16
Daily Reports of Intimate Partner Verbal Aggression by Self and Partner: Short-Term Consequences and Implications for Measurement Jaye L. Derrick, Maria Testa, and Kenneth E. Leonard University at Buffalo, The State University of New York Objective: Agreement within couples regarding the occurrence of aggression is surprisingly low. Survey research often collapses across partners’ reports to create a pooled estimate of aggression in the relationship. This method ignores possible differences in partners’ per- ceptions of the event, potentially weakening researchers’ ability to detect consequences of aggression. The current study examines both partners’ reports of verbal aggression to determine whether aggression reported by only 1 partner influences both partners’ short- term outcomes. Method: We used a 56-day daily diary to examine the effect of verbal aggression on short-term negative outcomes. We examined whether aggression reported by either partner is sufficient to predict consequences for both partners, or whether an individual must report aggression to experience consequences. Results: Victims’ reports of receiving verbal aggression were a better predictor of next day victim consequences than perpetrators’ reports. Perpetrators’ reports of perpetrating verbal aggression were a better predictor of next-day perpetrator consequences than victims’ reports. Days when partners agreed that aggression had occurred generally predicted the worst outcomes. Conclusions: People’s own reports of verbal aggression are the best predictor of short-term consequences. Pooling partner reports of aggression may make it more difficult to understand the consequences of intimate partner aggression. Keywords: daily diary, domestic violence, dyadic data analysis, hierarchical linear modeling, intimate partner aggression, intimate partner violence, multilevel modeling Agreement within couples regarding the occur- rence of aggression is surprisingly low. It is often assumed that this disagreement is a type of mea- surement error, reflecting underreporting and so- cial desirability concerns. Consequently, when re- ports of aggression are obtained from both partners in a couple, it has become standard prac- tice in survey research to collapse across partners’ reports to create a pooled estimate of aggression in the relationship. A critical but unexplored ques- tion is whether partner aggression has to be rec- ognized as aggression for it to have an impact on well-being. If the consequences of aggression de- pend on recognition by the recipient, this has important implications for self-report measure- ment of partner aggression and for the standard practice of combining partner reports. Using inde- pendent daily reports from both members of inti- mate couples, the current study examined whether the next-day consequences of verbal aggression depend on whether it is reported by the victim, the perpetrator, or both. We also explored gender and other individual difference variables as potential correlates of partner agreement regarding the oc- currence of aggression. Agreement Regarding the Occurrence of Aggression A large body of survey research has exam- ined the prevalence, predictors, and conse- This article was published Online First August 11, 2014. Jaye L. Derrick, Maria Testa, and Kenneth E. Leonard, Research Institute on Addictions, University at Buffalo, The State University of New York. Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) of the National Institutes of Health under award number R01AA016127. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We thank Joseph F. Lucke for his assistance during revision of the manuscript. Correspondence concerning this article should be ad- dressed to Jaye L. Derrick, Research Institute on Addictions, 1021 Main Street, Buffalo, NY 14203. E-mail: jderrick@ria .buffalo.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Psychology of Violence © 2014 American Psychological Association 2014, Vol. 4, No. 4, 416 – 431 2152-0828/14/$12.00 http://dx.doi.org/10.1037/a0037481 416

Daily Reports of Intimate Partner Verbal Aggression by Self and Partner: Short-Term Consequences and Implications for Measurement

  • Upload
    uh

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Daily Reports of Intimate Partner Verbal Aggression by Self andPartner: Short-Term Consequences and

Implications for Measurement

Jaye L. Derrick, Maria Testa, and Kenneth E. LeonardUniversity at Buffalo, The State University of New York

Objective: Agreement within couples regarding the occurrence of aggression is surprisinglylow. Survey research often collapses across partners’ reports to create a pooled estimate ofaggression in the relationship. This method ignores possible differences in partners’ per-ceptions of the event, potentially weakening researchers’ ability to detect consequences ofaggression. The current study examines both partners’ reports of verbal aggression todetermine whether aggression reported by only 1 partner influences both partners’ short-term outcomes. Method: We used a 56-day daily diary to examine the effect of verbalaggression on short-term negative outcomes. We examined whether aggression reported byeither partner is sufficient to predict consequences for both partners, or whether anindividual must report aggression to experience consequences. Results: Victims’ reports ofreceiving verbal aggression were a better predictor of next day victim consequences thanperpetrators’ reports. Perpetrators’ reports of perpetrating verbal aggression were a betterpredictor of next-day perpetrator consequences than victims’ reports. Days when partnersagreed that aggression had occurred generally predicted the worst outcomes. Conclusions:People’s own reports of verbal aggression are the best predictor of short-term consequences.Pooling partner reports of aggression may make it more difficult to understand theconsequences of intimate partner aggression.

Keywords: daily diary, domestic violence, dyadic data analysis, hierarchical linear modeling,intimate partner aggression, intimate partner violence, multilevel modeling

Agreement within couples regarding the occur-rence of aggression is surprisingly low. It is oftenassumed that this disagreement is a type of mea-surement error, reflecting underreporting and so-cial desirability concerns. Consequently, when re-ports of aggression are obtained from bothpartners in a couple, it has become standard prac-tice in survey research to collapse across partners’

reports to create a pooled estimate of aggression inthe relationship. A critical but unexplored ques-tion is whether partner aggression has to be rec-ognized as aggression for it to have an impact onwell-being. If the consequences of aggression de-pend on recognition by the recipient, this hasimportant implications for self-report measure-ment of partner aggression and for the standardpractice of combining partner reports. Using inde-pendent daily reports from both members of inti-mate couples, the current study examined whetherthe next-day consequences of verbal aggressiondepend on whether it is reported by the victim, theperpetrator, or both. We also explored gender andother individual difference variables as potentialcorrelates of partner agreement regarding the oc-currence of aggression.

Agreement Regarding the Occurrence ofAggression

A large body of survey research has exam-ined the prevalence, predictors, and conse-

This article was published Online First August 11, 2014.Jaye L. Derrick, Maria Testa, and Kenneth E. Leonard,

Research Institute on Addictions, University at Buffalo, TheState University of New York.

Research reported in this publication was supported by theNational Institute on Alcohol Abuse and Alcoholism (NIAAA)of the National Institutes of Health under award numberR01AA016127. The content is solely the responsibility of theauthors and does not necessarily represent the official views ofthe National Institutes of Health. We thank Joseph F. Lucke forhis assistance during revision of the manuscript.

Correspondence concerning this article should be ad-dressed to Jaye L. Derrick, Research Institute on Addictions,1021 Main Street, Buffalo, NY 14203. E-mail: [email protected]

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Psychology of Violence © 2014 American Psychological Association2014, Vol. 4, No. 4, 416–431 2152-0828/14/$12.00 http://dx.doi.org/10.1037/a0037481

416

quences of intimate partner aggression usingself-report measures. The most common ofthese measures is the CTS-2, or revised ConflictTactics Scales (Straus, Hamby, Boney-McCoy,& Sugarman, 1996). The measure includes aseries of behaviorally specific items assessingthe occurrence of aggressive acts over the pastyear (or other time period), both as perpetrator(e.g., I hit my partner) and as victim (e.g., Mypartner hit me). As with any self-report mea-sure, responses depend on the ability of therespondent to recall and interpret relevant be-haviors in memory that match the items(Schwarz, 2012). By definition, all of the CTSitems refer to dyadic behaviors that require thepresence of both partners at the time of occur-rence, but numerous studies have shown within-couple agreement regarding the occurrence ofrelationship aggression to be modest, withagreement about nonoccurrence far exceedingagreement about occurrence (Caetano, Field,Ramisetty-Mikler, & Lipsky, 2009; O’Leary &Williams, 2006; Schafer, Caetano, & Clark,2002; Szinovacz & Egley, 1995).

Daily diary studies provide an ideal method-ology for comparing partner perspectives on theoccurrence of relationship aggression. Recallbiases should be minimized by the short recallperiod (Follingstad & Rogers, 2013); thus, wemight expect superior partner agreement re-garding aggression with shorter relative to lon-ger (e.g., 1 year) intervals. However, in a re-cently completed study examining the impact ofalcohol use on partner aggression, we foundonly modest agreement between partners re-garding the occurrence of aggression in the past24 hours (Testa & Derrick, 2014). Of 414 dayson which male-perpetrated verbal aggressionwas reported by at least one partner, partnersagreed on only 97 days (23%). Similarly, of 482days on which female-perpetrated verbal ag-gression was reported by at least one partner,partners agreed on only 130 days (27%). Sur-prisingly, agreement was even poorer for phys-ical aggression (only 15.6% for male-perpe-trated and 12.5% for female-perpetrated).Women reported more victimization and moreperpetration than men, for both verbal and phys-ical aggression. This level of agreement is com-parable with that reported in survey studies(Caetano et al., 2009; Caetano, Schafer, Field,& Nelson, 2002; Schafer et al., 2002; Testa etal., 2012).

In light of this modest agreement, it is con-sidered accepted and best practice to pool thereports of aggression from both partners whenestimating the prevalence of aggression (Scha-fer et al., 2002; Straus et al., 1996). For exam-ple, if the wife reports that she hit her partnerbut her husband fails to report being hit, thecouple is classified as experiencing wife-to-husband aggression. Because most partner ag-gression is reported by only one partner(Caetano et al., 2002), using the maximum re-port generally results in an estimate of aggres-sion within the relationship that is higher thanthat based upon a single partner’s report. It iswidely assumed within this tradition that part-ner aggression is a socially undesirable behaviorthat is underreported. Thus, pooling the reportsof both partners (or multiplying the single part-ner’s report by a correction factor, e.g., Heyman& Schlee, 1997) is thought to correct for whatSzinovacz and Egley (1995) label a “reportingbias,” providing a better estimate of the truelevel of aggression within a relationship. Al-though there is some support for the role ofsocial desirability in suppressing reports of part-ner aggression (e.g., Sugarman & Hotaling,1997), other studies have found less clear evi-dence (Szinovacz & Egley, 1995). For example,O’Leary and Williams (2006) found that partneragreement regarding the occurrence of positiveacts was no better than that for the occurrence ofphysical aggression.

Implicit in the pooled partner approach is theassumption that reporting an act of aggressionreflects correct identification of an experience,whereas failing to report indicates reporting bi-as. In actuality, responses to the CTS or anyself-report measure require the respondent tomatch the items to representations in memory,and whether or not the respondent reports anitem as occurring depends on wording, context,memory, and interpretation (Schwarz, 2012).Thus, a respondent may fail to report an aggres-sive behavior because she or he does not rec-ognize that behavior as occurring, rather thanthrough a willful desire to distort (Hamby,2005). For example, a woman might report thatconflict involved heated disagreement but not“yelling.” Her partner, on the other hand, mayview the episode differently, and report thatyelling did indeed occur. Although these twoperspectives may be equally valid, the partnerviolence field has treated separate partner re-

417SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

ports as incomplete pieces, not of interest inthemselves.

The Role of Interpretation inUnderstanding Events

In contrast, an established body of literaturewithin the close relationships field views theseparate reports of each partner as meaningfulreflections of subjective reality (Baldwin, 1992;Collins, 1996; Kelley, 1979; Reis & Shaver,1988). According to this research tradition, thesame objective event can be interpreted verydifferently by different people based on theirworking models (e.g., their expectations, mo-tives, goals, and fears). These different interpre-tations of the same objective event can thenhave very real consequences, both positive andnegative, for the person and the relationship.For example, people who expect others to beresponsive to their needs perceive greater ac-ceptance and support and draw closer to theirpartner when they experience negative interac-tions (Murray, Bellavia, Rose, & Griffin, 2003;Srivastava, McGonigal, Richards, Butler, &Gross, 2006). Conversely, people who expectrejection feel worse and react more negativelyin response to negative relationship events(Campbell, Simpson, Boldry, & Kashy, 2005;Downey, Freitas, Michaelis, & Khouri, 1998;Murray et al., 2003). Extending the couples’framework to understanding partner aggressionsuggests that discrepancies in partner reports ofaggression may be more than nuisance variance;rather, each partner’s perspective may providedistinct information with differential predictiveability.

Consequences of Aggression

Experiencing partner aggression, both physi-cal and psychological, negatively impacts well-being (Anderson, 2002; Coker et al., 2002;Follingstad, 2009; Testa & Leonard, 2001).Within community samples, the consequencesof psychological aggression may be even moresevere than for physical aggression (Lawrence,Yoon, Langer, & Ro, 2009). A critical but un-explored question is whether psychological ag-gression has to be recognized by the recipientfor it to have an impact on well-being. Thecurrent study examines whether a report of ver-

bal aggression from either partner is sufficientto lead to negative short-term consequences, orwhether the victim must recognize that verbalaggression occurred to experience such conse-quences. Because our analyses are based onreports from a community sample, and not froma sample selected for high levels of aggression(e.g., battered women and their partners), wefocus on verbal aggression and not on physicalaggression in the current study.

Hypotheses

If failure to report aggression reflects mea-surement error, then negative effects on thevictim’s next day functioning should be similarregardless of which partner reports aggression;that is, if both partners report, only the victimreports, or only the perpetrator reports verbalaggression, the victim should experience worseconsequences than when neither partner reports.However, if recognizing aggression is impor-tant, then next-day functioning for victimsshould be worse when the victim reports receiv-ing aggression (i.e., both-partner reports andvictim-only reports) than when the victim doesnot report receiving aggression (i.e., perpetra-tor-only reports and neither-partner reports). Itis also possible that when partners agree on theoccurrence of aggression (i.e., both-partner re-ports), those incidents may be particularly sa-lient, and hence will result in the most negativeoutcomes.

Hypotheses regarding the adverse effects ofvictimization are straightforward, but we alsoconsidered the effects of perpetration using par-allel analyses. We tentatively offered hypothe-ses similar to those for victimization, reasoningthat recognizing that one has aggressed towardone’s partner should also be distressing. Perpe-trators might experience guilt, shame, or re-morse for their actions, increasing their negativemood. Because enacting aggression harms therelationship, we might also expect perpetratorsto report poorer relationship functioning follow-ing aggression (e.g., Leopold, 2012). We testedthe following hypotheses:

Hypothesis 1a: Victims will experienceworse mood after reporting verbal aggres-sion (both-partner/victim-only) than after notreporting (perpetrator-only/neither-partner).

418 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Hypothesis 1b: Perpetrators will experienceworse mood after reporting verbal aggression(both-partner/perpetrator-only) than after notreporting (victim-only/neither-partner).

Hypothesis 2a: Victims will experience poorerrelationship functioning after reporting aggres-sion (both-partner/victim-only) than after notreporting (perpetrator-only/neither-partner).

Hypothesis 2b: Perpetrators will experi-ence poorer relationship functioning afterreporting verbal aggression (both-partner/perpetrator-only) than after not reporting(victim-only/neither-partner).

Exploratory Analyses

Many studies have considered the role ofindividual characteristics in interpretation ofpartner behavior (e.g., Campbell et al., 2005;Downey et al., 1998; Murray et al., 2003), butonly a few have considered the role of individ-ual characteristics on agreement regarding theoccurrence of partner aggression. It is likely thatany characteristic that increases people’s sensi-tivity to negative partner behavior will increasethe likelihood that such behavior is seen asaggressive. For example, women report bothmore victimization and perpetration than is cor-roborated by their partner (e.g., Testa & Der-rick, 2014; Testa et al., 2012). Similarly, peoplewith lower relationship satisfaction report re-ceiving more psychological aggression thantheir partner reports perpetrating (Marshall,Panuzio, Makin-Byrd, Taft, & Holtzworth-Munroe, 2011). We examined individual char-acteristics that might be associated with greatersensitivity to perceiving behavior as aggressive,and thus, a greater likelihood of disagreeingregarding the occurrence of aggression. Specif-ically, we examine gender, marital satisfaction,depressive symptomatology, emotion regula-tion, and aggressive anger expression.

Method

Participants

Participants included 118 married and cohab-iting couples, between the ages of 21 and 45.They were initially recruited for a laboratorystudy involving alcohol administration (Testa,Crane, Quigley, Levitt, & Leonard, 2014)

through advertisements and targeted householdscreening. To be eligible, both partners drankfour or more drinks on an occasion at least onceper month; neither partner reported medicalcontraindications to alcohol consumption ormet criteria for alcohol dependence; and neitherpartner reported extremely severe partner vio-lence (e.g., use of weapon, injury requiringmedical attention). Additional details regardingrecruitment and screening are provided in Testaand Derrick (2014) and Testa et al. (2014).

Men averaged 33.9 (SD � 6.8) and women32.7 (SD � 6.9) years of age. Most participantswere white (91.5% of men and 95.8% ofwomen), had at least some college education(87% of men and 95% of women), and wereemployed at least part time (91.5% of men and77.1% of women). The median household in-come was $60,000 to $75,000. Most coupleswere married (76% vs. 24% cohabiting), andhad been living together for an average of 6.3years (SD � 5.1). The majority (63%) had chil-dren. Of those with children, the median num-ber was 2.

Procedures

Couples were mailed a packet of self-reportquestionnaires that included relationship satis-faction, emotion regulation, depression, and an-ger. All questionnaires used in the current studyare published measures with established valid-ity. They completed the packet and returned itthrough the mail before a 45-min training ses-sion for the daily report study at the university.After the training session, couples completeddaily reports using interactive voice responsetechnology (IVR) for 56 days. Participantscalled the IVR system each day to completetheir reports. The questions were asked viacomputer recording, and participants respondedusing the keypad on their phone. Participantswere instructed to complete their reports at ap-proximately the same time each day. Each part-ner was compensated $1 for each report, $10 foreach complete week, and a $30 bonus for 8complete weeks. Participants were permitted tosubmit late entries (resulting in nearly 100% ofdays complete), but mood and relationshipfunctioning were assessed only for the 87.9%and 87.2% of reports completed on time by menand women, respectively. Additional details re-garding procedures and compliance are pro-

419SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

vided in Testa and Derrick (2014). This studyreceived human subjects approval from the Uni-versity’s Social and Behavioral Sciences Insti-tutional Review Board.

Daily Report Measures

Each day, participants completed a series ofquestions regarding mood, relationship func-tioning, positive events, conflict events, and al-cohol use (see Testa & Derrick, 2014, for moredetails). The mood and relationship functioningitems referred to the day participants were com-pleting the reports, given their temporal insta-bility. Because we expected most couple inter-actions would occur in the evenings, possiblyafter participants had completed their reports forthat day, we assessed reports of conflict andaggression for the previous day, allowing us totest the hypothesized temporal ordering (Dayt�1 aggression predicting Day t outcomes).

Negative mood. Participants were asked toindicate, on a scale from 1 (not at all) to 5 (verymuch), the extent to which they currently feltirritable, stressed out, angry, anxious, over-whelmed, and sad. Items were averaged to cre-ate the overall score (� � .92 for both husbandsand wives). Final scores ranged from 1.00 to5.00 (M � 1.60, SD � 0.74 for men, and M �1.48, SD � 0.66 for women), with higher scoresindicating greater negative mood.

Negative relationship perceptions. Parti-cipants responded to two items on a scale from1 (not at all) to 5 (very much): “Today, howangry or irritated do you feel toward yourspouse or partner?”; “Today, how much did youand your spouse or partner argue or disagree?”The two items were averaged to create the over-all score (� � .77 for husbands, � � .78 forwives). Final scores ranged from 1.00 to 5.00(M � 1.30, SD � 0.61 for men, and M � 1.28,SD � 0.61 for women), with higher scoresindicating greater negative relationship percep-tions.

Verbal aggression. Each day, participantswere asked “At any time yesterday, did you andyour partner have a conflict, argument, or dis-agreement, whether major or minor?” If so, theyindicated whether the conflict included severalbehaviors derived from the CTS-2 (Straus et al.,1996). Each behavior was asked twice to assessbehaviors directed toward the partner and thepartner’s behavior toward the self. A positive

response to yelled, insulted, or made threats wasscored as verbal aggression (i.e., a score of 1);the lack of a positive response was scored as theabsence of verbal aggression (i.e., a score of 0).A positive response to threw/kicked/hit some-thing or pushed/grabbed/hit was scored as phys-ical aggression; no positive response was scoredas the absence of physical aggression. Becauseof the low rates of physical aggression in thiscommunity sample (only 32 male-perpetratedevents and 40 female-perpetrated events; seeTesta & Derrick, 2014), physical aggression isnot analyzed in the current study.

Self-Report Questionnaires

Marital satisfaction. We assessed maritalsatisfaction using the 32-item Dyadic AdjustmentScale (DAS; � � .90 [men] and .93 [women];Spanier, 1976). Items were summed to create theoverall scale. Final scores ranged from 28 to 145(M � 113.02, SD � 13.58 for men, M � 113.17,SD � 15.92 for women), with higher scores indi-cating greater satisfaction.

Depression. Depressive symptomatologywas assessed using the 20-item Center forEpidemiologic Studies – Depression Scale(CES-D; � � .88 [men] and .90 [women];Radloff, 1977). Participants responded on ascale from 1 (rarely or none of the time) to 4(most or all of the time). Items were averagedto create the overall scale. Final scores rangedfrom 1.00 to 3.40 (M � 1.47, SD � 0.37 formen, M � 1.44, SD � 0.40 for women), withhigher scores indicating greater depressivesymptomatology.

Difficulties in emotion regulation. Chronicproblems with emotion regulation were as-sessed using the 36-item Difficulties in EmotionRegulation Scale (DERS; � � .75 [men] and.78 [women]; Gratz & Roemer, 2004). Partici-pants responded on a scale from 1 (almostnever) to 5 (almost always). Items were aver-aged to create the overall scale. Final scoresranged from 1.03 to 3.81 (M � 1.97, SD � 0.47for men, M � 1.86, SD � 0.53 for women),with higher scores indicating greater difficultyregulating emotion.

Aggressive anger expression. The State-Trait Anger Expression Inventory-2 (STAXI-2;Spielberger, 1999) assesses the propensity toexperience anger and methods of handling an-ger. We focus in the current study on the 8-item

420 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Anger-Expression-Out subscale (AX-O). TheAX-O assesses tendencies toward aggressiveexpression of feelings of anger (� � .76 formen and .64 for women). Participants re-sponded on a scale from 1 (almost never) to 4(almost always). Items were summed to createthe overall scale. Summed scores ranged from 9to 28 (M � 14.70, SD � 3.41 for men, M �13.94, SD � 2.79 for women).

Results

Daily Diary Analyses: Verbal Aggressionand Negative Consequences

To analyze the couple daily diary data andexamine Hypotheses 1a–2b, we conducted mul-tilevel modeling analyses using the multivariatefeature in MLwiN 2.28 (Rasbash, Browne,Healy, Cameron, & Charlton, 2013). We used atwo-level nested structure: within couple as-sessments of verbal aggression (neither-partner/victim-only/perpetrator-only/both-partner) eachday made up Level 1, couple made up Level 2,and male and female reports of negative conse-quences the following day were treated as mul-tivariate outcomes. This approach estimates twosets of equations simultaneously, one set formale outcomes and one set for female out-comes, controlling for the correlation betweenthe two. The method is statistically equivalentto other methods of analyzing dyadic daily diarydata (Kenny, Kashy, & Cook, 2006), and hasbeen used in prior research (Derrick, Leonard,& Homish, 2013; Testa & Derrick, 2014). Thismethod allows for straightforward tests of gen-der differences (a 1 df �2 test). When the effectsdid not differ significantly by gender, we pooledthe coefficients; otherwise, we report the resultsfor men and women separately. We ran each setof analyses twice, once examining victim out-comes (male and female) and once examiningperpetrator outcomes (male and female). Weestimated the following equations simultane-ously:

YijMt � �0jM � �1jMYt�1 � �2jMPt�1

� �3jMDt � �4jMBRt�1 � �5jMVRt�1

� �6jMPRt�1 � u7jM (1)

YijWt � �0jW � �1jWYt�1 � �2jWPt�1 � �3jWDt

� �4jWBRt�1 � �5jWVRt�1 � �6jWPRt�1

� u7jW (2)

�0jM � �00M � �01MTBRM � �02MTVRM

� �03MTPRM � r0jM (3)

�0jF � �00W � �01WTBRW � �02WTVRW

� �03WTPRW � r0jW (4)

Equation 1 represents the Level 1 (i.e., dailylevel) effects on men’s outcomes (M); Equation2 represents the Level 1 effects on women’soutcomes (W). We predicted one day’s out-comes (Y) from a random intercept term (�0j,the average level of that outcome for that dayfor couple j), yesterday’s report on that outcome(�1j, to control for autocorrelation), the part-ner’s report on that outcome yesterday (�2j, tocontrol for partners’ interdependence), day ofthe week (�3j, dummy-coded), three dummy-codes representing today’s reports of yester-day’s aggression (both-partner reports [�4j],victim-only reports [�5j], and perpetrator-onlyreports [�6j]), and an error term (u7j) that re-flects each person’s deviation from his or hermean over time. All Level 1 slopes were treatedas fixed effects. Dummy-coded Level 1 predic-tors were left uncentered, and continuous Level1 predictors were person mean centered. Miss-ing Level 1 data were considered to be missingat random (MAR) and handled using maximumlikelihood estimation.

Equations 3 and 4 represent the Level 2 (i.e.,couple level) effects on male outcomes andfemale outcomes, respectively. These equationscontain an intercept term (�00, the average levelof that outcome for the sample), the total num-ber of both-partner reports (�01), victim-onlyreports (�02), and perpetrator-only reports (�03)per person over the diary period (to control forbetween-couple variability), and an error term(r0j) that reflects the deviation of each couple’soutcome from the average level of that outcomefor the overall sample. The Level 2 predictorswere grand mean centered.

Victim negative mood. Results for analy-ses examining the victim’s negative mood arepresented in the first column of Table 1. Thosecoefficients that differed significantly by gender

421SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Tab

le1

Per

petr

ator

and

Vic

tim

Out

com

esas

aF

unct

ion

ofV

erba

lA

ggre

ssio

n

Vic

timne

gativ

em

ood

Perp

etra

tor

nega

tive

moo

dV

ictim

nega

tive

rela

tions

hip

perc

eptio

nsPe

rpet

rato

rne

gativ

ere

latio

nshi

ppe

rcep

tions

b95

%C

Ib

95%

CI

b95

%C

Ib

95%

CI

Inte

rcep

t1.

431 M

���

[1.3

49,1

.513

] M1.

432 M

���

[1.3

51,1

.513

] M1.

108�

��

[1.0

74,1

.142

]1.

106�

��

[1.0

71,1

.140

]1.

310 W

���

[1.2

46,1

.373

] W1.

306 W

���

[1.2

43,1

.368

] WC

ovar

iate

sL

agge

dD

V–

vict

im.0

14[�

.003

,.03

2].0

13[�

.004

,.03

0]�

.048

���

[�.0

66,�

.031

].0

40���

[.02

2,.0

57]

Lag

ged

DV

–pe

rpet

rato

r.0

13[�

.004

,.03

1].0

14[�

.004

,.03

1].0

41���

[.02

3,.0

58]

�.0

49���

[�.0

66,�

.031

]W

eekd

ayvs

.w

eeke

nd�

.099

���

[�.1

26,�

.071

]�

.099

���

[�.1

27,�

.071

]�

.014

[�.0

40,.

012]

�.0

14[�

.040

,.01

1]T

otal

both

-par

tner

repo

rts

�.0

05[�

.052

,.04

2].1

03M

��

[.03

2,.1

73] M

.009

[�.0

21,.

038]

.070

M���

[.02

9,.1

11] M

�.0

39W

[�.0

89,.

011]

W�

.002

W[�

.034

,.03

0]W

Tot

alvi

ctim

-onl

yre

port

s.0

50���

[.02

6,.0

74]

�.0

04[�

.030

,.02

3].0

73M

���

[.04

4,.1

01] M

�.0

03[�

.019

,.01

3].0

26W

��

[.01

0,.0

43] W

Tot

alpe

rpet

rato

r-on

lyre

port

s�

.007

[�.0

35,.

022]

.042

���

[.01

8,.0

67]

�.0

02[�

.019

,.01

4].0

29���

[.01

4,.0

45]

Lev

el1

verb

alag

gres

sion

Nei

ther

-par

tner

repo

rts

——

——

——

——

Bot

h-pa

rtne

rre

port

s.3

78���

[.27

1,.4

85]

.367

���

[.26

0,.4

73]

.547

���

[.44

8,.6

46]

.609

���

[.51

0,.7

08]

Vic

tim-o

nly

repo

rts

.240

���

[.16

0,.3

21]

.051

[�.0

30,.

132]

.391

���

[.31

7,.4

65]

.129

���

[.05

5,.2

02]

Perp

etra

tor-

only

repo

rts

.013

[�.0

68,.

094]

.276

���

[.19

6,.3

56]

.126

���

[.05

3,.2

00]

.399

���

[.32

5,.4

72]

Not

e.T

hepo

ssib

lera

nge

for

nega

tive

moo

dan

dne

gativ

ere

latio

nshi

ppe

rcep

tions

ison

lyfr

om1

to5,

and

thus

the

test

ofth

ein

terc

ept’

ssi

gnifi

canc

eis

notm

eani

ngfu

l.W

eekd

ayis

adi

chot

omou

sva

riab

lesc

ored

0�

Mon

day

–T

hurs

day,

1�

Frid

ay–

Sund

ay.

Lag

ged

DV

refe

rsto

depe

nden

tva

riab

les

onD

ayt�

1.T

otal

both

-par

tner

repo

rts,

vict

im-o

nly

repo

rts,

and

perp

etra

tor-

only

repo

rts

are

gran

dm

ean

cent

ered

tota

lsre

cord

edfo

rea

chco

uple

duri

ngth

e56

-day

diar

ype

riod

.Lev

el1

verb

alag

gres

sion

isa

dum

my-

code

dca

tego

rica

lva

riab

lere

flect

ing

agre

emen

tin

verb

alag

gres

sion

onD

ayt�

1.N

eith

er-p

artn

erre

port

sse

rves

asth

ere

fere

nce

grou

p.C

oeffi

cien

tsth

atdi

ffer

edsi

gnifi

cant

lyby

gend

erar

em

arke

dw

ithsu

bscr

ipts

.M

�m

en;

W�

wom

en;

b�

unst

anda

rdiz

edre

gres

sion

coef

ficie

nt;

95%

CI

�95

%co

nfide

nce

inte

rval

.��

p�

.01.

���

p�

.001

.

422 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

are presented separately with subscripts desig-nating male and female values; otherwise, co-efficients were pooled across gender. Predictedvalues, pooled across gender, are depicted inFigure 1 in dark bars. Reports in which partnersagree regarding the occurrence of verbal aggres-sion are presented in the panel on the left;reports in which partners disagree are presentedin the panel on the right. Hypothesis 1a statesthat victims will experience greater negativemood on days when they report receiving verbalaggression the previous day (both-partner/victim-only) than on days when they do not(perpetrator-only/neither-partner). As can beseen in Table 1 and Figure 1, both-partner re-ports and victim-only reports predicted signifi-cantly greater victim negative mood than nei-ther-partner reports. Negative mood did notdiffer for perpetrator-only reports and neither-partner reports. In follow-up analyses with re-coded dummy variables, both-partner reportspredicted greater victim negative mood thanvictim-only reports, b � .137, 95% CI � [.007,.267], p � .038, but both-partner reports andvictim-only reports predicted significantlygreater victim negative mood than perpetrator-only reports, b � .365, 95% CI � [.235, .495],

p � .001, and b � .228, 95% CI � [.125, .331],p � .001, respectively. In other words, Hypoth-esis 1a was supported; victims reported worsemood on days when they reported receivingaggression the previous day. However, eventsthat both partners reported predicted the worstconsequences.

Perpetrator negative mood. Next we ex-amined perpetrator’s negative mood. Resultsare presented in the second column of Table 1.Predicted values are depicted in Figure 1 in lightbars. Hypothesis 1b states that perpetrators willexperience greater negative mood on days whenthey report perpetrating verbal aggression theprevious day (both-partner/perpetrator-only)than on days when they do not (victim-only/neither-partner). As can be seen in Table 1 andFigure 1, both-partner reports and perpetrator-only reports predicted significantly greaterperpetrator negative mood than neither-partnerreports. Negative mood did not differ for vic-tim-only reports and neither-partner reports. Infollow-up analyses, both-partner reports did notdiffer significantly from perpetrator-only re-ports, b � .091, 95% CI � [�.037, .219], p �.165. However, both-partner reports and perpe-trator-only reports were associated with signif-

Figure 1. Predicted values for analyses examining the effect of one day’s verbal aggressionon the next day’s negative mood (pooled across gender). The dark bars portray effects onvictim negative mood. The light bars portray effects on perpetrator negative mood. Reports inwhich partners agree regarding the occurrence of verbal aggression are depicted on the left.Reports in which partners disagree regarding the occurrence of verbal aggression are depictedon the right. Negative mood was assessed on a scale from 1 (not at all) to 5 (very much).

423SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

icantly greater perpetrator negative mood thanvictim-only reports, b � .316, 95% CI � [.186,.445], p � .001, and b � .225, 95% CI � [.122,.328], p � .001, respectively. In other words,Hypothesis 1b was supported. Perpetrators re-ported worse mood on days when they reportedperpetrating aggression the previous day. In thisanalysis, both-partner reports did not have morenegative consequences than perpetrator-only re-ports.

Victim negative relationship perceptions.Results for analyses examining the victim’snegative relationship perceptions are presentedin the third column of Table 1. Predicted valuesare presented in Figure 2 in dark bars. Hypoth-esis 2a states that victims will experiencegreater negative relationship functioning ondays when they report receiving verbal aggres-sion the previous day (both-partner/victim-only) than on days when they do not (perpetra-tor-only/neither-partner). As can be seen inTable 1 and Figure 2, both-partner, victim-only,and perpetrator-only reports were all associatedwith significantly greater victim negative rela-tionship perceptions than neither-partner re-ports. Follow-up analyses revealed that victims

reported more negative relationship perceptionson days after both-partner reports and victim-only reports than on days after perpetrator-onlyreports, b � .421, 95% CI � [.303, .539], p �.001, and b � .264, 95% CI � [.173, .356], p �.001, respectively. Both-partner reports werealso associated with significantly greater victimnegative relationship perceptions than victim-only reports, b � .157, 95% CI � [.038, .275],p � .01. In other words, Hypothesis 2a wassupported; victims reported perceiving their re-lationship more negatively on days when theyreported receiving aggression the previous day.Again, events that both partners reported pre-dicted the worst consequences. Although perpe-trator-only reports were associated with worserelationship functioning than neither-partner re-ports, the association was considerably weaker.

Perpetrator negative relationship perceptions.We repeated the analyses to examine the perpe-trator’s negative relationship perceptions (seeTable 1, fourth column). Predicted values arepresented in Figure 2 in light bars. Hypothesis2b states that perpetrators will experiencegreater negative relationship functioning ondays when they report perpetrating verbal ag-

Figure 2. Predicted values for analyses examining the effect of one day’s verbal aggressionon the next day’s negative relationship perceptions. The dark bars portray effects on victimrelationship perceptions. The light bars portray effects on perpetrator relationship perceptions.Reports in which partners agree regarding the occurrence of verbal aggression are depicted onthe left. Reports in which partners disagree regarding the occurrence of verbal aggression aredepicted on the right. Negative relationship perceptions were assessed on a scale from 1 (notat all) to 5 (very much).

424 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

gression the previous day (both-partner/perpetrator-only) than on days when they do not(victim-only/neither-partner). As can be seen inTable 1 and Figure 2, both-partner, perpetrator-only, and victim-only reports were all associ-ated with significantly greater perpetrator negativerelationship perceptions than neither-partner re-ports. Follow-up analyses revealed that perpetra-tors reported significantly greater negative rela-tionship perceptions for both-partner reports andperpetrator-only reports than for victim-only re-ports, b � .480, 95% CI � [.362, .598], p � .001,and b � .270, 95% CI � [.179, .361], p � .001,respectively. Both-partner reports were also as-sociated with significantly greater perpetratornegative relationship perceptions than perpetra-tor-only reports, b � .210, 95% CI � [.092,.328], p � .001. In other words, Hypothesis 2bwas supported, with results appearing as a mir-ror image of victimization findings. Perpetratorsreported perceiving their relationship more neg-atively on days when they reported perpetratingaggression the previous day. Events that bothpartners reported predicted the worst conse-quences. Although victim-only reports were as-sociated with worse relationship functioningthan neither-partner reports, the association wasconsiderably weaker.

Mutual Aggression

We hypothesized that the effects for perpe-tration would mirror those for victimization,reasoning that recognizing that one has perpe-trated aggression should be distressing. Yet vic-timization and perpetration often co-occur (e.g.,Straus, 2008), and this is the case for verbalaggression in the current data. Men reportedonly victimization on 77 days, only perpetrationon 25 days, and mutual aggression on 187 days.Women reported only victimization on 48 days,only perpetration on 98 days, and mutual ag-gression on 251 days. Given the high rates ofmutual aggression, it may be the case that thenegative short-term consequences for perpetra-tion observed here are an artifact of the associ-ation with victimization. To examine this pos-sibility, we conducted supplemental analysesexamining the effects of perpetration-only, vic-timization-only, and mutual aggression onmood and relationship perceptions. Mood andrelationship perceptions were significantlyworse when any type of aggression was re-

ported for the previous day than when it wasnot, all ps � .001. The effects for perpetration-only and victimization-only did not differ sig-nificantly from mutual aggression, all ps .223. Therefore, mutual aggression cannot com-pletely account for the association between per-petration and negative consequences in the cur-rent study.

Exploratory Analyses: IndividualDifferences in Perception

To better understand why partners might inter-pret the same event differently, we examined in-dividual characteristics that might influence sen-sitivity to negative behavior. Specifically, weexamined gender, marital satisfaction, depression,emotion regulation, and aggressive anger expres-sion as correlates of partner agreement regardingaggression. We conducted paired sample t tests todetermine whether the total number of episodes ofverbal aggression reported over the diary perioddiffered by gender. We found that there weresignificantly more victim-only reports of maleperpetration (vs. female perpetration), t(117) ��2.26, p � .026, and perpetrator-only reports offemale perpetration (vs. male perpetration),t(117) � 3.50, p � .001. In other words, as pre-viously described in Testa and Derrick (2014)using a different form of analysis, women reportedmore of both victimization and perpetration thanmen.

We examined associations with the continu-ous variables by conducting multivariate analy-ses in MLwiN 2.28 (Rasbash et al., 2013). Eachcouple’s total number of both-partner reports,victim-only reports, and perpetrator-only re-ports of verbal aggression over the 56-day diaryperiod were entered simultaneously to predicteach individual characteristic. We included thetotal number of conflicts reported by men andwomen as a covariate to control for individualdifferences in the likelihood of experiencingconflict. To be conservative, missing data in thediary were treated as zeroes when creating totalconflict and aggression reports. One husbands’report of depression was missing, but there wereno other missing data on the total scores for theself-report measures. Accordingly, we used list-wise deletion.

First, we describe correlates of greater sen-sitivity to negative behavior (i.e., victim char-acteristics that predict victim-only reports and

425SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

perpetrator characteristics that predict perpe-trator-only reports). There were more days onwhich the victim but not the perpetrator re-ported verbal aggression when victims wereless satisfied with their relationship, b � �1.988, 95% CI � [�2.826 �1.150], p � .001,were higher in depression, b � .049, 95%CI � [.022, .076], p � .001, had more diffi-culties in emotion regulation, b � .052, 95%CI � [.019, .085], p � .002, and were higherin aggressive anger expression, b � .261,95% CI � [.073, .449], p � .007. There weremore days on which the perpetrator but notthe victim reported aggression when perpetra-tors were less satisfied with their relationship,b � �1.280, 95% CI � [�2.428, �0.132],p � .029, were higher in depression, b �.048, 95% CI � [.011, .085], p � .012, andhad more difficulties in emotion regulation,b � .046, 95% CI � [.001, .091], p � .046.None of the effects for greater sensitivitydiffered by gender.

Next, we describe correlates of experiencingweaker sensitivity to negative behavior (i.e.,perpetrator characteristics that predict victim-only reports and victim characteristics that pre-dict perpetrator-only reports). There were moredays on which only the victim reported aggres-sion when perpetrators were less satisfied withtheir relationship, b � �1.429, 95% CI � [�2.190, �0.668], p � .001, when male (but notfemale) perpetrators had more difficulties inemotion regulation, b � .068, 95% CI � [.035,.101], p � .001, and when perpetrators werehigher in aggressive anger expression, b � .220,95% CI � [.018, .422], p � .033. There weremore days on which only the perpetrator re-ported aggression when victims were less satis-fied with their relationship, b � �1.071, 95%CI � [�2.127, �0.014], p � .047, had moredifficulties in emotion regulation, b � .044,95% CI � [.001, .087], p � .046, and werehigher in aggressive anger expression, b � .415,95% CI � .133, .697], p � .004. Only oneeffect (perpetrator emotion regulation) differedby gender. In brief, correlates of both types ofpartner discrepancy, greater and weaker sensi-tivity, were similar.

Discussion

The partner violence field has long recog-nized that partner reports of intimate partner

aggression do not show perfect agreement. Yet,there has been a lack of research into the rea-sons for and consequences of these disagree-ments. In the current study, we examined theconsequences of verbal aggression, reported byboth partners in a daily diary study, for next-dayindividual and relationship well-being. Findingsreveal that the short-term consequences of ver-bal aggression depend upon recognition andreporting of its occurrence. As predicted, vic-tims reported worse mood (Hypothesis 1a) andpoorer relationship functioning (Hypothesis 2a)on days when they reported verbal aggressionthe previous day (both-partner/victim-only)than on days when they did not (perpetrator-only/neither-partner). It was particularly strik-ing that victim mood on days when perpetratorsreported aggression the previous day and vic-tims did not (perpetrator-only reports) was nomore negative than days when neither partnerreported aggression the previous day. Unless avictim recognized and reported receiving verbalaggression, it had no discernible effect onmood.

Although several studies have documentedthe effects of partner aggression on victims’outcomes (e.g., Lawrence et al., 2009), its ef-fects on perpetrators have rarely been consid-ered. The current study’s dyadic design allowedus to consider the effects of aggression on theshort-term outcomes of perpetrators as well asvictims. We found that the pattern we observedfor victims was mirrored for perpetrators. Thatis, perpetrators reported worse mood (Hypoth-esis 1b) and poorer relationship functioning(Hypothesis 2b) when they reported verbal ag-gression (both-partners/perpetrator-only) thanwhen they did not (victim-only/neither-partner).We examined the possibility that these associ-ations were actually driven by victimization,given that mutual aggression was common. Wewere able to rule out this possibility, suggestingthat within this community sample, recognizingthat one has perpetrated aggression is distress-ing. It may be the case that perpetrators expe-rience guilt, shame, or remorse for their actions,increasing their negative mood. Additionally,they may report poorer relationship functioningbecause enacting aggression harms the relation-ship (e.g., Leopold, 2012).

We also found that events in which partnersagreed that verbal aggression had occurred re-sulted in particularly negative consequences.

426 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Victims experienced worse mood and poorerrelationship functioning when both partnersagreed that aggression had occurred than whenit was reported by only the victim. Similarly,perpetrators experienced poorer relationshipfunctioning when both partners agreed that ag-gression had occurred than when it was reportedby only the perpetrator. Thus, in three of foursets of analyses, participants experienced worseconsequences when both partners reported ag-gression than when only they, themselves re-ported that aggression had occurred the previ-ous day. Because we do not know why theseincidents predict worse outcomes, obtaining re-ports from only one partner or combining re-ports from both partners may result in the lossof important information.

Consistent with prior research suggesting thatindividuals with certain characteristics may beprone to greater sensitivity to conflict andnegative behavior (e.g., Campbell et al., 2005;Murray et al., 2003), we found several charac-teristics that predicted victim-only and perpe-trator-only reports: gender, depression, maritaldissatisfaction, poor emotion regulation, andaggressive anger expression. As reported previ-ously (Testa & Derrick, 2014), gender was dif-ferentially associated with sensitivity to nega-tive behavior in that women reported both morevictimization and more perpetration than men.Although we found these differences in terms offrequencies or mean levels of aggression, wedid not find any gender differences in terms ofthe associations between reports of aggressionand short-term negative consequences (i.e., nomoderation). Thus, even though we have repli-cated the work of previous research showingthat women report more aggression than men(e.g., Schafer et al., 2002; Testa et al., 2012), wealso show that men are not necessarily lessaffected by aggression when they recognize it assuch. Consistent with our exploratory hypothe-ses, depression was also associated with greatersensitivity (both victim and perpetrator) andwas not associated with weaker sensitivity (nei-ther victim nor perpetrator) to aggressive acts.These findings are consistent with previous re-search demonstrating that depression is associ-ated with greater bias for negative information(Mogg, Bradley, & Williams, 1995; Murphy etal., 1999). It is likely that people with greaterdepression have a lower threshold for labelingan ambiguous act as aggressive.

Unexpectedly, and contrary to previous re-search (Marshall et al., 2011), our results indi-cated that the other characteristics tended topredict both types of mismatches: reporting ag-gression when one’s partner did not and notreporting aggression when one’s partner did.Specifically, marital dissatisfaction and pooremotion regulation were associated with bothgreater and weaker victim sensitivity and bothgreater and weaker perpertrator sensitivity toaggressive acts. Aggressive anger expressionwas associated with both greater and weakervictim sensitivity and weaker (but not greater)perpetrator sensitivity to aggressive acts. Thus,people with these characteristics consistentlyinterpret events in a different manner from theirpartner. Rather than experiencing differentthresholds for interpretation (as with depres-sion), people in these relationships seem to be“out of sync” with each other. Perhaps thesecharacteristics impact the ability to accuratelyperceive and interpret the thoughts and emo-tions of others, leading people with these char-acteristics both to misinterpret their partner’sintentions and to blithely ignore how their ownbehavior affects their partner.

Interpretation, Memory,or Social Desirability?

The ability to report aggression depends on(a) interpretation and encoding of an event asaggression, (b) retrieval of the event from mem-ory, and (c) overcoming social desirability con-cerns surrounding the reporting of aggression.All three conditions have likely influenced par-ticipants’ reports of aggression in prior re-search. Throughout this article, we have dis-cussed partner agreement regarding theoccurrence of aggression as though interpreta-tion/encoding is the primary mechanism bywhich disagreement occurs. We can have someconfidence that, at least within our low violencecommunity sample, disagreement is indeed aresult of differential interpretation/encoding.First, we found that several individual charac-teristics were associated with both victim-onlyand perpetrator-only reports of aggression.Whether these associations are attributable todifferent thresholds (e.g., depression) or differ-ent person perception skills (e.g., difficultieswith emotion regulation), these characteristicsare associated with differential interpretation of

427SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

an event. Therefore, people with these charac-teristics appear to “overreport” and “underre-port” aggression relative to their partner, con-tributing to disagreement regarding theoccurrence of aggression.

Second, disagreement in the current studylikely is not solely attributable to forgetting.Obtaining daily reports of partner aggressionhas been recommended as a way of minimizingforgetting and increasing reporting accuracyrelative to retrospective surveys (e.g., Folling-stad & Rogers, 2013). Although participants inthe current study may have been required torecall events for the past 24–48 hours, thistimeframe is still a marked improvement oversurveys that require participants to recall eventsthat occurred over previous 1-, 6-, or 12-monthperiods. Finally, disagreement in the currentstudy likely is not attributable to social desir-ability. Participants only demonstrated negativeconsequences of aggression when they reportedthat aggression had occurred the previous day(whether their partner reported it or not). Ifsocial desirability were the primary mechanism,we would expect mood and relationship func-tioning to decline equally when either partnerreported aggression.

Although interpretation/encoding may be theprimary mechanism for disagreement regardingthe occurrence of aggression in our low vio-lence community sample, this likely is not al-ways the case. The behaviors assessed in thecurrent study could be somewhat ambiguous(yelling, threatening, insulting), allowing fordifferences in interpretation. Even physicallyaggressive behaviors like pushing and grabbingmay be somewhat ambiguous, and subject tointerpretation. More severe forms of aggression,such as choking or using a weapon, are lessambiguous and therefore should be interpretedand encoded as aggression. We cannot rule outthe possibility that discrepancies in partner re-ports of these more severe types of behaviorsreflect deliberate underreporting of aggression.

Limitations

Findings are based on a community sample ofcouples in which rates of conflict, verbal ag-gression, and particularly physical aggressionwere low. We were unable to consider physi-cally aggressive incidents within the currentframework because there were too few reports

(only 5 “both-partner reports” each for male-and female-perpetrated aggression). Therefore,the current findings may not generalize to phys-ical aggression. In addition, the sample wasoriginally recruited for an alcohol administra-tion study and hence, had to meet several selec-tion criteria (moderate drinking, not dependent,no medical contraindications, no extremely se-vere partner violence), which may have un-known effects on results. Thus, generalizabilityis uncertain and the findings need to be repli-cated in other community, and especially clini-cal samples. These results might not generalizeto battered women and their partners, for exam-ple.

Although daily diary methodology reducesretrospection, reports of aggression were madefor the previous day; thus, it is possible thatmemory degraded, and agreement might havebeen higher if reports were made at the time ofoccurrence. Furthermore, mood and relation-ship functioning at the time of reporting mayhave influenced recall of aggression on the pre-vious day, potentially inflating associations.Additionally, reports of aggression were madeonly if the participant reported a conflict, con-sistent with the wording and conceptualizationof the Conflict Tactics Scale. However, it ispossible that some experiences of verbal aggres-sion were not reported because the respondentdid not perceive that a conflict had occurred.

Another potential limitation is that the asso-ciations between victim reports and victim out-comes (or perpetrator reports and perpetratoroutcomes) were based on only one source,whereas the associations between perpetratorsreports and victim outcomes (or victim reportsand perpetrator outcomes) were based on twosources. The former associations are based onreports that share more method variance thanthe latter associations, potentially biasing theobserved associations (Orth, 2013). The extentto which shared method variance is a problem inany given study is often exaggerated, however(Spector, 2006). In the current study, we con-trolled for the total number of both-partner re-ports, victim-only reports, and perpetrator-onlyreports for each couple in all analyses. Accord-ingly, the most commonly cited biases for self-reports—acquiescence bias, negative affectiv-ity, and social desirability—are statisticallycontrolled. We cannot entirely rule out the con-tribution of shared method variance to the cur-

428 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

rent results, but this contribution is likely to beminimal.

Research Implications

Despite these limitations, our findings haveimportant implications for assessment of part-ner aggression, in particular, for the commonpractice of pooling partner reports of aggres-sion. Individual recognition that relationship ag-gression has occurred—distinct from partnerreports of aggression—is a critical determinantof short-term individual and relationship func-tioning. In light of these findings, the commonpractice of pooling partner reports requires care-ful consideration. If the goal is to estimate thetotal amount of aggressive behavior within acouple, or to predict the occurrence of aggres-sion within couples, pooling may be a reason-able approach. However, when considering theconsequences of verbal aggression for an indi-vidual, that individual’s perceptions, separatefrom those of the partner, may be preferable. Inaddition, many researchers obtain reports fromonly the victim. Although the victim’s report ofaggression might be considered more importantwhen identifying consequences of aggression,we find in the current study that incidents inwhich partners agree that aggression occurredhave more severe consequences than incidentsin which only the victim reports. Obtaining re-ports from both partners may be the best prac-tice even when the victim is the target of inter-est.

Clinical and Policy Implications

Although there are many research implica-tions of the current study, there are also poten-tial clinical implications for our findings. Spe-cifically, the results of the current study speakquite clearly to the fact that partners perceiveand experience events differently. When treat-ing distressed couples, therefore, it is importantto keep in mind that partners will not alwaysagree on what has occurred, possibly even whendiscussing relatively more objective events.Rather than reflecting defensiveness or willfuldistortion, the “underreporting” partner maysimply have experienced the event differently,and may in fact have experienced very fewnegative consequences for the event itself.

References

Anderson, K. L. (2002). Perpetrator or victim? Rela-tionships between intimate partner violence andwell-being. Journal of Marriage and Family, 64,851–863. doi:10.1111/j.1741-3737.2002.00851.x

Baldwin, M. W. (1992). Relational schemas and theprocessing of social information. PsychologicalBulletin, 112, 461–484. doi:10.1037/0033-2909.112.3.461

Caetano, R., Field, C., Ramisetty-Mikler, S., & Lip-sky, S. (2009). Agreement on reporting of physi-cal, psychological, and sexual violence amongWhite, Black, and Hispanic couples in the UnitedStates. Journal of Interpersonal Violence, 24,1318–1337. doi:10.1177/0886260508322181

Caetano, R., Schafer, J., Field, C., & Nelson, S. M.(2002). Agreement on reports of intimate partnerviolence among White, Black, and Hispanic cou-ples in the United States. Journal of InterpersonalViolence, 17, 1308 –1322. doi:10.1177/088626002237858

Campbell, L., Simpson, J. A., Boldry, J., & Kashy,D. A. (2005). Perceptions of conflict and support inromantic relationships: The role of attachmentanxiety. Journal of Personality and Social Psy-chology, 88, 510–531. doi:10.1037/0022-3514.88.3.510

Coker, A. L., Davis, K. E., Arias, I., Desai, S.,Sanderson, M., Brandt, H. M., & Smith, P. H.(2002). Physical and mental health effects of inti-mate partner violence for men and women. Amer-ican Journal of Preventive Medicine, 23, 260–268.doi:10.1016/S0749-3797(02)00514-7

Collins, N. L. (1996). Working models of attach-ment: Implications for explanation, emotion, andbehavior. Journal of Personality and Social Psy-chology, 71, 810–832. doi:10.1037/0022-3514.71.4.810

Derrick, J. L., Leonard, K. E., & Homish, G. G.(2013). Perceived partner responsiveness predictsdecreases in smoking during the first nine years ofmarriage. Nicotine & Tobacco Research, 15,1528–1536. doi:10.1093/ntr/ntt011

Downey, G., Freitas, A. L., Michaelis, B., & Khouri,H. (1998). The self-fulfilling prophecy in closerelationships: Rejection sensitivity and rejectionby romantic partners. Journal of Personality andSocial Psychology, 75, 545–560. doi:10.1037/0022-3514.75.2.545

Follingstad, D. R. (2009). The impact of psycholog-ical aggression on women’s mental health andbehavior: The status of the field. Trauma, Vio-lence, & Abuse, 10, 271–289. doi:10.1177/1524838009334453

Follingstad, D. R., & Rogers, M. J. (2013). Validityconcerns in the measurement of women’s andmen’s report of intimate partner violence. Sex

429SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Roles, 69, 149 –167. doi:10.1007/s11199-013-0264-5

Gratz, K. L., & Roemer, L. (2004). Multidimensionalassessment of emotion regulation and dysregula-tion: Development, factor structure, and initial val-idation of the difficulties in emotion regulationscale. Journal of Psychopathology and BehavioralAssessment, 26, 41–54. doi:10.1023/B:JOBA.0000007455.08539.94

Hamby, S. L. (2005). Measuring gender differencesin partner violence: Implications from research onother forms of violent and socially undesirablebehavior. Sex Roles, 52, 725–742. doi:10.1007/s11199-005-4195-7

Heyman, R. E., & Schlee, K. A. (1997). Toward abetter estimate of the prevalence of partner abuse:Adjusting rates based on the sensitivity of theConflict Tactics Scale. Journal of Family Psychol-ogy, 11, 332–338. doi:10.1037/0893-3200.11.3.332

Kelley, H. H. (1979). Personal relationships: Theirstructures and processes. Hillsdale, NJ: Erlbaum.

Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006).Dyadic data analysis. New York, NY: GuilfordPress.

Lawrence, E., Yoon, J., Langer, A., & Ro, E. (2009).Is psychological aggression as detrimental as phys-ical aggression? The independent effects of psy-chological aggression on depression and anxietysymptoms. Violence and Victims, 24, 20–35. doi:10.1891/0886-6708.24.1.20

Leopold, A. N. (2012). Enacted aggression and perpe-trator outcomes: The moderating roles of experi-enced aggression and perpetrator gender (Unpub-lished master’s thesis). Saint Mary’s University,Halifax, Nova Scotia.

Marshall, A. D., Panuzio, J., Makin-Byrd, K. N.,Taft, C. T., & Holtzworth-Munroe, A. (2011). Amultilevel examination of interpartner intimatepartner violence and psychological aggression re-porting concordance. Behavior Therapy, 42, 364–377. doi:10.1016/j.beth.2010.09.003

Mogg, K., Bradley, B. P., & Williams, R. (1995).Attentional bias in anxiety and depression: Therole of awareness. British Journal of Clinical Psy-chology, 34, 17–36. doi:10.1111/j.2044-8260.1995.tb01434.x

Murphy, F., Sahakian, B., Rubinsztein, J., Michael,A., Rogers, R., Robbins, T., & Paykel, E. (1999).Emotional bias and inhibitory control processes inmania and depression. Psychological Medicine,29, 1307–1321. doi:10.1017/S0033291799001233

Murray, S. L., Bellavia, G. M., Rose, P., & Griffin,D. W. (2003). Once hurt, twice hurtful: How per-ceived regard regulates daily marital interactions.Journal of Personality and Social Psychology, 84,126–147. doi:10.1037/0022-3514.84.1.126

O’Leary, K. D., & Williams, M. C. (2006). Agree-ment about acts of aggression in marriage. Journalof Family Psychology, 20, 656–662. doi:10.1037/0893-3200.20.4.656

Orth, U. (2013). How large are actor and partnereffects of personality on relationship satisfaction?The importance of controlling for shared methodvariance. Personality and Social Psychology Bul-letin, 39, 1359 –1372. doi:10.1177/0146167213492429

Radloff, L. S. (1977). The CES-D scale a self-reportdepression scale for research in the general popu-lation. Applied Psychological Measurement, 1,385–401. doi:10.1177/014662167700100306

Rasbash, J., Browne, W. J., Healy, M., Cameron, B.,& Charlton, C. (2013). MLwiN: Version 2.28. Lon-don, UK: Centre for Multilevel Modeling, Univer-sity of Bristol.

Reis, H. T., & Shaver, P. R. (1988). Intimacy as aninterpersonal process. In S. Duck, D. F. Hay, S. E.Hobfoll, W. Ickes, & B. M. Montgomery (Eds.),Handbook of personal relationships: Theory, re-search and interventions (pp. 367–389). Oxford,England: Wiley.

Schafer, J., Caetano, R., & Clark, C. L. (2002).Agreement about violence in U.S. couples. Jour-nal of Interpersonal Violence, 17, 457–470. doi:10.1177/0886260502017004007

Schwarz, N. (2012). Why researchers should thinkreal-time: A cognitive rationale. In M. R. Mehl &T. S. Conner (Eds.), Handbook of research meth-ods for studying daily life (pp. 22–42). New York,NY: Guilford Press.

Spanier, G. B. (1976). Measuring dyadic adjustment:New scales for assessing the quality of marriageand similar dyads. Journal of Marriage and theFamily, 38, 15–28. doi:10.2307/350547

Spector, P. E. (2006). Method variance in organiza-tional research: Truth or urban legend? Organiza-tional Research Methods, 9, 221–232. doi:10.1177/1094428105284955

Spielberger, C. D. (1999). State-Trait Anger Expres-sion Inventory–2 (STAXI-2) professional manual.Lutz, FL: Psychological Assessment Resources.

Srivastava, S., McGonigal, K. M., Richards, J. M.,Butler, E. A., & Gross, J. J. (2006). Optimism inclose relationships: How seeing things in a positivelight makes them so. Journal of Personality andSocial Psychology, 91, 143–153. doi:10.1037/0022-3514.91.1.143

Straus, M. A. (2008). Dominance and symmetry inpartner violence by male and female university studentsin 32 nations. Children and Youth Services Review, 30,252–275. doi:10.1016/j.childyouth.2007.10.004

Straus, M. A., Hamby, S. L., Boney-McCoy, S., &Sugarman, D. B. (1996). The Revised ConflictTactics Scales (CTS2): Development and prelimi-

430 DERRICK, TESTA, AND LEONARD

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

nary psychometric data. Journal of Family Issues,17, 283–316. doi:10.1177/019251396017003001

Sugarman, D. B., & Hotaling, G. T. (1997). Intimateviolence and social desirability: A meta-analyticreview. Journal of Interpersonal Violence, 12,275–290. doi:10.1177/088626097012002008

Szinovacz, M. E., & Egley, L. C. (1995). Comparingone-partner and couple data on sensitive maritalbehaviors: The case of marital violence. Journal ofMarriage and the Family, 57, 995–1010. doi:10.2307/353418

Testa, M., Crane, C. A., Quigley, B. M., Levitt, A., &Leonard, K. E. (2014). Effects of administeredalcohol on intimate partner interactions in a con-flict resolution paradigm. Journal of Studies onAlcohol and Drugs, 75, 249–258.

Testa, M., & Derrick, J. L. (2014). A daily processexamination of the temporal association between

alcohol use and verbal and physical aggression incommunity couples. Psychology of Addictive Be-haviors, 28, 127–138. doi:10.1037/a0032988

Testa, M., Kubiak, A., Quigley, B. M., Houston,R. J., Derrick, J. L., Levitt, A., . . . Leonard, K. E.(2012). Husband and wife alcohol use as indepen-dent or interactive predictors of intimate partnerviolence. Journal of Studies on Alcohol and Drugs,73, 268–276.

Testa, M., & Leonard, K. E. (2001). The impact ofmarital aggression on women’s psychological andmarital functioning in a newlywed sample. Journalof Family Violence, 16, 115–130. doi:10.1023/A:1011154818394

Received September 3, 2013Revision received May 14, 2014

Accepted June 16, 2014 �

431SPECIAL ISSUE: INTIMATE PARTNER VERBAL AGGRESSION BY SELF AND PARTNER

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.