You are on page 1of 11

Prev Sci (2016) 17:357–366

DOI 10.1007/s11121-015-0617-0

The Effects of the Moms and Teens for Safe Dates


Program on Dating Abuse: a Conditional Process Analysis
Vangie A. Foshee 1 & Thad Benefield 2 & May S. Chen 1 & Luz McNaughton Reyes 1 &
Kimberly S. Dixon 3 & Susan T. Ennett 1 & Kathryn E. Moracco 1 & J. Michael Bowling 1

Published online: 22 October 2015


# Society for Prevention Research 2015

Abstract Moms and Teens for Safe Dates (MTSD) is a dating effects on other mediators including teen feeling of family
abuse (DA) prevention program for teens exposed to domestic closeness and cohesion and mother-perceived susceptibility
violence. In a randomized controlled trial (RCT), MTSD of her teen to DA. As expected, all significant indirect effects
prevented certain types of DA victimization (psychological of MTSD on DA outcomes through mediators were for teens
and physical) and perpetration (psychological and cyber) with higher exposure to domestic violence. Findings have
among teens with higher, but not lower, exposure to domestic implications for developing DA victimization and perpetration
violence. We built on these findings by using moderated me- prevention programs for teens with high exposure to domestic
diation analysis to examine whether level of teen exposure to violence.
domestic violence conditioned the indirect effects of MTSD
on these types of DA through targeted mediators. MTSD Keywords DA prevention . Teens exposed to domestic
consisted of six mailed activity booklets. Mothers who had violence . Teen dating violence . Conditional indirect effects
been former victims of domestic violence delivered the pro-
gram to their teens. Mother and teen pairs were recruited into
the RCT through community advertising and completed base- Although children who have been exposed to domestic vio-
line and 6-month follow-up interviews (N=277 pairs). As ex- lence are at increased risk of being involved in dating abuse
pected, MTSD had significant favorable effects for teens with (DA) as teens and young adults (Ehrensaft et al. 2003; Exner-
higher but not lower exposure to domestic violence on several Cortens et al. 2013; Foshee et al. 1999; Magdol et al. 1998;
mediators that guided program content, including teen conflict O'Donnell et al. 2006), few DA prevention programs target
management skills and mother-perceived severity of DA, self- this high-risk group. One exception is the Moms and Teens for
efficacy for enacting DA prevention efforts, and comfort in Safe Dates (MTSD) program. MTSD is designed for mothers
communicating with her teen. MTSD had significant main who are former victims of domestic violence to deliver to their
teens who were exposed to the violence. The effects of MTSD
on eight DA outcomes (psychological, physical, sexual, and
cyber victimization and perpetration), and four a priori speci-
* Vangie A. Foshee
fied moderators of program effects (sex and race/ethnicity of
foshee@email.unc.edu the teen, mothers’ psychological health, and the amount of
teen exposure to domestic violence) were tested in a random-
1
ized controlled trial (RCT) (Foshee et al. 2015a). MTSD had
Department of Health Behavior, Gillings School of Global Public
Health, The University of North Carolina at Chapel Hill, Rosenau
significant effects in preventing psychological and physical
Hall, Chapel Hill, NC 27599-7400, USA DA victimization, and psychological and cyber DA perpetra-
2
Carolina Mammography Registry, Department of Radiology, The
tion among the teens who had higher, but not lower, exposure
University of North Carolina at Chapel Hill, Chapel Hill, NC, USA to domestic violence. There were no other main or condi-
3
Equal Opportunity and Compliance Office, The University of North
tioned effects of MTSD on DA outcomes.
Carolina at Chapel Hill, CB# 9196, 137 E. Franklin St., Suite 404, The purpose of the current study was to examine why
Chapel Hill, NC 27599, USA MTSD was effective at preventing DA among teens who
358 Prev Sci (2016) 17:357–366

had higher but not lower exposure to domestic violence. intended to change (1) factors that motivate mother’s engage-
Program content may have had greater salience for those at ment in DA prevention activities, (2) factors that facilitate
higher levels of domestic violence exposure, which may have mothers’ engagement in DA prevention activities, and (3) the-
resulted in greater program engagement. Evidence supporting oretically and empirically based risk and protective factors for
this possibility would be findings of greater impact of the teen DA.
program on the theoretically and empirically based mediators Mothers are the change agents in MTSD; therefore, moti-
that guided program content for those with higher levels of vating them to deliver intervention content to their teen and
domestic violence exposure. Thus, the current study examined facilitating their delivery of the program are imperative.
whether and how levels of teen exposure to domestic violence MTSD was designed to motivate mothers to engage in DA
conditioned the indirect effects of MTSD on DA outcomes prevention activities with their teen by increasing mothers’
through targeted mediators, including both mother- and teen- beliefs that their teen could become involved in DA (per-
targeted mediators. ceived teen susceptibility to DA), that specific negative con-
Beyond the specific purpose of examining evidence that sequences could accrue to them and their teen if the teen
might explain the moderated effects of MTSD on DA out- became involved in DA (perceived severity of DA), and that
comes, examining the process through which the program they could potentially influence their teen’s DA behavior (re-
had effects on intended outcomes is crucial for program de- sponse efficacy for preventing DA). Protection motivation
velopment. Constructs found to be mediators of program ef- theory (PMT) (Rogers 1983), which explicates conditions
fects can be targeted for change in future programs, and those needed to motivate a prevention action, informed selection
found to not mediate effects can be discarded from further of the three motivation factors to target for change.
consideration, resulting in more efficacious and efficient pro- The program was designed to facilitate mother engagement
graming (Kraemer et al. 2002). Additionally, identifying the in DA prevention by increasing mothers’ self-efficacy, or con-
mediators of program effects contributes to greater under- fidence, in participating in DA prevention activities with their
standing of the development of problematic behaviors and teen, improving their communication with their teen, and in-
can thus inform future etiological research and interventions creasing their knowledge of DA. Self-efficacy for performing
for primary prevention of DA (Kraemer et al. 2002; a preventive action is a key factor of PMT and is described as
MacKinnon and Dwyer 1993). Yet, despite the importance important for enabling enactment of prevention behaviors
of assessing mediators of program effects, only one evaluation (Rogers 1983). Poor parent communication skills and lack
of a DA prevention program has examined mediation process- of knowledge are commonly noted barriers to parent–teen
es (Foshee et al. 2005). communication about sensitive topics (Jackard et al. 2002).
Thus, improving mother communication skills with the teen
and her knowledge of DA were expected to facilitate delivery
The MTSD Program and Targeted Mediators of the program and optimize the magnitude and duration of the
program impact.
MTSD consisted of six booklets of DA prevention informa- The risk and protective factors for teen DA targeted for
tion and interactive activities (role plays, puzzles, games, sce- change included teen acceptance of DA, teen conflict manage-
nario analyses, guided discussions). The first booklet, Getting ment skills, mother monitoring of the teen, and family close-
Started, was for mothers only, and the other five provided ness and cohesion. Theory and empirical findings suggest that
activities for mothers to do with their teens in their homes at these are particularly salient constructs to target in DA preven-
times convenient for the family. Based on characterizations of tion programs targeted at teens exposed to domestic violence.
teen abusive dating relationships (Mulford and Giordano Social learning theory (SLT) posits that aggression is learned
2008), the activities depict both boys and girls as perpetrators from observing the positive consequences of aggression by
and victims of DA and provide examples of both mutually and others (Bandura 1977). Because violence is a powerful means
unilaterally abusive relationships. The program was devel- of coercion and control, children who observe domestic vio-
oped by adapting an evidence-based DA prevention program lence may witness positive consequences from aggression,
for the general population, Families for Safe Dates (Foshee et al. which promotes learning of aggression and development of
2012), for teens exposed to domestic violence. The adaptation an accepting attitude toward or even preference for aggression
process is described in detail in Foshee et al. (2015b). as a means of interacting. Also, children exposed to domestic
The overall goals of MTSD were to motivate and facilitate violence often do not have the opportunity to observe the
the mother’s engagement with her teen in DA prevention ac- positive consequences of constructive conflict management
tivities (i.e., completing the MTSD booklets) that decrease skills, and thus they lack opportunities to learn skills that could
risk factors and enhance protective factors for DA, ultimately help them avoid aggression in situations of conflict.
leading to decreases in DA victimization and perpetration. Adolescents who have been exposed to domestic violence
Specifically, MTSD included information and activities have been found to be more accepting of DA and demonstrate
Prev Sci (2016) 17:357–366 359

poorer skills in response to conflict than those who have not Methods
been exposed to domestic violence (Jouriles et al. 2012), and
these variables, in turn, have been found to predict both vic- Design
timization and perpetration of DA (Foshee et al. 1999; Lichter
and McCloskey 2004; Reitzel-Jaffe and Wolfe 2001). Data were drawn from a randomized controlled trial (RCT)
Spillover theory suggests that Bemotions, affect, and that assessed MTSD program effects 6 months after program
mood generated in the marital realm transfer to the par- exposure. Mothers who had been in an abusive relationship
ent–child relationship^ (Krishnakumar and Buehler 2000; but were currently living apart from the abusive partner were
p.26). According to this theory, parent preoccupation with recruited into the study, along with their 12- to 15-year-old
relationship problems can deplete their emotional resources teen who had been exposed to the domestic violence. Details
for engaging with their children, leading to inadequate on the variety of community-based recruitment strategies used
monitoring of the child’s activities and friends, lack of in- can be found in Foshee et al. (2015b). Eligible mothers pro-
volvement in the child’s activities, and emotional unavail- vided consent for their teen’s and assent for their own partic-
ability for the child (Gerard et al. 2006; Krishnakumar and ipation; eligible teens provided assent for their own participa-
Buehler 2000). Parental monitoring has been found to be tion. Mothers and teens who consented and assented to par-
protective against DA in longitudinal studies (Capaldi and ticipate then completed a 25-min baseline telephone interview
Clark 1998; Lavoie et al. 2002). Lack of family cohesion and were subsequently randomly allocated to the treatment or
and closeness has also been found to predict DA perpetra- control condition. Families in the treatment group were mailed
tion (Gorman-Smith et al. 2001; Magdol et al. 1998) and a program booklet every 2 weeks. Families in the control
victimization (Magdol et al. 1998; Pflieger and Vazsonyi group were not sent any program materials. Nine months after
2006), and mother–child closeness has been found to buffer baseline (6 months post-intervention completion), the mothers
the negative effects of domestic violence on children and teens were interviewed again by telephone. Study proce-
(Wekerle and Wolfe 1998). Thus, promoting parental mon- dures were approved by the University of North Carolina
itoring of the teen and family closeness and cohesion are Institutional Review Board.
particularly important goals for a program targeting teens
exposed to domestic violence. Sample Description

A total of 409 families (mother and teen pairs) were recruited


into the study. Families were recruited from 17 states with the
Study Hypotheses majority (51 %) from North Carolina. The average age of the
teens was 13.6 years. Approximately 36 % were males;
Because MTSD had stronger effects on DA outcomes 54.8 % were black, 26.9 % were white, and the rest were of
for teens with higher than lower levels of domestic vi- another race. The average age of the mothers was 38.1 years.
olence exposure, we posited that the proposed mediation Approximately 20 % had less than a high school education,
process described above may have occurred only for 59.7 % were unemployed, 64.1 % were single, and 84 %
those teens with the highest amount of exposure, pro- received public assistance. Approximately 66 % of the teens
ducing conditional indirect effects or moderated had been exposed to domestic violence perpetrated by their
mediation. Moderated mediation arises when the associ- biological father. At baseline, 35.7 and 23.2 % of the teens had
ation between an independent variable and a mediator ever been a victim of psychological and physical DA, respec-
variable (the a path) varies by a moderator, or the effect tively. These percentages are higher than the 24.5 % (in the
of a mediator variable on an outcome (b path) varies by past 18 months) and 10 % (in the past year) for psychological
a moderator, or both (Hayes 2013). We posited that the and physical DA victimization, respectively, found in repre-
effect of MTSD on the mediators (i.e., the a paths) sentative samples of the general population (Centers for
would be moderated by teen exposure to domestic vio- Disease Control and Prevention 2008; Halpern et al. 2001,
lence, producing conditional indirect effects of the pro- 2004). At baseline, 27.4 and 26.2 % reported ever having
gram on DA outcomes through the mediators (Fig. 1). perpetrated psychological and cyber DA; national estimates
Specifically, we hypothesized that (1) MTSD would be from general population samples are not available for these
associated with changes in the proposed mediators types of DA perpetration.
among teens with higher, but not lower, exposure to Of the 409 recruited families, 305 mothers (75 %) and 295
domestic violence and, as a result, (2) the indirect ef- teens (72 %) completed follow-up interviews. There was no
fects of MTSD on DA outcomes through the proposed significant difference between the treatment and control
mediators would be significant for those with higher, groups in the amount of attrition between baseline and fol-
but not lower, exposure to domestic violence. low-up. Also, there were no significant interactions between
360 Prev Sci (2016) 17:357–366

Fig. 1 Conceptual model of


Motivate Program Engagement
study hypotheses
Mother perceived susceptibility of teen to DA
Mother perceived severity of DA
Mother response-efficacy for preventing DA

Facilitate Program Engagement

Mother self-efficacy for enacting DA prevention efforts


Mother communication skills DA Victimization
MTSD
Mother comfort in communicating with teen and Perpetration
Mother knowledge of DA

Teen Risk and Protective Factors

Teen acceptance of DA
Amount of Teen Teen conflict management skills
Exposure to Mother monitoring of teen
Domestic Violence Teen feeling of family closeness and cohesion

Note. DA = Dating abuse

treatment condition and study variables on loss to follow-up, up. Response options to all of the DA questions ranged from
and no main effects of these variables on loss to follow-up. Bnever^ to Bmore than 4 times.^

Program Exposure Psychological DA Victimization and Perpetration


Victimization from and perpetration of psychological DA
Each of the five MTSD activity booklets included a card were assessed with items from the Safe Dates Psychological
that asked a series of questions that could be answered only Dating Abuse Scales (Foshee 1996). To assess victimization,
if the family had done that booklet’s activities. The card also teens were asked how many times a date had done the follow-
included a space for filling in a Bsecret code^ that was re- ing to them: (1) insulted them in front of others, (2) not let
vealed throughout the booklet as families carried out activ- them do things with other people, (3) threatened to hurt them,
ities and thus served as an indicator of booklet completion. (4) hurt their feelings on purpose, and (5) said mean things to
After finishing a booklet, families answered the questions, them. Parallel questions were asked to assess perpetration by
filled in the code, and then mailed the card to the study asking teens how many times they had done these things to a
office. The teens were then sent a $15 visa gift card for date. Responses were summed to create psychological DA
completion of the booklet. Booklet completion in the treat- victimization and perpetration composite variables.
ment group ranged from 80 % for the first to 62 % for the
last booklet. The amount of teen exposure to domestic vio- Physical DA Victimization Victimization from physical DA
lence (measure described below) was not associated with the was assessed with items from the Safe Dates Physical Dating
number of booklets completed. Abuse Scale (Foshee 1996). Teens were asked how many
times the following had been done to them that was not done
Measures in self-defense: (1) been slapped or scratched by a date, (2)
had their arm physically twisted or their fingers bent back by a
Dating Abuse Outcomes date, (3) been pushed, grabbed, shoved, or kicked by a date,
(4) been hit by a date with a fist or something hard, and/or (5)
Prior to answering the questions about DA, the teens were been assaulted by a date with a knife or gun. Teens were
asked if they had ever been on a date, defined as informal instructed not to count acts perpetrated against them that were
activities like going out in a group and then pairing up with done in self-defense. Responses were summed to create the
someone in whom they were interested; meeting someone physical DA victimization composite variable.
they were interested in at a mall, park, or sporting event; or
more formal activities like going out to eat or to a movie Cyber DA Perpetration The perpetration of cyber DA was
together. If teens responded Byes,^ then they completed the measured using a modified version of the Tech Abuse in Teen
DA questions. If they responded Bno,^ they were assigned a Relationships Scale (Picard 2007). To assess perpetration,
value of 0 on the DA outcomes. The reference period for DA teens were asked how many times they had done the following
questions was lifetime at baseline and past 3 months at follow- things to a person they were dating using a cell phone, e-mail,
Prev Sci (2016) 17:357–366 361

IM, text messaging, Web chat, a blog, or a networking site like condition and teen exposure to domestic violence. Post hoc
MySpace or Facebook: (1) spread rumors about someone they analyses of significant interactions were conducted to deter-
were dating, (2) called the person bad names, put him/her mine the significance of the association between treatment
down, or said really mean things to the person, (3) showed condition and the proposed mediator at high (1 standard devi-
private or embarrassing pictures/videos of the person to ation above the mean), medium (the mean), and low (1 stan-
others, and/or (4) repeatedly checked up on the person to see dard deviation below the mean) levels of teen exposure to
where he or she was. Responses were summed to create a domestic violence (simple slopes). If the interaction was not
cyber DA perpetration composite variable. significant, it was dropped from the model and the main effect
of MTSD on the mediator was examined.
Moderator Variable Next, analyses were conducted to test the second set of
hypotheses, that the indirect effects of the program on DA
Amount of Teen Exposure to Domestic Violence To mea- outcomes through the mediators would be conditioned by
sure the amount of teen exposure to domestic violence, the amount of exposure the teen had to domestic violence.
mothers were first asked to think about the abusive partner For these analyses we used the SAS PROCESS computational
that the teen was around for the longest period of time. The macro which was developed specifically to test for conditional
mothers indicated how many times this partner had used the indirect effects (Hayes 2013). Separate models for each DA
following abusive acts against them, Bused a knife or gun on outcome were specified to examine conditional indirect ef-
you,^ Bbeat you up,^ Bhit you with a fist or with something fects of MTSD on the DA outcome through the proposed
else hard,^ Bpushed, grabbed, or shoved you,^ Bslapped or mediators, controlling for the baseline measures of the medi-
scratched you,^ Bthreatened you with physical harm,^ ators, the baseline measure of the specific DA outcome
Binsulted you,^ and Bdid something to humiliate you,^ and (referencing a lifetime time period), and the control variables.
then were asked the percentage of these times the teen had Each model was specified to accommodate our expectations
heard or witnessed the abusive act. The number of times the that the associations between treatment condition and the pro-
teen had witnessed or heard each of these acts was then cal- posed mediators would be moderated by teen exposure to
culated and summed. As in the earlier paper (Foshee et al. domestic violence (conditional a paths). The models also in-
2015a), this score was coded into even quartiles, with the corporated our previous findings that the associations between
highest score (3) indicating the greatest amount of exposure treatment condition and DA outcomes were conditioned by
to domestic violence and the lowest score (0) indicating the teen exposure to domestic violence. The indirect effects of
least amount of exposure. MTSD on each DA outcome through each mediator at the
three previously defined levels of teen exposure to domestic
Targeted Mediating Variables Measurement of the targeted violence (high, medium, low) were estimated using
mediating variables is described in Table 1. All of the media- bootstrapping, a non-parametric method that provides greater
tors, except for teen acceptance of DA, teen conflict manage- power for detecting indirect effects than other methods (Hayes
ment skills, and teen feeling of family closeness and cohesion, 2013; Preacher and Hayes 2008). PROCESS addresses miss-
which were measured with established scales, were created ing data through listwise deletion. Therefore, all analyses, in-
specifically for the MTSD RCT as needed to match program cluding the regression analyses, were conducted with a sample
content. of 277 teens and mothers who completed baseline and follow-
up data collection and were not missing data on any of the
Analytical Strategy study variables.

An intent-to-treat approach was used in all analyses; that is,


analyses were based on the treatment condition families were Results
assigned to, not on the amount of treatment received (Wright
and Sim 2003). Linear regression was performed using PROC Main and Conditioned Effects of MTSD on Proposed
GLM in SAS 9.3. (SAS Institute Inc. 2012) to test the first set Mediators
of hypotheses, that the effects of MTSD on each of the
targeted mediators would be conditioned by teen exposure to Table 2 presents the findings from the regression analyses
domestic violence. Each proposed mediator was regressed on testing the first set of hypotheses that the effects of MTSD
treatment condition (0=control and treatment=1), teen level on the proposed mediators would be conditioned by teen ex-
of exposure to domestic violence, the control variables (teen posure to domestic violence. As expected, the effects of
sex, race/ethnicity, age in years, and follow-up dating status, MTSD on several of the proposed mediators were conditioned
and mother’s education level), the baseline value of the pro- by teen exposure to domestic violence in the hypothesized
posed mediator, and the interaction between treatment direction. Specifically, MTSD was significantly associated
362

Table 1 Measurement of mediating variables

Variable No. of items Response categories Item or example item


(follow-up alpha)

Factors motivating mother engagement in DA prevention activities


Mother perceived susceptibility of 4 (.85) 1=very unlikely to 4=very likely BCompared to other teens [TEEN’S NAME] age, how likely or unlikely is it
teen to DA that [HE/SHE] will become a victim of psychological or emotional DA
in the future?^
Mother perceived severity of DA 3 (.60) 1=strongly agree to 4=strongly disagree BIf a teen is frequently insulted by a date it is not that big a deal as long as there
is no physical violence.^
Mother response efficacy for 5 (.68) 1=strongly disagree to 4=strongly agree BTalking to [TEEN’S NAME] about DA will decrease [HIS/HER] chances of
preventing DA abusing dates^
Factors facilitating mother engagement in DA prevention activities
Mother self-efficacy for enacting 5 (.70) 1=not confident at all to 4=very confident BHow confident do you feel in your ability to talk to [TEEN’S NAME] about
teen DA prevention teen DA?^
Mother communication skills 5 (.65) 1=never to 4=very often BDuring the last 3 months when you and [TEEN’S NAME] had a disagreement,
with teen how often did you suggest a possible solution?^
Mother comfort in communicating 2 1=very hard to 4=very easy BIn general, how hard or easy is it for you to talk to [TEEN’S NAME] about
with her teen things that are personal to [HIM/HER]?^
Mother knowledge of DA 5 (N/A) 1=true, 2=false Coded to be proportion correct BAbout 5 % of teens have been physically abused by a date.^
Risk and protective factors
Teen acceptance of DA 11 (.84) 0=strongly disagree to 4=strongly agree BIt is OK for a girl to hit her boyfriend if he insulted her in front of friends.^
(Foshee et al. 2005) BA girl who makes her boyfriend jealous on purpose deserves to be hit.^
Teen conflict management skills 13 (.69) 1=never to 4=very often BDuring the last 3 months, when you were angry with someone, how often
(Foshee et al. 2005) would you say that you explained to the person why you were angry?^
Maternal monitoring of teen 7 (.78) 1=not like you to 4=very like you BYou find out where [TEEN’S NAME] is going when [HE/SHE] goes out.^
BYou try to meet [TEEN’S NAME] friends.^
Teen feeling of family closeness and 11 (.90) 1=never to 4=very often BHow often does your mother cheer you up when you are sad?^
cohesion (Office of Applied Studies 2000; 1=not close at all to 5=extremely close BIn the past 3 months, how often was there a feeling of togetherness in
Simpson and McBride 1992) your family?^
BHow close do you feel towards your mother?^

DA dating abuse, N/A not applicable


Prev Sci (2016) 17:357–366
Prev Sci (2016) 17:357–366 363

Table 2 Effects of MTSD on the proposed mediating variables (N=277)

Targeted outcome Main effect coef (se) Interactiona coefficient if significant

Factors motivating mother engagement in DA prevention activities


Mother perceived susceptibility of teen to DA −0.191 (0.09)*
Mother perceived severity of DA −0.055 (0.058) 0.071 (0.031)*
Mother response efficacy for preventing DA 0.088 (0.077) –
Factors facilitating mother engagement in DA prevention activities
Mother self-efficacy for enacting DA prevention efforts −0.139 (.089) 0.150 (0.048)**
Mother communication skills with teen 0.047 (0.044) –
Mother comfort in communicating with teen −0.154 (0.147) 0.162 (0.079)*
Mother knowledge of DA 0.005 (0.019) –
Risk and protective factors for DA
Teen acceptance of DA −0.117 (0.068) ‡ –
Mother monitoring of teen 0.061 (0.065) 0.108 (0.557) ‡
Teen conflict management skills −0.047 (.087) 0.133 (.047) **
Teen feeling of family closeness and cohesion 0.177 (.085)* –

All analyses control for teen age in years, sex (0=female; 1=male), race/ethnicity (white and other with the reference for both being black) and dating
status (ever dated or not), mother’s education level (exactly high school and more than high school with the reference for both being not graduating from
high school), and baseline levels of the outcome. Bold indicates significance at p<.05
DA dating abuse

p<.10; *p<.05;**p<.01;***p<.001
a
Significant interaction between treatment condition and level of teen exposure to domestic violence

with greater mother perceived severity of DA when teen ex- MTSD had a significant main effect on teen feeling of
posure to domestic violence was high (b=.1576; p=.0107), family closeness and cohesion such that teens exposed
but not when teen exposure was medium (b=.0514; p=.1638) to the program reported feeling more family closeness
or low (b=−.0548; p=.3482). MTSD exposure was signifi- and cohesion at follow-up than those not receiving the
cantly related to greater mother self-efficacy when teen expo- program. Additionally, there was a marginally signifi-
sure was high (b=.3103; p=.0011), but not when it was me- cant negative effect of MTSD on teen acceptance of
dium (b=.0856; p=.1299) or low (b=−.1392; p=.1200). The DA (b = −0.117; p = .07); teens receiving the program
program was also significantly related to greater mother com- had less acceptance of DA at follow-up than those not
fort in communicating with her teen when teen exposure to receiving the program.
domestic violence was high (b=.3322; p=.0319), but not
when it was medium (b = .0889; p = .3373) or low (b = Conditioned Indirect Effects
−.1544; p=.2929). Additionally, MTSD exposure was signif-
icantly associated with improved teen conflict management The mediators that were included in the four PROCESS
skills among teens with high (b=.3506; p=.0002) and medi- models testing conditional indirect effects were those
um (b=.1517; p=.0064), but not low (b=−.0473; p=.5895), that were found in the regression analyses to be affected
exposure to domestic violence. Finally, there was a marginal by MTSD (at p<.10) as either main or conditioned ef-
conditional effect of MTSD on mother monitoring (b=.1080; fects (mother perceived susceptibility of teen to DA,
p=.053) and the trend was in the direction hypothesized; perceived severity of DA, self-efficacy, comfort in com-
MTSD was associated with more monitoring among teens of municating with her teen, and monitoring, and teen ac-
high (b = .2300; p = .0350), but not medium (b = .0680; ceptance of DA, conflict management skills, and feeling
p=.2979) or low (b=−.0941; p=.3629), exposure. of family closeness and cohesion).
Contrary to expectations, the effects of MTSD on As expected, all of the indirect effects (IE) that were signifi-
other proposed mediators were not conditioned by teen cant were significant only for teens with higher exposure to
exposure to domestic violence. There was a main effect domestic violence. The indirect effects of MTSD on psycho-
of MTSD on mother perceived susceptibility of teen to logical DA victimization through teen conflict management
DA, but the association was opposite that expected; skills and teen feeling of family closeness and cohesion were
MTSD was associated with lower rather than higher significant at higher but not lower levels of teen exposure.
mother perceived susceptibility of her teen to DA. Specifically, the indirect effect of MTSD on psychological
364 Prev Sci (2016) 17:357–366

DA victimization through teen conflict management skills was Discussion


significant when teen exposure was high (IE = −0.180,
CI = −0.4484, −0.0350) and medium (IE = −0.0792, CI = In earlier analyses, MTSD was found to be effective in
−0.2143, −0.0141), but not when it was low (IE=0.0211, preventing psychological and physical DA victimization and
CI=−0.0678, 0.2013). The indirect effect of MTSD on psy- psychological and cyber DA perpetration among teens with
chological DA victimization through teen feeling of family higher but not lower levels of exposure to domestic violence
closeness and cohesion was significant when teen exposure (Foshee et al. 2015a). The current study examined a possible
was high (IE = −0.0885, CI= −0.2457, −0.0041), but not explanation for those findings: that MTSD may have been
when it was medium (IE=−0.0382, CI=−0.1365, 0.0150) more effective among teens with higher domestic violence
or low (IE=0.0122, CI=−0.0660, 0.1077). exposure in changing the theoretically and empirically based
As with psychological DA victimization, the indirect effect mediators that guided program content, which led to stronger
of MTSD on physical DA victimization through teen conflict program effects on DA for high exposure teens. In general, we
management skills was significant when teen exposure was found support for this explanation in that MTSD had signifi-
high (IE=−0.1192, CI=−0.2831, −0.0297) and medium (IE= cant effects on several of the proposed mediators for teens
−0.0531, CI=−0.1504, −0.0110), but not when it was low with higher but not lower exposure to domestic violence,
(IE=0.0131, CI=−0.0429, 0.1267). Unlike with psychologi- and all of the significant indirect effects of MTSD on DA
cal DA victimization, the indirect effect of MTSD on physical outcomes through mediators were for teens with higher expo-
DA victimization through teen feeling of family closeness and sure to domestic violence.
cohesion was not significant at any level of teen exposure. MTSD was effective among teens with higher, but not
The indirect effects of MTSD on psychological DA perpe- lower, exposure to domestic violence at increasing the
tration through teen conflict management skills, teen feeling mothers’ perceptions of the severity of DA, self-efficacy
of family closeness and cohesion, and teen acceptance of DA for enacting DA prevention efforts, and comfort in commu-
were significant at higher, but not lower, levels of teen expo- nicating with the teen and at improving the teens’ conflict
sure. More specifically, the indirect effect of MTSD on psy- management skills. However, MTSD had significant effects
chological DA perpetration through conflict management on mother-perceived susceptibility of her teen to DA and
skills was significant when teen exposure was high (IE= teen feeling of family closeness and cohesion, and margin-
−0.1395, CI=−0.3572, −0.0352), and medium (IE=−0.0633, ally significant effects on teen acceptance of DA regardless
CI=−0.1644, −0.0133), but not when low (IE=0.0128, CI= of the level of exposure to domestic violence. The effects of
−0.0560, 0.1326). The indirect effect of MTSD on psycholog- MTSD on all of the mediators were in the expected direc-
ical DA perpetration through teen feeling of family closeness tion, except for the effects on mother perceived susceptibil-
and cohesion was significant when teen exposure was high ity of the teen to DA; mothers who received MTSD per-
(IE=−0.0652, CI=−0.1876, −0.0002), but not when exposure ceived their teen to be less rather than more susceptible to
was medium (IE=−0.281, CI=−0.1112, 0.0086) or low (IE= DA than those who did not receive MTSD. Perhaps these
0.0089, CI=−0.0475, 0.847). Likewise, the indirect effect of mothers believed that the program Binoculated^ their teen
MTSD on psychological DA perpetration through teen accep- against future involvement in DA.
tance of DA was significant when teen exposure was high Teen acceptance of DA, teen conflict management skills,
(IE=−0.1174, CI=−0.3426, −0.0052), but not when exposure and teen feeling of family closeness and cohesion significantly
was medium (IE=−0.0720, CI=−0.2314, 0.0009) or low (IE= mediated the effects of MTSD on DA outcomes among teens
−0.0266, CI=−0.2127, 0.0738). with higher exposure to domestic violence; the specific medi-
The indirect effect of MTSD on cyber DA perpetration ator(s) varied depending on the DA outcome. There were no
through teen acceptance of DA was significant at higher, but significant indirect effects of the program on DA outcomes
not lower, levels of teen exposure. More specifically, the in- through the mediators for teens with lower exposure to do-
direct effect of MTSD on cyber DA perpetration through teen mestic violence. Although there were significant main or con-
acceptance of DA was significant when teen exposure was ditioned effects of MTSD on several mother motivation and
high (IE = −0.0732, CI =−0.2331, −0.0088), and medium facilitating variables, none of these variables were significant
(IE=−0.0452, CI= −0.1449, −0.0010), but not when low mediators of the MTSD and DA relationships for teens at any
(IE=−0.0172, CI=−0.1255, 0.0490). level of exposure to domestic violence. However, changes in
Finally, contrary to expectations, we found no indi- these constructs due to the program may have indirectly af-
rect effects (conditional or unconditional) of the pro- fected DA by enhancing delivery of program content that
gram on DA outcomes through mother-perceived sus- targeted changes in teen risk and protective factors, which
ceptibility of teen to DA, perceived severity of DA, did mediate program effects.
self-efficacy, comfort in communicating with her teen, The findings have implications for practice and future re-
or monitoring. search. Key constructs were identified (teen acceptance of
Prev Sci (2016) 17:357–366 365

DA, teen conflict management skills, and family closeness violence variable was limited by the possibility that
and cohesion) that can be targeted for change in DA preven- mothers may have under- or over-reported teen expo-
tion programs designed for teens with high levels of exposure sure. Teen reports were not collected but there are lim-
to domestic violence. Our findings demonstrated that those itations to assessing exposure to domestic violence from
constructs can be changed through intervention, even in this children/teens as well, and there is no consistency in
very high risk group, and result in the prevention of DA. recommendations about whether exposure reports are
These constructs link to theoretical explanations, specifically more accurate from the parent or the child (Edleson
social learning theory (Bandura 1977) and spillover theory et al. 2007).
(Krishnakumar and Buehler 2000) for why exposure to do- The study had many strengths. The RCT design eliminated
mestic violence may increase risk for DA. Tenants of the the- many potential threats to the internal validity of the findings.
ories could be drawn on in developing future DA prevention Contamination between the treatment and control groups was
programs targeting teens with high exposure to domestic vio- minimal because the families in the trial were spread out over
lence. These constructs should also inform areas to focus on in a large geographic area; thus, the integrity of the design was
future theory-based etiological research on the development maintained. Participation in the program was relatively high,
of DA among high-exposure teens. Study findings also poten- especially given the high-risk sample. The mediators guiding
tially inform more efficient programing in that they inform program content were based on theory and empirical findings
how to limit the scope of what is targeted for change in pro- and were specified a priori. The study statistically tested for
grams so that unnecessary program content (targeting ineffec- mediation, which has rarely been done in evaluations of DA
tual constructs) is not developed and delivered. Other RCTs of prevention programs.
MTSD are needed to determine if the findings can be replicat- In conclusion, our findings suggest that the greater effects
ed; such trials need to include process data, beyond simply the of MTSD on certain types of DA victimization and perpetra-
number of booklets completed, that would allow for examin- tion for teens with higher exposure to domestic violence that
ing why MTSD was more effective at changing the proposed were found in earlier analyses (Foshee et al. 2015a) may have
mediators for teens with higher than lower domestic violence been attained because the program was also more effective at
exposure, if that finding should be replicated, and also why the changing targeted mediator variables that guided program
changes that occurred in some mediators (teen acceptance of content for those teens. Aiming to improve teen conflict man-
DA and family closeness and cohesion) for all teens resulted agement skills, increase teen feeling of family closeness and
in prevention of DA for only those at the highest exposure. cohesion, and decrease teen acceptance of DA in future DA
The primary study limitation relates to the external validity, prevention programs for teens with high domestic violence
or generalizability, of study findings. MTSD is designed for exposure offers promise because we demonstrated that these
administration by mothers who no longer live with an abuser constructs are modifiable through intervention and that mod-
because of the potential dangers inherent in delivering a teen ification of them led to the prevention of certain types of DA
DA prevention program in homes where domestic violence is among the high exposure teens. Although limitations of the
currently occurring. Whether the findings are applicable for MTSD program are that it was not effective at preventing DA
teens whose mothers still live with the abuser is not known. for teens at all levels of exposure, and it was not effective at
Also, the sample for the RCT was a convenience sample that preventing certain types of DA for any teens, a strength of the
was primarily of low socio-economic status (SES) families. program is that it was effective at preventing DA among the
Thus, the study findings may not generalize to teens at higher highest-risk teens because there have been noted challenges to
SES levels who are exposed to domestic violence. However, preventing problem behaviors among very high risk teens
there are few sampling options other than convenience sam- (Dutton et al. 2003).
pling for identifying teens exposed to domestic violence.
Also, attrition in the RCT was fairly high, which may influ-
Acknowledgements This research was supported by a grant from the
ence the generalizability of study results. However, none of
National Institute of Justice [NIJ 2008-WG-BX-1003; PI Vangie A.
the study variables were associated with attrition, suggesting Foshee] which was administered through the Injury Prevention Research
that the analytic sample represented the sample of families Center at the University of North Carolina at Chapel Hill (UNC-CH) and
recruited into the trial. by a grant from the Centers for Disease Control and Prevention [CDC
R01CE001867-01; PI Vangie A. Foshee] which was administered
Another limitation is that most of the measures of the
through the Department of Health Behavior at UNC-CH. The study
mediators were created for the study and therefore had was reviewed and approved by the Public Health-Nursing Institutional
not been subjected to prior validity and reliability test- Review Board for the Protection of Human Subjects at UNC-CH.
ing. However, creating measures was necessary when
Compliance with Ethical Standards
standardized measures for constructs were not available.
Also, creating measures allowed for tailoring them to Conflict of Interest None of the authors have any conflicts of interest
the content of the program. The exposure to domestic associated with this study.
366 Prev Sci (2016) 17:357–366

References sexual relationships in a national sample of adolescents. Journal of


Adolescent Health, 35, 124–131.
Hayes, A. F. (2013). Introduction to mediation, moderation, and condi-
Bandura, A. (1977). Social learning theory. Englewood Cliffs: Prentice-
tional process analysis: a regression-based approach. New York:
Hall.
Guilford Press.
Capaldi, D. M., & Clark, S. (1998). Prospective family predictors of Jackard, J., Dodge, T., & Dittus, P. (2002). Parent teen communication
aggression toward female partners for at-risk young men. about sex and birth control: A conceptual framework. In S. Feldman
Developmental Psychology, 34, 1175–1188. & D. A. Rosenthal (Eds.), Talking sexuality: parent-teen
Centers for Disease Control and Prevention. (2008). Surveillance communication (pp. 9–41). San Francisco: Jossey-Bass.
Summaries. MMWR. Morbidity and Mortality Weekly Report, 57, Jouriles, E. N., McDonald, R., Mueller, V., & Grych, J. H. (2012). Youth
1–131. experiences of family violence and teen dating violence perpetra-
Dutton, M. A., Holtzworth-Monroe, A., Jouriles, E., McDonald, R., tion: Cognitive and emotional mediators. Clinical Child and Family
Krishnan, S., McFarlane, J., & Sullivan, C. (2003). Recruitment Psychology Review, 15, 58–68.
and retention in intimate partner violence research. Washington Kraemer, H. C., Wilson, G. T., Fairburn, C. G., & Agras, W. S. (2002).
DC: National Institute of Justice. Mediators and moderators of treatment effects in randomized clini-
Edleson, J. L., Ellerton, A. L., Seagren, E. A., Kirchberg, S. L., Schmidt, cal trials. Archives of General Psychiatry, 59, 877–883.
S. O., & Ambrose, A. T. (2007). Assessing child exposure to adult Krishnakumar, A., & Buehler, C. (2000). Interparental conflict and par-
domestic violence. Children and Youth Services Review, 29, 961– enting behaviors: A meta-analytic review. Family Relations, 49, 25–
971. 44.
Ehrensaft, M. K., Cohen, P., Brown, J., Smailes, E., Chen, H., & Johnson, Lavoie, F., Hebert, M., Tremblay, R., Vitaro, F., Vezina, L., & McDuff, P.
J. G. (2003). Intergenerational transmission of partner violence: A (2002). History of family dysfunction and perpetration of dating
20-year prospective study. Journal of Consulting and Clinical violence by adolescent boys: A longitudinal study. The Journal of
Psychology, 71, 741–753. Adolescent Health, 30, 375–383.
Exner-Cortens, D., Eckenrode, J., & Rothman, E. (2013). Longitudinal Lichter, E. L., & McCloskey, L. A. (2004). The effects of childhood
associations between teen dating violence victimization and adverse exposure to marital violence on adolescent gender-role beliefs and
health outcomes. Pediatrics, 71, 71–78. dating violence. Psychology of Women Quarterly, 28, 344–357.
Foshee, V. (1996). Gender differences in adolescent dating abuse MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in
prevalence, types, and injuries. Health Education Research, prevention studies. Evaluation Review, 17, 144–158.
11, 275–286. Magdol, L., Moffitt, T. E., Caspi, A., & Silva, P. A. (1998).
Foshee, V. A., Bauman, K. E., & Linder, G. F. (1999). Family violence Developmental antecedents of partner abuse: A prospective-
and the perpetration of adolescent dating violence: Examining social longitudinal study. Journal of Abnormal Psychology, 107, 375–389.
learning and social control processes. Journal of Marriage and the Mulford, C., & Giordano, P. C. (2008). Teen dating violence: A closer
Family, 61, 331–342. look at adolescent romantic relationships. NIJ Journal, 261, 34–40.
Foshee, V. A., Bauman, K. E., Ennett, S. T., Suchindran, C., Benefield, T., O'Donnell, L., Stueve, A., Myint-U, A., Duran, R., Agronick, G., &
& Linder, G. F. (2005). Assessing the effects of the dating violence Wilson-Simmons, R. (2006). Middle school aggression and subse-
prevention program BSafe Dates^ using random coefficient regres- quent intimate partner physical violence. Journal of Youth and
sion modeling. Prevention Science, 6, 245–258. Adolescence, 25, 693–703.
Office of Applied Studies. (2000). National household survey on drug
Foshee, V. A., Reyes, H. L. M., Ennett, S. T., Cance, J. D., Bauman, K. E.,
abuse: CAI specs for programming. Rockville: Substance Abuse
& Bowling, J. M. (2012). Assessing the effects of Families for Safe
and Mental Health Services Administration.
Dates, a family-based teen DA prevention program. Journal of
Pflieger, J. C., & Vazsonyi, A. T. (2006). Parenting processes and dating
Adolescent Health, 51, 349–356.
violence: The mediating role of self-esteem in low-and high-SES
Foshee, V. A., Benefield, T., Dixon, K. S., Chang, L, Senkomago, V.,
adolescents. Journal of Adolescence, 29, 495–512.
Ennett, S. T., … Bowling, J. M. (2015). The effects of Moms and
Picard, P. (2007). Tech abuse in teen relationships study: a report pre-
Teens for Safe Dates, a dating abuse prevention program for adoles-
pared for Liz Clairborne Inc. Northbrook: Teenage Research
cents exposed to domestic violence. Journal of Youth and
Unlimited (TRU).
Adolescence, 44, 995–1010.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strat-
Foshee, V. A., Dixon, K. S., Ennett, S. T., Moracco, K. E., Bowling, J. M., egies for assessing and comparing indirect effects in multiple medi-
Chang, L., & Moss, J. L. (2015b). The process of adapting a uni- ator models. Behavior Research Methods, 40, 879–891.
versal DA prevention program to adolescents exposed to domestic Reitzel-Jaffe, D., & Wolfe, D. A. (2001). Predictors of relationship abuse
violence. Journal of Interpersonal Violence, 30, 2151–2173. among young men. Journal of Interpersonal Violence, 16, 99–115.
Gerard, J. M., Krishnakumar, A., & Buehler, C. (2006). Marital conflict, Rogers, R. (1983). Cognitive and physiological processes in fear based
parent–child relations, and youth maladjustment: A longitudinal in- attitude change: A revised theory of protection motivation. In J.
vestigation of spillover effects. Journal of Family Issues, 27, 951– Cacioppo & R. Petty (Eds.), Social psycho-physiology: a
975. sourcebook (pp. 153–176). New York: Guilford Press.
Gorman-Smith, D., Tolan, P. H., Shiedow, A. J., & Henry, D. B. (2001). SAS Institute Inc. (2012). SAS Software, Version 9.3. Cary: SAS Institute
Partner violence and street violence among urban adolescents: Do Inc.
the same family factors relate? Journal of Research on Adolescence, Simpson, D. D., & McBride, A. A. (1992). Family, Friends, and Self
11, 273–295. (FFS) assessment scales for Mexican American Youth. Hispanic
Halpern, C. T., Oslak, S. G., Young, M. L., Martin, S. L., & Kupper, L. L. Journal of Behavioral Sciences, 14, 327–340.
(2001). Partner violence among adolescents in opposite-sex roman- Wekerle, C., & Wolfe, D. A. (1998). The role of child maltreatment and
tic relationships: Findings from the National Longitudinal Study of attachment style in adolescent relationship violence. Development
Adolescent Health. American Journal of Public Health, 91, 1679– and Psychopathology, 10, 571–586.
1685. Wright, C. C., & Sim, J. (2003). Intention-to-treat approach to data from
Halpern, C. T., Young, M. L., Waller, M., Martin, S. L., & Kupper, L. randomized controlled trials: A sensitivity analysis. Journal of
(2004). Prevalence of partner violence in same-sex romantic and Clinical Epidemiology, 56, 833–842.
Copyright of Prevention Science is the property of Springer Science & Business Media B.V.
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.

You might also like