violence Evaluation) is a research-based response to their suggestions.
A review of the literature and mediation practice reveals that DOVE is not the only
response to the challenge of domestic violence between separating/divorcing partners
participating in divorce mediation. Violence screening instruments have been created by
Erickson and McKnight (1990), Girdner (1990), and the Woman Abuse Council of Toronto
(2000) implemented by 17 court-connected mediation services in Ontario, and 149 U.S.
court-based mediation programs surveyed by Pearson (1996). Collectively, these responses
Correspondenc: desellis@yorku.ca or stuckles@yorku.ca

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 659
provided a foundation for the creation of DOVE. At the same time, they present three problems
that a more effective instrument would have to solve.
First, findings from a number of studies indicate that there are, minimally, two major
types of male partner violence: control motivated and conflict instigated (Ellis & Stuckless,
1996; Ellis, Stuckless, & Wight, 2006; Johnson, 1979; Johnson & Leone, 2005). Yet, questions
measuring both types of violence are either not included in assessment instruments (e.g.,
Erickson & McKnight, 1993; Woman Abuse Council of Toronto, 2000), or male partner
violence is defined as control motivated and measured using an instrument (Conflict
Assessment Protocol) designed to measure conflict instigated violence (e.g., Girdner, 1990).
Second, Safety Plan interventions were either not integrated or not fully integrated with
assessment data. For example, the assessment instrument used by mediators in Ontario’s
Kawartha Family Court does not include a Safety Plan. Girdner’s triage instrument includes
a Safety Plan but interventions other than exclusion from mediation are not identified for
males in the highest risk category, “those most likely to experience harm” (p. 374).
Third, none of the instruments we reviewed had been actually tested and validated by
research findings. As a result, “males who have threatened suicide in relation to marital
separation/divorce” or who have been “convicted of assaulting anyone” are excluded from
mediation in the absence of any evidence indicating that these factors are statistically
significant predictors of violence when the parties were living together and after they lived
apart (Girdner, 1960, p. 374).
Creators of DOVE deal with these problems in the following ways. First, questions
measuring control motivated and conflict instigated violence are included in the assess-
ment segment (Part 1). Second, a Safety Plan with interventions focusing on safety during
and following participation in mediation are fully integrated with assessment risk scores.
Third, DOVE has been empirically validated by the results of a two-year field study
of couples participating in divorce mediation in two family courts (Ellis, Stuckless, &
Wight, 2006).
The primary objective of this paper is to describe DOVE in sufficient detail for it to be
used by divorce mediators, family lawyers, and family court judges to reduce or prevent
violence between partners who are separating or are separated.
RATIONALE FOR USING DOVE
Using a violence screening instrument such as DOVE to achieve this objective is
warranted by the relatively high level of risk associated with separating (high base rates).
Findings reported by Ellis, Stuckless and Wight (2006) indicate that a relatively high
proportion of separated women experienced male partner violence and abuse after they
separated. Specifically, 71 percent (
n
= 80) of them reported being assaulted, seriously
harmed physically, emotionally abused, and/or seriously harmed emotionally. Post-separation
rates for each type and level of violence and abuse are presented in Ellis, Stuckless and
Wight (2006).
The use of DOVE is also warranted by its demonstrated ability to identify 19 statistically
significant predictors of male partner violence and abuse against female partners after they
separated (Table 1).
A third reason for using DOVE is that it links violence prevention interventions with
(a) level of risk; (b) the presence of specific types of predictors; and (c) types and levels of
violence and abuse.

660 FAMILY COURT REVIEW
DOVE—PART 1
Dove is a two-part, 19 item instrument designed to assess (Part 1) and manage (Part 2)
the risk of domestic violence during and following participation in divorce mediation. The
earliest version of DOVE included 33 items. The results of a pilot project indicated that
the number of items (hypothesized predictors) should be reduced to 19. These items were
validated by data collected from three independent sources: surveys of women in four
shelters for battered women, a review of the literature, and the authors’ own theory and
research on linkages between separation and violence.
The 19 items are statistically significant predictors of assaults, assaults that result in seri-
ous physical injury, sexual assaults, emotional abuse, and emotional abuse that results in
serious emotional harm. More specifically, the significant predictors refer to conduct, events,
and experiences that occurred or were present when the partners were living together. For
example, male partner assaults against female partners when the couples lived together
predicts male partners assaults against female partners after they separated (
n
= 72, r. 386
p
< .001). As another example, male partner control (item 18) predicts female partner
reports of being seriously hurt physically following separation (
n
= 56, r. 297
p
< .026)
Tab le 1
Statistically significant predictors of post-separation violence and abuse against female partners by male ex-partners
Predictors
P
ast violence
1. assaulted
2. seriously hurt physically
3. sexually assaulted
4. left home because of partner’s violence
5. called police because of partner’s violence
P
ast abuse
6. emotionally abused
7. seriously hurt emotionally
E
motional dependency
8. threats to harm/kill self if partner left
9. threats to harm/kill partner if partner left
10. possessive/jealous
R
elationship problems
11. hard to get along with
12. communication deficits
13. blame
14. anger
M
ental health problems
15. taking medication
Control
16. tried to control partner
17. used violence/abuse to control partner
Substance abuse
18. drinking
19. drugs
* For a more detailed description of these predictors see the Appendix

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 661
(Ellis et al., 2006). In addition to different types of assault and levels of physical and
emotional abuse, the items identified in Table 1 refer to seven different subsets of predictors.
They are: past violence, past abuse, emotional dependency, relationship problems, mental
health problems, control, and substance abuse.
RISK ASSESSMENT
Part 1 of DOVE is administered by the mediator to each partner privately at intake. Both
partners are asked to respond as victims, observers, or knowledgeable persons. As a result,
Part 1 completed by female partners provides information about male partners and
vice versa.
Risk assessment items included in Table 1 are measured using answers to the questions
and responses to the statements described in the Appendix. The questions and statements
refer to two time periods—when the partners lived together and after they separated. Answers
and responses relevant to the first time period are used to predict answers and responses to
the same questions and statements relevant to the second time period.
An examination of the Appendix will reveal that DOVE items were measured in terms
of their frequency when they referred to the conjugal period and dichotomously when they
refer to the period following separation. This was done because, in pre-testing DOVE, we
found that partners were reluctant to take the time to answer all questions when they had
to provide frequency answers for all items covering both time periods. A number of them
responded by skipping questions. The number of skipped questions was markedly reduced
when the time taken to complete DOVE was reduced to about 12 minutes by replacing fre-
quency answers with categorical yes/no answers for items relevant to the period following
separation. In those cases where questions were not answered, the “80% rule” was invoked
in calculating the risk scores used to locate individuals in risk categories.
The “80% rule” was used to: (a) eliminate cases where individuals did not respond to 80%
or more of the questions and statements identified in Table 1, and (b) calculate risk scores for
individuals who responded to 80% or more but not all (100%) of the questions and statements.
A noteworthy feature of the procedure used in placing individuals in risk categories is
the use of frequency (TOTFREQ), frequency (SINSEP, since separation) and seriousness of
physical and emotional injuries (ELEVATED & ELSINSEP) in calculating risk scores.
Risk scores vary in the weight attached to them. Pilot project findings suggested that risk
scores should be weighted as follows: more recent (SINSEP) scores are weighted more
heavily than pre-separation (conjugal) scores; predictors of serious physical injury (ELE-
VATED) are weighted more heavily than predictors that are not significantly associated
with this outcome; elevated predictors present since separation (ELSINSEP) are weighted
most heavily (Appendix).
Four risk categories are identified in the Appendix. These are: Low, Moderately High,
High, and Very High. Placement of individuals in one of these four risk categories is dependent
upon their risk scores. Risk category cut-off points were empirically generated so as to pro-
duce roughly equal numbers in each category. Findings from an ongoing study of domestic
violence during the high risk four-month period following the closing of the mediation file
will provide a more valid empirical basis for maintaining or changing cut-off points.
Two methods of calculating TOTDOVE scores and using them to place individuals in
risk categories are described in detail in the Appendix. Method A, the simplest and quickest,
can be used where partners answer all 19 questions and statements. We found that mediators
can ensure that all 19 questions/statements are responded to by administering DOVE

662 FAMILY COURT REVIEW
themselves instead of handing DOVE to partners and asking them to complete it. Partner
responses to DOVE questions/statements can be directly entered into a computer and an
Excel-based software package is available for immediately calculating total DOVE scores
(TOTDOVE) and assigning partners to risk categories.
How do we know that partners’ answers to questions/statements that we enter into the
computer or record in other ways are accurate? Criterion related validation is normally used
to answer this question, but we could not use it because medical, police, and child welfare
data that would corroborate partner responses are treated as confidential. In other cases,
corroborative data sources could not be used because they were incomplete. For example,
family court files could not be used because assault charges laid by the police were not
included in them. One step we did take was to establish reliability by comparing male and
female partner responses to specific questions. When we compared male and female
answers to questions such as “called police because my partner physically assaulted me” we
found consistent answers. In addition, we looked for consistent answers by male and female
partners to different questions. Thus, all female partners who reported calling the police
because they were physically assaulted by their male partners (see Item 7 of Appendix) also
reported being assaulted by their male partners (see Item 1 of Appendix).
INTERPRETING AND COMMUNICATING RISK
Using DOVE to assess risk may yield accurate predictions of relatively frequently
occurring, specific types and levels of male partner violence and abuse over a relatively short,
high-risk period of time, yet the way in which risk is interpreted and communicated to others
“can render risk assessment . . . completely useless or even worse than useless, if it gives
[assessors themselves and others] the wrong impression” (Heilbrun, Dvoskin, Hart, & McNeil,
1999, p. 94). Impressions created by risk assessment formats influence decision-making,
and wrong impressions or interpretations can lead to bad or improper decisions about
Safety Plan interventions that may adversely affect the quality of life of both partners.
The impressions and interpretations of mediators and other assessors are strongly influ-
enced by the amount and kind of information produced by the risk assessment format they
are using. Violence risk assessment formats that produce more information without inflating
or deflating the risk posed by individuals being assessed yield more accurate impressions
and interpretations than those that provide relatively little information and exaggerate the risk
posed by individuals being assessed. Of the three risk assessment formats most frequently
cited in the literature, categorical formats (high, medium, low) yield less information than
frequency (20 out of 100) and probability (20%) formats. Moreover, for organizational,
policy or fund-raising purposes, categorical formats can be used to inflate or deflate risk by
creating cut-off points that increase the number of individuals in different risk categories
(Slovic, Monahan, & MacGregor, 2000, p. 293).
DOVE assesses risk by using a categorical format (Low, Moderately high, High and
Very high) but it provides more information by providing an empirically generated range
of risk scores for each risk category. Moreover, DOVE risk categories are linked with
Safety Plan interventions aimed at managing risk. Still, this format suffers from its inability
to provide quantitative assessments of risk for individuals in each of the four categories.
Frequency and probabilistic formats provide this information. Findings on “serious physical
injury since separation” for partners in the family court sample reported by Ellis, Stuckless
and Wight (2000) can be used to illustrate the point.

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 663
The risk category format provides four risk category names and the range of DOVE
scores linked with them. The following additional information is provided by the frequency
format or scale: 5 out of 42 partners in the low risk category, 5 out of 23 partners in the
moderately high risk category, 14 out of 37 partners in the high risk category, and 13 out
of 24 partners in the very high risk category for seriously injuring an ex-partner. The prob-
abilistic format/scale provides the following additional information: 12% of the partners in
the low, 23% of partners in the moderately high, 38% of partners in the high, and 54% of
the partners in the very high risk category for seriously injuring an ex-partner.
In interpreting the categories/figures presented above, it is important to note that they
refer to a median of three years since separation. Assuming that the serious injuries reported
by separated women are equally distributed across the three years, the percentages must be
divided by three to yield a “past 12 months” rate. Thus, the yearly rates reported by women
in the low and very high risk categories would be 4% and 18% respectively.
Frequency and probability formats provide additional information but the latter are less
likely than the former to exaggerate the risk posed by partners (Slovic et al., 2000).
These researchers found that “at any given level of likelihood, a patient [partner]
was judged as posing a higher risk if that likelihood was derived from a frequency
scale (e.g., 10 out of 100) than if it was derived from a probability scale (e.g., 10%)
(p. 271). Why? Because assessors, and the people to whom they communicate risk
using a frequency scale tend to focus on the numerator, the number of individuals likely to
behave violently and not on the denominator, “out of how many” individuals. Focusing on
the numerator elicits “frightening images” (p. 292) and fear leads to higher perceptions of
risk.
For this reason, we suggest a dual categorical /probability format approach. To this end,
we included risk estimates for assaults, assaults resulting in serious injuries, emotional
abuse and emotional abuse resulting in serious emotional harm against female ex-partners
on the last page of the Appendix. As data from the second, post-mediation phase of this
study are collected and analyzed the percentages reported will become more stable and will
include data on male partner victims because the sample size will be much greater.
RISK MANAGEMENT
Risk is managed by invoking interventions suggested by the Safety Plan. The DOVE
Safety Plan is grounded in two assumptions. The first one is that partners who do not inter-
act with each other in person, or who only interact with each other in the presence of third
parties who can act as guardians or serve as credible witnesses, are unlikely to physically
harm each other. Separating and separated partners who do not communicate with each
other, or who only communicate with each other through third parties who are trusted by
recipients, are unlikely to abuse each other emotionally.
Serious physical injuries
Risk category Dove scores N No Yes % yes
Low 0 to 19 42 37 5 12
Moderately high 19.1 to 27 23 18 5 23
High 27.1 to 37 37 23 14 38
Very high 37.1 to 100 24 11 13 54

664 FAMILY COURT REVIEW
The second assumption is that individuals who are highly motivated to harm their ex-
partners are more likely to do so than individuals whose motivation is low. The relation
between the first (opportunity) and second (motivation) assumption is
additive.
That is, vio-
lence and abuse can occur when the motivation to harm is low or absent, but the situation
in which both partners are present instigates violence and/or abuse, and also when the
opportunity for violence/abuse is low or absent, but the motivation to harm is high.
An example of the first kind of additive relation is provided by the case of a divorced
mother who comes to her ex-husband’s home to pick up their seven-year old child after his
weekend stay. The exchange had always been cordial, but she is verbally attacked when she
asks him for the child-support money he had promised to pay when she arrived. She is
slapped by him when she threatens to report him to the government agency responsible for
enforcing child support orders.
An example of the second additive relation is provided by the case of an ex-wife who
has gone into hiding with her two young children because she is afraid the husband will
harm her. She is found with a serious stab wound administered by the ex-husband who
spent a lot of time and money tracking her down.
The Safety Plan, interventions invoked by mediators, family lawyers, or judges using
DOVE, are dependent upon the individuals risk category placement. With respect to par-
ticipation in divorce mediation, interventions are cumulative. That is, interventions invoked
for individuals in a higher risk category include those invoked for individuals in a lower risk
category plus additional interventions appropriate for the higher risk category. Interven-
tions invoked for individuals placed in the four risk categories are presented below.
LOW RISK
1. Clearly stated written “rules of civility” that encourage respectful communications and
specifically exclude coercive conduct during and between mediation sessions.
2. Parties agree in writing to terminate mediation if the mediator obtains credible evidence of
threatened or actual violence and/or abuse.
3. Face-to-face mediation.
4. Referrals to appropriate treatment interventions.
MODERATE RISK
5. Mediator carefully monitors compliance with violence/abuse prevention rules during private
interviews with partners, and/or by communicating with third parties identified as trusted
contact persons by partners.
6. Partners arrive and leave at different times or routes and do not wait in the same room.
7. Mediators provide both partners with a list of community resources such as shelters, men’s
programs, health services, male and female support groups, and legal information.
8. Face-to-face mediation with advocate or supporter present, or shuttle mediation.
9. Referrals to appropriate treatment interventions.
HIGH RISK
10. Partners given safety warnings in writing.
11. Interpersonal contact only takes place in public places, or with trusted third parties present.
12. Arrange for third party to be present during exchanges of children, or third party transports
children.

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 665
13. Communication only through trusted third parties or through journals exchanged with children
and subject to monitoring by mediator.
14. Partners escorted to and from premises where mediation is being conducted.
15. Shuttle, telephone, or on-line mediation.
16. Referral to appropriate treatment interventions.
VERY HIGH RISK
17. Referral to appropriate treatment interventions.
18. Telephone or on-line mediation if referrals produce credible evidence of positive personal
and/or situational change.
Within each risk category DOVE scores may vary with respect to past violence and
abuse, emotional dependency, mental health, control, substance abuse, relationship problems,
and conflict sub-scores. Appropriate treatment interventions are those that are explicitly
designed to treat individuals with high scores on one or more of these predictors (Adams,
1988; O’Leary, Neidig, & Heymann, 1995; Ptacek, 1998). Although it was not identified as
a statistically significant predictor of post-separation violence, conflict is included because
it predicts emotional abuse and also because children are harmed by high levels of conflict
between separating and separated parents (Ellis & Anderson, 2006; Special Joint Committee
on Custody and Access, 1998).
We conclude this article by repeating two points made in the conclusion of an earlier
article (Ellis et al., 2000). First, DOVE is not a substitute for but a supplement to the
professional judgment and intuition of divorce mediators and any other relevant information
to which they may have access.
Second, safety interventions should be requested from female victims of male partner
violence and their recommended interventions should be taken seriously. They know their
partners very well and are highly motivated to promote their own safety as well as the safety
of their children.
ASSESSING RISK
PROCEDURE
1. [a] Count the total number of risk markers circled/
checked [0, 1, 2, 3, 4 or 5]
excluding
d/k and n/a.
Enter total number of MARKERS line. MARKERS ___________
Sub-total scores
[b] Assign
1
point for each risk marker present [1, 2, 3,
4, or 5 circled/checked].
Add
them to calculate
total
present score
[PRESENT]. 1. PRESENT ___________
2.
Add
the numbers circled/checked for each risk marker.
Thus, if 3, 4 and 5 are circled for questions 16, 17 and
18, add 3, 4 and 5. Add the numbers circled for all 19
questions to obtain a
total frequency score
[TOTFREQ]. 2. TOTFREQ ___________
3. Assign
2
points for each risk marker present since separation.
Add
points to calculate total
since separation score
[SINSEP] 3. SINSEP _____________

666 FAMILY COURT REVIEW
4. Assign
3
points for each ELEVATED RISK MARKER
present. These are identified on the next page.
Add
points to calculate
total elevated risk marker total
score
[ELEVATED] 4. ELEVATED __________
5. Assign
5
points for each ELEVATED risk marker
present SINCE SEPARATION.
Add
points to
calculate total
elevated risk marker since separation
score
[ELSINSEP] 5. ELSINSEP ___________
6. There are two methods [a] and [b] of calculating
TOTSCORE depending upon the MARKERS score.
[a] If MARKERS score is 19,
add
five sub-total scores
to calculate
total risk score
[TOTSCORE] TOTSCORE ___________
[b] If MARKERS score is less than 19 then:
Add
five sub-total scores. Enter total on
Line _____
Divide
total by MARKERS score
Multiply
by 19 to calculate [TOTSCORE] TOTSCORE ___________
Example-Method [b]
MARKERS = 16
5 sub-total scores (PRESENT to ELSINSAP) = 124
Divide 124 by 16 = 7.8
Multiply 7.8 by 19 = 148 = TOTSCORE
7.
Divide
TOTSCORE by 2.25 to calculate
total DOVE
score
[TOTDOVE] TOTDOVE ____________
8. Use TOTDOVE scores to assign clients to RISK
CATEGORIES by placing a check mark on the
appropriate line. The TOTDOVE score ranges for
different risk categories are:
Scores Risk Category
0 to 19 LOW ___
19.1 to 27 MODERATE ___
27.1 to 37 HIGH ___
37.1 to 100 VERY HIGH ___
Elevated Risk Markers
1. Physically assaulted
2. Seriously hurt physically
6. Medications for mental health problems
7. Called police
9. Complained about drinking
12. Outbursts of anger
13. Poor communication/social skills
17. Possessive/jealous
18. Control

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 667
Categorical/probability Risk Assessment*
ASSAULTS
Risk category n % yes
Low 18 0
Moderately high 15 7
High 20 13
Very high 19 16
ASSAULTS RESULTING IN SERIOUS INJURY
Low 18 28
Moderately high 14 36
High 21 62
Very high 16 63
EMOTIONAL ABUSE
Low 19 37
Moderately high 17 71
High 20 85
Very high 17 88
SERIOUSLY HURT EMOTIONALLY
Low 18 28
Moderately high 14 36
High 21 62
Very high 16 63
* For violence and abuse against female ex-partners

668 FAMILY COURT REVIEW

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 669

670 FAMILY COURT REVIEW
REFERENCES
Adams, D. (1988). Treatment models of men who batter: A pro-feminist analysis. In K. Yllo & M. Bograd (Eds.),
Feminist perspectives on wife abuse
(pp. 90–113). Newbury Park, CA: Sage.
American Bar Association. (2000).
Report to the House of Delegates
. Commission on Domestic Violence section
of Dispute Resolution.
Ellis, D. (2000). Safety, equity and human agency: Contributions of divorce mediation.
Violence Against Women
,
6
, 1012–1027.
Ellis, D., & Anderson, D. (2005).
Conflict resolution: An introductory text
. Toronto, Canada: Emond-Montgomery.
Ellis, D., & Stuckless, N. (1992). Pre-separation abuse, marital conflict mediation and post-separation abuse.
Mediation Quarterly
,
9
, 205–225.
Ellis, D., & Stuckless, N. (1996).
Mediating and negotiating marital conflicts
. Thousand Oaks, CA: Sage.
Ellis, D., Stuckless, N. & Wight, L. (in press). Separation, domestic violence and divorce mediation.
Conflict
Resolution Quarterly
.
Erickson, S., & McKnight, M. (1990). Mediating spousal abuse divorces.
Mediation Quarterly
,
7
, 377–388.
Fischer, K., Vidmar, N., & Ellis, R. (1993). The culture of battering and the role of mediation in domestic violence
cases.
SMU Law Review
,
46
, 2117–2174.
Fund for Dispute Resolution. (1993).
Report for the Toronto Forum on Woman Abuse and Divorce Mediation
.
Waterloo, Ontario.
Girdner, L. (1990). Mediation triage: Screening for spouse abuse in divorce mediation.
Mediation Quarterly
,
7
,
365–376.
Heilbrun, K., Dvoskin, J., Hart, S., & McNeil, D. (1999). Violence risk communication: Implications for research,
policy and practice.
Health, Risk and Society
,
1
, 91–106.
Johnson, M. P. (1995). Patriarchal terrorism and common couple violence: Two forms of violence against women.
Journal of Marriage and the Family
,
57
, 283–294.
Johnson, M. P., & Leone, P. M. (2005). The differential effects of intimate terrorism and situational couple violence:
Findings from the National Violence Against Women Survey.
Journal of Family Issues
,
26
, 322–349.
O’leary, K. D., Neidig, P. H., & Heymann, R. E. (1995). Assessment and treatment of partner abuse: A synopsis
for the legal profession.
Albany Law Review
,
58
, 1215–1234.
Ontario Association of Interval and Transition Houses. (1989).
Stop the violence against women
. Background
report. Toronto, Ontario.
Pearson, J. (1996).
Divorce mediation & domestic violence
. Center for Policy Research, Denver, Colorado.
Ptacek, J. (1998). Why do men batter? In K. Yllo & M. Bograd (Eds.),
Feminist perspectives on wife abuse
(pp. 133–157). Newbury Park, CA: Sage.
Slovic, P., Monahan, J., & MacGregor, D. G. (2000). Violence risk assessment and risk communication; The
effects of using actual cases, providing information and employing probability versus frequency formats.
Law
and Human Behavior
,
24
, 271–296.
Special Joint Committee on Custody and Access. (1988).
For the sake of the children
. Parliament of Canada,
Ottawa.
Transition House Association of Nova Scotia. (2000).
Abused women in family mediation: A Nova Scotia snap-
shot
. Halifax, Nova Scotia.
Woman Abuse Council of Toronto. (2000).
A tool for risk assessment in woman abuse situations
. Toronto, Ontario.
Desmond Ellis received his BA in sociology from Leicester University in England. He received his Ph.D.
from Washington University, St. Louis, and taught at UNC, Chapel Hill, NC. He has been a professor at
York University, Toronto for over 20 years. Presently, he is collecting data on postmediation violence and
abuse from a sample of ex-partners participating in a court-annexed divorce mediation service. Recently,
he published a textbook entitled
Conflict Resolution: An Introductory Text
and an article on separation/
divorce entitled “Domestic Violence and Divorce Mediation” in
Conflict Resolution Quarterly
, in July
2006. He is based in the La Marsh Research Centre on Violence and Conflict Resolution, York University,
Tor onto, Canada.
Noreen Stuckless completed her Ph.D. in Psychology at York University, Toronto, in 1996, with her disser-
tation examining trauma faced by families of homicide and drunk driving victims. Her postgraduate
SSHRC Fellowship, conducted at the University of Toronto, looked at difficulties faced by domestic victims

Ellis and Stuckless/DOMESTIC VIOLENCE, DOVE, AND DIVORCE MEDIATION 671
of violence and other separating couples. She held the position of Research Scientist for 7 years in the
Department of Psychiatry at Women’s College Hospital in Toronto and currently is the Undergraduate
Director of Psychology at York University. Her work generally has been focused on issues related to violence
against women, divorce and mediation, and scale development. She has coauthored a book and authored
and coauthored eighteen articles in these areas (e.g., Ellis, D., Stuckless, N. & Wight, L. (in press),
“Separation, Domestic Violence and Divorce Mediation” in
Conflict Resolution Quarterly
).