BC

Xa

b

c

a

ARRA

KFCSBEC

1

sUanaiq1Crr(esifipC

KC

0h

Early Childhood Research Quarterly 28 (2013) 83– 93

Contents lists available at SciVerse ScienceDirect

Early Childhood Research Quarterly

idirectional longitudinal relations between father–child relationships andhinese children’s social competence during early childhood

iao Zhanga,b,c,∗

Department of Early Childhood Education, The Hong Kong Institute of Education, Hong Kong, ChinaSchool of Psychology, Beijing Normal University, ChinaDepartment of Psychology, University of Jyväskylä, Finland

r t i c l e i n f o

rticle history:eceived 16 November 2010eceived in revised form 4 May 2012ccepted 25 June 2012

a b s t r a c t

Using a two-year and three-wave cross-lagged design with a sample of 118 Chinese preschoolers, thepresent study examined bidirectional longitudinal relations between father–child relationships andchildren’s social competence. The results of structural equation modeling showed bidirectional effectsbetween father–child conflict and social competence. Higher conflict in father–child relationships at threemonths after preschool entry predicted lower levels of children’s social competence at the end of the first

eywords:ather–child relationshipsonflictocial competenceidirectional effectsarly childhood

preschool year over and above continuity in competence. Lower levels of social competence at threemonths after preschool entry also predicted higher father–child conflict at the end of the first preschoolyear over and above continuity in conflict. These bidirectional relations did not vary across child genderand proved robust to the inclusion of potential common factors.

© 2012 Elsevier Inc. All rights reserved.

hina

. Introduction

Historically, father ideals have evolved through nearly theame stages in China (Chuang, Moreno, & Su, in press) as in thenited States (Pleck & Pleck, 1997): from the ‘moral overseer’nd disciplinarian, through the economic provider and breadwin-er, and the modern involved dad and ‘buddy,’ to the father as

nurturing co-parent. Over the past three decades, increasingnterest has been shown in the nature, antecedents, and conse-uences of father–child relationships (e.g., Black, Dubowitz, & Starr,999; Cabrera, Hofferth, & Chae, 2011; Tamis-LeMonda, Shannon,abrera, & Lamb, 2004). There is a growing consensus amongesearchers that the involvement of fathers, or positive father–childelationships, is important for children’s development and welfareCabrera, Tamis-LeMonda, Bradley, Hofferth, & Lamb, 2000). How-ver, most of these studies have been carried out in Euro-Americanettings, with only limited research on father–child relationshipsn Chinese societies. Moreover, there is a growing awareness in the

eld that more studies need to be conducted on the transactionalrocess between fathers and children (Coley & Medeiros, 2007;oley, Votruba-Drzal, & Schindler, 2008; Ream & Savin-Williams,

∗ Correspondence address: Department of Early Childhood Education, The Hongong Institute of Education, 10 Lo Ping Road, Tai Po, New Territories, Hong Kong,hina.

E-mail address: xiao@graduate.hku.hk

885-2006/$ – see front matter © 2012 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.ecresq.2012.06.005

2005). The purpose of the present study is to examine bidirec-tional longitudinal relations between father–child relationships(i.e., closeness and conflict) and Chinese children’s social compe-tence during early childhood. Specifically, we assess whether thequality of father–child relationships is predictive of later socialcompetence among Chinese young children, and whether the chil-dren’s social competence is predictive of their later relationshipswith their fathers. To avoid awkward and cumbersome terminol-ogy, causal terms (e.g., effect, impact, and influence) are used in anoncausal sense to refer to predictive relations in this article.

1.1. The effect of father–child relationships on social competence

Positive growth in social competence is typical during earlychildhood. A variety of perspectives share the view that chil-dren’s social competence is at least partly grounded in theirrelationships with their fathers. First, from an attachment per-spective, father–child relationships reflect the quality of emotionalbonds between fathers and children (Lamb, 2002). Bowbly’s (1982)attachment theory posits that the internal working models derivedfrom early attachment guide children’s expectations, feelings, andcompetencies in later social interactions, especially when the cir-cumstances and the peer’s intent are ambiguous (Cassidy, Kirsh,

Scolton, & Parke, 1996). Children with positive father–child rela-tionships may develop models of caregivers as trustworthy andsupportive, and later approach others with positive attitudesand expectations. Consequently, they are likely to be socially

8 search

cdlttatsnhaehittdvpsso

fdasci&ts(WsfpCtbt

abs(isc1ccsntZfltip

mlrh(

4 X. Zhang / Early Childhood Re

ompetent. In contrast, children with negative relationships mayevelop models of the social world as untrustworthy and are less

ikely to be competent. Second, social-cognitive theorists arguehat fathers can influence their children’s social competence skillshrough modeling (Parke & Buriel, 1998). Fathers who are warmnd directive in their interactions may provide children with adap-ive models of how to resolve conflicts and solve problems. Withinuch positive interactions, children can learn about and practiceew social skills in a systematic and meaningful context, whichelps them develop social competence. By contrast, hostile andggressive fathers may provide children with maladaptive mod-ls. Children who engage in such negative interactions may viewostility and aggression as effective ways to settle disputes, which

n turn is likely to undermine their development of social compe-ence. Last but not least, the social capital perspective also suggestshat father–child relationships are one of the contexts in which chil-ren’s social competence can be developed (Parke, 2004). On thisiew, fathers may create opportunities for children to interact witheers and provide advice about peer choices and appropriate playtrategies, which is particularly important for the development ofocial competence in early childhood when children have limitedpportunities to seek out peers on their own.

A growing body of research supports the notion that closeather–child relationships contribute to the prediction of chil-ren’s social competence skills. For example, closeness (e.g., securettachment, warmth, and positive affect) in father–child relation-hips has been found to predict children’s self-regulatory abilities,ompetence behavior with peers, and peer acceptance and popular-ty during early childhood (e.g., Main & Weston, 1981; Verschueren

Marcoen, 1999). Moreover, the literature has further revealedhe substantial impact of father–child closeness on children’social competence independent of that of mother–child closenessVerschueren & Marcoen, 1999; Youngblade, Park, & Belsky, 1993).

ithin the body of research on Chinese father–child relation-hips, evidence has also emerged in support of the role that closeather–child relationships play in children’s social competence andeer liking (Zhang, 2011). For example, in a longitudinal study ofhinese toddlers, Chen, Wang, Chen, and Liu (2002) demonstratedhat paternal warmth negatively predicted children’s aggressiveehavior two years later, independent of maternal parenting prac-ices.

During the toddler and preschool years, children are able tossert their own will, and parents begin to assert control over theirehavior (Laible, Panfile, & Makariev, 2008). Conflict is thus con-idered a large part of parent–child relationships beyond infancyKlimes-Dougan & Kopp, 1999). Although the common perceptions that close relationships involve less conflict, recent research hasuggested the need to systematically differentiate conflict fromloseness in parent–child relationships (Klimes-Dougan & Kopp,999; Laible et al., 2008). For example, conflict was found toorrelate only at low to moderate levels with closeness in earlyhildhood (Laible et al., 2008; Pianta, 1992). Moreover, a recenttudy of Chinese young children showed that conflict and close-ess in parent–child relationships contributed independently tohe prediction of children’s behavior problems (Zhang, Chen, Zhang,hou, & Wu, 2008b). These findings suggest that closeness and con-ict may not be simply two end points on the same continuum. Inhe literature, however, little attention has been paid to conflictn young children’s father–child relationships. Consequently, theresent study focused on conflict as well as closeness.

Most researchers consider extreme and frequent conflict thearker of dysfunctional relationships across childhood and ado-

escence (e.g., Rubin, Burgess, Dwyer, & Hastings, 2003). The extantesearch has also converged in suggesting that parent–child conflictas a negative impact on child and adolescent social adjustmente.g., Rubin et al., 2003; Waschbusch, 2002). In this rich literature,

Quarterly 28 (2013) 83– 93

however, “parent–child conflict” is nearly always operationalizedusing measures focused solely on mothers or on the parents asa unit, failing to assess father–child conflict independently. Thus,little is known about the role that father–child conflict plays inchildren’s socio-emotional development.

Conflict in Chinese father–child dyads may be manifested whenfathers use authoritarian parenting styles such as physical coercion,verbal hostility, and non-reasoning/punitive behavior (Lau, Lew,Hau, Cheung, & Berndt, 1990). A recent study found that Chinesefathers were more authoritarian than their US counterparts (Porteret al., 2005). Moreover, a growing number of studies have docu-mented that Chinese fathers’ use of authoritarian parenting stylesis associated with externalizing behavior on the part of their chil-dren, such as aggression across early and middle childhood (Chenet al., 2002; Nelson, Hart, Yang, Olsen, & Jin, 2006; Xu, Farver, &Zhang, 2009; Yang et al., 2004). Hence, there are good reasons toassume that father–child conflict in the Chinese context may havean impact on young children’s social competence.

1.2. Influences of children’s social behavior on father–childrelationships

Given the importance of father–child relationships for children’ssocial development, further examination of the factors associ-ated with these relationships is warranted. Relationships betweenfathers and children develop as the two parties interact with eachother within families. It is not surprising that the attributes ofeach party contribute to the quality of these relationships. In theliterature, much attention has been paid to the role of fathers’attributes (e.g., education, income, job satisfaction) in father–childrelationships (Cabrera et al., 2011; Marsiglio, 1991; Pleck, 1997).The few studies that have examined the role of child attributesin father–child relationships have focused mainly on gender, age,temperament, and health status (e.g., Belsky, 1996; Marsiglio, 1991;Porter et al., 2005). Only three studies were located which investi-gated how children’s social behavior influenced their father–childrelationships (Coley & Medeiros, 2007; Coley et al., 2008; Ream &Savin-Williams, 2005); however, all these studies focused solely onadolescents.

Notably, the studies of young children have documented the roleof child behavior in both closeness and conflict in mother–childrelationships (Crockenberg, 1981; Laible et al., 2008). Becausefathers’ behavior may be influenced more by contextual factorsthan mothers’ (Doherty, Kouneski, & Erickson, 1996), father–childcloseness and conflict may be also influenced by children’s socialbehavior. In addition, historical views of fathers as disciplinarianssuggest that fathers tend to engage in disciplinary practices if chil-dren show problems or difficulties in adapting to the social world(Coley & Medeiros, 2007). Such practices often involve fathers’ con-trol over their children’s behavior, which in turn may lead to conflictin their interactions (Laible et al., 2008). It is thus possible thatsocially difficult children engage more in conflictual and less inclose relationships with fathers than their socially competent peers.

The studies of linkages between child temperament and father-ing in the Chinese context also suggest that Chinese children’s socialcompetence may have an impact on their father–child relation-ships. For example, it has been documented in Chinese preschoolersthat negative emotionality tends to elicit authoritarian parentingapproaches from their fathers (Porter et al., 2005; Yang et al., 2004).In Chinese girls, sociability has been found to correlate positivelywith fathers’ authoritative parenting, as manifested in warmth

and autonomy granting (Porter et al., 2005). Such findings sug-gest that social ineptness is very likely to induce conflict in Chinesefather–child relationships. Taken together, there are good reasonsto believe that Chinese children’s social competence skills may

search

cw

1a

tRramb(es(2etur

1

natl

bhpuiaaatacaphsTftthi

sspcceaitBcci

X. Zhang / Early Childhood Re

ontribute to the prediction of the quality of their relationshipsith their fathers.

.3. Bidirectional relations between father–child relationshipsnd child development

Transactional models of parent–child interactions posit bidirec-ional effects between parents and children (Bell, 1968; Patterson,eid, & Dishion, 1992; Sameroff, 1975). Such models argue for recip-ocal influences in which children are affected by parents’ behaviorsnd also elicit particular reactions from parents. Transactionalodels have been applied extensively to examine mutual effects

etween mother–child relationships and children’s social behaviore.g., Buist, Dekovic, Meeus, & van Aken, 2004). The limited researchxamining the reciprocal effects between father–child relation-hips and child development has focused mainly on adolescentsColey & Medeiros, 2007; Coley et al., 2008; Ream & Savin-Williams,005). To our knowledge, no studies have examined bidirectionalffects between father–child relationships and the social compe-ence of young children. Moreover, all three studies of adolescentssed unidirectional analyses to test each direction of effects sepa-ately and did not apply bidirectional analyses.

.4. Overview of the present study

The main goal of the present study is to examine longitudi-al relations between Chinese children’s father–child relationshipsnd social competence during early childhood by means of bidirec-ional analyses. To this end, we used a two-year and three-waveongitudinal sample of Chinese preschool children.

Since the late 1970s, traditional Chinese families have facedoth internal and external challenges. During this period, Chinaas implemented a reform and opening-up policy and a one-childolicy. As a result, Chinese family structures and relationships havendergone momentous changes. On the one hand, Chinese people,

ncluding parents, have become more open to Western individu-listic values and ideologies such as liberty, individual freedom,nd independence (Chen, Cen, Li, & He, 2005). Many families havelso begun to expand their educational goals to include helpingheir children develop social skills such as independence, self-ssertiveness, and individual autonomy. On the other hand, onlyhildren receive extraordinarily high levels of responsiveness andttention from their parents (Wu, 1992). Presumably due to thesearenting practices, the effect size was found to be particularlyigh for the influence of early attachment on Chinese children’social behavior (Schneider, Atkinson, & Tardif, 2001; Wu, 1992).hese practices, however, may have also resulted in potential harm-ul effects. A study of only children in Beijing documented thathey were less persistent, more egocentric, and less cooperativehan children with siblings (Jiao, Ji, & Jing, 1986). The question ofow families change to accommodate only children has become an

mportant concern in China.Three competing hypotheses were generated for the present

tudy: (1) father–child relationships would predict children’s laterocial competence (i.e., the father-driven effect), (2) social com-etence would predict later father–child relationships (i.e., thehild-driven effect), and (3) father–child relationships and socialompetence would be reciprocally related (i.e., the bidirectionalffects). To test these hypotheses, we applied a model comparisonpproach within a structural equation modeling (SEM) frameworkn a cross-lagged design. We expect bidirectional longitudinal rela-ions between father–child relationships and social competence.

ecause such relations have rarely been investigated in younghildren, we did not want to assume that both dimensions (i.e.,loseness and conflict) of father–child relationships would be sim-larly reciprocally related to social competence.

Quarterly 28 (2013) 83– 93 85

The robustness of the findings for our final model was subjectedto follow-up analyses in two key respects. First, we tested gen-der invariance in the final model to evaluate whether child gendermoderated the associations between father–child relationships andsocial competence. Although girls generally have stronger socialcompetence than boys (Golombok & Fivush, 1994), and fatherstend to have closer relationships with girls than boys (Zhang, inpress), gender differences have rarely been found in the associ-ations between father–child relationships and social competence(Coley & Medeiros, 2007; Coley et al., 2008; Ream & Savin-Williams,2005). Therefore we did not expect child gender to moderate thehypothesized relations tested in this study. Second, we examinedwhether potential candidates for “common causes” implicated inprevious research (Ahnert, Gunnar, Lamb, & Barthel, 2004; Belsky,1996; Marsiglio, 1991; Zhang & Chen, 2010) could account for anyobserved effects. To this end, we added child gender, age, temper-ament, parental education, and mother–child relationships to thefinal model, with the expectation that any cross-sectional or cross-lagged effect would prove robust to the inclusion of these controlvariables.

2. Method

2.1. Participants

Children were recruited from six nursery classes in three urbanpreschools (two public and one private) in Beijing, China and wereidentified with parental consent as they entered preschool (a min-imum age of two and a maximum age of three are required forChinese children to attend nurseries). Neither families nor teach-ers were compensated for their participation. At three months afterthe children’s preschool entry (Time 1 [T1]), 118 children (agerange = 27–39 months) participated in the study with their parents.Most of the children (95%) were only children, and those who werenot had one sibling. Data from three mothers and five fathers werenot obtained due to various reasons (e.g., divorce). Table 1 presentsthe sample characteristics of the children and their parents. Theparents can generally be considered highly educated as 54.3% of thefathers and 44.4% of the mothers had a bachelor’s or higher degree.The mean per capita monthly household income of the sample wasalso considerably above the urban mean in China ($106; NationalBureau of Statistics of China, 2006) at the time of the study.

At nine months after T1 (Time 2 [T2]), 11 children hadtransferred to other preschools and no longer participated. Mother-reported data were obtained on 106 children and father-reporteddata on 104 children. At one year after T2 (Time 3 [T3]), 19children had transferred to other preschools and no longer par-ticipated. Mother-reported data were obtained on 85 children andfather-reported data on 81 children. Attrition analyses revealed nosignificant difference in social competence, father–child relation-ships, or any of the control variables measured at T1.

2.2. Measures

The measures of father–child relationships and social compe-tence used here were initially developed in the United States.We translated them and then asked an independent translator toback-translate them. Discrepancies between the translation andback-translation were resolved through discussion. Using a vari-ety of formal and informal strategies (e.g., interviews with parents,repeated discussion in the research group, psychometric analy-

ses), three members of our research team carefully examined theitems in order to ensure cultural appropriateness. All the itemswere adopted, but item wording inappropriate for use in China wasmodified.

86 X. Zhang / Early Childhood Research

Table 1Demographic variables, child temperament, and mother–child relationships at T1.

Variables M (SD) (%)

Demographic variablesChild characteristics

Age (months) 33.3 (3.1)Gender

Boys 47.5%Girls 52.5%

EthnicityHan Chinese 92.4%Others (i.e., Man, Hui, or Mongolian

Chinese)7.6%

Only children or notOnly children 94.9%Others (i.e., children with one sibling) 5.1%

Family characteristicsPaternal age (years) 33.7 (3.4)Maternal age (years) 31.4 (3.1)Per capita monthly household income ($) 304 (219)Family structure

Nuclear families 97.4%Others (i.e., parents separated and living

alone or divorced)2.6%

Paternal education 3.55 (1.04)Secondary school or below .9%High school or vocational school degree 18.1%Vocational college degree 26.7%Bachelor’s degree 33.6%Master’s degree or above 20.7%

Maternal education 3.29 (1.02)Secondary school or below 3.4%High school or vocational school degree 19.7%Vocational college degree 32.5%Bachelor’s degree 33.3%Master’s degree or above 11.1%

Child temperament (approach/withdrawal) 23.43 (3.09)Mother–child relationships

Closeness 41.18 (4.27)Conflict 27.31 (6.77)

Note. All the variables were measured via maternal report. To assess per capitamonthly household income, mothers were asked to divide monthly householdip

2

t(mdttasita(tnsfsfltcrtca

gesting that the nested nature (children nested within classrooms;

ncome (including not only wages and salaries, but items such as bribe money anderformance bonus) by family size.

.2.1. Father–child relationshipsAt each of the three time points, fathers rated their own rela-

ionships with their child using the Child–Parent Relationship ScaleCPRS; Pianta, 1992). The CPRS is a 26-item self-report instru-

ent rated on a 5-point rating scale ranging from 1 (definitelyoes not apply) to 5 (definitely applies) to assess a parent’s percep-ion of his or her relationship with a target child. The items onhe scale were based on attachment theory, the Attachment Q-Set,nd observations of parent–child interactions (Pianta, 1992). Thecale contains three subscales: closeness (10 items), conflict (12tems), and dependence (4 items). Factor analysis confirmed thehree orthogonal factors in the Chinese version of the CPRS among

cross-sectional sample of 443 preschool children and their fathersZhang, Chen, Zhang, Zhou, & Wu, 2008a). Because the reliability ofhe dependence scale was low in both the US (Pianta, 1992) and Chi-ese samples (Zhang et al., 2008a), only the closeness and conflictcales were used in the present study. Closeness measures a father’seelings of affection and open communication with his child (e.g., “Ihare an affectionate, warm relationship with my child.”), and con-ict measures a father’s perception of negativity and conflict withhe child (e.g., “My child sees me as a source of punishment andriticism.”). Higher scores on the closeness scale indicate a closeelationship between father and child, whereas higher scores on

he conflict scale indicate a conflictual relationship. Cronbach’s ˛oefficients were .76, .80, and .82 in the closeness scale and .75, .71,nd .78 in the conflict scale at T1, T2, and T3, respectively.

Quarterly 28 (2013) 83– 93

2.2.2. Social competenceAt each of the three time points, mothers rated their child’s social

competence using the social competence subscale in the EarlySchool Behavior Rating Scale – Parent Form (ESBRS-P; Caldwell &Pianta, 1991). The subscale contains 16 items rated on a 4-point rat-ing scale to assess a parent’s perception of his or her child’s socialcompetence. Factor analysis confirmed the single factor in the Chi-nese version of the ESBRS-P among a cross-sectional sample of 443preschool children (Zhang, Ke, & Wang, in press). Mothers wereasked to rate how each behavior listed in the scale (e.g., “Playswell with other children”) described her child (1 = hardly ever,2 = sometimes, 3 = much of the time, 4 = almost always). Cronbach’s

coefficients were .75, .78, and .79 at T1, T2, and T3, respectively.Higher scores indicate stronger social competence.

2.2.3. Control variablesAt T1, mothers provided data on seven control variables,

including child gender, age, maternal and paternal education,temperamental approach/withdrawal, and mother–child closenessand conflict (income was not used as a control because it con-tained too many missing values). The first four control variableswere measured using a demographic questionnaire. Maternal andpaternal education were coded as follows: 5 = Master’s degree orabove, 4 = Bachelor’s degree, 3 = vocational college degree, 2 = highschool or vocational school degree, and 1 = secondary school or below.Approach/withdrawal (8 items, = .67) was assessed using theChinese version of Thomas and Chess’s (1977) Child Tempera-ment Questionnaire (CTQ). We focused on approach/withdrawalbecause it is related to children’s extraversion, which has beenlinked to father–child relationships in previous research (Belsky,1996). Moreover, approach/withdrawal has been found to be par-ticularly relevant to children’s social adaptation during the criticaltransition to preschool (Ahnert et al., 2004). High scores on the scaleindicate approach behaviors. Mother–child closeness and conflictwere assessed using the same scales (i.e., the closeness and conflictsubscales in the CPRS) as those used to assess closeness and con-flict in father–child relationships. Higher scores on the closenessscale ( = .70) indicate a close relationship between mother andchild, whereas higher scores on the conflict scale ( = .76) indicatea conflictual relationship.

2.3. Procedure

Data were collected during the fall (T1), the summer of thefollowing year (T2), and the summer of the second year (T3). Ateach time point, families were contacted by mail delivered by thechildren’s teachers and asked if they would be willing to partici-pate in a study of child development. After agreeing to participate,mothers and fathers were asked independently to fill out severalquestionnaires and a consent form and to return the completedquestionnaires to the teachers. The data collections at each timepoint were completed within a month. The recruitment and datacollection procedures were approved by the institutional reviewboard of the author’s university.

2.4. Data analytical strategies

Statistical analyses were conducted using Mplus 6 (Muthén &Muthén, 1998–2010). Missing data were handled using maximumlikelihood estimation with robust standard errors (MLR). For eachof the variables in this study, the intraclass correlation (ICC) wassmaller than .01, and the design effect was smaller than 1.05, sug-

classrooms nested within preschools) of the data was not a prob-lem (Muthén & Muthén, 1998–2010). We therefore did not includeclassroom or preschool effects in the analyses. To evaluate the

search Quarterly 28 (2013) 83– 93 87

m(a(TtlMdAfi

2

tifcsltcltr

(tfsrcs

2

nafcawCeibwpemctcM

3

3g

ssve

Table 2Means and standard deviations of father–child relationships and social competenceat each time point.

Variables Time points

T1 T2 T3

M SD M SD M SD

Father–child relationshipsCloseness 38.74 4.88 38.13 4.96 38.95 4.63Conflict 28.40 6.47 28.85 5.70 27.51 5.93

X. Zhang / Early Childhood Re

odel fit, we used chi-square values, the comparative fit indexCFI), the Tucker–Lewis index (TLI), the root mean square error ofpproximation (RMSEA), and the standardized root mean residualSRMR). A nonsignificant chi-square indicates good model-data fit.he general cutoffs for accepting a model were equal to or greaterhan .95 for CFI and TLI, equal to or less than .05 for the RMSEA, andess than .08 for the SRMR (Hu & Bentler, 1999). Because we used

LR, Satorra–Bentler-scaled chi-square difference tests were con-ucted when comparing nested models (Satorra & Bentler, 2001).

significant chi-square change indicates a significant difference int between two compared models.

.4.1. Bidirectional analysesTo assess the bidirectional relations between father–child rela-

ionships and social competence, we applied an SEM approachn a cross-lagged model. We tested and compared the followingour SEM models (see Fig. 1): (1) a stability model without anyross-lagged structural paths (M1), (2) a model with cross-laggedtructural paths from prior father–child closeness and conflict toater social competence (M2), (3) a model with cross-lagged struc-ural paths from prior social competence to later father–childloseness and conflict (M3), and (4) a model with both cross-agged structural paths representing reciprocal effects (M4). Notehat paths were only estimated between adjacent time points foreasons of statistical power and the total number of parameters.

The effect of father–child relationships on social competencei.e., the father-driven effect) would be supported if M2 had a bet-er fit to the data than M1; the effect of social competence onather–child relationships (i.e., the child-driven effect) would beupported if M3 better accounted for the data than M1; and theeciprocal effects between father–child relationships and socialompetence (i.e., the hypothesized bidirectional effects) would beupported if M4 had the best fit to the data among the four models.

.4.2. Follow-up analysesFollow-up analyses were conducted to examine the robust-

ess of the observed effects. First, to assess gender invariance, wepplied a multi-group SEM approach. We tested and compared theollowing two models: (1) an unconstrained model with all pathoefficients (i.e., stability, cross-sectional, and cross-lagged paths)llowed to be unequal across gender, and (2) a constrained modelith all path coefficients constrained to be equal across gender.omparable fit between the constrained and unconstrained mod-ls indicates gender invariance, whereas a significant differencen fit indicates significant gender difference in the associationsetween father–child relationships and social competence. Second,e tested whether the observed cross-sectional and cross-laggedaths remained significant after accounting for the shared influ-nces of child gender, age, temperamental approach/withdrawal,aternal and paternal education, and mother–child closeness and

onflict. Statistically, this was accomplished by including these con-rols as a set of correlated variables in the SEM that predicted eachomponent of a significant cross-sectional or cross-lagged path (seeasten et al., 2005).

. Results

.1. Descriptive statistics, correlations, mean-level changes, andender differences

Table 2 shows the descriptive statistics for father–child relation-

hips and social competence at each time point. The descriptivetatistics of the control variables are shown in Table 1. For all theariables, skewness was less than three, and kurtosis was less thanight, suggesting that non-normality of the data was not a problem

Social competence 40.66 5.92 41.16 5.43 43.28 5.82

Note. Ns = 81–118.

(Kline, 2005). Table 3 shows the Pearson correlations among all thevariables at each time point.

As shown in Table 3, father–child closeness correlated signifi-cantly with social competence within T1 but not within T2 or T3.Father–child conflict correlated significantly with social compe-tence within T2 but not within T1 or T3. T1 father–child closenesscorrelated significantly with social competence at both T2 and T3.T1 father–child conflict correlated significantly with T2 compe-tence, and T2 conflict correlated significantly with T3 competence.T1 competence also correlated significantly with T2 father–childconflict. The test-retest stability coefficients in social competenceand father–child closeness and conflict were at moderate levels.

To assess mean-level changes and gender differences infather–child relationships and social competence, we performeda series of 3 (Time: T1 vs. T2 vs. T3) × 2 (Gender: boys vs. girls)GLM (General Linear Modeling) repeated measures. The resultsrevealed significant main effects of Gender on father–child close-ness, F(1, 79) = 3.85, p = .05, �2

p = .05, and on social competence,F(1, 80) = 8.42, p = .005, �2

p = .10. Girls had closer relationshipswith fathers than boys (girls: M = 39.44, SD = .61; boys: M = 37.70,SD = .65). They were also more socially competent than boys (girls:M = 42.93, SD = .70; boys: M = 39.94, SD = .76). The main effect ofTime on social competence was also significant, F(2, 80) = 13.26,p < .001, �2

p = .14. Social competence increased significantly fromT1 (M = 40.29, SD = .64) and T2 (M = 40.91, SD = .58) to T3 (M = 43.11,SD = .62). No other significant main effects or interaction effectswere found in the analyses.

3.2. Bidirectional relations between father–child relationshipsand social competence

To examine the bidirectional relations between father–childrelationships and social competence, we tested and compared thefour models listed in Fig. 1. First, we tested the stability model(i.e., M1 without cross-lagged paths); second, we tested M2 (i.e.,the father-driven model); third, we tested M3 (i.e., the child-drivenmodel); and finally, we tested M4 (i.e., the bidirectional model).Table 4 shows the fit indices of M1–M4 and the differences in modelfit (i.e., Satorra–Bentler-scaled chi-square statistics).

As shown in Table 4, the chi-square differences between M1 andM2 and between M1 and M3 were both significant, suggesting thatM2 and M3 had better fit to the data than M1. The chi-square differ-ences between M1 and M4, between M2 and M4, and between M3and M4 were all significant, indicating that M4, which representedthe bidirectional longitudinal relations between father–child rela-tionships and children’s social competence, had the best fit to thedata among the four models. Table 4 also showed that M4 had a verygood fit to the data. Therefore, M4 was chosen as our final model.

Fig. 2 presents the final model (M4) with a completely standard-

ized solution. The paths from T1 father–child conflict to T2 socialcompetence and from T1 social competence to T2 father–childconflict were both statistically significant. The other father- andchild-driven cross-lagged paths were not significant. Father–child

88 X. Zhang / Early Childhood Research Quarterly 28 (2013) 83– 93

T1 Closeness

T1 Competence

T2 Closeness

T1 Conflict T2 Conflict

T3 Closeness

T3 Competence

T3 Conflict

T2 Competence

M3

T1 Closeness

T1 Competence

T2 Closeness

T1 Conflict T2 Conflict

T3 Closeness

T3 Competence

T3 Conflict

T2 Competence

M4

T1 Closeness

T1 Competence

T2 Closeness

T1 Conflict T2 Conflict

T3 Closeness

T3 Competence

T3 Conflict

T2 Competence

M2

M1

T1 Closeness T2 Closeness

T1 Conflict T2 Conflict

T3 Closeness

T3 Competence

T3 Conflict

T1 Competence T2 Competence

F hild ref

cwaT

3

sastfisr

ig. 1. Tested models for the evaluation of the bidirectional effects between father–cather–child conflict; Competence, social competence.

loseness was associated significantly with social competenceithin T1 but not within T2 or T3; father–child conflict was associ-

ted significantly with social competence within T2 but not within1 or T3.

.3. Gender invariance analyses

To assess whether the relations between father–child relation-hips and social competence differed significantly between girlsnd boys, we tested and compared the aforementioned uncon-trained and constrained models for the final model. Table 5 shows

he fit indices of the tested models and the difference in modelt. The chi-square difference between the unconstrained and con-trained models was not significant, suggesting that the observedelations did not differ significantly by gender.

lationships and social competence. Note. Closeness, father–child closeness; Conflict,

3.4. Control variable analyses

Our final analyses examined to what extent the cross-sectional and cross-lagged paths in the final model wouldremain significant when the control variables (i.e., child gender,age, approach/withdrawal, maternal and paternal education, andmother–child closeness and conflict) were added to the model. Wereran the final model four times, separately testing the robustnessof each significant cross-sectional or cross-lagged path by addingthe set of correlated control variables representing potential com-mon causes. The results showed that, for the cross-lagged path

from T1 father–child conflict to T2 social competence, the mag-nitude decreased to −.183 and remained significant (p = .008). Forthe cross-lagged path from T1 social competence to T2 father–childconflict, the magnitude increased to −.197 and remained significant

X. Zhang / Early Childhood Research Quarterly 28 (2013) 83– 93 89

Table 3Pair-wise Pearson intercorrelations among all variables.

Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

T11. Gendera –2. Age .05 –3. Maternal education .02 .01 –4. Paternal education .08 −.01 .63* –5. Approach/withdrawal .12 .00 .09 .12 –6. Mother–child closeness .02 .01 .06 .05 .26* –7. Mother–child conflict −.07 .09 −.14 −.05 .12 −.22* –8. Father–child closeness −.18* −.05 .06 −.09 .01 .17 −.16 –9. Father–child conflict .07 .13 −.02 −.03 .07 .13 .17 −.11 –10. Social competence −.25* −.09 .24* .10 .11 .48* −.36* .21* −.04 –

T211. Father–child closeness −.19* .13 .10 −.10 −.05 .09 −.01 .53* .09 .09 –12. Father–child conflict .09 .12 −.02 .03 .07 −.21* .18 −.14 .59* −.21* −.11 –13. Social competence −.28* −.03 .06 −.04 −.05 .40* −.33* .21* −.20* .56* .07 −.43* –

T314. Father–child closeness −.16 .03 −.03 −.22* −.10 .12 .03 .48* −.02 .09 .58* −.16 .16 –15. Father–child conflict .05 .09 .05 .08 .18 −.08 .08 −.11 .38* .04 −.10 .51* −.09 −.13 –16. Social competence −.31* −.11 .14 .01 −.04 .41* −.22* .26* −.03 .52* .19 −.36* .62* .11 −.14 –

Note. Ns = 81–118.a Effect coded: 1 = boys, −1 = girls.* p < .05.

Table 4Evaluation of the bidirectional effects between father–child relationships and social competence.

Tested models Fit indexes

�2 df CFI TLI RMSEA (90% CI) SRMR

No cross-lagged paths (M1) 35.455, p = .025 21 .935 .898 .076 (.027–.119) .101Cross prior relationships – later competence (M2) 24.851, p = .098 17 .965 .931 .063 (.000–.112) .074Cross prior competence – later relationships (M3) 25.710, p = .080 17 .961 .924 .066 (.000–.115) .073Both cross-lagged paths (M4) 15.052, p = .304 13 .991 .977 .037 (.000–.102) .050

Model comparisons ��2 �df

M2 vs. M1 10.996, p = .027 4M3 vs. M1 9.654, p = .047 4M4 vs. M1 20.718, p = .008 8M4 vs. M2 9.783, p = .044 4M4 vs. M3 11.145, p = .025 4

N ean sqc

(flabe(

T

Fp

ote. N = 118. CFI, comparative fit index; TLI, Tucker–Lewis index; RMSEA, root monfidence interval; ��2, �Satorra–Bentler-scaled �2.

p = .007). For the cross-sectional path between T2 father–child con-ict and T2 social competence, the magnitude decreased to −.317nd remained significant (p < .001). For the cross-sectional pathetween T1 father–child closeness and T1 social competence, how-

ver, the magnitude dropped to .053 and became nonsignificantp = .569).

Because the path between T1 father–child closeness and1 social competence fell below statistical significance, we

T1 Closeness

T1 Competence

T2 Closene

T1 Conflict T2 Confli

T2 Compete

R2 = .266 *

R2 = .362 *

R2 = .398 *

.593*

.521*

.550*

.211*

–.046

–.109

–.191*

–.184*

–.026

.052

ig. 2. The final model (M4) with a completely standardized solution. Note. N = 118. The conaths. Closeness, father–child closeness; Conflict, father–child conflict; Competence, soci

uare error of approximation; SRMR, standardized root mean square residual; CI,

examined the contribution of the control variables to that path inmore detail. The purpose of these analyses was to clarify concep-tually the processes involved with our putative control variables.We examined the effect of adding each control variable sepa-

rately. Age reduced the magnitude of the path to .205 (p = .035),approach/withdrawal reduced the magnitude to .209 (p = .030),maternal education reduced the magnitude to .201 (p = .032),and paternal education increased the magnitude of the path to

ss

ct

T3 Closeness

T3 Competence

T3 Conflict

nce

R2 = .371 *

R2 = .427 *

R2 = .269 *

.562*

.560*

.585*

–.339*

.017–.186 –.120

–.061

–.019–.125

–.126

.123

.128

trol variables were not included in this model. Dashed lines indicate nonsignificantal competence. *p < .05.

90 X. Zhang / Early Childhood Research Quarterly 28 (2013) 83– 93

Table 5Evaluation of gender invariance.

Tested models Fit indexes

�2 df CFI TLI RMSEA (90% CI) SRMR

Unconstrained model 29.542, p = .287 26 .984 .960 .048 (.000–.117) .076Constrained model 59.098, p = .153 49 .955 .939 .059 (.000–.108) .193

Model comparison ��2 �df

Unconstrained vs. constrained models 29.628, p = .160 23

N ean sqc

.mpw(cowblftcTs

4

orebcttv

rcfswcoycidAhisaswfcahtat

ote. N = 118. CFI, comparative fit index; TLI, Tucker–Lewis index; RMSEA, root monfidence interval; ��2, �Satorra–Bentler-scaled �2.

220 (p = .021). The magnitude of the path lost significance whenother–child closeness ( = .150, p = .123) or conflict ( = .174,

= .070) was added in the model. Although T1 social competenceas predicted by both closeness ( = .479, p < .001) and conflict

= −.360, p < .001) in mother–child relationships, T1 father–childloseness was not predicted by either closeness ( = .165, p = .082)r conflict ( = .132, p = .175) in mother–child relationships. Finally,hen gender was added in the model, the magnitude of the path

etween T1 father–child closeness and T1 social competence alsoost significance ( = .171, p = .078), and gender predicted both T1ather–child closeness ( = −.186, p = .042) and T1 social compe-ence ( = −.245, p = .004). Hence, child gender appeared to be theommon cause of father–child closeness and social competence at1, while mother–child closeness and conflict contributed to T1ocial competence but not to T1 father–child closeness.

. Discussion

Using a two-year and three-wave cross-lagged design, we setut to examine longitudinal associations between father–childelationships and Chinese children’s social competence duringarly childhood. The results of the SEM model comparisons showedidirectional relations between father–child conflict and socialompetence. The follow-up analyses suggested that the bidirec-ional relations operated similarly for boys and girls. Moreover,hese relations proved robust to the inclusion of potential “third-ariable causes” implicated in previous research.

The present study indicates that the influences of father–childelationships on children’s social competence may exist in earlyhildhood. Our final model showed a negative link betweenather–child conflict at three months after preschool entry andocial competence at the end of the first preschool year. In otherords, fathers who had conflictual father–child relationships had

hildren who showed relatively low levels of social competencever a nine-month follow-up period during the first preschoolear. This finding highlights the potential power of father–childonflict to engender change in social competence over and abovets continuity. It is possible that father–child conflict reflects chil-ren’s difficulties in forming emotional bonds with their fathers.ccording to Bowbly’s (1982) attachment theory, children whoave trouble bonding to their parents may develop internal work-

ng models of the social world as untrustworthy and rejecting andee themselves as unworthy of love, especially in ambiguous situ-tions (Cassidy et al., 1996). Consequently, they are less likely to beocially competent than peers who have strong, affectionate bondsith their parents. It may also be that father–child conflict involves

athers’ hostile and aggressive parenting behavior. From a social-ognitive perspective (Parke & Buriel, 1998), fathers who are hostilend aggressive may provide children with maladaptive models of

ow to solve social problems. Consequently, children may modelheir fathers’ negative behavior and view hostility and aggressions effective ways to settle social disputes, which in turn underminesheir social development. In contrast, fathers who are warm,

uare error of approximation; SRMR, standardized root mean square residual; CI,

directive, and rarely engage in conflictual interactions are likely toteach appropriate social skills that encourage children’s develop-ment of social competence. Taken together, our finding underscoresthe significance that father–child relationships may have for chil-dren’s social development during early childhood, and is consistentwith the findings of previous studies of young children in both Euro-American (Main & Weston, 1981; Verschueren & Marcoen, 1999)and Chinese contexts (Chen et al., 2002; Nelson et al., 2006; Zhang,2011). Moreover, because previous research has rarely paid specificattention to father–child conflict in early childhood, our findingscontribute to the existing body of knowledge by suggesting thepotential role that conflict in father–child relationships may playin young children’s social development.

The results also confirmed a negative longitudinal effect ofsocial competence on conflict in father–child relationships overand above continuity in conflict. Specifically, over a nine-monthperiod in their first preschool years, children who had lower levelsof social competence were more likely to form conflictual relation-ships with their fathers than their peers who had higher levels ofcompetence. This finding suggests that lack of adequate social skillsin the early preschool years may cause young children to experiencenegative interactions with their fathers. Fathers, as disciplinarians(Cabrera et al., 2000; Chuang et al., in press), may resort to disci-plinary practices and seek to assert control over their children’sbehavior when the latter display social problems or difficulties.Consequently, these children are likely to form negative relation-ships with their fathers. Again, given the scarcity of research onconflict in early father–child relationships, our findings furthercontribute to the literature by emphasizing children’s own con-tributions to their conflictual father–child relationships.

In contrast to past research that predominantly investigatedeither the effect of father–child relationships on child developmentor the effect of child characteristics on father–child relationships,the present study used transactional models of parent–child inter-actions to examine bidirectional effects. Recent research has begunto examine the reciprocal effects between fathers and adolescents(Coley & Medeiros, 2007; Coley et al., 2008; Ream & Savin-Williams,2005). To our knowledge, the present study represents the firstendeavor to explicitly delineate the transactional process betweenfather–child relationships and children’s social competence duringearly childhood. Thus, our results extend past research in highlight-ing the reciprocal manner in which the links between father–childrelationships and social competence function during children’searly preschool years.

No bidirectional longitudinal relations were found betweenfather–child closeness and children’s social competence. Children’sclose relationships with fathers did not prospectively predict theirsocial competence, and social competence did not prospectivelypredict father–child closeness either. Although social competence

and father–child closeness were reciprocally related within T1, ourresults indicated that this concurrent relation was accounted for bythe shared influence of child gender. That is, the relation betweensocial competence and father–child closeness at T1 seems not to be

search

abeaccnsssc1dIa

fcncwyfaucdcll(t2td

fatdsffciZbo

grritrafodgst

phn

X. Zhang / Early Childhood Re

causal link; instead, child gender emerged as a common cause ofoth competence and closeness. These findings suggest that differ-nt dimensions of father–child relationships may not be similarlyssociated with children’s social competence and that distinct pro-esses may be involved in the antecedents and consequences ofloseness and conflict in father–child relationships. Notably, close-ess and conflict in father–child relationships did not correlateignificantly either within or across time points in the presenttudy. Other studies of young children have also clearly demon-trated that conflict correlates only at low to moderate levels withloseness in parent–child relationships (Laible et al., 2008; Pianta,992; Zhang, in press) and that these two dimensions indepen-ently predict children’s behavior problems (Zhang et al., 2008b).

t would be useful for future studies systematically to examine hownd why conflict and closeness function in differing ways.

It is also noteworthy that the bidirectional effects betweenather–child relationships and social competence were found inhildren’s first but not second preschool year. Neither closenessor conflict in father–child relationships was predictive of socialompetence during the second preschool year, and competenceas also not predictive of father–child relationships during that

ear. These findings suggest that the mutual influences betweenather–child relationships and social competence may be morepparent in children’s first than second preschool year. They alsonderscore the importance of father–child transactions during theritical transition from home to preschool. For Chinese young chil-ren, the transition to preschool marks a qualitative change inontext with accompanying physical, cognitive, and social chal-enges. Many Chinese children have been found to exhibit highevels of social maladjustment when navigating this transitionZhang et al., 2008b). Because fathers generally assume an impor-ant role in children’s adaptation to the social world (Paquette,004), there is good reason to believe that fathers’ parenting prac-ices are likely to affect and be affected by children’s social behavioruring this transition.

The observed continuity of and bidirectional relations betweenather–child relationships and social competence in this study werelso tested for differential gender effects. Our analyses revealedhat both the continuity and bidirectional relations showed gen-er invariance, indicating that the observed effects were operatingimilarly across boys and girls in this sample. However, weound significant gender differences in the population means ofather–child closeness and social competence: girls had strongerompetence and closer father–child relationships than boys, whichs consistent with previous findings (Golombok & Fivush, 1994;hang, in press). It thus seems that the transactional processetween fathers and children is similar for girls and boys, althoughverall levels differ across gender.

Our common cause analyses explored how inclusion of childender, age, temperament, parental education, and mother–childelationships would affect the observed bidirectional effects. Theesults suggested that the inclusion of broad individual and fam-ly SES variables had no influence on the bidirectional effects andhat the bidirectional effects also survived control for mother–childelationships. In contrast, common cause effects seemed to play

greater role in the relations between social competence andather–child closeness. These findings likely reflect the importancef conflict, instead of closeness, in father–child transactions, at leasturing the early preschool years. Overall, our control analyses sug-est that the bidirectional effects between father–child conflict andocial competence are robust to several alternative explanationshat are implicated in previous research.

The extent to which the findings derived from the present sam-le apply to children elsewhere is not clear. First, families fromigher SES backgrounds were overrepresented in this study. In aumber of studies, low SES has been linked with children’s social

Quarterly 28 (2013) 83– 93 91

maladjustment (see a review in Bradley & Corwyn, 2002); it hasalso been found that low SES often limits fathers’ engagement withchildren (Black et al., 1999; Cabrera et al., 2011) and underminescloseness in father–child relationships (Belsky, 1996; Zhang, inpress). However, we do not know whether the transactional processbetween children and fathers functions similarly across differentSES backgrounds or not. Thus, it remains to be seen whether the pat-tern of our results is unique to urban Chinese samples or applicablealso to rural Chinese samples.

Second, it is also noteworthy that our sample consisted of pre-dominantly only children. Analyses of the only children in oursample revealed results very similar to those presented in thisarticle. For the sibling children, however, we were not able todemonstrate any conclusive results due to the extremely smallsample size. In a meta-analysis, Schneider et al. (2001) found thatthe effect size was particularly high for the influence of early attach-ment on Chinese children’s peer relations. They speculated thatthis might result from the one-child policy in China, because onlychildren receive extraordinarily high levels of attention and respon-siveness from their parents (Wu, 1992). Hence, it also remains tobe seen whether the pattern of our results is applicable to siblingchildren living in China or children growing up in a predominantlysibling society.

Finally, Chinese culture may also play a role in the observedbidirectional relations in this study. Traditional Chinese culturevalues the role of the father in disciplining children and helpingthem learn proper social skills. The famous “Three-Character Clas-sic” (Mo, 1996) states that: “It is the father’s fault if a child is notadequately educated.” Moreover, Lau et al. (1990) found that theChinese proverb “strict father, kind mother” continues to prevailin Chinese families. Therefore, Chinese fathers may be particularlysensitive or susceptible to social ineptness on the part of theirchildren and thus strive to influence their children’s social devel-opment. It would be valuable for future studies to use samples fromother cultures to replicate the present study and test whether thepattern of our results is unique to Chinese samples or universal tosamples across different cultures.

Several limitations of the present study should be noted. First,we did not test for the longitudinal measurement equivalenceof our core measures due to the small sample size. Because thesame instrument and informant were used three times for eachcore measure within a relatively short period (21 months), weassumed measurement equivalence over time. If a larger samplewere obtained in the future, longitudinal measurement invari-ance should be established before conducting any cross-laggedanalyses. Second, father–child relationships were assessed throughself-report questionnaires, which might have led to social desir-ability bias. Future replication studies might be advised to usealternative methods (e.g., laboratory or home observation). Third,we relied on mothers as the sole informants of social competence.It is widely acknowledged that mothers contribute valuable infor-mation to the examination of childhood functioning because theyare able to observe their child in multiple settings (Renk & Phares,2004). Yet rather than parents, it is teachers and peers who seea fuller variety of child behaviors manifested in school settings.Hence it would be valuable for future research to use other methods(e.g., teacher reports, observation) to complement mothers’ rat-ings. Fourth, although the mothers and fathers were asked to fillout the questionnaires independently, they might have consultedeach other, leading to contamination between parents’ ratings and,in turn, contributing to the observed bidirectional relations. Lastbut not least, this study considered only a small number of third-

variable causes. Alongside these variables, other individual (e.g.,negative emotionality, effortful control) or family process (e.g.,parents’ mental health status, marital/coparenting conflict) vari-ables may also contribute to father–child relationships and social

9 search

cfrvrr

fsncOfs(a2tai

frccsdrisad(pfi(sasbctsnepfocirdtr

A

STcrgl

2 X. Zhang / Early Childhood Re

ompetence (Belsky, 1996; Cabrera et al., 2011). It would be help-ul for future studies to assess whether the observed bidirectionalelations would also prove robust to control for such potential thirdariables. Moreover, because the effects of third variables cannot beuled out, drawing any causal inferences on the basis of the presentesults must be approached with caution.

Despite these caveats, this study provides strong evidenceor bidirectional linkages between father–child relationships andocial competence during children’s early preschool years. Ourested model comparisons allowed tests of the transactional pro-ess on the basis of both unidirectional and bidirectional analyses.ur results in favor of bidirectional longitudinal relations between

ather–child relationships and social competence are broadly con-istent with the transactional models of parent–child interactionsBell, 1968; Patterson et al., 1992; Sameroff, 1975). Our findingslso extend earlier research (Coley & Medeiros, 2007; Coley et al.,008; Ream & Savin-Williams, 2005) and suggest that such bidirec-ional relations function in early childhood. These relations werelso found similarly for boys and girls and proved robust to thenclusion of several potential common factors.

The results of this study also have important implicationsor parent education and child interventions. First, given theeciprocal relations between father–child relationships and socialompetence, programs that combine father education with ahild intervention may be more effective than programs focusingolely on one aspect. In his transactional model, Sameroff (1975)escribed an intervention model in terms of 3 Rs – remediation,edefinition, and reeducation – directed at the three interlock-ng parts of the system. Based on our findings, we suggest thatuch an intervention model could usefully be applied in the familynd preschool settings. More specifically, on the one hand, chil-ren’s social competence can be improved through remediationi.e., child intervention); this in turn may change the way fatherserceive their children, inducing a redefinition. Redefining a childor a father can change the way a father interacts with his child and,n turn, reduce father–child conflict. On the other hand, reeducationi.e., father education) may reduce conflict in father–child relation-hips and change the way fathers interact with their children. Such

change in fathers’ interactive behavior may enhance children’social competence and change fathers’ attributions of children’sehavior, again resulting in a redefinition. Obviously, programs thatombine father education with a child intervention are very likelyo produce the most pronounced effect. Second, the results of thistudy have important implications for helping Chinese childrenavigate the transition to preschool. Compared to those in West-rn nations, current family- or school-based programs in Chinalace little emphasis on child interventions or reducing conflict inather–child relationships. Instead, greater emphasis has been laidn encouraging parents to establish close relationships with theirhildren, which might be misleading in light of the present resultsndicating that father–child closeness may be much less stronglyelated to children’s social competence than conflict. To help chil-ren successfully navigate the transition to preschool, future parentraining programs in China might be advised to focus more oneducing conflict in father–child relationships.

cknowledgements

This study was supported by grants from the National Naturalcience Foundation of China (No. 31100751) and the National Keyechnologies R&D Program of China (No. 2012BAI36B00). This arti-

le was partially prepared while Xiao Zhang was a post-doctoralesearcher at University of Jyväskylä, with support provided by arant from the Academy of Finland to the Finnish Center of Excel-ence in Learning and Motivation Research (No. 213486).

Quarterly 28 (2013) 83– 93

References

Ahnert, L., Gunnar, M. G., Lamb, M. E., & Barthel, M. (2004). Transition to childcare: Associations with infant–mother attachment, infant negative emotion, andcortisol elevations. Child Development, 75, 639–650.

Bell, R. (1968). A reinterpretation of the direction of effects in studies of socialization.Psychological Review, 75, 81–95.

Belsky, J. (1996). Parent, infant and social-contextual antecedents of father–sonattachment security. Developmental Psychology, 32, 905–913.

Black, M. M., Dubowitz, H., & Starr, R. H. (1999). African American fathers in low-income, urban families: Development, behavior, and home environment. ChildDevelopment, 70, 967–978.

Bowlby, J. (1982). (2nd ed.). Attachment and loss: Attachment New York, NY: BasicBooks.

Bradley, R. H., & Corwyn, R. F. (2002). Socioeconomic status and child development.Annual Review of Psychology, 53, 371–399.

Buist, K. L., Dekovic, M., Meeus, W., & van Aken, M. A. (2004). The reciprocal relation-ship between early adolescent attachment and internalizing and externalizingproblem behavior. Journal of Adolescence, 27, 251–266.

Cabrera, N. J., Hofferth, S. L., & Chae, S. (2011). Patterns and predictors of father–infantengagement across race/ethnic groups. Early Childhood Research Quarterly, 26,365–375.

Cabrera, N. J., Tamis-LeMonda, C. S., Bradley, B., Hofferth, S. L., & Lamb, M. E. (2000).Fatherhood in the twenty-first century. Child Development, 71, 127–136.

Caldwell, C. B., & Pianta, R. C. (1991). A measure of young children’s problem andcompetence behaviors: The Early School Behavior Scale. Journal of Psychoeduca-tional Assessment, 9, 32–44.

Cassidy, J., Kirsh, S. J., Scolton, K. L., & Parke, R. D. (1996). Attachment and represen-tations of peer relationships. Developmental Psychology, 32, 892–904.

Chen, X., Cen, G., Li, D., & He, Y. (2005). Social functioning and adjustment in Chinesechildren: The imprint of historical time. Child Development, 76, 182–195.

Chen, X., Wang, L., Chen, H., & Liu, M. (2002). Noncompliance and childrea-ring attitudes as predictors of aggressive behavior: A longitudinal studyin Chinese children. International Journal of Behavioral Development, 26,225–233.

Chuang, S. S., Moreno, R. P., & Su, Y. Moving fathers from the “sidelines”: An explo-ration of contemporary Chinese fathers in Canada and China. In K. B. Chan & N.C. Hung (Eds.), Advances in research in Chinese families: A global perspective. NewYork, NY: Springer, in press.

Coley, R. L., & Medeiros, B. L. (2007). Reciprocal longitudinal relations between non-resident father involvement and adolescent delinquency. Child Development, 78,132–147.

Coley, R. L., Votruba-Drzal, E., & Schindler, H. (2008). Trajectories of parenting pro-cesses and adolescent substance use: Reciprocal effects. Journal of Abnormal ChildPsychology, 36, 613–635.

Crockenberg, S. B. (1981). Infant irritability, mother responsiveness, and social sup-port influences on the security of infant–mother attachment. Child Development,52, 857–865.

Doherty, W. J., Kouneski, E. F., & Erickson, M. F. (1996). Responsible fathering: Anoverview and conceptual framework. Report prepared for the Administration forChildren and Families and the Office of the Assistant Secretary for Planningand Evaluation of the United States Department of Health and Human Services,Washington, D.C.

Golombok, S., & Fivush, R. (1994). Gender development. New York, NY: CambridgeUniversity Press.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance struc-ture analysis: Conventional criteria versus new alternatives. Structural EquationModeling, 6, 1–55.

Jiao, S., Ji, G., & Jing, Q. (1986). Comparative study of behavioral qualities of onlychildren and sibling children. Child Development, 57, 357–361.

Klimes-Dougan, B., & Kopp, C. (1999). Children’s conflict tactics with mothers: Alongitudinal investigation of the toddler and preschool years. Merrill-PalmerQuarterly, 45, 226–242.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.).New York, NY: Guilford Press.

Laible, D., Panfile, T., & Makariev, D. (2008). The quality and frequency of mother-toddler conflict: Links with attachment and temperament. Child Development,79, 426–443.

Lamb, M. E. (2002). Nonresidential fathers and their children. In C. S. Tamis-LeMonda,& N. Cabrera (Eds.), Handbook of father involvement (pp. 169–184). Mahwah, NJ:Erlbaum.

Lau, S., Lew, W. J. F., Hau, K. T., Cheung, P. C., & Berndt, T. J. (1990). Relations amongperceived parental control, warmth, indulgence, and family harmony of Chinesein Mainland China. Developmental Psychology, 26, 674–677.

Main, M., & Weston, D. R. (1981). The quality of the toddler’s relationship to motherand to father: Related to conflict behavior and the readiness to establish newrelationships. Child Development, 52, 932–940.

Marsiglio, W. (1991). Paternal engagement activities with minor children. Journal ofMarriage and the Family, 53, 973–986.

Masten, A. S., Long, J. D., Roisman, G. I., Burt, K. B., Obradovic, J., Roberts, J. M., &Tellegen, A. (2005). Developmental cascades: Linking academic achievement,

externalizing and internalizing symptoms over 20 years. Developmental Psychol-ogy, 41, 733–746.

Mo, Z. (1996). Readings for children. Xinjiang, PRC: Youth Publication House.Muthén, L. K., & Muthén, B. O. (1998–2010). Mplus user’s guide (6th ed.). Los Angeles,

CA: Muthén & Muthén.

search

N

N

P

P

P

PP

P

P

P

R

R

R

S

S

S

T

X. Zhang / Early Childhood Re

ational Bureau of Statistics of China. (2006). China statistical yearbook 2006. Beijing,China: China Statistics Press.

elson, A. D., Hart, C. H., Yang, C., Olsen, J. A., & Jin, S. (2006). Aversive parent-ing in China: Associations with child physical and relational aggression. ChildDevelopment, 77, 554–572.

aquette, D. (2004). Theorizing the father–child relationship: Mechanisms anddevelopmental outcomes. Human Development, 47, 193–219.

arke, R. D. (2004). Development in the family. Annual Review of Psychology, 55,365–399.

arke, R. D., & Buriel, R. (1998). Socialization in the family: Ethnic and ecologicalperspectives. In W. Damon (Ed.), Handbook of child psychology: Social, emotional,and personality development (pp. 463–552). New York, NY: Wiley.

atterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Eugene, OR: Castalia.ianta, R. C. (1992). Child–Parent Relationship Scale (CPRS). Charlottesville, VA: Uni-

versity of Virginia.leck, J. H. (1997). Paternal involvement: Levels, sources, and consequences. In M. E.

Lamb (Ed.), The role of the father in child development (3rd ed., Vol. 3, pp. 66–103).New York, NY: Wiley.

leck, E. H., & Pleck, J. H. (1997). Fatherhood ideals in the United States: Historicaldimensions. In M. E. Lamb (Ed.), The role of the father in child development (3rded., pp. 33–48). New York, NY: Wiley.

orter, C. L., Hart, C. H., Yang, C., Robinson, C. C., Olsen, S. F., Zeng, Q., & Jin, S. (2005).A comparative study of child temperament and parenting in Beijing, China andthe western United States. International Journal of Behavioral Development, 29,541–551.

eam, G. R., & Savin-Williams, R. C. (2005). Reciprocal associations between adoles-cent sexual activity and quality of youth–parent interactions. Journal of FamilyPsychology, 19, 171–179.

enk, K., & Phares, V. (2004). Cross-informant ratings of social competence in chil-dren and adolescents. Clinical Psychology Review, 24, 239–254.

ubin, K. H., Burgess, K. B., Dwyer, K. M., & Hastings, P. D. (2003). Predictingpreschoolers’ externalizing behaviors from toddler temperament, conflict, andmaternal negativity. Developmental Psychology, 39, 164–176.

ameroff, A. (1975). Transactional models in early social relations. Human Develop-ment, 18, 65–79.

atorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic formoment structure analysis. Psychometrika, 66, 507–514.

chneider, B. H., Atkinson, L., & Tardif, C. (2001). Child–parent attachment and chil-dren’s peer relations. Developmental Psychology, 37, 86–100.

amis-LeMonda, C. S., Shannon, J. D., Cabrera, N., & Lamb, M. E. (2004). Fathers andmothers at play with their 2- and 3-year-olds: Contributions to language andcognitive development. Child Development, 75, 1806–1820.

Quarterly 28 (2013) 83– 93 93

Thomas, A., & Chess, S. (1977). Temperament and development. New York, NY:Brunner-Mazel.

Verschueren, K., & Marcoen, A. (1999). Representation of self and socioe-motional competence in kindergartners: Differential and combinedeffects of attachment to mother and to father. Child Development, 70,183–201.

Waschbusch, D. A. (2002). A meta-analytic examination of comorbid hyperactive-impulsive-attention problems and conduct problems. Psychological Bulletin, 128,118–150.

Wu, F. (1992). Social competency of only-children in China: Associations among mater-nal attachment, teacher attachment, and peer relationships (Unpublished doctoraldissertation). University of California, Los Angeles.

Xu, Y., Farver, J. M., & Zhang, Z. (2009). Temperament, harsh and indulgent parenting,and Chinese children’s proactive and reactive aggression. Child Development, 80,244–258.

Yang, C., Hart, C. H., Nelson, D. A., Porter, C. L., Olsen, S. F., Robinson, C. C., & Jin, S.(2004). Fathering in a Beijing, Chinese sample: Associations with boys’ and girls’negative emotionality and aggression. In R. Day, & M. E. Lamb (Eds.), Reconcep-tualizing and measuring fatherhood (pp. 185–215). Mahwah, NJ: Erlbaum.

Youngblade, L. M., Park, K. A., & Belsky, J. (1993). Measurement of young children’sclose friendship: A comparison of two independent assessment systems andtheir associations with attachment security. International Journal of BehavioralDevelopment, 16, 563–588.

Zhang, X. (2011). Parent–child and teacher–child relationships in Chinesepreschoolers: The moderating role of preschool experiences and the medi-ating role of social competence. Early Childhood Research Quarterly, 26,192–204.

Zhang, X. The effects of parent education and family income on mother–child rela-tionships, father–child relationships, and family environment in the People’sRepublic of China. Family Process, in press.

Zhang, X., & Chen, H. (2010). Reciprocal influences between parents’ perceptionsof mother–child and father–child relationships: A short-term longitudi-nal study in Chinese preschoolers. The Journal of Genetic Psychology, 171,22–34.

Zhang, X., Chen, H., Zhang, G., Zhou, B., & Wu, W. (2008a). Reliability and valid-ity of Early Father–Child Relationship Scale in China. Chinese Journal of ClinicalPsychology, 16, 13–21.

Zhang, X., Chen, H., Zhang, G., Zhou, B., & Wu, W. (2008b). A longitudinal study ofparent–child relationships and problem behaviors in early childhood: Transac-tional models. Acta Psychologica Sinica, 40, 571–582.

Zhang, X., Ke, X., & Wang, X. Reliability and validity of the Parent Form of the SocialCompetence Scale in Chinese preschoolers. Psychological Reports, in press.