You are on page 1of 15

Psychology of Violence © 2014 American Psychological Association

2014, Vol. 4, No. 2, 128 –142 2152-0828/14/$12.00 DOI: 10.1037/a0034536

Emotional Abuse and Its Unique Ecological Correlates Among


Military Personnel and Spouses

Heather M. Foran Richard E. Heyman


University of Braunschweig and Amy M. Smith Slep
New York University

United States Air Force Family Advocacy Research Program


This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Lackland Air Force Base, San Antonio, Texas


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Objective: Identify unique correlates of clinically significant emotional abuse (CS-EA) in a


large representative U.S. sample of men and women. Method: Active duty members of the
U.S. Air Force in relationships (N ⫽ 42,744) and civilian spouses (N ⫽ 17,266) from 82
bases worldwide completed an anonymous online survey on CS-EA, individual, family,
community, and workplace risk factors. Results: Relationship dissatisfaction, poor self-
efficacy, financial stress, and alcohol problems were among the strongest correlates of
emotional abuse among the 21 factors examined. In addition, community factors such as
support from neighbors and community cohesion independently related to men’s CS-EA,
whereas workplace factors were uniquely related to victimization among active duty and
civilian women. The strength of bivariate associations with CS-EA for several family,
workplace, and community factors differed by military/civilian status, gender, and marital
status, but overall ecological models replicated across gender. Conclusions: Although
many workplace and community factors were related to CS-EA bivariately, only a select
few were related after accounting for individual and family factors. CS-EA is an under-
studied but important public health problem and the current study helps to identify key
correlates of CS-EA that can help inform prevention and treatment efforts aimed at reducing
partner violence.

Keywords: domestic violence, psychological abuse, intimate partner violence, marriage, socio-
ecological model

Emotional abuse among intimate partners is female victims reported that the emotional
prevalent and a strong predictor of physical abuse had a more negative impact on them than
abuse in longitudinal studies (Murphy & physical abuse (Follingstad, Rutledge, Berg,
O’Leary, 1989; Schumacher & Leonard, 2005). Hause, & Polek, 1990). In more recent studies,
In one of the first studies comparing the relative emotional abuse was associated with poorer
impact of emotional and physical abuse, mental and physical health after controlling for
Follingstad and colleagues found that 72% of physical abuse in both male and female victims

Heather M. Foran, Institute for Psychology, University of for their help with preparation of this article. This research
Braunschweig, Braunschweig, Germany; Richard E. Heyman was supported by Centers for Disease Control R01 Grant
and Amy M. Smith Slep, Department of Cardiology and 1R49CE00091901. The opinions expressed in this article
Comprehensive Care, New York University; United States are solely those of the authors and do not necessarily
Air Force Family Advocacy Research Program, Lackland Air represent the official views of the U.S. Government, the
Force Base, San Antonio, Texas. Department of Defense, or the Department of the Air
We would like to thank the personnel at the local Air Force. The authors have no financial or other conflicts of
Force sites who were responsible for promoting the survey, interest regarding this article.
to Caliber Associates (especially Dr. Chris Spera) for ad- Correspondence concerning this article should be ad-
ministering the Web survey, and to the active duty mem- dressed to Heather M. Foran, Institute for Psychology,
bers and their spouses who took time to complete it. University of Braunschweig, 33 Humboldtstr., Braun-
Additionally, we thank Ava Berman and Lisanne Bulling schweig, Germany. E-mail: hforan@tu-braunschweig.de

128
EMOTIONAL ABUSE CORRELATES 129

(Taft et al., 2006). Emotional abuse also pre- that results in significant and impairing fear,
dicted mental symptoms after controlling for stress, or sadness/depression (e.g., Foran, Slep
physical abuse in a longitudinal study of cou- et al., 2011).
ples followed for 3 years (Lawrence, Yoon,
Langer, & Ro, 2009). Further, emotional abuse, Identify Correlates Within and Across
but not physical abuse, predicted treatment Ecological Levels Among Military Families
dropout in a program designed to reduce both
forms of abuse (Brown, O’Leary, & Feldbau, Individual, family, workplace, and commu-
1997). Emotional abuse is associated with a nity level correlates are examined to provide a
variety of negative health problems including more comprehensive evaluation of key corre-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

depressive symptoms, posttraumatic stress dis- lates for CS-EA using very large samples of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

order, marital problems, substance abuse disor- active duty military personnel (N ⫽ 42,744) and
ders, suicide, and physical health problems (e. civilian spouses (N ⫽ 17,226). An ecological
g., Coker et al., 2002; Stuart & Holtzworth- perspective to understanding key correlates of
Munroe, 2005). Forms of emotional abuse from violence has been widely utilized as a way to
family members (i.e., expressed emotion) have integrate findings by leading violence preven-
also been associated with increased risk for tion agencies (e.g., Centers for Disease Control,
relapse from a number of mental health prob- 2011) and the rationale for this theoretical per-
lems (Hooley, Miklowitz, & Beach, 2006). spective has been reviewed elsewhere (e.g.,
Hence, emotional abuse is an important risk Heise, 1998; Krug, Dahlberg, Mercy, Zwi, &
factor for a variety of mental health problems Loranzo, 2002; Stith, Smith, Penn, Ward, &
that could be targeted with prevention and in- Tritt, 2004). Given the lack of epidemiological
tervention work. Emotional abuse may repre- literature on CS-EA, the ecological approach
sent an etiological precursor, not only to phys- provides an attractive, broad exploratory frame-
ical abuse, but to a wide array of other mental work to integrate the key correlates of CS-EA
health problems associated with it. detected. Thus, although there have been other
studies that examined emotional abuse corre-
Study Goals lates, most previous studies have relied on small
samples and assessed a limited number of cor-
Examine Correlates of Clinically relates restricting implications for understand-
Significant Emotional Abuse With New ing how correlates relate from an ecological
Measurement Criteria framework. This information can be used to
inform interventions operating at different eco-
Given the clinical importance of emotional logical levels (e.g., individual therapy, commu-
abuse, further systematic identification of key nity prevention).
correlates is needed. A number of studies have In particular, what is needed is more un-
identified correlates of emotional abuse in both derstanding of the role of how more distal
clinical and community samples, but implica- ecological factors may contribute to under-
tions of these studies have been hampered by standing victimization. Much research has ex-
limitations in the measurement of emotional amined individual and family level factors,
abuse (Schumacher, Slep, & Heyman, 2001). but less is known about workplace and com-
Mild and severe acts of emotional abuse are munity factors. Examining these types of fac-
clustered together and impact is not assessed. tors in an intact organization such as the U.S.
Mild emotionally aggressive acts occur in Air Force is particularly informative for help-
nearly all couples (75% of couples each year) ing leaders identify broader organization or
and estimates of severe forms occur in 6%–9% community-wide factors that could be tar-
of couples each year (Foran, Slep et al., 2011; geted.
Straus & Sweet, 1992). Severe emotional abuse, The quality of the workplace/organization
rather than mild forms, is a clearer public health environment has been associated with risk for
target. The current study improves on past stud- mental health problems and family conflict
ies of severe emotional abuse by using new (Ford, Heinen, & Langkamer, 2007; Wright,
criteria to operationalize “clinically significant” Foran, Wood, Eckford, & McGurk, 2012).
emotional abuse (CS-EA)— emotional abuse Among military families, deployments and
130 FORAN ET AL.

long work hours may increase stress and fam- Test Gender Differences and
ily conflict (e.g., McCarroll et al., 2000). Generalizability Across Marital Status
Quality of leadership, overall job satisfaction,
and relationships with coworkers may also The current study examined ecological cor-
place strain on families, leading to more re- relates of CS-EA in a large sample of Air
lationship conflict, and in turn, more emo- Force active duty members and civilian
tional victimization by partners. Although spouses. There has been little research on
these workplace factors have been identified CS-EA among military personnel and spouses
as potential risk factors for family strain, and it is unclear whether similar ecological
there has little research on which factors are factors will play a role for active duty mem-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

bers and spouses. Thus, a range of correlates


most important to emotional victimization
This document is copyrighted by the American Psychological Association or one of its allied publishers.

identified in past civilian samples (e.g., alco-


among military personnel and spouses.
hol problems, relationship dissatisfaction,
In addition to specific workplace factors,
community safety) and correlates specific to
the broader support structure of the military the military environment were included (e.g.,
community may provide valuable resources spouse support for deployments, weeks de-
that help military families reduce risk for ployed, years in the military).
emotional victimization. This community Among active duty members, significant
support may come in the form or general unique correlates of CS-EA detected in the cur-
social support, military-specific support from rent study were tested for generalizability across
formal agencies or for youth, or support from marital status given some evidence that rela-
neighbors (Bowen, Mancini, Martin, Ware, & tionship violence may differ for married and
Nelson, 2003). Other factors such as sense of nonmarried couples (Campbell et al., 2003). All
belonging and cohesion within the military results were computed separately for men and
community may lead to improved coping with women due to previous empirical and theoreti-
military stressors and less family conflict and cal literature that suggests gender differences in
victimization. Further, similar to studies with victimization (e.g., Coker et al., 2002; Yllo,
civilians, low levels of community safety and 1993). Further, there is limited literature on
high community stressors may increase vic- male victims of CS-EA and it is unclear whether
timization rates (Little & Kaufman, 2002; the same factors will be important for both men
Pinchevsky & Wright, 2012). and women.
Taken together, a main goal of this study
was to examine which community and work- Method
place factors contribute after accounting for
well-known individual (e.g., alcohol prob- Participants
lems) and family level (e.g., relationship dis-
satisfaction) correlates of victimization. No- Active duty (AD) members (N ⫽ 128,950) of
tably, many individual and family factors the United States Air Force (AF) and all civilian
spouses of AF AD members (N ⫽ 157,455)
such as relationship dissatisfaction, self-
were invited to complete the 2006 Community
efficacy, and alcohol problems, may function
Assessment (CA); this is a biennial, anonymous
as a risk factor, consequence, or both. Given survey conducted at 82 Air Force sites world-
the cross-sectional nature of the study, we are wide. A total of 54,543 AD members and
not able to address the question of direction- 19,722 spouses responded to the invitations,
ality, but rather address the question of which resulting in a response rate of 44.7% for AD
factors appear to relate uniquely to CS-EA members and 12.3% for spouses. The AD sam-
using large samples and testing within and ple and civilian spouse sample represented two
across ecological levels. Which work and separate samples of individuals rather than cou-
community factors relate uniquely in the con- ples. Prior to analyses, one member of all iden-
text of other individual and family factors can tifiable couples (dual AD couples or AD mem-
help identify what factors deserve further at- ber/civilian couples n ⫽ 1595) were removed
tention in terms of prevention research and from the dataset due to their nonindependence).
longitudinal studies. For the present study, only those in current
EMOTIONAL ABUSE CORRELATES 131

relationships were eligible; this resulted in an and impact. First, participants were asked
analytical sample of AD members that included whether they experiencing significant depres-
34,713 men and 8,031 women in current rela- sion/sadness, stress, and/or fear in the past 12
tionships with the majority of participants being months as a result of their partners’ behavior
married (88.4%). The AD sample was weighted (11 items; e.g., “During the past 12 months,
on marital status, base, ethnicity, religion, mil- were you so stressed that it affected you almost
itary rank, and job type to be representative of every day for 2 weeks?” “How much of this
the U.S. Air Force and account for oversam- stress was related to things your partner said or
pling at small bases and differential response did?”). If they endorsed experiencing one of the
rates (see Foran, Slep et al., 2011 for un- impacts due to their partner’s behavior, they
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

weighted demographic characteristics). were then presented with a list of 10 specific


This document is copyrighted by the American Psychological Association or one of its allied publishers.

After data cleaning, the civilian spouse sam- acts and asked how often (in the past year) their
ple included 879 men and 16,347 women. The partners had committed them (e.g., “put me
mean age was 38.37 (SD ⫽ 9.95) for men and down or humiliated me,” “stalked me,” “grilled
33.08 (SD ⫽ 7.67) for women. Employment or interrogated me about where I had been, what
status of spouses was as follows: full-time I had done, etc.”). If at least one act was re-
(45.7% men; 12.1% women), part-time (13.5% ported, participants were presented with a list of
men; 32.4% women), and unemployed (40.7% all endorsed acts and asked whether the acts had
men; 51.3% women). The majority of the sam- caused or contributed to the depression, stress,
ple had at least one child living with them and/or fear they had reported earlier. Responses
(76.3% men; 74.2% women). The spouse sam- to the “write-in” item (i.e., “Did another similar
ple was not weighted due to a lack of informa- behavior not listed here. . .”) were later coded
tion on population marginals for spouses of AD for whether they described an act of emotional
members. aggression. CS-EA was operationalized as (a) at
As reported in detail previously (Slep, Foran, least one reported emotionally aggressive act
Heyman, Snarr, & U.S. Air Force Family Ad- that caused (b) significant stress, depression, or
vocacy Program, 2011), over 75% of partici- fear (for the victim’s own safety or that of
pants completed the full survey and the remain- someone s/he cared about) that interfered with
der had missing survey responses on at least one the functioning. Internal consistency of the 10
question. Fifty iterations of multiple imputation acts of partner emotional abuse among active
were conducted, with every tenth resulting da- duty members (.80 for men and .76 for women)
taset saved, resulting in five imputed datasets and spouses (.76 men and .76 women) was
for analysis. comparable with the Conflict Tactics Scale-2
(CTS2; e.g., O’Leary & Williams, 2006).
Procedure This measure assesses victimization, rather
than perpetration; emotional/psychological im-
The 2006 CA was administered by Caliber pacts on victims are not reliably reported by
Associates. The survey was entirely Web- those perpetrating the emotional violence be-
based; each AD member selected to participate cause they may not be aware of the impacts on
was sent an e-mailed invitation containing the their partner. Understandably, individuals who
Web link to the survey, as well as an access are victims of CS-EA may not want to disclose
code. Weekly e-mails were sent reminding AD the emotional impact to their abusive partner for
members to participate; each base also con- fear that it could be used as ammunition for
ducted community-wide campaigns to encour- future emotional abuse.
age participation. All civilian spouses of AF AD Overall, the results obtained by Heyman et al.
members were also sent three postcard invita- (2012) largely supported the validity of the
tions. measure. Respondents were invited to record
Measures oral descriptions/explanations of the behaviors
and impacts (if any) they had endorsed on the
Clinically significant emotional intimate questionnaire. Independent ratings of these au-
partner violence (CS-EA; Heyman & Slep, dio recordings, made by trained coders, indi-
2009). CS-EA was assessed with a series of cated no “false negatives” on the questionnaire
questions to assess both acts of emotional abuse for CS-EA. Heyman et al., 2012 also demon-
132 FORAN ET AL.

strated that the questionnaire converges well (Conger et al., 1993). The items (e.g., “How
with the CTS2 (Straus, Hamby, Boney-McCoy, much difficulty do you have paying your bills
& Sugarman, 1996). Specifically, yes/no agree- each month?”) were rated on a Likert scale from
ment between the CS-EA measure and CTS2 1 to 5 with higher scores indicating more finan-
for emotionally aggressive acts was 81%. Fur- cial stress. This scale has a stable one factor
ther, in a 41-site field trial of the CS-EA oper- structure (Snarr, Heyman, & Slep, 2006) and
ational definition, there was high agreement be- good internal consistency (␣ ⫽ .90 AD mem-
tween field site reviewers (primarily clinical bers; ␣ ⫽ .84 spouses).
social workers) and master reviewers (93%, Alcohol problems. The 10-item Alcohol
Heyman & Slep, 2009). Use Disorders Identification Test (AUDIT;
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Notably, most participants who reported ele- Rumpf, Hampke, Meyer, & John, 2002) was
This document is copyrighted by the American Psychological Association or one of its allied publishers.

vated depressive symptoms (as defined by Cen- administered. Scores range from 0 to 40 with
ter for Epidemiological Studies Depression higher scores indicating more alcohol problems.
Scale; Radloff, 1977, scores greater than 14) did This measure has been validated against other
not meet criteria for CS-EA (AD members: clinical assessments of alcohol problems
69.8% of men and 68.6% of women; spouses: (Rumpf et al., 2002; ␣ ⫽ .85 AD members; ␣ ⫽
66.1% men, 66.0% women), and many partici- .80 spouses).
pants who experiencing CS-EA did not endorse Family income. Active duty member’s in-
the depression/sadness impact item (AD mem- come was calculated using salary statistics for
bers: 70.0% of men, 50.8% of women; spouses: individuals with particular pay grades and years
77.5% of men, 76.6% of women). In other of military experience provided by the U.S. Air
words, the majority of the sample endorsed Force for 2006. Additional income from second
other impacts of emotional abuse such as stress jobs, housing allowances, and subsistence al-
and fear rather than depression/sadness.
lowances were added to the base salary. Spouse
Self-efficacy. Nine items assessed partici-
income was estimated from information pro-
pants’ ability to cope with stress and to manage
vided on average income based on spouse gen-
work and family demands (e.g., “I can usually
der and whether the spouse worked full- or
handle whatever comes my way”). This mea-
part-time (U.S. Census Bureau, 2006). This es-
sure included the six highest loading items from
the General Self-Efficacy Scale (Scholz, Guti- timated spouse income was added to AD mem-
errez, Sud, & Schwarzer, 2002), and three items ber’s income to create an estimate of overall
developed for an earlier CA (␣ ⫽ .84 AD mem- family income.
bers; ␣ ⫽ .85 spouses). This measure has been Relationship satisfaction. Four items from
used in other studies with military samples and the widely used Quality of Marriage Index
is negatively associated with mental health (QMI; Norton, 1983) were administered. Three
symptoms (Foran, Heyman, Slep, & U.S. AF items asked AD members to rate how much
Family Advocacy Program., 2011). they agree with statements about their relation-
Physical well-being. The physical health ship on a 6-point scale. The fourth item asked
subscale of the Short Form-8 Health Survey AD members to rate their level of happiness in
(Ware, Kosinski, Dewey, & Gandek, 2001) their relationship, all things considered, on a
adapted for use in military populations assessed scale ranging from 1 to 9. Scores across the four
overall current health, pain, energy levels, sleep, items were averaged to yield a range of scores
diet, and exercise patterns (six items, ␣ ⫽ .65 from 1 to 6.5 (␣ ⫽ .92 AD members; ␣ ⫽ .95
AD members; ␣ ⫽ .67 spouses). The Short spouses).
Form Health Survey is a widely used and well- Career support from significant other.
validated measure of health status and has been This was assessed with three face-valid ques-
used for epidemiological research in many tions asking respondents to rate their partners’
countries. levels of understanding and support for the re-
Perceived financial stress. Five items spondents’ AF jobs and careers (e.g., “My part-
were drawn from the Social Change in Canada ner is understanding of the demands of my Air
Survey (Krause & Baker, 1992), a financial Force job” from almost never to most of the
strain scale (Vinokur, Price, & Caplan, 1996), time; ␣ ⫽ .77 AD members; ␣ ⫽ .77 spouses).
and a measure of family economic pressure This measure is moderately correlated with
EMOTIONAL ABUSE CORRELATES 133

other indices of relationship functioning (Slep members). This measure correlates with other
et al., 2011). indices of work quality and negatively corre-
Ability of spouse to cope with deployment. lates with time deployed (Slep et al., 2011).
Two items inquired about spouses’ preparation Support from leadership. Seventeen
to handle family matters in the case of the AD items, developed and evaluated as part of the
member’s deployment, as well as difficulty the CA, assessed the level of support received from
spouses would likely experience in the case of various AF leaders and the preparation received
deployment (see Spera, 2009, for more details before and support received after a deployment.
on this measure among military personnel; ␣ ⫽ For spouses, the same questions were asked in
.63 AD members; ␣ ⫽ .66 spouses). reference to leaders in their husband’s/wife’s
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Family coping. Three items assessed fam- unit but only included the questions before and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ilies’ ability to work together as a team, keep after the deployment (14 items; ␣ ⫽ .94 AD
positive perspectives during rough times, and members; ␣ ⫽ .90 spouses). Low support from
directly confront problems or challenges rated leadership is associated with poorer mental
on a 6-point scale from almost never to almost health among military personnel (Slep et al.,
always (e.g., “When my family faces a chal- 2011).
lenge or difficulty, we confront the problem Support from formal agencies. Six items
directly”). This measure was developed and val- measured satisfaction with the abilities of for-
idated as part of the 2003 CA (Bowen et al., mal community agencies to meet member and
2003); it correlates with other measures of fam- family needs, and with official base programs
ily functioning, has a stable one factor structure and services. This scale, developed and refined
across four validation samples (all items load- based on a previous CA, has excellent internal
ing greater than .77; Snarr et al., 2006), and has consistency, a stable factor structure, and cor-
good internal consistency (␣ ⫽ .89 AD mem- relates with other community measures of sup-
bers; ␣ ⫽ .88 spouses). port (Snarr et al., 2006; ␣ ⫽ .95 AD members;
Weeks deployed. Respondents were asked ␣ ⫽ .96 spouses).
to report how much time they spent deployed or Social support. Social support was as-
on temporary duty assignments in the 12 sessed with a series of five questions, adapted
months prior to survey participation. from the 1989 Army Soldier and Family Survey
Work group cohesion/preparedness. This (Research Triangle Institute, 1990), that in-
scale, completed by AD members only, com- quired about the availability of tangible support
prised six questions, adapted from an Army from various sources (e.g., friends, neighbors,
Family Research Program individual readiness coworkers, relatives) across a number of differ-
measure (U.S. Army Community & Support ent realms (␣ ⫽ .94 AD members; ␣ ⫽ .95
Center, 1989) that elicited ratings of various spouses).
indicators of squadron cohesion (e.g., sticking Community safety. Six items assessed
together, working together as a team, having perceived safety. Four of the items were
high morale). This measure correlates with adapted from the Knight Community Indicators
other work measures (Slep et al., 2011) and had Survey (Princeton Survey Research Associates,
good internal consistency in this sample (␣ ⫽ 1999). This measure has good structural validity
.88 AD members). and relates to other measures of community
Satisfaction with the Air Force. Five health (Foran, Heyman et al., 2011; Snarr et al.,
items assessed satisfaction with the Air Force as 2006; ␣ ⫽ .77 AD members; ␣ ⫽ .78 spouses).
a way of life for AD members and their families Community stress. Sixteen questions, de-
(␣ ⫽ .88 AD members). This measure was veloped and validated in a previous CA (Snarr
developed for the 2003 CA and expanded to et al., 2006), assessed the availability, quality,
improve validity in the 2006 CA (Snarr et al., and affordability of community resources (e.g.,
2006). housing, health care, child care, transportation;
Work relationships. Three items assessed ␣ ⫽ .89 AD members; ␣ ⫽ .91 spouses).
the quality of AD members’ relationships with Community cohesion. Twenty-one items
coworkers, supervisors, and supervisees. This assessed members’ senses of shared mission,
measure, developed for this study, was com- teamwork, unity, and connectedness in the com-
pleted by AD members only (␣ ⫽ .85 AD munity (Bowen et al., 2003). Low community
134 FORAN ET AL.

cohesion is a predictor of neighborhood disor- Significant differences are shown in Table 1 as


der and a risk factor of family violence (e.g., indicated by paired superscripts. Significant
Bowen et al., 2003; Slep et al., 2011). Items are gender differences were found between AD
rated on a 6-point scale from strongly disagree men and AD women for three family level
to strongly agree (e.g., “Civilian spouses/active variables and one organization/work level vari-
duty members find it easy to make connections able. Poorer family coping and having more
with other families”; ␣ ⫽ .95 AD members; children were more strongly associated with
␣ ⫽ .96 spouses). CS-EA for women (family coping: z ⫽ 2.83,
Support from neighbors. Six items as- p ⫽ .005; number of children: z ⫽ ⫺2.61, p ⫽
sessed support from people in the neighbor- .009) and lower family income and poorer work
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

hood, and one question (adapted from the Social relations were more strongly associated with
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Capital Benchmark Survey, 2000) asked how CS-EA for men (family income: z ⫽ ⫺2.65,
often neighbors get together to fix or improve p ⫽ .008; work relations ⫽ z ⫽ ⫺2.53, p ⫽
something. This measure is moderately corre- .12). There were no significant gender differ-
lated with other support measures (Slep et al., ences between civilian men and women, but
2011) and has excellent internal consistency sample sizes for all comparisons between civil-
(␣ ⫽ .93 AD members; ␣ ⫽ .93 spouses). ian men and other groups were unbalanced,
limiting ability to detect differences. Similarly,
among AD men and civilian men, there was
Results
only one significant difference; poor family
The prevalence of CS-EA (as reported previ- coping was more strongly associated with
ously, Foran, Slep et al., 2011) was 6% for men CS-EA among AD men compared to civilian
and 8.5% for women AD women. The preva- men (z ⫽ 1.97, p ⫽ .049).
lence among spouses was 6.5% for men and In contrast, there were several differences in
10.7% for women. strength of associations between AD and civil-
ian women. Poor physical well-being (z ⫽ 2.52,
Analytical Strategy p ⫽ .012) and lower family income (z ⫽ 4.38,
p ⫽ .000) were more strongly associated with
The active duty sample was randomly split more CS-EA among civilian than AD women.
into a development sample and validation sam- Not surprisingly, women’s report of received
ple to permit cross-validation analyses. This deployment support from their husbands (AD
resulted in a development sample of 17,247 women) was more related to their own lower
men and 4,016 women and a validation sample emotional victimization than women’s report of
of 17,466 men and 4,015 women. The spouse provided deployment support to their husbands
sample was analyzed separately to elucidate (civilian women; z ⫽ ⫺2.36, p ⫽ .018). Fur-
potential differences between spouses and AD ther, among AD women, having more children
member populations; it consisted of 879 men was more strongly related to higher CS-EA than
and 16,347 women. for civilian women (z ⫽ 4.49, p ⫽ .013).

Bivariate Level Analyses Ecological Models for Active Duty


Members
Analyses of bivariate relationships between
CS-EA and each of the study variables were Backward stepwise logistic regression analy-
conducted using logistic regression with the de- ses were conducted for each ecological level for
velopment sample of AD members and the men and women on the development sub-
civilian spouse sample separately. Logistic re- sample. The specific variables entered for each
gression coefficients, odds ratios, and confi- ecological level are listed in Table 1. The final
dence intervals are presented in Table 1. models of the variables that made a unique
To explore potential gender differences and contribution in accounting for variance in
differences between same gender AD and civil- CS-EA within the work and community level in
ian samples further, coefficients were tested for the development subsamples are presented in
differences in strength of associations using the Table 2. The unique variables at the work and
formula provided by Altman and Bland (2003). community level were then added to a com-
EMOTIONAL ABUSE CORRELATES 135

Table 1
Bivariate Logistic Regression Analyses of CS-EA Victimization
Military personnel Civilian spouses
Men Women Men Women
OR CI OR CI OR CI OR CI
Individual level
Alcohol problems 1.34 [1.26, 1.33] 1.34 [1.17, 1.53] 1.51 [1.12, 2.04] 1.35 [1.27, 1.43]
Financial stress 1.58 [1.49, 1.67] 1.37 [1.21, 1.56] 1.54 [1.25, 1.91] 1.57 [1.50, 1.63]
Self-efficacy 0.61 [0.57, 0.65] 0.67 [0.61, 0.75] 0.59 [0.46, 0.76] 0.60 [0.57, 0.63]
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Physical well-being 0.65 [0.60, 0.69] 0.70a [0.63, 0.79] 0.66 [0.53, 0.82] 0.60a [0.57, 0.63]
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Years in the military 0.84 [0.77, 0.92] 0.87 [0.73, 1.02] 0.83 [0.59, 1.18] 0.78 [0.73, 0.83]
Family level
Support from sign. other 0.64 [0.60, 0.68] 0.62 [0.55, 0.69] 0.62 [0.49, 0.79] 0.66 [0.63, 0.69]
Relationship satisfaction 0.39 [0.36, 0.42] 0.41 [0.35, 0.47] 0.38 [0.29, 0.49] 0.39 [0.37, 0.42]
Number of children 1.01a [0.93, 1.09] 1.23ab [1.08, 1.39] 1.19 [0.90, 1.57] 1.03b [0.98, 1.09]
Family income 0.86a [0.78, 0.95] 1.04ab [0.94, 1.16] 0.89 [0.58, 1.36] 0.79b [0.75, 0.84]
Marital length 0.82 [0.74, 0.91] 0.81 [0.65, 1.01] 0.89 [0.66, 1.18] 0.77 [0.72, 0.83]
Family coping 0.58ab [0.54, 0.62] 0.47a [0.42, 0.54] 0.42b [0.30, 0.58] 0.45 [0.42, 0.48]
Spouse deploy. support 0.64 [0.59, 0.69] 0.65a [0.58, 0.73] 0.73 [0.58, 0.92] 0.76a [0.73, 0.80]
Work level
Support from leadership 0.74 [0.70, 0.78] 0.77 [0.69, 0.86] 0.79 [0.64, 0.96] 0.75 [0.72, 0.79]
Satisfaction with AF 0.71 [0.66, 0.77] 0.76 [0.68, 0.85] — — — —
Workgroup cohesion 0.76 [0.71, 0.81] 0.82 [0.72, 0.94] — — — —
Work relations 0.74a [0.69, 0.80] 0.89a [0.79, 1.01] — — — —
Weeks deployed 0.97 [0.90, 1.04] 0.98 [0.84, 1.14] 0.88 [0.65, 1.17] 0.97 [0.92, 1.02]
Hours worked 1.11 [1.05, 1.18] 1.21 [1.08, 1.36] 0.91 [0.55, 1.07] 1.13 [1.07, 1.18]
Community level
Community cohesion 0.72 [0.68, 0.76] 0.82 [0.72, 0.93] 0.76 [0.61, 0.94] 0.78 [0.74, 0.81]
Support from neighbors 0.76 [0.71, 0.80] 0.85 [0.76, 0.94] 0.82 [0.63, 1.06] 0.80 [0.76, 0.83]
Support formal agencies 0.76 [0.71, 0.82] 0.82 [0.73, 0.92] 0.75 [0.57, 0.97] 0.81 [0.77, 0.85]
Social support 0.78 [0.73, 0.83] 0.87 [0.78, 0.98] 0.83 [0.64, 1.08] 0.79 [0.76, 0.83]
Community safety 0.81 [0.76, 0.86] 0.88 [0.79, 0.98] 0.84 [0.65, 1.10] 0.79 [0.75, 0.83]
Community stressors 0.76 [0.71, 0.81] 0.86 [0.76, 0.98] 1.31 [1.06, 1.62] 1.31 [1.25, 1.37]
Note. ORs ⫽ Odds Ratios. CI ⫽ 95th percentile Confidence Interval. Military personnel sample in development sample:
n ⫽ 17,247 men and n ⫽ 4,016 women for all variables except those that were only answerable by married individuals
(marital length, family coping, spouse deployment support: n ⫽ 14,909 men and n ⫽ 2,937 women). Spouse sample: n ⫽
879 men; n ⫽ 16,347 women.
a,b
⫽ paired subscripts indicate parameter estimates are significantly different between men and women or between same
gender military personnel and spouses, p ⬍ .05.

bined model with the unique individual and work variable—number of hours worked—
family level variables. The final model across accounted for additional variance over and
ecological level for the full sample and those above individual and family factors among
who were married is also shown in Table 2. married women such that more work hours
Models were tested separately across genders, was associated with more CS-EA. For men,
but all variables that made a contribution to the two community factors also added uniquely to
model for men or women are included so that the overall model even after accounting for
the model can be compared across gender. individual and family factors (more support
Several work and community level vari- from neighbors and community cohesion, less
ables were uniquely related to CS-EA for men CS-EA). As expected, the strongest correlates
and women. When individual and family level tended to be individual and family level vari-
variables were also included in the model (see ables (e.g., lower relationship satisfaction,
Across Ecological Levels in Table 2), one more alcohol problems, more CS-EA).
136 FORAN ET AL.

Table 2
Logistic Regression Analyses of Emotional Victimization and Tests for Cross-Validation and
Generalizability Across Marital Status and Gender for Active Duty Members
Men Women
Development sample CV MS Development sample CV MS Gender
Wald’s Wald’s
b SE statistic Rubin’s Fa b SE statistic Rubin’s Fa
Work level
Satisfaction with AF ⫺0.23 0.04 27.11ⴱⴱⴱ 0.42 4.88ⴱⴱⴱ ⫺0.21 0.06 11.28ⴱⴱ 0.42 2.56 2.37
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Work relations ⫺0.16 0.04 14.53ⴱⴱⴱ ⫺0.03 0.07 0.18


9.79ⴱⴱ 10.78ⴱⴱ
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Hours worked 0.10 0.03 0.19 0.06


Support from leadership ⫺0.15 0.04 15.95ⴱⴱⴱ ⫺0.19 0.06 9.22ⴱⴱ
Community level
Community cohesion ⫺0.20 0.04 20.56ⴱⴱⴱ 0.81 2.08 ⫺0.15 0.05 9.40ⴱⴱ 0.19 0.57 2.51
Support from neighbors ⫺0.10 0.04 7.23ⴱⴱ ⫺0.12 0.03 12.78ⴱⴱⴱ
Social support ⫺0.09 0.04 5.37ⴱ ⫺0.10 0.06 3.00
Community safety ⫺0.09 0.03 7.41ⴱⴱ ⫺0.08 0.04 4.43ⴱ
Support formal agencies 0.06 0.05 1.37 0.03 0.05 0.38
Across ecological levels
Relationship satisfaction ⫺0.81 0.04 510.10ⴱⴱⴱ 0.12 1.66 ⫺0.78 0.08 103.17ⴱⴱⴱ 0.47 0.67 1.04
Financial stress 0.23 0.04 43.78ⴱⴱⴱ 0.07 0.08 0.79
Self-efficacy ⫺0.18 0.04 20.02ⴱⴱⴱ ⫺0.14 0.07 4.10ⴱ
Alcohol problems 0.11 0.03 11.49ⴱⴱ 0.21 0.08 8.14ⴱⴱ
Physical well-being ⫺0.11 0.05 6.22ⴱ ⫺0.13 ⫺.07 3.77
Support from neighbors ⫺0.08 0.04 4.94ⴱ ⫺0.05 0.06 0.63
Support from sign. other ⫺0.07 0.03 3.87ⴱ ⫺0.12 0.06 3.55
Number of children 0.01 0.04 0.02 0.22 0.07 8.88ⴱⴱ
Across ecological levels
married sample
Relationship satisfaction ⫺0.87 0.04 485.65ⴱⴱⴱ 0.37 NA ⫺0.82 0.91 81.44ⴱⴱⴱ 1.29 NA 2.45
Financial stress 0.23 0.04 31.99ⴱⴱⴱ ⫺0.03 0.09 0.00
Self-efficacy ⫺0.18 0.04 18.28ⴱⴱⴱ ⫺0.20 0.08 7.05ⴱⴱ
Spouse deploy. support ⫺0.18 0.04 17.18ⴱⴱⴱ ⫺0.23 0.07 9.88ⴱⴱ
Community cohesion ⫺0.13 0.04 10.35ⴱⴱ 0.02 0.11 0.04
Alcohol problems 0.11 0.04 9.34ⴱⴱ 0.23 0.09 6.99ⴱⴱ
Hours worked ⫺0.06 0.05 1.60 0.21 0.10 4.54ⴱ
Note. Development Sample: n ⫽ 17,247 men, n ⫽ 4,016 women. Validation Sample: n ⫽ 17,466 men, n ⫽ 4,015. Across
Ecological Level ⫽ all significant unique predictors for men and women in relationships. Across Ecological Levels
Married ⫽ all significant unique predictors for married men and women (n ⫽ 29,992 men, n ⫽ 5,861 women). Support from
Sign. Other ⫽ Career Support from Significant Other. Spouse Deploy. Support ⫽ Spouse Support for Deployment. CV ⫽
cross-validation of model results across development and validation samples. MS ⫽ comparisons of model results across
married versus nonmarried individuals. Gender ⫽ comparison of model results across gender.
a
Significant Rubin’s F indicates strength of risk factors are different across groups.

p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.

Cross-Validation and Generalizability Mplus 5.1 statistical software (Muthén and


Across Marital Status and Gender—Active Muthén, 2007) was used for the analyses. The
Duty Members final logistic regression models were tested con-
straining the regression coefficients to be equal
After the logistic regressions were performed across validation or generalizability groups (e.g.,
on the development subsamples, the final models married vs. unmarried, development vs. validation
were tested to see whether the results cross- samples, men vs. women). The models were run a
validated in the independent sample and to see second time with the regression coefficients free
whether results were generalizable across marital to vary across validation or generalizability
status and gender among active duty members. groups. The chi-square difference test was used to
EMOTIONAL ABUSE CORRELATES 137

calculate whether the chi-square from the con- tors indicated that satisfaction with the Air
strained to be equal analyses differed significantly Force and support from leadership were not
from the chi-square from the unconstrained anal- significant predictors for unmarried men’s emo-
yses. A significant scaled Satorra-Bentler chi- tional abuse but were for married men such that
square difference test indicates there is a signifi- married men who reported higher satisfaction
cant difference in model fit across validation with the Air Force and support from leadership
groups, gender, or marital status. The chi-square reported lower emotional abuse (Rubin’s F ⫽
values across the five imputed datasets were com- 9.07, p ⬍ .001). The difference between men’s
bined using the formula provided by Rubin and hours worked and work relations association
Schenker (1991) that takes into account differ- with victimization was not significantly differ-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ences in variance across imputed datasets. ent for unmarried or married men (Rubin’s F ⫽
This document is copyrighted by the American Psychological Association or one of its allied publishers.

As can be seen in Table 2, the chi-square 1.27, p ⫽ .31).


difference test was not significant (p ⬎ .05) in
the tests for cross-validation, indicating that Ecological Models for Civilian Spouses
constraining the coefficients across the develop-
ment and validation samples did not signifi- Lastly, to identify which ecological variables
cantly worsen model fit. Hence, all the models related to CS-EA among the separate samples of
cross-validated in the hold-out samples for men civilian spouses, backward stepwise regression
and women. The regression coefficients were analyses were run on the civilian spouse sam-
also not significantly different across gender, ples of men and women and are presented in
indicating that the work, community, and across Table 3. As apparent from the bivariate analyses
the ecological level models all generalized presented in Table 1, patterns of association for
across gender (p ⬎ .05). AD members and civilian spouses were not
The work level model for men did not, how- identical and, therefore, testing the models in-
ever, generalize across married and unmarried dependently was preferred. Notably, there were
men. Pairwise comparison of work level predic- fewer unique correlates of CS-EA for civilian

Table 3
Logistic Regression Analyses of Civilian Spouses’ Emotional Victimization
Men Women
b SE Wald’s statistic b SE Wald’s statistic
Work level
Support from leadership ⫺0.23 0.10 5.33ⴱ ⫺0.28 0.03 123.10ⴱⴱⴱ
Hours worked 0.10 0.03 15.08ⴱⴱⴱ
Community level
Community cohesion ⫺0.12 0.03 16.06ⴱⴱⴱ
Support from neighbors ⫺0.07 0.03 6.67ⴱ
Social support ⫺0.10 0.03 12.60ⴱⴱⴱ
Community safety ⫺0.11 0.03 13.62ⴱⴱⴱ
Community stress 0.26 0.11 5.98ⴱ 0.10 0.03 11.58ⴱⴱ
Across ecological levels
Relationship satisfaction ⫺0.88 0.14 40.96ⴱⴱⴱ ⫺0.77 0.04 408.40ⴱⴱⴱ
Alcohol problems 0.28 0.14 3.92ⴱ 0.22 0.03 44.34ⴱⴱⴱ
Financial stress 0.23 0.11 4.43ⴱ 0.18 0.03 38.54ⴱⴱⴱ
Marital length ⫺0.18 0.04 15.40ⴱⴱⴱ
Family coping ⫺0.19 0.05 16.68ⴱⴱⴱ
Spouse deploy. support ⫺0.09 0.03 9.59ⴱⴱ
Family income ⫺0.09 0.04 6.09ⴱ
Physical well-being ⫺0.22 0.03 48.71ⴱⴱⴱ
Support from leadership ⫺0.07 0.03 4.68ⴱ
Note. n ⫽ 879 men, n ⫽ 16,347 women. Spouse Deploy. Support ⫽ Spouse Support for Deployment. Across Ecological
Levels ⫽ all significant unique predictors for civilian spouses.

p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
138 FORAN ET AL.

men than for women, but the power to detect geted in prevention planning, may be similar
differences was also much stronger for the across genders and for different relationship
women sample due to the larger sample size. types (dating or married). However, in terms of
Gender differences for the models in Table 3 the strength of the bivariate relationships, some
were not tested due to the large differences in factors were stronger for specific subgroups.
sample size for men and women and dissimi- Physical well-being was more related to CS-EA
larity in the models found. Similar to the results of civilian women than AD women. This may
for AD members, individual and family level relate to differences between the two popula-
variables were the strongest correlates of CS- tions in that a minimum level of physical health
EA. Among women civilians, the work level is required for active duty service. It may be that
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

variable—support from leadership—was also the emphasis on being physically fit related to


This document is copyrighted by the American Psychological Association or one of its allied publishers.

uniquely associated with reduced CS-EA. their occupation protects them from victimiza-
tion common to civilian women related to phys-
Discussion ical health and weight (e.g., being called fat by
their partner). Alternatively, it may also be that
The current study examined correlates of female spouses of AD men are more susceptible
men’s and women’s clinically significant emo- to emotional abuse related to health issues (such
tional abuse using a large epidemiological mil- as weight gain) due to their partners being more
itary sample. Most previous work on emotional focused on physical health as a part of their
abuse has not examined clinically significant occupation.
levels and little is known about which factors Another interesting finding was that high
are most related to men’s and women’s CS-EA family income was related to less CS-EA
victimization within and across ecological lev- among AD men and civilian women than for
els. Findings of this study contribute to the AD women whereas having children was asso-
literature on severe emotional abuse and inform ciated with more CS-EA among AD women
prevention and treatment operating at different than either AD men or civilian women. Previ-
ecological levels. ous research has suggested that AD women, in
At the bivariate level, nearly all individual, particular, may be vulnerable to marital conflict
family, workplace, and community factors were and dissolution compared with other groups
significant related to CS-EA. However, only a (Karney & Crown, 2007). It may be that they
subset of these factors accounted for unique face risk for victimization due to existing gen-
variance in CS-EA. Consistent with expecta- der biases against women not conforming to the
tions, most factors that accounted for unique traditional gender roles (e.g., working rather
variance were individual and family level fac- than staying at home with the children). Al-
tors such as relationship dissatisfaction, per- though family income tended to ease the risk of
ceived financial stress, alcohol problems, and CS-EA for other groups, why the positive im-
self-efficacy. In addition, a set of work and pact of family income was less for AD women
community factors were uniquely related to needs to be explored further. It may be related
CS-EA even after accounting for the individual to a suppressing effect of gender biases against
and family factors, highlighting potential targets women earners or other reasons not assessed.
for prevention at the organization and military
community levels. More support from neigh- Limitations
bors and community cohesion were uniquely
related to reduced risk of CS-EA for AD men Exploratory approaches (stepwise regression
and less hours worked was uniquely related to analyses) have been criticized methodologically
reduced risk of CS-EA for AD women. Among for lack of replicability, limiting confidence in
civilian women, more support from leadership the findings derived from such methods without
related to lower CS-EA across ecological levels. cross-validation (e.g., Greenland, 2008). Al-
Interestingly, all multivariate models for ac- though the use of exploratory techniques is a
tive duty members were not significantly differ- limitation of the current study, cross-validation
ent across gender and only the work factors of the results, analyses across four separate
differed across marital status. This suggests that samples of AD men, AD women, civilian men,
factors that relate to CS-EA and could be tar- and civilian women, and examination of gener-
EMOTIONAL ABUSE CORRELATES 139

alizability across marital status at least partially Research Implications


addresses concerns about replicability of the
findings. Clearly, results will need to be con- Theoretical models of intimate partner vio-
firmed with other community samples. lence (IPV; e.g., Yllo, 1993) have primarily
Further, there may be differences not ac- focused on women’s victimization given wom-
counted for that limit generalizability outside a en’s heightened risk for severe physical injury
military context. In particular, the sample in the (Breiding, Black, & Ryan, 2008), but to more
current study was younger than the U.S. popu- fully understand IPV for both men and women,
lation and findings may differ for older couples. theories should incorporate more consideration
This sample also differed from the U.S. popu- of gender similarities and differences to provide
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

lation in that all households had at least one a framework for future empirical studies (e.g.,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

member with a full-time job. Thus, the study Hamel, 2009).


findings are representative of couples with one Given that most research has focused on
or both members of the dyad working, but not physical, rather than emotional, IPV there is
necessarily families in which both members are only a limited research literature to compare
unemployed. Additionally, the sample only as- with the current findings. In a previous study of
sessed current CS-EA of those in relationships clinically significant physical abuse perpetra-
and does not assess those who may have expe- tion with the same AD sample used in the
rienced CS-EA in the past, but are no longer in current study (Slep et al., 2011), only individual
that relationship. The sample also only included and family factors contributed across ecological
heterosexual couples and the generalizability to levels to explaining perpetration, whereas work-
place and community factors were found to
same-sex military couples in unknown.
explain unique variance in CS-EA in the current
Causality cannot be inferred from the cur-
study. This highlights the importance of assess-
rent cross-sectional study and some of the
ing different types of IPV separately (perpetra-
factors examined may precede the abuse,
tion vs. victimization, emotional vs. physical) to
whereas other factors may be consequences of
best inform prevention efforts rather than using
the abuse or have bidirectional effects (e.g.,
a composite measure that collapses IPV into one
Kim, Laurent, Capaldi, & Feingold, 2008). category. The need for research on emotional
Longitudinal research would be needed to abuse is apparent, as there are a growing num-
address directionality and, thus, these results ber of studies that have found emotional abuse
can only highlight which factors relate to is related to negative mental and physical health
CS-EA for further study. For example, alco- outcomes over and above physical IPV (e.g.,
hol problems may lead to relationship conflict Coker et al., 2002; Lawrence, Yoon, Langer, &
about drinking behavior and victimization, Ro, 2009; Taft et al., 2006). Clearly, treatment
but may also increase following the victim- and prevention programming aimed at victims
ization as a way to cope with the victimiza- and perpetrators needs to be tailored to the
tion. Individuals with low self-efficacy may unique factors that relate to each. Thus, the
be more likely to stay in a relationship in current exploratory study fills a gap in the lit-
which their victimized and the victimization, erature by identifying, cross-validating, and
may in turn, further decrease their self- testing the generalizability of key unique corre-
efficacy. lates of CS-EA within and across ecological
Finally, there are notable differences between levels.
military branches that should be considered in
interpreting the generalizability of the findings. Clinical and Policy Implications
Although approximately half of the active duty
members in this sample had deployed in the The current study has potential to inform
past, the deployments in the Air Force are usu- clinical practice and prevention efforts by pro-
ally shorter than in other branches and less viding information on which factors should be
frequently see direct ground combat. Air Force assessed among those with CS-EA. Large epi-
military personnel also have lower mental demiological studies of CS-EA, such as this
health symptoms than found in other branches one, are a particularly important first step in
(e.g., alcohol problems, Bray et al., 2003). informing community level public health inter-
140 FORAN ET AL.

ventions. Although there are many studies that Violence, 12, 365–387. doi:10.1023/A:
examine forms of emotional abuse and correlates, 1021906825073
fewer studies have taken an epidemiological per- Campbell, J. C., Garza, M. A., Carlson Gielen, A.,
spective relevant to large-scale intervention plan- O’Campo, P., Kub, J., Dienemann, A., . . . Jafar, E.
(2003). Intimate partner violence and abuse among
ning (e.g., using large samples, cross-validating,
active duty military women. Violence Against
and assessing generalizability). Women, 9, 1072–1092. doi:10.1177/
Results tended to generalize across marital 1077801203255291
status and gender, but there were also differ- Centers for Disease Control. (2011). Intimate partner
ences that may need to be considered in devel- violence: Risk and protective factors. Retrieved
oping interventions and prevention planning. from http://www.cdc.gov/ViolencePrevention/
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

For example, a set of workplace factors were intimatepartnerviolence/riskprotectivefactors.html


This document is copyrighted by the American Psychological Association or one of its allied publishers.

more strongly related to CS-EA particularly for Coker, A. L., Davis, K. E., Arias, I., Desai, S.,
married AD men compared with unmarried Sanderson, M., Brandt, H., & Smith, P. H. (2002).
men. Further, individual level variables tended Physical and mental health effects of intimate part-
to relate to CS-EA similarly across genders, but ner violence for men and women. American Jour-
nal of Preventive Medicine, 23, 260 –268. doi:
several gender differences in the role of fam- 10.1016/S0749-3797(02)00514-7
ily and workplace level variables were de- Conger, R. D., Conger, K. J., Elder, G. H., Lorenz,
tected. Taken together, these results suggest F. O., Simons, R. L., & Whitbeck, L. B. (1993).
the interplay between occupational role (AD, Family economic stress and adjustment of early
civilian), family status, and gender needs to adolescent girls. Developmental Psychology, 29,
be considered in addressing CS-EA among 206 –219. doi:10.1037/0012-1649.29.2.206
military families. Although community-wide Follingstad, D. R., Rutledge, L. L., Berg, B. J.,
programs designed to reduce CS-EA could Hause, E. S., & Polek, D. S. (1990). The role of
target some risk factors that appear universal emotional abuse in physically abusive relation-
(e.g., relationship dissatisfaction via marital ships. Journal of Family Violence, 5, 107–120.
doi:10.1007/BF00978514
therapy and prevention programs), they may
Foran, H. M., Heyman, R. E., Slep, A. M. S., & U.S.
also need to be tailored to address other risk Air Force Family Advocacy Program. (2011). A
factors for specific groups (e.g., married men, model of hazardous drinking and military commu-
AD women, civilian spouses). nity functioning—Identifying mediating risk fac-
tors. Journal of Consulting and Clinical Psychol-
References ogy, 79, 521–532. doi:10.1037/a0024110
Foran, H. M., Slep, A. M. S., Heyman, R. E., & U.S.
Altman, D. G., & Bland, J. M. (2003). Interaction Air Force Family Advocacy Program. (2011).
revisited: The difference between two estimates. Prevalences of intimate partner violence in a rep-
British Journal of Medicine, 326, 219. doi: resentative U.S. Air Force sample. Journal of Con-
10.1136/bmj.326.7382.219 sulting and Clinical Psychology, 79, 391–397. doi:
Bowen, G. L., Mancini, J. A., Martin, J. A., Ware, 10.1037/a0022962
W. B., & Nelson, J. P. (2003). Promoting the Ford, M. T., Heinen, B. A., & Lankamer, K. L.
adaptation of military families: An empirical test (2007). Work and family satisfaction and conflict:
of a community practice model. Family Relations, A meta-analysis of cross-domain relations. Jour-
52, 33– 44. doi:10.1111/j.1741-3729.2003.00033.x nal of Applied Psychology, 92, 57– 80. doi:
Bray, R. M., Hourani, L. L., Rae, K. L., Dever, J. A., 10.1037/0021-9010.92.1.57
Brown, J. M., Vincus, A. A., . . . Vandermaas- Greenland, S. (2008). Introduction to regression
Peller, R. (2003). 2002 Department of Defense modeling. In K. L. Rothman, S. Greenland, & L.
survey of health related behaviors among military Lash (Eds.), Modern epidemiology (3rd ed., pp.
personnel. Research Triangle Park, NC: Research 418 – 455). Philadelphia, PA: Lippincott Williams
Triangle Institute. & Wilkins.
Breiding, M. J., Black, M., & Ryan, G. W. (2008). Hamel, J. (2009). Toward a gender-inclusive concep-
Prevalence and risk factors of intimate partner tion of intimate partner violence research and the-
violence in eighteen U.S. states/territories, 2005. ory: Pt. 2, New directions. International Journal of
American Journal of Preventive Medicine, 34, Men’s Health, 8, 41–59. doi:10.3149/jmh.0801.41
112–118. doi:10.1016/j.amepre.2007.10.001 Heise, L. L. (1998). Violence against women: An
Brown, P. D., O’Leary, K. D., & Feldbau, S. R. integrated, ecological framework. Violence
(1997). Dropout in a treatment program for self- Against Women, 4, 262–290. doi:10.1177/
referring wife abusing men. Journal of Family 1077801298004003002
EMOTIONAL ABUSE CORRELATES 141

Heyman, R. E., & Slep, A. M. S. (2009). Reliability Norton, R. (1983). Measuring marital quality: A crit-
of family maltreatment diagnostic criteria: 41 site ical look at the dependent variable. Journal of
dissemination field trial. Journal of Family Psy- Marriage and the Family, 45, 141–151. doi:
chology, 23, 905–910. doi:10.1037/a0017011 10.2307/351302
Heyman, R. E., Snarr, J. D., Foran, H. M., Makin- O’Leary, K. D., & Williams, M. C. (2006). Agree-
Byrd, K., Slep, A. M. S., & U.S. Air Force Family ment about acts of aggression in marriage. Journal
Advocacy Program. (2012). High sensitivity and of Family Psychology, 20, 656 – 662. doi:10.1037/
specificity screening for clinically significant inti- 0893-3200.20.4.656
mate partner violence. Manuscript submitted for Pinchevsky, G. M., & Wright, E. M. (2012). The
publication. impact of neighborhoods on intimate partner vio-
Hooley, J. M., Miklowitz, D. J., & Beach, S. R. H. lence and victimization. Trauma, Violence, and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(2006). Expressed emotion and DSM-V. In S. R. Abuse, 13, 112–132.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

H. Beach, M. Z. Wamboldt, N. J. Kaslow, R. E. Princeton Survey Research Associates. (1999). The


Heyman, M. B. First, L. G. Underwood, & D. Community Indicators Survey–National: Topline
Reiss (Eds.), Relational processes and DSM-V: results. Retrieved April 10, 2008 from http://www
Neuroscience, assessment, prevention, and treat- .knightfoundation.org/research_publications/
ment (pp. 175–191). Arlington, VA: American community_indicators/nat/topline.pdf
Psychiatric Publishing. Radloff, L. S. (1977). The CES-D Scale: A self-
Karney, B. R., & Crown, J. S. (2007). Families under report depression scale for research in the general
stress: An assessment of data, theory, and research population. Applied Psychological Measurement,
on marriage and divorce in the military. Santa 1, 385– 401. doi:10.1177/014662167700100306
Monica, CA: RAND Corporation. doi:10.1037/ Research Triangle Institute. (1990). AFRP analysis
e660632007-001 plan (Volume II: Appendices). Research Triangle
Kim, H. K., Laurent, H. K., Capaldi, D. M., & Park, NC: Author.
Feingold, A. (2008). Men’s aggression toward Rubin, D. B., & Schenker, N. (1991). Multiple im-
women: A 10-year panel study. Journal of Mar- putation in health-care databases: An overview and
riage and Family, 70, 1169 –1187. doi:10.1111/j some applications. Statistics in Medicine, 10, 585–
.1741-3737.2008.00558.x 598. doi:10.1002/sim.4780100410
Krause, N., & Baker, E. (1992). Financial strain, Rumpf, H-J., Hampke, U., Meyer, C., & John, U.
economic values, and somatic symptoms in later (2002). Screening for alcohol use disorders and
life. Psychology and Aging, 7, 4 –14. doi:10.1037/ at-risk drinking in the general population: Psycho-
0882-7974.7.1.4 metric performance of three questionnaires. Alco-
Krug, E. G., Dahlberg, L. L., Mercy, J. A., Zwi, hol and Alcoholism, 37, 261–268. doi:10.1093/
A. B., & Loranzo, R. (2002). World report on alcalc/37.3.261
violence and health. Geneva, Switzerland: World Scholz, U., Gutierrez, D. B., Sud, S., & Schwarzer,
Health Organization. R. (2002). Is general self-efficacy a universal con-
Lawrence, E., Yoon, J., Langer, A., & Ro, E. (2009). struct? European Journal of Psychological Assess-
Is psychological aggression as detrimental as phys- ment, 18, 242–251. doi:10.1027//1015-5759.18.3
ical aggression? The independent effects of psy- .242
chological aggression on depression and anxiety Schumacher, J. A., & Leonard, K. E. (2005). Hus-
symptoms. Violence and Victims, 24, 20 –35. doi: bands’ and wives’ marital adjustment, verbal ag-
10.1891/0886-6708.24.1.20 gression, and physical aggression as longitudinal
Little, L., & Kaufman, K. (2002). Using ecological predictors of physical aggression in early marriage.
theory to understand intimate partner violence and Journal of Consulting and Clinical Psychology,
child maltreatment. Journal of Community Health 73, 28 –37. doi:10.1037/0022-006X.73.1.28
Nursing, 19, 133–145. Schumacher, J. A., Slep, A. M. S., & Heyman, R. E.
McCarroll, J. E., Ursano, R. J., Liu, X., Thayer, L. E., (2001). Risk factors for male-to-female partner
Newby, J. H., Norwood, A. E., & Fullerton, C. S. physical abuse. Aggression and Violent Behavior,
(2000). Deployment and the probability of spousal 6, 255–268. doi:10.1016/S1359-1789(00)00025-2
aggression by U.S. Army soldiers. Military Medi- Slep, A. M. S., Foran, H. M., Heyman, R. E., Snarr,
cine, 165, 41– 44. J. D., & U.S. Air Force Family Advocacy Pro-
Murphy, C. M., & O’Leary, K. D. (1989). Psycho- gram. (2011). Risk and protective factors for clin-
logical aggression predicts physical aggression in ically significant intimate partner violence in a
early marriage. Journal of Consulting and Clinical population survey of the U.S. Air Force. Journal of
Psychology, 57, 579 –582. doi:10.1037/0022-006X Marriage and Family, 73, 486 –501. doi:10.1111/
.57.5.579 j.1741-3737.2010.00820.x
Muthén, L. K., & Muthén, B. O. (2007). Mplus Snarr, J. D., Heyman, R. E., & Slep, A. M. S. (2006).
Version 5.1. Los Angeles, CA: Muthén & Muthén. The 2006 Air Force Behavioral Health Problem
142 FORAN ET AL.

Review: Assessing the prevalence of Force health U.S. Army Community and Support Center. (1989).
challenges. Stony Brook, NY: Stony Brook Uni- 1989 Army Soldier and Family Survey: Soldier
versity, Family Translational Research Group. Survey (SCN: ATNC-AO-89-10A). Cary, NC: Re-
Spera, C. (2009). Spouses’ ability to cope with de- search Triangle Institute, Caliber Associates, and
ployment and adjust to Air Force family demands: Human Resources Research Organization.
Identification of risk and protective factors. Armed U.S. Census Bureau. (2006). Current population sur-
Forces and Society, 35, 286 –306. doi:10.1177/ vey, annual social and economic Suppl. Retrieved
0095327X08316150 from http://pubdb3.census.gov/macro/032006/hhinc/
Stith, S. M., Smith, D. B., Penn, C. E., Ward, D. B., new01_001.htm
& Tritt, D. (2004). Intimate partner physical abuse Vinokur, A. D., Price, R. H., & Caplan, R. D. (1996).
perpetration and victimization risk factors: A
Hard times and hurtful partners: How financial
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

meta-analytical review. Aggression and Violent


strain affects depression and relationship satisfac-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Behavior, 10, 65–98. doi:10.1016/j.avb.2003.09


.001 tion of unemployed persons and their spouses.
Straus, M. A., Hamby, S. L., Boney-McCoy, S., & Journal of Personality and Social Psychology, 71,
Sugarman, D. (1996). The Revised Conflict Tac- 166 –179. doi:10.1037/0022-3514.71.1.166
tics Scale (CTS2): Development and preliminary Ware, J. E., Jr., Kosinski, M., Dewey, J. E., & Gan-
psychometric data. Journal of Family Issues, 17, dek, B. (2001). How to score and interpret single-
283–316. doi:10.1177/019251396017003001 item health status measures: A manual for users of
Straus, M. A., & Sweet, S. (1992). Verbal/symbolic the SF-8TM Health Survey. Lincoln, RI: Quality-
aggression in couples: Incidence rates and relation- Metric Incorporated, 2001.
ships to personal characteristics. Journal of Mar- Wright, K. M., Foran, H. M., Wood, M. D., Eckford,
riage and the Family, 54, 346 –357. doi:10.2307/ R. K., & McGurk, D. (2012). Alcohol problems,
353066 aggression, and other externalizing behaviors after
Stuart, G. L., & Holtzworth-Munroe, A. (2005). Test- return from deployment: Understanding the impact
ing a theoretical model of the relationship between of combat exposure, internalizing symptoms, and
impulsivity, mediating variables, and husband vi- social environment. Journal of Clinical Psychol-
olence. Journal of Family Violence, 20, 291–303. ogy, 1–19. doi:10.1002/jclp.21864
doi:10.1007/s10896-005-6605-6 Yllo, K. A. (1993). Through a feminist lens: Gender,
Taft, C. T., O’Farrell, T. J., Torres, S. E., Panuzio, J., power, and violence. In R. J. Gelles & D. R.
Monson, C. M., Murphy, M., & Murphy, C. M.
Loseke (Eds.), Current controversies on family
(2006). Examining the correlates of psychological
violence (pp. 47– 62). Newbury Park, CA: Sage.
aggression among a community sample of couples.
Journal of Family Psychology, 20, 581–588. doi:
10.1037/0893-3200.20.4.581
The Social Capital Benchmark Survey. (2000). Re- Received June 18, 2012
trieved April 8, 2008, from http://www.cfsv.org/ Revision received June 18, 2013
communitysurvey/docs/survey_instrument.pdf Accepted June 19, 2013 䡲

E-Mail Notification of Your Latest Issue Online!


Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!

You might also like