How Much Positivity Is Needed to Buffer the Impact of Parental Negativity on Children?

14
Martina Zemp University of Zurich Christine E. Merrilees State University of New York at Geneseo Guy Bodenmann University of Zurich ∗∗ How Much Positivity Is Needed to Buffer the Impact of Parental Negativity on Children? Previous research suggests that a couple’s abil- ity to balance adverse interactions with positiv- ity is a better predictor of relationship outcomes than negativity per se. The purpose of this study was to extend the findings on couples to par- ents by examining the positive-to-negative ratio in parental interaction as a predictor of child adjustment. Three hundred seventy-five parents completed questionnaires on the negative and positive interactions in their intimate relation- ship and their child’s functioning online. Differ- ent interparental positive – negative ratios were identified by latent class analysis and then linked to children’s adjustment. Children whose par- ents were characterized by more negativity rel- ative to positivity scored higher in externalizing problems compared to children whose parents’ negativity was outbalanced by positive interac- tions. Girls with parents yielding the highest positive – negative ratios were reported to show most prosocial behavior. The findings suggest that parental positivity should be expressed at Department of Psychology, University of Zurich, Binz- muehlestrasse 14/23, 8050 Zurich, Switzerland (martina. [email protected]). Department of Psychology, State University of New York at Geneseo, 1 College Circle, Geneseo, NY 14454. ∗∗ Department of Psychology, University of Zurich, Binz- muehlestrasse 14/23, 8050 Zurich, Switzerland. Key Words: interparental conflict, balance theory, positive– to-negative ratio, family relations, emotional security. least twice as much as parental negativity for the sake of children’s well-being. Introduction Marital conflict has been highlighted as a form of stress that deserves particular consideration in the context of children’s development (e.g., Barletta & O’Mara, 2006). However, there is increasing evidence from couple research that, rather than adverse interactions per se, the cou- ple’s capabilities to compensate negativity with positivity may be a stronger predictor of marital outcomes (Gottman, 1994). Remarkably, to the best of our knowledge, there is limited research investigating the meaning of Gottman’s findings for children’s well-being in the family context. The purpose of this survey study was to address this gap by examining the positive-to-negative ratio in parental interaction as a predictor of child adjustment. In recent decades couple research has been increasingly concerned with studying interac- tion patterns in marriages. It was shown that pos- itivity and negativity in intimate relationships may not be considered as poles of one bipo- lar dimension but as two independent dimen- sions that constitute marital quality (Fincham & Linfield, 1997). According to this assumption, happy and unhappy couples do interact pos- itively and negatively but, notably, in differ- ent proportions. Gottman and Levenson (1992) 602 Family Relations 63 (December 2014): 602 – 615 DOI:10.1111/fare.12091

Transcript of How Much Positivity Is Needed to Buffer the Impact of Parental Negativity on Children?

Martina Zemp University of Zurich

Christine E. Merrilees State University of New York at Geneseo∗

Guy Bodenmann University of Zurich∗∗

How Much Positivity Is Needed to Buffer the Impact

of Parental Negativity on Children?

Previous research suggests that a couple’s abil-ity to balance adverse interactions with positiv-ity is a better predictor of relationship outcomesthan negativity per se. The purpose of this studywas to extend the findings on couples to par-ents by examining the positive-to-negative ratioin parental interaction as a predictor of childadjustment. Three hundred seventy-five parentscompleted questionnaires on the negative andpositive interactions in their intimate relation-ship and their child’s functioning online. Differ-ent interparental positive – negative ratios wereidentified by latent class analysis and then linkedto children’s adjustment. Children whose par-ents were characterized by more negativity rel-ative to positivity scored higher in externalizingproblems compared to children whose parents’negativity was outbalanced by positive interac-tions. Girls with parents yielding the highestpositive – negative ratios were reported to showmost prosocial behavior. The findings suggestthat parental positivity should be expressed at

Department of Psychology, University of Zurich, Binz-muehlestrasse 14/23, 8050 Zurich, Switzerland ([email protected]).∗Department of Psychology, State University of New Yorkat Geneseo, 1 College Circle, Geneseo, NY 14454.∗∗Department of Psychology, University of Zurich, Binz-muehlestrasse 14/23, 8050 Zurich, Switzerland.

Key Words: interparental conflict, balance theory, positive–to-negative ratio, family relations, emotional security.

least twice as much as parental negativity for thesake of children’s well-being.

Introduction

Marital conflict has been highlighted as a formof stress that deserves particular considerationin the context of children’s development (e.g.,Barletta & O’Mara, 2006). However, there isincreasing evidence from couple research that,rather than adverse interactions per se, the cou-ple’s capabilities to compensate negativity withpositivity may be a stronger predictor of maritaloutcomes (Gottman, 1994). Remarkably, to thebest of our knowledge, there is limited researchinvestigating the meaning of Gottman’s findingsfor children’s well-being in the family context.The purpose of this survey study was to addressthis gap by examining the positive-to-negativeratio in parental interaction as a predictor ofchild adjustment.

In recent decades couple research has beenincreasingly concerned with studying interac-tion patterns in marriages. It was shown that pos-itivity and negativity in intimate relationshipsmay not be considered as poles of one bipo-lar dimension but as two independent dimen-sions that constitute marital quality (Fincham &Linfield, 1997). According to this assumption,happy and unhappy couples do interact pos-itively and negatively but, notably, in differ-ent proportions. Gottman and Levenson (1992)

602 Family Relations 63 (December 2014): 602 – 615DOI:10.1111/fare.12091

How Much Positivity Is Needed for Children? 603

reported that couples characterized by more neg-ative than positive interactions (i.e., unregulatedcouples) were at greater risk to start the cascadetoward divorce in comparison to regulated cou-ples outperforming negativity by positive inter-action. Additional studies based on observa-tional data supported Gottman’s (1993, 1994)balance theory of marriage that states that theequation of positivity to negativity is crucial topredict marital quality and stability, suggesting aratio of about 5:1 in regulated couples. Holmanand Jarvis (2003) found that Gottman’s (1993)couple typology could be reliably verified insurvey data too. Subsequently, European surveystudies found a positive – negative ratio of about2:1 in satisfied couples compared to unsatisfiedcouples or couples seeking for marital therapyyielding a ratio of smaller than 1 (Bertoni &Bodenmann, 2010; Bodenmann, Meyer, Binz, &Brunner, 2004).

The ratio of a couple’s positivity to negativitymay also help to explain why the vast major-ity of children from conflictual homes do notdevelop psychological problems (Fincham &Osborne, 1993). The emotional security theory(EST; Davies & Cummings, 1994) supports theidea that the extent to which parental negativityis buffered by their positive interaction is criticalfor child adjustment. EST holds that parentaldiscord harms children’s development by under-mining children’s emotional security; that is,their primary goal of felt safety in the family andconfidence in their parents’ abilities to preservefamily stability. However, the implications ofnegative parental interaction on a child’s emo-tional security and welfare hinges on the qualityof the broader parental and family function-ing (Cummings & Davies, 2010). Supportingthis hypothesis, Davies and colleagues (2002)found that interparental conflict was a weakerpredictor of children’s emotional insecurity andinsecurity was less associated with children’spsychological problems in families where par-ents scored high in emotional expressiveness.Parents’ ability to discuss their emotions ina direct and constructive manner seemed toprotect children from developing emotionalinsecurity and subsequent maladjustment in theface of parental conflict. In a similar vein, itwas reported that family functioning providesa salient backdrop for children’s reactivity tomarital conflict (Davies, Cummings, & Winter,2004). Children from cohesive families per-ceived simulated parental arguments as less

threatening to their well-being than childrenfrom other (less functional) family profiles.

It was also shown that the emotional cli-mate in the family setting affects children’sappraisals of interparental conflict. For instance,children’s self-blaming attributions for their par-ents’ arguments seem to depend on parenting;supportive parenting mitigated the impact ofparental discord on children by decreasing thoseappraisals (DeBoard-Lucas, Fosco, Raynor,& Grych, 2010). Strikingly, children whoseparents described their emotional climate inthe family by high levels of negative affect andlow positive affect were more likely to blamethemselves for parental discord in comparisonto all other groups, including children fromfamilies high in positive and negative emo-tional expression (Fosco & Grych, 2007). Itwas, notably, the combination of negativityand positivity that best explained variation inself-blaming appraisals. In the study by Lindahland Malik (2011), high levels of observationalmeasures of family cohesion were associatedwith relatively low child-perceived threat andself-blame in parental conflict situations, inde-pendent of conflict style. Under conditions oflow family cohesion, children from parentsof the conflictual-hostile group (who wereusually not able to resolve their arguments)reported higher threat compared to childrenfrom conflictual-expressive couples charac-terized by intensely angry interactions which,however, generally closed in a positive manner.These results demonstrate that a lack of familycohesion does potentiate the negative impact ofparental negativity, but only in conditions wherechildren perceive their parents’ negativity notbeing balanced by positivity.

In summary, a number of studies reveal thatinterparental negative interactions have far lessdetrimental to negligible impact on children’swelfare when they occur in a broader familycontext marked by high levels of positivity.We think that the balance theory of Gottman(1993) adds a significant contribution in thiscontext. That is, parental negativity may predictadverse child outcomes only when it crosses athreshold in proportion to the number of positiveinteractions. The first attempt to address thisidea was reported by Katz and Woodin (2002).They examined the relations between Gottman’stypes of couples and their children’s adjustmentin terms of mother-reported behavior problemsand observed negative affect/noncompliance in

604 Family Relations

a peer interaction. They found that the maritaltypology significantly contributed to variance inchild adjustment, even after controlling for par-enting and family-level processes, and above andbeyond marital satisfaction or marital violence.For mother-reported child behavior problems,differences in the absolute degree of negativitybetween the couple types did not account forthe differences in child outcomes. This studysupported that the positive-to-negative ratio inparental interaction may be crucial to children’swell-being in the family, but they focused onnegativity in couples (i.e., hostility and detach-ment) and did not answer the question of howmuch positivity is needed for children to bufferthe negativity.

The Current Study

To the best of our knowledge, this is the firststudy to investigate child outcomes as a functionof the ratio of their parents’ positivity to negativ-ity. The first aim of the study was to identify dis-tinct classes of ratios of interparental positivityto negativity in a European online survey sam-ple of 375 mothers and fathers (individual data).We relied on latent class analysis (LCA) forthis purpose given previous evidence of the use-fulness of cluster analytic approaches to applyGottman’s (1993) theory in survey data (Holman& Jarvis, 2003). In contrast to Katz and Woodin(2002), we primarily focused on the numericalpositive – negative ratio rather than on qualita-tive descriptions of different couple types. Thesecond goal was to examine whether the classesof parents’ interactivity were linked with the par-ents’ perception of their children’s psychologi-cal functioning (i.e., internalizing and external-izing problems, prosocial behavior). Third, weinvestigated buffering effects of parental posi-tivity on their negative interactions in respect tochild adjustment.

In contrast to the majority of studies inthis field that have focused on family vari-ables going beyond the parental dyad (e.g.,parent – child relation, family cohesion), weexamined more specifically three positivedyadic interactions of parents that are assumedto be key dimensions to counterbalance thesequelae of interparental negativity on children.First, parents’ constructive communication inconflict situations (e.g., support, problem solv-ing) was shown to elicit positive responding inchildren in a study based on home diary reports

(Cummings, Goeke-Morey, & Papp, 2003).Children’s exposure to constructive communi-cation seems to foster their emotional security,which is likely to elevate their prosocial behavior(McCoy, Cummings, & Davies, 2009). Second,reciprocal positive everyday interaction (e.g.,affection, validation) between partners is beingincreasingly discussed as a unique predictor ofrelationship outcomes over and above couple’snegativity (Graber, Laurenceau, Miga, Chango,& Coan, 2011). Its significance for broader fam-ily interactions has been reported recently (Ack-erman, Kashy, Donnellan, & Conger, 2011).Third, consideration of the couple’s dyadiccoping is warranted because it has emerged asa powerful predictor of relationship satisfactionand stability (Bodenmann, 2000). Dyadic copingrefers to the partners’ abilities to work jointlyand mutually as a couple when they contend witheveryday stress. There is growing evidence thatfunctional dyadic coping might be a vital mod-erator mitigating the impact of parental stresson children (Gabriel & Bodenmann, 2006).

Hypotheses

Children from parents exhibiting a higher pro-portion of positivity to negativity were expectedto be better adjusted (i.e., fewer internalizing andexternalizing problems, more prosocial behav-ior) than children from parents with smallerratios. We further hypothesized that the threepositive dyadic dimensions were potent protec-tive factors alleviating the negative impact ofinterparental negativity on children. Because thisstudy was not designed to provide rigorous test-ing of a given theory, a conservative analyticstance was adopted; that is, two-tailed tests ofsignificance were used throughout.

Method

Participants

The participants constituted a convenience sam-ple of 375 parents, of which 308 mothers and67 fathers served as respondents independentof their partner (individual data). Inclusion cri-teria for participation were good knowledge ofGerman, cohabiting with spouse or partner, andhaving at least one biological child age 4 to 18years. If participants had more than one child inthe respective age range parents were asked toselect one child at random. This age range was

How Much Positivity Is Needed for Children? 605

selected as the questionnaire used to assess thechild’s adjustment (i.e., the Strengths and Diffi-culties Questionnaire [SDQ]; Klasen, Woerner,Rothenberger, & Goodman, 2003; originallydeveloped by Goodman, 1997) is applicableto children not younger than age 4 years. Therespondents had to be the biological father ormother of the selected child and to have primarycustody in cases where they were living with anew partner who was not the biological parentof the child.

The average age of participants was39.53 years (SD= 7.06). Most of them wereSwiss (68%), 22% were Germans, and 10%were from other countries, but all were fluent inGerman. Eighty-four percent were married, and16% cohabitated with their partner. Mean rela-tionship duration was 14.58 years (SD= 6.54,range= 1 – 38 years). Participants’ educationalachievement was high school in 4%, grammarschool in 56%, and university in 40%, respec-tively. The age of children who were evaluatedby their parents averaged 9.25 years (SD= 4.10,range= 4 – 18 years). One hundred ninety-sevenchildren (53%) were females and 178 (47%)were males.

Procedure

Participants were recruited by means of fly-ers distributed in local community centers andvia the Internet by posting a link on familyforums. Eligible mothers and fathers completedquestionnaires on their current intimate rela-tionship and their child’s adjustment online bymeans of a license-required web survey software(see http://www.unipark.com/). Unipark is partof Globalpark, which provides panel, commu-nity, and survey software. It is a well-establishedonline research survey tool for European uni-versities. Participation in this study was notrewarded financially.

Measures

Interparental Communication Quality. Theshort version of the Marital Communica-tion Questionnaire (MCQ; see Bodenmann,2000) was used. This is a self-report question-naire based on the affective communicationcategories identified in the Specific AffectCoding System (SPAFF; Gottman, 1994). Thisquestionnaire assesses with eight items neg-ative marital communication behaviors (i.e.,

interparental negativity) in conflict situationssuch as contempt, defensiveness, domineering,belligerence, and stonewalling (e.g., “I criticizemy partner; I deny responsibility or blamemy partner”). Constructive marital conflictcommunication was measured by four items(e.g., “I validate my partner’s opinion and feel-ings; I try to understand my partner”). Itemswere reported on a 6-point Likert-type scale(1= very rarely, 6= very often). The MCQ hasbeen widely used (e.g., Bodenmann, Bradbury,& Pihet, 2008), comprises good convergentvalidity, and has good psychometric proper-ties with reliabilities of 𝛼 = .80 for negativecommunication and 𝛼 = .83 for positive com-munication (Bodenmann, 2000). In this study,the internal consistency of negative communi-cation was 𝛼 = .76 and 𝛼 = .85 for constructivecommunication.

Positive Everyday Interaction. Four items weredeveloped assessing positive everyday interac-tion in the relationship (e.g., showing appreci-ation; exchange of tenderness) based on manyyears of research on constructive interactivity inintimate relationships (Ruffieux & Bodenmann,2011). Participants rated their own as well astheir partners’ behavior using a Likert-type scale(1= very rarely, 6= very often). Internal consis-tencies were 𝛼 = .80 for the own behavior towardthe partner and 𝛼 = .84 regarding the behavior ofthe partner toward the respondent. Relationshipsatisfaction assessed by the Relationship Assess-ment Scale (RAS; Hendrick, 1988) correlatedsignificantly with the respondents’ own posi-tive behavior (r = .52, p< .01) and with the per-ceived positive behavior of the partner (r = .58,p< .01).

Dyadic Coping. The short version of the DyadicCoping Inventory (DCI; Bodenmann, 2008) wasused to measure the participants’ perspective onhow they cope together with their partners withcommon, everyday stresses as a couple by threeitems (e.g., “We try to cope with the problemtogether and seek solutions”). Each question wasanswered using a Likert-type scale ranging from1 (very rarely) to 5 (very often). The psychomet-rics of this questionnaire have been examined inprior research with good construct, convergentand discriminant validity, and satisfying reliabil-ity ranging from 𝛼 = .71 to .93 through the sub-scales (Bodenmann, 2008). Cronbach’s alpha inthis study was 𝛼 = .90.

606 Family Relations

Child Adjustment. Child’s adjustment wasassessed by parent report of the German versionof the SDQ (Klasen et al., 2003; originallydeveloped by Goodman, 1997). Followingrecent recommendations to rely on the broaderthree factor structure in low-risk samples (vs.clinical or high-risk samples) instead of usingthe original five subscales (Goodman, Lamping,& Ploubidis, 2010), we measured internalizingproblems with 10 items (e.g., Many fears, easilyscared; Often unhappy, depressed or tearful),externalizing problems with 10 items (e.g.,Often fights with other children or bullies them;Easily distracted, concentration wanders), andprosocial behavior with five items (e.g., Oftenoffers to help others [parents, teachers, children];Helpful if someone is hurt, upset or feeling ill).Parents’ rated each subscale on a 3-point Lik-ert scale (0= not true, 1= somewhat true,2= certainly true). The SDQ is being used as aresearch tool throughout the world and is welladapted particularly for studies of the generalpopulation. It is a well-established measure inthe German-speaking area that comprises goodvalidity and reliability ranging from 𝛼 = .58 to.82 through the subscales in the German version(Klasen et al., 2003). The internal consistencyof the subscales in this study were 𝛼 = .75for internalizing problems, 𝛼 = .80 concerningexternalizing problems, and 𝛼 = .76 with regardto children’s prosocial behavior, respectively.

Data Analysis

Latent class analysis (LCA) was used to addressthe first objective of the study; that is, theidentification of distinct classes of ratios ofinterparental positivity to negativity. LCA is aperson-centered approach assigning individualsto homogenous subpopulations (latent classes)by virtue of their similarity across a number ofrelevant variables; that is, interparental negativ-ity, constructive communication, positive every-day interaction of the respondent and his or herpartner, and dyadic coping in this study. In accor-dance with Holman and Jarvis (2003), we con-sidered cluster analysis as a convenient tool insurvey data for distinguishing couples (in thisinstance, individual data of parents) due to dis-tinct relations of negativity to positivity in theirrelationship. Referring to previous reports (e.g.,Bertoni & Bodenmann, 2010), we computed anumerical proportion of perceived negativity topositivity based on the variable means whereby

positivity constitutes the average of the meansof the three positive dyadic dimensions for thesake of simplicity. The mean of positivity wasdivided by the mean of negativity to obtain thepositive – negative ratio for each latent class.Because items assessing dyadic coping wereanswered on a different range compared to theother variables, all measures that were includedin the calculation of the ratio were transformedto scales ranging from 0 to 5 to make them com-parable.

The LCA was computed using mixture mod-eling in Mplus 6 (Muthén & Muthén, 1998).Parameters are estimated by the maximumlikelihood criterion. To avoid local solutionsof maximum likelihood we specified 500random sets of starting values for the initialstage and 50 final stage optimizations to beused. Residual variances across classes wereheld equal as by default. The optimal num-ber of latent classes that best fit the data wasdetermined by evaluating fit of models withprogressively more groups (beginning withn= 1) using Schwarz’s Bayesian InformationCriterion (BIC), with lower values indicatingbetter fit. The BIC can be regarded as supe-rior to other relative fit indices as it correctsfor the number of parameters to be estimated(Field, 2009); hence, models with fewer groupsare favored. Notwithstanding, we decided thatthe number of classes were chosen at whichthe BIC value was smallest, on condition thateach group constituted at least 15 individuals toavoid overfitting. In adopting this approach, wewarranted to provide meaningful groups whilestill establishing parsimony. Posterior proba-bility of class membership was calculated foreach individual, and they were then identifiedas belonging to one group on the basis of theirmaximum posterior probability.

Hierarchical multiple regression analyseswere conducted to test (a) whether parents’ classmembership of different positivity-to-negativityproportions predicted child adaptation and (b)whether positive dyadic functioning of parentsbuffered the adverse effect of negative inter-action on children. For purpose one, parent’sgender, child’s gender, and child’s age werecontrolled for potential confounds by enteringthem in the first step of the regression analyses.Parents’ membership to the classes resultingfrom the LCA was entered in the second stepin the form of dummy-coded variables. Sincedummy variables only allow dichotomous

How Much Positivity Is Needed for Children? 607

Table 1. Means, Standard Deviations, and Correlations Among Study Variables (N= 375)

1 2 3 4 5 6 7 Mean (SD)

1. Negativity — 1.03 ( .55)2. Constructive communication – .38∗∗∗ — 3.15 ( .85)3. Positive everyday interaction (respondent) – .31∗∗∗ .53∗∗∗ — 2.45 ( .86)4. Positive everyday interaction (partner) – .14∗∗ .42∗∗∗ .71∗∗∗ — 2.26 ( .99)5. Common dyadic coping – .32∗∗∗ .48∗∗∗ .50∗∗∗ .56∗∗∗ — 3.35 (1.21)6. Internalizing problems of children .19∗∗∗ – .05 – .05 – .13∗ – .15∗∗ — .33 ( .31)7. Externalizing problems of children .28∗∗∗ – .21∗∗∗ – .14∗∗ – .15∗∗ – .15∗∗ .36∗∗∗ — .51 ( .35)8. Prosocial behavior of children – .12∗ .25∗∗∗ .11∗ .06 .11∗ – .21∗∗∗ – .31∗∗∗ 1.53 ( .36)

∗p< .05, ∗∗p< 0.01, ∗∗∗p< .001.

comparisons of a number of categories againstone chosen baseline group (i.e., the referencegroup), we determined the reference groupsin due consideration of maximal interest andplausibility in terms of content. Thus, the refer-ence group for computing the dummy variableswas the class with the lowest (least functional)ratio of parental positivity relative to negativityregarding the two SDQ scores assessing childmaladjustment (i.e., internalizing and external-izing problems). Children from this class wereexpected to score lowest on these two dimen-sions, thus comparison of each group against thisbaseline group was most interesting. In contrast,the class with the highest (most functional) ratioserved as the reference group concerning thepositive child outcome (i.e., children’s prosocialbehavior). Multivariate analyses of variance(MANOVA) would have been the method ofchoice comparing means of a categorical vari-able with more than two categories pairwise andmultivariate, but this method was not tenablein this study because the assumption of homo-geneity of covariance matrices was violated andcell sizes were highly unequal. As robustnessof statistics cannot be assumed in such cases(Field, 2009), we have opted for a regressionanalytic approach rather than MANOVA.

For the second purpose, hierarchical multipleregression analyses were computed to examinepositive interactions between parents as moder-ators in the link between parental negativity andchild psychological well-being. Again, parent’sgender as well as child’s gender and age werecontrolled for in the first step of the regressionanalyses. Parental negativity and the respectivemoderator (as main effects) formed the sec-ond step and the two-way interaction betweenthe main predictor and the moderator wereentered in the third step. Referring to Cohen,

Cohen, West, and Aiken (2003), only interactionterms of theoretical interest were included tomaximize the possibility to detect significanteffects by reducing statistical distortions. Allnumerical predictors were grand mean centered.

Results

Descriptive Statistics

Means, standard deviations, and correlationsbetween study variables are listed in Table 1. Asexpected, all indicators of the dyadic functioningwere significantly correlated with each other, aswell as the measures of child outcome. Dyadicadjustment was further strongly associated withchildren’s welfare in the expected direction (i.e.,negative associations between constructive cou-ple behaviors with child problem behaviors butpositive relations with prosocial behavior andvice versa in respect of parental negativity),albeit more consistently with children’s external-izing problems than with their internalizing orprosocial behaviors.

Latent Classes of Dyadic Positivity in Relationto Negativity

The results of the LCA indicated that thefour-class solution outperformed alternativeclass solutions across our two domains ofevaluation; that is, parsimony and model fit.The BIC decreased from the three-class solu-tion (BIC= 4376.73) to the four-class model(BIC= 4340.78), and the smallest class did notcome below the threshold of 15 individuals.The BIC value dropped further in the five-classsolution (BIC= 4307.28), but the smallest groupdid then include only nine individuals (2.4%),hence the four classes were adopted in the wholesample. In a second step, we examined whether

608 Family Relations

parents’ gender significantly predicted differentclass solutions. Therefore, a mixture modelingwith known classes (in this instance, femalesvs. males) was computed (i.e., a multiple groupanalysis; see Muthén & Muthén, 1998). A con-strained model in which means of the specificvariables of the four latent classes were heldequal across the classes of gender while thevariances were held equal across the latentclasses was compared to an unconstrained classsolution. The BIC of the constrained model was4734.84 and provided a better fit than the uncon-strained (BIC= 4784.30). However, the cogencyof this result had to be considered as limitedowing to the high disproportion in sample sizesbetween females (n= 308) and males (n= 67).Consequently, we computed two LCA sepa-rately for women and men and compared theselatent groups with the four-class solution result-ing from the total sample. As expected due to thehigh percentage of females constituting our sam-ple, class solution of women closely resembledthe overall LCA. Analogous to the former analy-sis, the four-class model best fit the data yieldinglowest BIC value (3634.36) without minimalgroup size falling short of 15 individuals of thefemale subsample. Furthermore, a comparisonof the four-class solution constrained by fix-ing the means at the values arising from theoverall LCA with the unconstrained model indi-cated better fit for the former (BIC= 3527.57).According to these results, latent structure infemales can be assumed to be highly congruent

to the total sample and separate analyses weretherefore not necessary from this point of view.

Next, the same approach was appliedfor the male subsample. Independent LCAof men revealed that a three-class solution(BIC= 758.82) fit data better than the two-classmodel (BIC= 766.13) or the four-class solution(BIC= 762.78). Visual inspection of class pat-tern suggested that the three groups of fathers’reports approached the classes identified in thesample as a whole. Again, the unconstrainedmodel was tested against a constrained modelwith fixed means at the values resultant from theoverall LCA and the latter provided a lower BICvalue (729.71). Thus, latent classes of malescould successively be merged into the overallLCA solution. These preliminary analyses indi-cated mere quantitative rather than significantqualitative differences in latent groups betweenwomen and men. Hence, we concluded thattreating the sample as a whole without consid-eration of gender effects in latent classes wasjustified.

Table 2 shows estimated parameters of thelatent classes for the total sample. In three outof the four classes the order of the means weresimilar (dyadic coping> constructive communi-cation> own positive everyday interaction>positive everyday interaction of partner).In the fourth group couple’s dyadic, cop-ing was markedly lower, but rankings of theother positive interactions were comparable.Hence, averaging all positive dimensions to

Table 2. Parameter Estimates of Latent Classes of Interparental Negative and Positive Interactions

Means and Residual Variances of Dimensions of Parental Interactivity

Class NegativityConstructive

Communication

PositiveEveryday

Interaction(Respondent)

PositiveEveryday

Interaction(Partner)

CommonDyadicCoping n (%)

Positive-to-Negative

Ratio

1 Very highpositive – lownegative

.66 (.27) 4.18 (.45) 3.91 (.28) 3.85 (.37) 4.64 (.72) 36 ( 9.60%) 6.3:1

2 High positive– low negative

.95 (.27) 3.36 (.45) 2.74 (.28) 2.62 (.37) 3.85 (.72) 172 (45.87%) 3.3:1

3 Moderatepositive – lownegative

1.15 (.27) 2.82 (.45) 1.91 (.28) 1.63 (.37) 2.79 (.72) 149 (39.73%) 2.0:1

4 Lowpositive – moderatenegative

1.54 (.27) 1.83 (.45) 1.18 (.28) 0.67 (.37) 0.83 (.72) 18 ( 4.8%) 0.7:1

Note: All parameter estimates are significant at p< .01.

How Much Positivity Is Needed for Children? 609

compute the value of positivity in respect tothe positive-to-negative ratio seemed viable intotal. The four groups engendered by the LCAcan be distinguished between three functionalclasses in which dyadic negativity was out-weighed by positive interactivity to differentdegrees and one dysfunctional group yieldinglower positivity than negativity. The very highpositive – low negative group (9.60%) yieldedthe highest ratio, outperforming negativity morethan six times by positive interactions. In thehigh positive – low negative class (45.87%)the positive-to-negative ratio was 3.3:1 and itwas 2.0:1 for couples of the moderate posi-tive – low negative group (39.73%). A minority(18 individuals), which were labeled as the lowpositive – moderate negative subgroup, showedless positive interaction relative to negativityrevealing a ratio of 0.7:1.

Links Between Parental Classesof Positive – Negative Ratios and Child

Adjustment

To examine if the positive-to-negative ratiosof parental interactions were associated withchild functioning, the three measures of chil-dren’s psychological adaptation were submittedas dependent variables to a series of hierarchicalmultiple regression analyses. As described ear-lier, the low positive – moderate negative classserved as reference group concerning children’sproblem scores and the very high positive – lownegative class served as the reference grouppertaining to analyses of children’s prosocialbehavior. For predictions of children’s internal-izing symptoms only child’s age emerged to besignificantly related with the outcome variable(𝛽 = .13, p< .05). Nonsignificant coefficients ofthe dummy variables suggested that there wereno group differences in terms of latent classes ofparental interactivity (see Table 3).

Regarding children’s externalizing problemschildren from the dysfunctional parental pat-tern (i.e., the low positive – moderate negativesubgroup) were reported to score signifi-cantly higher in comparison to the very highpositive – low negative (𝛽 =−.32, p< .001),the high positive – low negative (𝛽 =−.50,p< .001), and the moderate positive – lownegative (𝛽 =−.41, p< .001) groups. Becausechild’s gender remained significant in the sec-ond step of the analysis as depicted in Table 3(𝛽 =−.14, p< .01; indicating boys showing

Table 3. Classes in Parental Interactivity as Predictors of

Children’s Internalizing and Externalizing Problems

InternalizingProblems

ExternalizingProblems

Predictors 𝛽 ΔR2 R2 𝛽 ΔR2 R2

Step 1 .02∗ .02 .04∗∗∗ .04Child’s age .13∗ – .13∗

Child’s gender – .06 – .13∗

Parent’s gender .04 .07Step 2 .01 .03 .05∗∗∗ .09

Class 4 vs. 1 – .09 – .32∗∗∗

Class 4 vs. 2 – .11 – .50∗∗∗

Class 4 vs. 3 – .04 – .41∗∗∗

Note: Standardized regression coefficients of the last stepare reported. Class 1= very high positive – low negative;class 2= high positive – low negative; class 3=moderatepositive – low negative; class 4= low positive – moderatenegative.

∗p< .05, ∗∗p< .01, ∗∗∗p< .001.

more externalizing problems than girls), a fur-ther step was entered to examine interactionterms between dummy variables representingclass membership and child’s gender. None ofthese interactions was significant, these resultsare thus not presented.

Table 4 shows that children’s prosocial behav-ior was significantly predicted by child’s gender(𝛽 = .12, p< .05), with girls being better adjustedon average compared to boys, and two dummyvariables depicting higher prosocial function-ing in children from very high positive – lownegative parents in relation to children from themoderate positive – low negative (𝛽 =−.21,p< .05) and the low positive – moderate nega-tive (𝛽 =−.14, p< .05) groups. Again, becausechild’s gender remained a significant predic-tor (𝛽 = .12, p< .05) over and above classmembership in the second step of the regres-sion analysis, two-way interactions betweendummy variables and gender of the child wereexamined in the third step, yielding significanteffects (see Table 4). We therefore conductedgender-specific regression analyses (not pre-sented) that did not provide any significant groupdifferences in boys. For girls, however, the mostfunctional ratio (very high positive – low neg-ative) in parental interactivity was associatedwith higher scores in their prosocial behav-ior in comparison to the high positive – lownegative (𝛽 =−.34, p< .05), the moderate posi-tive – low negative (𝛽 =−.45, p< .01), and thelow positive – moderate negative (𝛽 =−.32,

610 Family Relations

Table 4. Classes in Parental Interactivity as Predictors of

Children’s Prosocial Behavior

Prosocial behavior

Predictors 𝛽 ΔR2 R2

Step 1 .01 .01Child’s age .01Child’s gender .12∗

Parent’s gender .03Step 2 .02† .03

Class 1 vs. 2 – .15†

Class 1 vs. 3 – .21∗

Class 1 vs. 4 – .14∗

Step 3 .02∗ .06Class 1 vs. 2 ∗child’s gender – .55†

Class 1 vs. 3 ∗child’s gender – .68∗

Class 1 vs. 4 ∗child’s gender – .54∗∗

Note: Standardized regression coefficients of the last stepare reported. Class 1= very high positive – low negative;class 2= high positive – low negative; class 3=moderatepositive – low negative; class 4= low positive – moderatenegative.

†p< .10, ∗p< .05, ∗∗p< .01, ∗∗∗p< .001.

p< .001) groups, respectively. Overall, 8% ofthe variance in girls’ prosocial adjustment couldbe explained solely by membership of the veryhigh positive – low negative class relative to allother groups (ΔR2 = 0.08, p< .01).

Moderation of Impact of InterparentalNegativity on Children by Positive Dyadic

Functioning

For examination of moderating effects of inter-parental positivity in the association betweenparents’ negativity and children’s functioning,one regression analysis was computed for eachcombination of the four positive dimensionsas moderators and the three outcomes of chil-dren’s adjustment. It was found that the positiveeveryday interaction of the respondent’s partner(𝛽 =−.11, p< .05) and the dyadic coping of thecouple (𝛽 =−.10, p< .05) moderated effects oftheir negative interactions on children’s exter-nalizing behavior problems. Consistent with theformer analyses, interactions of child’s genderwith the main predictors (i.e., negativity, posi-tivity) were further explored in each regression,but results were nonsignificant in all cases. Fol-lowing Aiken and West (1991), the significanttwo-way interactions were interpreted by plot-ting the simple regression lines for high and low

values of the moderators (not presented), and thesimple slopes were examined whether they weresignificantly different from 0. According to therecommendation of Cohen et al. (2003), lowerand higher levels of moderators were definedas one standard deviation above (+1 SD) andbelow (−1 SD) the mean, respectively. High lev-els of the partner’s positive everyday interactionemerged to buffer the detrimental impact of highparental negativity on children. The simple slopetest revealed that negativity was significantlylinked with children’s externalizing behaviorunder conditions of low positive behavior ofthe respondent’s partner, b= .19, t(367)= 4.26,p< .001, but not in the case of high levels of pos-itive behavior, b= .07, t(367)= 1.59, ns. Like-wise, parental negativity significantly predictedpoor adjustment when dyadic coping of the par-ents were low, b= .16, t(367)= 3.23, p< .01, butnot when it was high, b= .04, t(367)= .80, ns.

Concerning children’s prosocial behavior asthe outcome variable, results indicate that theeffect of interparental negativity was moderatedby the parents’ dyadic coping (𝛽 = .10, p< .05).Children showed better prosocial adjustmentwhen high values of parents’ negativity werecombined with high scores in dyadic coping incomparison to low dyadic coping. The reversepattern revealed when parental negativity waslow. However, the simple slopes were not sig-nificant when dyadic coping was low, b=−.06,t(367)=−.93, ns, or high, b= .07, t(367)= 1.13,ns. Given the evidence of strong gender effectsin the analyses related to children’s proso-cial behavior, child’s gender was consistentlyconsidered in interaction terms. However, thethree-way product term of parental negativity,the respective moderator, and child’s gender didnot become statistically significant in any of theanalyses.

Discussion

The family provides a primary context tounderstand child development and research hasrepeatedly demonstrated marital functioningas a cornerstone in children’s socialization.Indeed, interparental conflict is well establishedas a powerful risk factor for an array of adjust-ment problems in children (e.g., Cummings& Davies, 2010). It must also be taken intoaccount, however, that marital conflict is a com-mon occurrence even in harmonious familiesand that only the minority of children raised

How Much Positivity Is Needed for Children? 611

in conflictual families experiences significantenduring problems (Fincham & Osborne, 1993).Gottman’s balance theory (1993) might add animportant contribution in this field; it suggeststhat conflict outcomes are more a result of theratio of positivity to negativity than the absolutenumber of each. To the best of our knowledge,this study was the first to examine how muchpositivity is needed to buffer the impact ofparental negativity on children.

In a first step, an LCA was computed to iden-tify different patterns of the expression of posi-tive behaviors in relation to negativity in couples(reported by one partner). One subgroup resultedin which negative interactions were outbalancedmore than 6 times. This exceeds Gottman’s 5:1equation (1993) and is likely a result of thestatistical method applied in the current study.The LCA-defined round 10% of the sample as ahigh-functioning group inductively from empir-ical data is not comparable to a deductive group-ing approach on the basis of a given theory.Furthermore, though Gottman only studied pos-itive behavior in a couple’s conflict discussion,in this study we also considered positive interac-tions and dyadic coping in the couple’s daily rou-tine, which could bloat the positivity part of theratio. The majority of respondents correspondedto the positive-to-negative ratio of about 2 to 3:1found in satisfied couples in comparable surveystudies (Bertoni & Bodenmann, 2010; Boden-mann et al., 2004). A minority obtained a ratioof less than 1 (in which the negativity exceedsthe positivity) in accordance with the unstablecouples studied by Gottman (1993). These arethe couples who most likely seek marital therapyaccording to Bertoni and Bodenmann (2010).

Examining the positive – negative ratio as apredictor of child adjustment in a second step,our results provided strong evidence that the pro-portion of interparental positivity to negativitydoes matter for children – for better or worse.That is, girls from parents yielding the highestpositive – negative ratio were reported to showmore prosocial behavior than girls from all othergroups. On the downside, when interparentalnegativity outweighed positivity, children scoredsignificantly higher in externalizing problemscompared to children from all other samplesubgroups. This is a solid basis to claim thatparental conflict does not occur in isolation fromother aspects of marital functioning, and thusits impact on children must always be weighedin terms of the balancing amount of positivity

between parents. Because children’s emotionalsecurity hinges on the quality of the broader fam-ily functioning (Cummings & Davies, 2010),parental negativity may have fewer adverseeffects on children when it takes place in a pos-itive family climate. We found that interparentalpositivity ought to exceed marital conflict by atleast double for the sake of the children. Childrenfrom families above this critical threshold (a)may be protected against negativity (i.e., fewerexternalizing problems) and (b) can even benefitfrom observed positivity by learning appro-priate ways to handle their own relations (i.e.,enhanced prosocial behavior). That negativityappears to be dysfunctional for children’s devel-opment only when it is not outweighed by posi-tivity is an important and disburdening messagethat should be disseminated to community fami-lies. Interparental conflict is an unavoidable partof normal family life, but parents should seekto outnumber them by instances of supportivebehavior. Our findings further support the notionthat parents should not repress their differencesof opinion, but they should learn to communicateand work through their disagreements within asupporting and positive family environment.

The reported gender differences in the effectson children’s prosocial functioning might beexplained by conventional gender-differentiatedsocialization emphases. Girls are commonlysocialized to value interdependence and con-nectedness in close relationships whereas boysare often supported to develop greater indepen-dence and autonomy. Hence, gender differencesare reasonable with girls being more likelyto incorporate prosocial behavior from theirparents than boys (Snyder, 1998). The lackof significant findings regarding internalizingproblems in contrast to the strong effects inexternalizing outcomes is consistent with earlierreports that (a) psychosocial risk factors inchildhood played a very prominent role in theprediction of prospective externalizing prob-lems whereas risk factors occurring in earlyadulthood had greater predictive power forinternalizing disorders (Ihle, Esser, Schmidt, &Blanz, 2002), and that (b) etiological factorsof externalizing disorders are in general betteridentifiable (Blanz, Remschmidt, Schmidt, &Warnke, 2006). Moreover, parents may be morereliable raters of their children’s externalizingproblems because they are inherently moreobservable. The child’s perspective might bethe most appropriate measure to understand

612 Family Relations

the impact of family problems on children’sinternalizing symptoms and the lack thereofin this study precludes definitive conclusionsconcerning this matter.

Strikingly, results from the moderationanalyses indicate that detrimental impacts ofinterparental negativity were buffered by pos-itive everyday interaction of the respondent’spartner and by dyadic coping in terms of childexternalizing problems. Likewise, dyadic copingmoderated the impact of parents’ negativityon children’s prosocial behavior. Children’swell-being can therefore be assumed to dependlargely on parental functioning beyond conflictcommunication, with the couple’s positive reci-procity and abilities to utilize the partner as abase of support to alleviate stress as promisingbuffers. The adult pair bond and the couple’sdyadic coping may not merely serve couple inti-macy but might also foster children’s emotionalsecurity that can mitigate the harmful effectsof parental negativity (cf. McCoy et al., 2009).Interestingly, only the positive everyday inter-action of the respondent’s partner, but not thebehavior of the respondent person itself, alle-viated the respondent’s negativity on children’sexternalizing symptoms. This finding indicatesthat the positive behavior of one parent is apotential buffer of the other parent’s negativity,which underscores the necessity of a dyadicperspective in this context. However, it is alsoconceivable that the positive behavior of one’spartner is a more objective and reliable measurethan the report of one’s own behavior.

In sum, this study stresses the significanceof disseminating the importance of managingparental conflict in appropriate, well-modulatedways to clinicians and the community. Multi-ple avenues of preventive programs aimed atenhancing positive parental behavior have beentaken to date, but most of them give priorityto parenting (Cowan & Cowan, 2002). Ourresearch supports previous impetus for thenotion that parenting as the exclusive focus infamily treatment omits an important piece of thepuzzle, a shortcoming that may be overcome bymore explicitly addressing the parents’ relation-ship difficulties themselves (Emery, Fincham, &Cummings, 1992). Prevention programs focusedon helping mothers and fathers make their rela-tionship less troubled and more satisfying maybe highly beneficial for child development.For instance, a recent finding indicates that theenhancement of parental dyadic coping may be

particularly efficacious in improving relation-ship quality but also in reducing dysfunctionalchild behavior, albeit less strongly compared toan evidence-based parenting program (Boden-mann, Cina, Ledermann, & Sanders, 2008). Ourfindings further suggest that the emphasis onmarital conflict (“conflict is king”; Bradbury,Rogge, & Lawrence, 2001, p. 62) oversimplifiesthe determinants of parental functioning andconsequently limits intervention approachesthat can be derived for the sake of the children.That is, a focus on resources in prevention ortreatment approaches for parents seems timelyand more promising than just reducing nega-tivity. This certainly remains one main targetbut should be augmented by an attempt toenhance positive interparental interactions (e.g.,constructive communication, positive everydayinteraction, and dyadic coping as examined inthis study) to create a positive family environ-ment for child development that is powerful todampen the adverse effects of parental conflict.

Several limitations of this study merit discus-sion. First, we acknowledge the methodologicallimitations as all study variables were assessedby self-report measures based on one parent’sperspective exclusively. This entails the riskthat unhappy marriages lead parents to perceivetheir children as being more poorly adjusted orthat effect sizes are inflated because of sharedmethod variance. The lack of assessing bothpartners’ perspectives (i.e., dyadic data) and thechildren’s view in this study emphasizes theurgency of replication and expansion of our find-ings. Second, albeit comparisons were madewith the couple typology proposed by Gottman(1993, 1994), this study is based on surveydata exclusively. This limitation reduces com-parability with the original work by Gottmanwhich was mostly based on observational data,though previous research supported the appli-cability of his balance theory on data assessedby self-report (Holman & Jarvis, 2003). Addi-tional investigation based on observational data(cf. Katz & Woodin, 2002) is strongly requiredto replicate our results. Third, the sample stud-ied appeared well adjusted with a high coupleratio of positivity to negativity on average pre-sumably owing to self-selection and social desir-ability biases. This raises concerns about theextent to which our results are representativeof community families. Furthermore, inclusioncriterion was the respondent being the biologicalparent with the primary custody for the child for

How Much Positivity Is Needed for Children? 613

whom they responded on the measures of childoutcome. However, we did not control whetherthe respondent’s cohabiting partner was the bio-logical parent and, if not, how long the stepfam-ily was living together in one household. Thislimitation further constrains the generalizabilityof our findings and provides additional fodderfor replication of these results in families of dif-ferent structures. Fourth, the measure assessingpositive everyday interaction in the relationshipwas self-developed by Ruffieux and Bodenmann(2011) and definite statements about its validityare precluded. Fifth, because the parental rela-tionship usually precedes parent – child relationsit is tempting to examine marital functioningas the predictor of child adjustment. Neverthe-less, most contemporary conceptualizations ofsystem theories approve reciprocal family trans-actions. Future research on bidirectional effectsbetween children’s and parents’ functioning isurgently needed in this field. Sixth, caution ininterpretation of findings is warranted as theindirect pathway of marital interaction on childoutcomes by parenting was not a major con-cern in this study and was thus not examined.We cannot establish from our data whether thereported effects were mediated by parent – childinteraction.

Conclusion

With these caveats in mind, our findings indicatethat children in homes where positivity outnum-bers negativity in the interparental relationshipare less vulnerable to the effects of parentalnegativity. But how much positivity is neededfor children? Two points merit emphasis fromour work. First, parental conflict is unavoidablein family life; what matters may be less maritalnegativity by itself, but rather that it is balancedby at least twice the amount of positivity. Sec-ond, the more positivity relative to negativity thebetter; children’s risk of adjustment problemsis substantially reduced and their prosocialfunctioning is enhanced when developing infamilies where interparental positivity exceedsnegativity many times over. Hence, this studyrevealed that multiple aspects of positive andnegative dyadic functioning in parents combineto explain variance in children’s welfare. Thetendency to focus on unitary causation in thisfield (e.g., between marital conflict and childoutcomes) is necessarily limiting and futureresearch is urgently needed to address that

parental negativity does not occur in an inter-personal vacuum. If replicated elsewhere, ourfindings are of particular practical significanceand raise important clinical implications.

References

Ackerman, R. A., Kashy, D. A., Donnellan, M. B., &Conger, R. D. (2011). Positive-engagement behav-iors in observed family interactions: A social rela-tions perspective. Journal of Family Psychology,25, 719 – 730. doi:10.1037/a0025288

Aiken, L. S., & West, S. G. (1991). Multiple regres-sion: Testing and interpreting interactions. New-bury Park, CA: Sage.

Barletta, J., & O’Mara, B. (2006). A review of theimpact of marital conflict on child adjustment.Australian Journal of Guidance and Counselling,16, 91 – 105. doi:10.1375/ajgc.16.1.91

Bertoni, A., & Bodenmann, G. (2010). Satisfied anddissatisfied couples: Positive and negative dimen-sions, conflict styles, and relationships with familyof origin. European Psychologist, 15, 175 – 184.doi:10.1027/1016-9040/a000015

Blanz, B., Remschmidt, H., Schmidt, M. H., &Warnke, A. (2006). Psychische Störungen imKindes- und Jugendalter: Ein entwicklungspsy-chopathologisches Lehrbuch [Mental disordersin childhood and adolescence: A developmentalpsychopathology textbook]. Miesbach, Germany:Schattauer Verlag.

Bodenmann, G. (2000). Stress und coping bei paaren[Stress and coping in couples]. Göttingen, Ger-many: Hogrefe.

Bodenmann, G. (2008). Dyadic Coping Inventar:Testmanual [Dyadic Coping Inventory: Test man-ual]. Bern, Switzerland: Huber.

Bodenmann, G., Bradbury, T. N., & Pihet, S.(2008). Relative contributions of treatment-relatedchanges in communication skills and dyadic cop-ing skills to the longitudinal course of marriagein the framework of marital distress prevention.Journal of Divorce & Remarriage, 50, 1 – 21.doi:10.1080/10502550802365391

Bodenmann, G., Cina, A., Ledermann, T., & Sanders,M. R. (2008b). The efficacy of the Triple P-positiveparenting program in improving parenting andchild behavior: A comparison with two other treat-ment conditions. Behaviour Research and Ther-apy, 46, 411 – 427. doi:10.1016/j.brat.2008.01.001

Bodenmann, G., Meyer, J., Binz, G., & Brunner,L. (2004). Eine deutschsprachige Replikationder Paartypologie von Gottman [A replication ofGottman’s couple typology in a representativeSwiss sample]. Zeitschrift Für Familienforschung,16, 178 – 193.

Bradbury, T. N., Rogge, R., & Lawrence, E. (2001).Reconsidering the role of conflict in marriage. In

614 Family Relations

A. Booth, A. C. Crouter, & M. Clements (Eds.),Couples in conflict (pp. 59 – 81). New York: Psy-chology Press.

Cohen, J., Cohen, P., West, S. G., & Aiken, L.S. (2003). Applied multiple regression/correlationanalysis for the behavioral sciences. Mahwah, NJ:Lawrence Erlbaum Associates.

Cowan, P. A., & Cowan, C. P. (2002). Interventions astests of family systems theories: Marital and familyrelationships in children’s development and psy-chopathology. Development and Psychopathology,14, 731 – 759. doi:10.1017/S0954579402004054

Cummings, E. M., & Davies, P. T. (2010). Maritalconflict and children: An emotional security per-spective. New York: Guilford.

Cummings, E. M., Goeke-Morey, M. C., & Papp, L.M. (2003). Children’s responses to everyday mari-tal conflict tactics in the home. Child Development,74, 1918 – 1929.

Davies, P. T., & Cummings, E. M. (1994). Mari-tal conflict and child adjustment: An emotionalsecurity hypothesis. Psychological Bulletin, 116,387 – 411. doi:10.1037/0033-2909.116.3.387

Davies, P. T., Cummings, E. M., & Winter, M. A.(2004). Pathways between profiles of familyfunctioning, child security in the interparentalsubsystem, and child psychological problems.Development and Psychopathology, 16, 525 – 550.doi:10.1017/S0954579404004651

Davies, P. T., Harold, G. T., Goeke-Morey, M. C.,Cummings, E. M., Shelton, K., Rasi, J. A., &Jenkins, J. M. (2002). Child emotional security andinterparental conflict. Monographs of the Societyfor Research in Child Development, 67, 1 – 115.doi:10.2307/3181513

DeBoard-Lucas, R. L., Fosco, G. M., Raynor, S. R., &Grych, J. H. (2010). Interparental conflict in con-text: Exploring relations between parenting pro-cesses and children’s conflict appraisals. Journalof Clinical Child and Adolescent Psychology, 39,163 – 175. doi:10.1080/15374410903532593

Emery, R. E., Fincham, F. D., & Cummings, E. M.(1992). Parenting in context: Systemic thinkingabout parental conflict and its influence on chil-dren. Journal of Consulting and Clinical Psychol-ogy, 60, 909 – 912. doi:10.1037/0022-006X.60.6.909

Field, A. (2009). Discovering statistics using SPSS(3rd ed.). London, England: Sage.

Fincham, F. D., & Linfield, K. J. (1997). A new lookat marital quality: Can spouses feel positive andnegative about their marriage? Journal of FamilyPsychology, 11, 489 – 502.

Fincham, F. D., & Osborne, L. N. (1993). Mar-ital conflict and children: Retrospect andprospect. Clinical Psychology Review, 13, 75 – 88.doi:10.1016/0272-7358(93)90009-B

Fosco, G. M., & Grych, J. H. (2007). Emotionalexpression in the family as a context for

children’s appraisals of interparental conflict.Journal of Family Psychology, 21, 248 – 258.doi:10.1037/0893-3200.21.2.248

Gabriel, B., & Bodenmann, G. (2006). Elter-liche Kompetenzen und Erziehungskonflikte.Eine ressourcenorientierte Betrachtung vonfamiliären Negativdynamiken [Parental com-petencies and conflicts related to parenting: Aresource-oriented view of negative dynamics infamilies]. Kindheit und Entwicklung, 15, 9 – 18.doi:10.1026/0942-5403.15.1.9

Goodman, A., Lamping, D. L., & Ploubidis, G.B. (2010). When to use broader internalis-ing and externalising subscales instead of thehypothesised five subscales on the Strengths andDifficulties Questionnaire (SDQ): Data fromBritish parents, teachers and children. Journalof Abnormal Child Psychology, 38, 1179 – 1191.doi:10.1007/s10802-010-9434-x

Goodman, R. (1997). The Strengths and Difficul-ties Questionnaire: A research note. Journal ofChild Psychology and Psychiatry, 38, 581 – 586.doi:10.1111/j.1469-7610.1997.tb01545.x

Gottman, J. M. (1993). The roles of conflict engage-ment, escalation, and avoidance in marital inter-action: A longitudinal view of 5 types of couples.Journal of Consulting and Clinical Psychology, 61,6 – 15.

Gottman, J. M. (1994). What predicts divorce? Therelationship between marital processes and mar-ital outcomes. Hillsdale, NJ: Lawrence ErlbaumAssociates.

Gottman, J. M., & Levenson, R. W. (1992). Maritalprocesses predictive of later dissolution: Behavior,physiology, and health. Journal of Personality andSocial Psychology, 63, 221 – 233.

Graber, E. C., Laurenceau, J.-P., Miga, E., Chango,J., & Coan, J. (2011). Conflict and love: Predict-ing newlywed marital outcomes from two interac-tion contexts. Journal of Family Psychology, 25,541 – 550. doi:10.1037/a0024507

Hendrick, S. S. (1988). A generic measure of relation-ship satisfaction. Journal of Marriage and Family,50, 93 – 98. doi:10.2307/352430

Holman, T. B., & Jarvis, M. O. (2003). Hostile,volatile, avoiding, and validating couple-conflicttypes: An investigation of Gottman’scouple-conflict types. Personal Relationships,10, 267 – 282. doi:10.1111/1475-6811.00049

Ihle, W., Esser, G., Schmidt, M. H., & Blanz, B.(2002). Die Bedeutung von Risikofaktoren desKindes- und Jugendalters für psychische Störun-gen von der Kindheit bis ins frühe Erwachse-nenalter [Childhood and adolescents predictors ofmental disorders from childhood to early adult-hood]. Kindheit und Entwicklung, 11, 201 – 211.doi:10.1026//0942-5403.11.4.201

Katz, L. F., & Woodin, E. M. (2002). Hostility, hostiledetachment, and conflict engagement in marriages:

How Much Positivity Is Needed for Children? 615

Effects on child and family functioning. ChildDevelopment, 73, 636 – 652.

Klasen, H., Woerner, W., Rothenberger, A., &Goodman, R. (2003). Die deutsche Fassungdes Strengths and Difficulties Questionnaire(SDQ-Deu): Übersicht und Bewertung ersterValidierungs- und Normierungsbefunde [TheGerman version of the Strengths and DifficultiesQuestionnaire (SDQ-Deu): Overview over firstvalidation and normative studies]. Praxis DerKinderpsychologie und Kinderpsychiatrie, 52,491 – 502.

Lindahl, K. M., & Malik, N. M. (2011). Marital con-flict typology and children’s appraisals: The mod-erating role of family cohesion. Journal of FamilyPsychology, 25, 194 – 201. doi:10.1037/a0022888

McCoy, K. P., Cummings, E. M., & Davies,P. T. (2009). Constructive and destructivemarital conflict, emotional security and chil-dren’s prosocial behavior. Journal of ChildPsychology and Psychiatry, 50, 270 – 279.doi:10.1111/j.1469-7610.2008.01945.x

Muthén, L. K., & Muthén, B. O. (1998). Mplus user’sguide (6th ed.). Los Angeles, CA: Author.

Ruffieux, M., & Bodenmann, G. (2011). Fragebo-gen zu Alltagspositivität [Questionnaire of positiveeveryday interaction]. Unpublished manuscript,University of Zurich, Switzerland.

Snyder, J. R. (1998). Marital conflict and child adjust-ment: What about gender? Developmental Review,18, 390 – 420. doi:10.1006/drev.1998.0469