Preschool Emotional Competence: Pathway to Social Competence?

Preschool Emotional Competence: Pathway to Social Competence?

Susanne A. Denham, Kimberly A. Blair, Elizabeth DeMulder, Jennifer Levitas, Katherine Sawyer,Sharon Auerbach-Major, and Patrick Queenan

Preschoolers’ (N5 143) patterns of emotional expressiveness, emotion regulation, and emotion knowledge wereassessed. Their contributions to social competence, as evidenced by sociometric likability and teacher ratings,were evaluated via latent variable modeling, both concurrently and across time. Moderation of key results byage and sex was also explored. Emotional competence assessed at 3 to 4 years of age contributed to bothconcurrent and kindergarten social competence. Even early in the preschool period, contributions of emotionalcompetence to social competence have long-term implications.

Emotional competence is crucial to children’s abilityto interact and form relationships with others (Parke,1994; Saarni, 1990). As Saarni (1990) stated, ‘‘We aretalking about how [children] can respond emotion-ally, yet simultaneously and strategically apply theirknowledge about emotions and their expression torelationships with others, so that they can negotiateinterpersonal exchanges and regulate their emo-tional experiences’’ (p. 116). Although these aspectsof emotional competence and others continue todevelop throughout the lifespan, preschool–agedchildren already are adept at several componentskills of emotional competence (Dunn, 1994). Theybegin to express a variety of vivid, but notincapacitating, emotions. As well, preschoolers arebecoming able to discern their own and others’emotional states, and to talk about them fluently.Finally, they are beginning to ‘‘up-’’or ‘‘down’’-

regulate emotions, depending on their goals(Denham, 1998; Saarni, 1990).

Each of the constituent elements of preschoolers’emotional competence contributes to the crucial taskof social competence from 2 to 5 years: successfulinitiation of peer relationships (Howes, 1987; Parker& Gottman, 1989). During this period, childrenbecome more skilled at interacting and at managingthe emotions so prominent during these initialforays into the peer arena. The young child whonegotiates this developmental task is in goodposition to continue thriving in a social world. Infact, successful, independent interaction with age-mates is a central predictor of later mental healthand well-being, beginning during preschool andcontinuing through during the grade school yearswhen peer reputations solidify (Denham & Holt,1993; Parker & Asher, 1987; Robins & Rutter, 1990).

It is important to specify the ways emotional andsocial competence are highly related but still separ-able constructs. Rose-Krasnor’s (1997) recent theo-rizing is useful in this regard. At the most abstractlevel, she defined the construct of social competenceas effectiveness in interaction, the result of organizedbehaviors that meet short- and long-term develop-mental needs. In the case of preschoolers, sociallycompetent behaviors would be organized, as alreadynoted, around the central developmental tasks ofpositive engagement and self-regulation during peerinteraction. Within this theoretical view of socialcompetence, it also is necessary to decide whether tofocus on self or otherFAre we interested inaccessing the child’s success in meeting personalgoals, or their interpersonal connectedness? Ourfocus here is on adaptation with peers and in theschool environment, the ongoing social relations andinteractions of children in their social group (Rubin

r 2003 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/2003/7401-0017

Susanne A. Denham, Department of Psychology, George MasonUniversity; Kimberly Blair, Department of Counseling, Psychol-ogy, and Special Education, Duquesne University; ElizabethDeMulder, Initiatives in Educational Transformation, GeorgeMason University; Jennifer Levitas, Katherine Sawyer, and SharonAuerbach-Major, Department of Psychology, George MasonUniversity; Patrick Queenan, Devereux Foundation.

The National Institutes of Mental Health grant R01MH54019supported the research discussed here while all authors were atGeorge Mason University. A preliminary version was presented atthe 1997 meetings of the Society for Research in Child Develop-ment. The authors thank the many children, as well as nurseryschool and daycare directors and teachers, who gave so muchtime and goodwill so that we could learn about emotionalcompetence, as well as Krysti Batt, Cameron Caswell, SarahCaverly, Rene Hackney, Teresa Mason, Rebecca Sears, andMeredith Vickery, who assisted in gathering data presented here.Wynne Chin was an invaluable source of support with respect toLVPLS analyses.

Correspondence concerning this article should be addressed toSusanne A. Denham, Department of Psychology, George MasonUniversity, 4400 University Drive, Fairfax VA 22030-4444 (703-993-4081). Electronic mail may be sent to [email protected].

Child Development, January/February 2003, Volume 74, Number 1, Pages 238–256

& Ross, 1988). Finally, at the most detailed level,Rose-Krasnor’s model of social competence includesspecific social, emotional, and cognitive abilities,behaviors, and motivations that are primarily in-dividual. It is at this level that specific componentsof emotional competence may contribute to moregeneral social competence.

The importance of these social competence out-comes should not be underestimated. Along withmarking successful development and predictinglater well-being, social competence is increasinglyrecognized as vital to school readiness (Carlton &Winsler, 1999). For example, socially competentkindergartners are more successful than their lesscompetent counterparts in developing positive atti-tudes about and adjusting to school, and they getbetter grades and achieve more (Birch & Ladd, 1997;Ladd, 1990; Ladd, Birch, & Buhs, 1999; Ladd,Kochenderfer, & Coleman, 1996). More specifically,social-emotional indicators, including positive inter-actions with teachers, positive representations of selfderived from attachment relationships, emotionknowledge, emotion regulatory abilities, social skills,and nonrejected peer status, often uniquely predictacademic success when other pertinent variables,even earlier academic success, are already taken intoaccount (e.g., Carlton, 2000; Howes & Smith, 1995;Izard et al., 2001; Jacobsen & Hofmann, 1997; O’Neil,Welsh, Parke, Wang, & Strand, 1997; Pianta, 1997,Pianta, Steinberg, & Rollins, 1995; Shields et al.,2001). For example, when children enter school withfriends, are well liked, are able to make and sustainnew friendships, and are able to initiate positiverelationships with their teachers, they also feel morepositive about school, participate in school more,and achieve more than children who are notdescribed this way. Kindergartners who are victi-mized by peers or are aggressive, in contrast, havemore school-adjustment problems and are at risk fora potential cascade of problems, including schooldifficulties, delinquency, and drug abuse (Gagnon,Craig, Tremblay, Zhou, & Vitaro, 1995; Haapasalo &Tremblay 1994; Kochenderfer & Ladd, 1996; Trem-blay, Pagani-Kurtz, Masse, Vitaro, & Pihl, 1995).

Because social competence is so important, andapparently so intertwined with aspects of emotionalcompetence, we need to better understand itsemotional prerequisites. To maximize social compe-tence, researchers and others must carefully scruti-nize how elements of emotional competence worktogether to allow preschoolers to mobilize personaland environmental resources. First, children’s emo-tional expressiveness is a central aspect of theiremotional competence. For example, positive affect

is important in the initiation and regulation of socialexchanges; sharing positive affect may facilitate theformation of friendships and render one morelikable (Denham, McKinley, Couchoud, & Holt,1990; Sroufe, Schork, Motti, Lawroski, & LaFreniere,1984). Conversely, negative affect, especially anger,can be problematic in social interaction (Denhamet al., 1990; Rubin & Clark, 1983; Rubin & Daniels–Byrness, 1983). Children who express and experi-ence relatively more positive than negative emotionsare rated higher by teachers on dimensions such asfriendliness and assertiveness, and lower on dimen-sions such as aggressiveness and sadness. Suchchildren also respond more prosocially to peers’emotions and are seen as more likable by their peers(Bower, 1985; Denham, 1986; Denham et al., 1990;Denham, Renwick, & Holt, 1991; Eisenberg, Fabes,Murphy et al., 1996; Sroufe et al., 1984; Strayer, 1980).

From a functionalist perspective (Campos &Barrett, 1984), it is easy to envision why children’senduring patterns of emotional expressiveness aresuch potent intrapersonal supports for, or road-blocks to, interacting with age-mates. If a child isoften emotionally negative, especially angry, it is nowonder when his peers flatly assert, as did one ofour 3-year-old participants, ‘‘He hits. He bites. Hekicked me this morning. I don’t like him.’’ But thehappier child’s smile and body language are likebeacons signaling, ‘‘Come join me’’ to adults andage-mates alike.

Emotion knowledge is the second key componentof young children’s emotional competence. Childrenwho understand emotions also are more prosociallyresponsive to their peers, and rated as more sociallycompetent by teachers, and more likable by theirpeers (Denham, 1986; Denham et al., 1990; Strayer,1980). Children who can identify the expression on apeer’s face or comprehend the emotions elicited bycommon social situations are more likely to reactprosocially to their peers’ displays of emotion. Forexample, if a preschooler sees one peer bickeringwith another and correctly deduces that the peersuddenly experiences sadness or fear, rather thanintensified anger, she may comfort her friend ratherthan retreat or enter the fray. Interactions with suchan emotionally knowledgeable age-mate wouldlikely be viewed as satisfying, rendering thatplaymate more likable. Similarly, teachers are likelyto be attuned to the behavioral evidence of suchemotion knowledgeFthe use of emotion language,the sympathetic reactionFand to evaluate it posi-tively. Emotion knowledge allows preschoolers toreact appropriately to others, thus bolstering theirrelationships. In earlier work, we found links

Preschool Emotional Competence 239

between emotion knowledge and expressivenessand prosocial behavior and peer status (Denham,1986; Denham et al., 1990; Denham & McKinley,1993).

Emotion regulation is the third vital aspect ofemotional competence. When the intensity, duration,or other parameters of the experience and expressionof emotion are ‘‘too much’’ or ‘‘too little’’ to meetgoals and expectations of the child or social partners,emotion regulation is needed. During the preschoolperiod, emotion regulation becomes both necessary,because of the increasing complexity of children’semotionality and the demands of their social world,and possible, because of their increased comprehen-sion and control of their emotionality (Denham,1998; Lewis, Sullivan, & Vasen, 1987).

Thompson (1994) suggested the following defini-tion: ‘‘Emotion regulation consists of the extrinsicand intrinsic processes responsible for monitoring,evaluating, and modifying emotional reactions y toaccomplish one’s goals’’ (pp. 27–28). Given our focuson emotion regulation’s contribution to evaluationsof preschoolers’ social competence made by poten-tial social partners, we focused on children’s abilityto overtly modify their emotional reactions: that is,their coping (Denham, 1998; Gross, 1998; Losoya,Eisenberg, & Fabes, 1998).

Young preschoolers often need external supportto become skilled at such coping regulation; care-givers’ support allows their strategies to be maxi-mally effective (Kopp, 1989). As they become moreautonomous and more capable of cooperation, olderpreschoolers can collaborate with caregivers’ effortsto regulate their emotions. But, failures of emotionregulation can still be seen throughout the preschoolperiod, in outbursts of temper and distress that mayimpede social interaction (Bridges & Grolnick, 1994;Cole, Michel, & Teti, 1994).

Maternal and teacher reports of such emotionregulatory coping are associated with social compe-tence (e.g., Eisenberg et al., 1995). For example, someinvestigators have discovered negative associationsof ineffective emotion regulation and high emotionalnegativity with preschoolers’ and older children’sadult-rated social competence (Eisenberg et al., 1995;Eisenberg, Fabes, Murphy, et al., 1996, Eisenberg,Fabes, Guthrie, et al., 1996; Eisenberg, Fabes, et al.,1997; Eisenberg, Guthrie, et al., 1997). In particular,sympathetic responsiveness is related to both dysre-gulation and negativity, and their interaction (Eisen-berg, Fabes, Murphy, et al., 1996; Murphy, Shepard,Eisenberg, Fabes, & Guthrie, 1999). Because of earlierresearch findings, we focused on lapses of emotionregulation via emotional venting and the interaction

of such venting and children’s expressed anger. Wereasoned that social competence would be mostevident when venting and expressed anger werelow, both separately and in combination.

Extant research has examined the constituents ofemotional competence piecemeal, one or two at atime. None that we know have looked at all threecomponents of emotional competence outlined here,to assess their unique contributions to multipleassessments of social competence. We need to betterspecify how these critical aspects of emotionalcompetence contribute, in concert, to social compe-tence, particularly to make findings useful forparents and early childhood educators who wish tofoster social competence in their youngsters. Accord-ingly, our main goal in this study was to examine amodel of the prediction of social competence byyoung children’s own emotional competence, toboth replicate and extend our earlier work (seeFigure 1 for the overall hypothesized model).

Enduring patterns of emotional expressiveness,emotion situation knowledge, and emotion regula-tory coping (i.e., the absence of dysregulated ventingin response to experienced problem situations, aswell as the absence of dysregulated negative reac-tions to the emotional expression of peers) were usedto predict indices of social competence. We at-tempted to gain a full picture of each participant’ssocial competence by assessing multiple dimensions,including likability, oppositionality, isolation, andsensitive cooperation, from the differing perspec-tives of peers and teachers, and across periods frompreschool to kindergarten. Both concurrent predic-tion of social competence at age 3 to 4 andlongitudinal prediction of social competence inkindergarten were assessed, with age 3 to 4emotional competence components as predictors.The longitudinal analyses allow us to make clearerstatements about direction of effect in the emotionalcompetence and social competence linkage. Weevaluated our model via latent variable partial leastsquares (LVPLS) modeling so that we could not onlyexamine the separate contributions of latent vari-ables for each aspect of emotional competence, butalso their combined contributions. Finally, weevaluated a competing model, in which age 3 to 4social competence predicted aspects of emotionalcompetence, which in turn predicted kindergartensocial competence; perhaps early capabilities tointeract with teachers and peers concurrently pro-mote aspects of emotional competence.

We also examined possible age and sex differ-ences in the contribution of preschoolers’ emotionalcompetence to their social competence. Thus, we

240 Denham et al.

evaluated moderation of the contribution of ele-ments of emotional competence by child sex andage. Regarding gender moderation, the developmentof boys’ social competence seems particularlysensitive to individual differences in emotionalcompetence (Murphy, 1999; Denham, Mitchell-Cope-land, Strandberg, Auerbach, & Blair, 1997; Eisenberg,Fabes, Guthrie, et al., 1996). Also, based on oursupposition that sex of child would moderate thoseaspects of emotional competence that differ most interms of gender, we were interested especially inhow child sex moderated the contributions ofindices of emotion regulation. However, it couldalso be that patterns of expressiveness wouldcontribute to others’ evaluations of boys’ socialcompetence differently from those of girls; perhaps

boys are expected to exhibit some level of dominantnegative emotion, so that such expressivenesspatterns would be most notable in evaluations ofgirls’ social competence. Along the same lines,perhaps strengths in emotion knowledge particu-larly bolster the social competence of boys, whosepathways to social competence appear less multiplydetermined.

As well, based on our supposition that age wouldmoderate those aspects of emotional competence onwhich it exerts effects, we were interested in how itmoderated the contributions of emotion knowledgeand emotion regulation indices. The nascent socialcompetence of younger, as opposed to older,preschoolers also might be supported or compro-mised by unique aspects of emotional competence or

Sex × Emotion

Regulation

EmotionalExpressiv. ×

EmotionRegulation

Emotion Knowledge

Emotion Regulation

Kindergarten Social

Competence

EmotionalExpressiv.

Age 3 to 4 Social

Competence

Sex × Emotion

Knowledge

Sex × Emotional Expressiveness

Age × Emotion

Knowledge

Age × Emotion

Regulation

Figure 1. Overall hypothesized model, including cross-time relations for social competence, with main effect contributions of emotionalcompetence to social competence, as well as contributions moderated by sex, age, and, for the contribution of emotion regulation,emotional expressiveness.

Preschool Emotional Competence 241

its lack (see Denham, 1998, for age- and gender-dependent patterns development for emotionalcompetence). In particular, because both emotionknowledge and emotion regulation become moresophisticated with age (with emotional expressive-ness changing more in quality than quantity;Denham, 1998), we could expect that the contribu-tions of these two aspects of emotional competencemight afford the younger preschooler with special‘‘boosts’’ toward effective social interaction.

Method

Participants

We examined the emotional and social compe-tence of 143 predominantly Caucasian, middle-income 3- and 4-year-olds (75 boys) in a multisetting,multimethod design (age: M5 46 months, SD5 4.88months; range5 32–59 months) at the time ofobservation during preschool. The median annualincome range for these families was $30,000 to$50,000. Of the total sample, 74% were described asCaucasian. The modal level of education for motherswas graduation from college, and 88% of thechildren lived in two-parent homes. By the kinder-garten year, we were able to reconnect with only 104of these children, but there were no differences onany study variables measured when children wereage 3 to 4 between children who continued the studythrough kindergarten and those who did not.

Procedure

To obtain the data discussed here we bothinterviewed and observed children, and adminis-tered questionnaires to their parents at home and toteachers in preschool or daycare settings. We thussought to view various aspects of the children’semotional and social competence at two time points,during preschool and kindergarten.

Various preschool and daycare centers in theWashington, D.C., metropolitan area were targetedbased on past liaison relationships and directorwillingness to participate. Recruitment was a multi-tiered process. First, graduate research assistantsand their research advisor met with center directorsand teachers to inform them of the role researcherswere to play in their classrooms during the year, aswell as to build rapport. Second, letters brieflyexplaining the research study and encouragingparents to be involved were sent home to thefamilies of each child in the appropriate age range.Third, team members attended parent meetings and

open houses to further personalize the researchefforts. Approximately 25% of families recruiteddecided to participate in the study. More than 95% ofthose evidencing interest in this family componentcontinued on to complete participation, includingthe elements reported on here. Because of this lowinitial interest rate, we compared these families’ withfamilies from two earlier samples, one of which alsoinvolved heavy family participation (Denham et al.,1997) and one of which did not involve the burden ofhome visits, with a greater than 75% participation(Denham et al., 1990). T tests comparing the currentsample and these two earlier samples on demo-graphic and study variables showed virtually nosignificant differences in mean levels across thesamples. Thus, we concluded that the low participa-tion rate for this study did not result in biases. Activeinformed consent was obtained from the parents ofall children in the study.

Mothers received a packet of questionnairesduring a home visit in which observations unrelatedto this study were performed. Home visitorscollected the questionnaires in person later, provid-ing clarification as needed. Children were observedin their preschool for their expression of emotionsand reactions to peer emotions. Teachers alsocompleted questionnaires, one of which was ger-mane to this study, and classmates provided socio-metric ratings for each participant. Emotionalcompetence measures were administered whenchildren were 3 to 4 years old, with social compe-tence measured both at 3 to 4 years and 5 to 6 years(i.e., kindergarten).

Measure of Emotional Competence: Preschool, Ages 3 to 4

Previously developed observational methodolo-gies were used to examine (a) the emotionsexpressed by children during free play, and (b) theirreactions to their peers’ emotions during free play.Semistructured interviews were used to assesschildren’s knowledge about their emotions. Carewas taken for assistants familiar to the children (butnot those who observed them in the home) toadminister all measures in the classroom. A pre-viously developed questionnaire was used formaternal report of children’s typical styles ofemotion regulation.

Naturalistic observation of emotions in the preschool

Children were observed in their classroom set-ting, during free play, by coders using laptop

242 Denham et al.

computers and software designed by Roberts (2002).Researchers made event-based recordings for 5-mintrials. Participants were each observed for twelvesessions of two 5-min trials over approximately 6weeks. Observers obtained twenty-four 5-min ob-servations for each child (12 with the child as focal,counting his or her emotions, and 12 with the childas target, counting his or her reactions to peers’emotions).

Each participating child was viewed as the focalchild in the first 5-min interval of each session.During these intervals, the focal child’s happy, sad,angry, afraid, tender, hurt, other, and neutral emo-tional displays were recorded. These emotions wereoperationally defined by facial, vocal, and motorindices (see Denham, 1986; Denham et al., 1990). Theaverage kappa for categorization of emotion dis-plays, across raters, was .85 (see also Denham et al.,1997). Each emotion was quantified as its percentageof total emotion displays. Because afraid, hurt,tender, and ‘‘other’’ emotions each accounted forless than 3% of emotion displays, these emotionswere not considered further.

Understanding of emotion. Children’s understand-ing of emotion (identification of emotions unequi-vocally appropriate to certain situations, andinferences of emotions in equivocal situations) wasassessed using puppets with detachable faces thatdepicted happy, sad, angry, and afraid expressions.Because these measures have been described indetail elsewhere (Denham, 1986; Denham &Couchoud, 1990a, 1990b; Denham, Zoller, &Couchoud, 1994), they are summarized as follows:In the emotional situation identification tasks, thepuppeteer made standard facial and vocal expres-sions of emotions while enacting an emotion–ladenstory. Each child was asked to place on the puppet theface that depicted the puppet’s feeling in the situation.

The first emotional situation identification taskexplored how well children know others’ feelings ineight common situations that elicited unequivocalemotional reactions, such as happiness at being givenan ice cream cone or fear at having a nightmare(Borke, 1971; Denham, 1986). The second emotionalsituation identification task measured how wellchildren identified others’ feelings in situations wherethe other feels differently from the child under similarcircumstances. All situations could elicit one of twoemotions, as in feeling happy or afraid to get into aswimming pool (Denham, 1986; Denham & Cou-choud, 1990b). Children’s mothers had reported, via aforced–choice questionnaire, how their childrenwould feel in 12 such vignettes. For each vignette,their response determined the emotion expressed by

the puppet. That is, the emotion expressed by thepuppet did not match what the mother had reportedthe child would feel. Emotion situation items weresummed for the unequivocal emotion situationsaggregate (a5 .79 for eight items), equivocal emotionsituations pitting positive and negative emotions(a5 .84 for six items), and equivocal emotion situa-tions pitting two negative emotions (a5 .69 for sixitems).

Lapses of emotion regulation: Maternal reports ofcoping behavior via venting. Children’s coping beha-vior when faced with emotionally difficult situationswith peers was assessed with coping items devel-oped by Eisenberg, Fabes, Nyman, Bernzweig, andPinuelas (1994). Mothers indicated on a 7-pointscale, with 1 indicating never and 7 indicatingusually, how often the child would engage in eachof several types of coping behavior when confrontedwith an emotionally arousing problem situation. Asalready noted, for this study we were interested onlyin dysregulated emotion regulationFthat is, copingvia venting emotional expressiveness or behavior.For this index, item content reflected: instrumentalaggression (e.g., hitting another child to obtain agoal), emotional intervention (e.g., crying to elicithelp from others), emotional outbursts (e.g., cryingto release frustration), and emotional aggression(using aggression to release frustration). Cronbach’salpha for the four-item summed venting scaleequaled .67.

Lapses of emotion regulation: Naturalistic observationof coping behavior via negative reactions to peers’emotions. Our rationale in including the followingindex was that others’ emotions, particularly at thisage, can be dysregulating and thus often requirecoping efforts. Accordingly, in the second 5-mininterval of each observational period describedearlier, children within approximately 3 ft of thestudy child were observed as the focal entity whilethe study child was observed for his or her reactions.Again, the context of these observations was alwaysfree play. Target reactions that denoted dysregulatedcoping included: opposite affect match (e.g., happywhen focal is sad), matching negative affect, dis-playing hurt feelings, and antisocial (e.g., volition-ally exacerbating focal child’s problem; Denham,1986; Denham et al., 1990).

For indices of children’s reactions to emotions, thetotal for each reaction was divided by the number ofemotions reacted to, to generate a proportion ofoccurrence for each. These reactions were thensummed to produce an aggregate parallel to thematernally reported data on venting coping. Aver-age kappa for categorization of reactions to peers’

Preschool Emotional Competence 243

emotion displays, across raters, was .82 (see alsoDenham et al., 1997). Cronbach’s alpha for this four-item rationally derived aggregate equaled .44.

Measures of Social Competence: Preschool (Ages 3 to 4)and Kindergarten (Ages 5 to 6)

The present study assessed social competence viathree methods, each addressing unique aspects ofthis multifaceted construct (Rose-Krasnor, 1997). Allof the children in the class were asked to rate theirpeers on a continuum of likability to obtain ameasure of popularity. Teachers also rated eachchild’s social competence in the classroom, asviewed from their perspective. All measures wereadministered both in preschool and in kindergartenyears.

Teacher-rated social competence. Teachers and day-care providers rated children’s social competence inthe classroom via questionnaires. They were unin-formed of the study’s hypotheses and completed thequestionnaires after they had known children for atleast 2 months. The Social Competence and BehaviorEvaluation Short Form (SCBE) was used (LaFreniere& Dumas, 1996). On the 30-item version of this well-validated measure, teachers rate children on 6-pointscales that address different social behaviors asso-ciated with the child’s capacity to modulate affect.Three scale scores may be computed from theteacher ratings on specific items: (a) an angry-aggressive scale (e.g., often angry, defiant whenreprimanded), (b) an anxious-withdrawn scale (e.g.,avoids new situations; sad, unhappy, depressed),and (c) a sensitive-cooperative scale (e.g., comfortsor assists children in difficulty, takes other children’sviewpoints into account).

As seen in the previous examples, both the angry-aggressive scale and the anxious-withdrawn scalecontain items that overlap conceptually with ourobserved emotion and dysregulated emotion vari-ables. Thus, because of possible problems of includ-ing items referring to anger and aggressive ventingin measurement of both emotional and socialcompetence, we first created a new oppositionalsubscale using only items from the angry-aggressivescale that did not refer directly or solely to anger andaggression (e.g., the first angry-aggressive itemlisted earlier). Our means of limiting items wasderived from our work on emotion language(Denham, Cook, & Zoller, 1992); wherever an itemcontained a word that we would code as an emotionword, it was deleted. Using this reasoning, we alsocreated an isolated subscale using only items from theanxious-withdrawn scale that did not refer directly

or solely to sadness and anxiety (e.g., the secondanxious-withdrawn item referred to earlier).

Cronbach’s alphas for the oppositional subscaleequaled .83 and .78 for preschool and kindergarten,respectively. Cronbach’s alphas for the isolatedsubscale equaled .78 and .77 for preschool andkindergarten, respectively. Cronbach’s alphas for theunaltered sensitive-cooperative scale equaled .90and .82 for preschool and kindergarten, respectively.

Peer-rated social competence. All children in theclassroom with parental consent (study participantsand nonparticipants) were taken out of the class-room by a familiar adult and presented with a set ofphotographs of all of their classmates (Asher,Singleton, Tinsley, & Hymel, 1979). Children firstnamed all classmates’ photographs, to assure recog-nition. Then, using ample vocal and facial emotioncues to model the task, the researcher placed oneplastic play person in each of three small boxesmarked with sketched drawings. The like-a-lot boxwas marked with a smiling face, the kinda like boxwas marked with a flat-mouthed face, and the do notlike box was marked with a frowning face. After thisdemonstration rating the likability of the plastic playpeople, each child was asked to place each of his orher peers’ pictures into the happy face, neutral face,or unhappy face boxes, according to how much he orshe liked to play with each peer. At the completionof picture placement, each child chose a stickerreward and was asked to make up his or her ownstories with puppets. Concluding with this neutral,constructive activity ensured that the peer rating hadno negative effect on the child’s social interactionsafter returning to the classroom.

Based on the sociometric component of the task,each child in the study received scores for both thenumber of positive ratings and the number ofnegative ratings from their classmates. One-yeartest–retest reliabilities for these scores were .38 and.34, respectively, pso.01 (df5 141). A peer likabilityscore for each child was computed by subtracting thenumber of negative ratings from the number ofpositive ratings and dividing by the total number ofratings (see also Denham & Holt, 1993; Denhamet al., 1990; Denham, et al., 1997; a5 .75).

In kindergarten, the procedure was changedslightly. Because of the new cognitive abilities oftheir age, children were presented with 1� 5 in.cards with the name of each classmate, one per card.Children matched these cards to the appropriateplacement on a 1- to 5-point graphic ‘‘smiley face’’liking scale. Thus, a 5-point Likert rating scaleresulted; we recorded the number of times eachparticipant child’s name was placed on each of the

244 Denham et al.

five way points. Then, a weighted sum (withweights of 5 to 1 for each way point, from liked mostto liked least) was created for kindergarten peerlikability.

Results

Preliminary Analyses

First, Kendall nonparametric correlations examin-ing the relations between individual demographicvariables (i.e., family income, race, parental educa-tion, perinatal complications, and child’s past orcurrent use of medications) and the predictor andcriterion variables were conducted. Because thenumber of significant results was less than thenumber expected by chance, none of these demo-graphic variables were used as covariates in lateranalyses.

Descriptive statistics. The group of participants asa whole appeared positive in terms of both emo-tional and social competence (see Table 1). Thechildren’s emotional profiles showed a predomi-nance of happiness, with some anger and little

sadness in the preschool. They showed good levelsof emotion understanding for their ages. It is notsurprising that mothers reported some emotionalventing. In terms of social competence indices,teachers rated children at both ages as relativelysensitive/cooperative, and not very oppositional orisolated; children were moderately well liked. Thereappeared to be adequate variation for correlationalanalyses.

Correlational analyses. Associations of manifestvariables for predictors and criteria are shown inTable 2. These preliminary analyses are followed byreports of associations among latent variablescalculated by LVPLS.

Taking relations among predictors first, aspects ofemotional expressiveness were related as expected,as were elements of emotion understanding. Thetwo indices of emotion regulation were, however,unrelated. Next, examining relations between sets ofpredictors, previously noted associations betweenemotional expressiveness and understanding ofemotion were replicated (Denham, 1986), andobserved regulation was related to emotional ex-pressiveness (happiness and anger).

Moving onto correlations among criteria, associa-tions among the SCBE scales and sociometriclikability were mostly significant both at ages 3 to4 and in kindergarten. Cross-time correlations fortwo of the four indices also were significant, albeitsmall in magnitudeFprobably due to differences inratersFacross the 2-year period.

We next examined correlations between predictorsand criteria. Emotional expressiveness patterns werelargely unrelated to the social competence evalua-tions of important others. In contrast, ability tounderstand equivocal negative emotional situationswas related to teachers’ evaluations of socialcompetence during kindergarten. Thus, one type ofemotion understanding was related in an interpre-table way to later social competence evaluations.Regarding emotion regulation, 3-year-olds who weredepicted by mothers as resorting to dysregulatedventing were seen by teachers as more oppositionaland by peers as less likable at age 3 to 4. The sameassociations held true for observed emotion regula-tion. The same observational index predicted con-current teachers’ ratings of oppositionality, as well aslower peer ratings of likability.

Evaluation of Longitudinal Models

To continue exploring the links between aspectsof emotional competence and social competence, wesought to best summarize the patterns of covariation

Table 1

Descriptions of Measures

Measure M SD Possible

range

Emotional competence

Expressiveness

Happy displays .79 .16 0–1

Sad displays .05 .12 ‘‘

Angry displays .11 .11 ‘‘

Understanding

Emotion knowledge: unequivocal 14.24 2.40 0–16

Emotion knowledge: Equivocal 10.41 2.53 0–12

Positive vs. negative

Emotion knowledge: equivocal 9.75 2.63 0–12

Negative vs. negative

Emotion regulation: maternal report

Emotional venting 15.06 3.70 4–28

Emotion regulation: observed

Emotional venting .10 .11 0–1

3- to 4-year-old social competence

SCBE oppositional 14.71 7.95 10–42

SCBE isolated 16.05 7.61 10–42

SCBE sensitive-cooperative 42.10 9.48 10–60

Sociometric likability aggregate .33 .37 � 1–1

Kindergarten social competence

SCBE oppositional 11.44 6.16 10–42

SCBE isolated 15.16 6.19 10–42

SCBE sensitive-cooperative 44.87 7.63 10–60

Sociometric likability aggregate 3.41 0.44 1–5

Preschool Emotional Competence 245

Table 2

Intercorrelations Among Elements of Longitudinal Model

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

Emotional competence

Expressiveness

1. Happy y

2. Sad � .64nnn y

3. Angry � .53nnn � .16 y

Understanding

4. Emotion knowledge: unequivocal .31nn � .26nn � .07 y

5. Emotion knowledge: equivocal, pos./neg. .18 � .20n � .01 .58nnn y

6. Emotion knowledge: equivocal, neg./neg. .20n � .19n .02 .60nnn .63nnn y

Emotion regulation: maternal report

7. Venting .02 � .01 � .01 .04 .04 � .05 y

Emotion regulation: observed

8. Venting � .31nnn .05 .49nnn .00 .01 � .01 .06 y

3 to 4 year-old social competence

9. SCBE oppositional � .05 � .02 .16 � .11 � .03 � .08 .20n .30nn y

10. SCBE isolated .15 .00 � .21n � .07 � .06 � .15 .03 � .11 .16 y

11. SCBE sensitive-cooperative � .01 .03 � .04 .11 .04 .14 � .18 � .19 � .58nnn � .42nnn y

12. Sociometric likability .16 � .13 � .05 .04 .03 .07 � .22n � .29nn � .18 � .25n .20n y

Kindergarten social competence

13. SCBE oppositional � .09 .05 .11 � .14 � .16 � .22n .19 � .11 .33nnn .14 � .17 � .18 y

14. SCBE isolated .12 .01 � .12 � .05 � .04 � .11 .08 � .05 .12 .15 � .18 .10 .28nn y

15. SCBE sensitive-cooperative .09 � .16 � .08 .14 .11 .21n � .18 � .02 � .28nn � .14 .23n .10 � .48nnn � .33

16. Sociometric likability .12 � .11 � .08 .12 .06 .08 � .08 � .14 � .17 � .01 .13 .13 � .12 � .12 .29nnn

npo.05.nnpo.01.nnnpo.001. (two-tailed)

24

6D

en

ha

me

ta

l.

in a structural model, in which the three aspects ofemotional competence (along with moderators of sexand age, and the interaction of emotional expres-siveness and emotion regulation) predicted socialcompetence concurrently and at each subsequentyear.

We conducted LVPLS analysis (Falk & Miller,1992; Lohmoller, 1989). We chose this method, whichis becoming more widely known, because it allowsus to explore hypothesized relations among con-structs without some of the restrictions of LISRELstructural modeling techniques. In particular, LVPLSis appropriate for use with relatively small groups ofparticipants, although it does require a reasonablelatent variable: participant ratio. Given the correla-tion matrix of interest here, a special advantage isthat the prediction of latent variables is based onshared variance of the manifest variables. Individualvariable residuals and the unreliability associatedwith measurement error are minimized. Because ofits limited assumptions, however, LVPLS yields onlyapproximate estimates of invariant structural para-meters.

Measurement model. LVPLS allows specification ofa measurement model as well as a structural model.Using multiple measures (manifest variables), con-structs (latent variables) are created, with thevariance shared by the measures taken as a singleindex of the latent variable. That is, scores for thecomposite latent variables are computed fromprincipal components weights, derived from ana-lyses of the manifest variables. Thus, a smaller set oftheoretical variables is created, whose relations canbe investigated without sacrificing information fromthe larger group of manifest variables.

For our study, 11 latent variables were created,including emotional expressiveness, emotion under-standing, emotion regulation (i.e., dysregulatedcoping indices negatively weighted), and socialcompetence at each age (see Table 3). Before analysis,we centered all manifest variables because thisprocedure helps avoid computational errors bylowering the correlations between the interactionterms and their individual components. Withoutsuch a procedure, multicollinearity could lead to aninability for the LVPLS procedure to accuratelyestimate the underlying interaction construct.

Chin, Marcolin, and Newsted (1996) outlined amethod to include moderator variables in LVPLSmodels via multiplicative-term manifest variables(e.g., Sex�Emotional Expressiveness), as also seenin Table 3. This advantageous method allowed us toexamine moderation as part of the full model, beforedecomposition into separate regression equations,

with proportionally smaller ns, for each level of themoderator.

Examination of the latent to manifest variableloadings in Table 3 reveals none with an absolutevalue less than .4, one benchmark for noting amanifest variable with very little in common withother measures of a particular construct (before thesefinal results, however, we did have to eliminate thevariables for Age�Emotion Regulation reported bymother and Expressed Anger�Emotion Regulationobserved because of low loadings). Twenty-four of29 retained manifest variables (excluding Age�Emotion Regulation observed, and Expressed Anger�Emotion Regulation reported by mother, both ofwhich were single indicators of their respectivelatent variable) loaded significantly on their latentvariables, with the remaining three loading at theborderline significant level.

Finally, to assess how accurate the path estimateswere to the ‘‘true’’ effects, composite reliabilitiesbased on latent variables’ communalities werecalculated (Werts, Linn, & Joreskog, 1974). Thesereliabilities ranged from .46 to .75. Given the smallnumber of items (i.e., manifest variables) per latentvariable, these reliabilities were acceptable accordingto Nunnally’s (1967) guidelines; that is, those thatdidn’t meet or exceed a .70 criterion appeared to doso because of a small number of items (i.e., if numberof items for each scale were multiplied by a factor of1 to 3, resulting in 4 to 12 items, the .70 criterionwould be met).

Correlations among latent variables. Associationsamong latent variables are shown in Table 4. One-tailed probability levels were used because direc-tions of associations were hypothesized. Emotionalexpressiveness was related to emotion knowledgeand emotion regulation. Every predictor was relatedto social competence at one age (if not both), exceptfor Age�Emotion Knowledge. Finally, despite themanifest variables’ low cross-time associations (seeTable 2), there was a highly significant correlationbetween ages 3 to 4 and kindergarten socialcompetence.

Overview of structural path model. We estimatedthe latent variables as an exact linear combination oftheir indicators with the goal of maximizing ex-plained variance for both indicators and latentvariables, using PLSGraph software (Chin, 2001).Figure 2 depicts the final structural model, in whichpath coefficients can be interpreted as standardizedbeta weights, each estimated after all other paths’effects have been controlled. To assess whether thepaths for interaction and main effects were signifi-cant, a jackknife resampling procedure (Efron &

Preschool Emotional Competence 247

Table 3

Latent Variable Loading for Manifest Variables in Longitudinal Model

Variable Loading t value Composite reliability

(communality)

Age 3 to 4

Emotional expressiveness

Observed happiness .946 23.94nnn .57

Observed sadness � .493 � 3.72nnn

Observed anger � .754 � 8.86nnn

Emotion knowledge

Emotion knowledge: unequivocal .875 16.50nnn .73

Emotion knowledge: equivocal .815 7.12nnn

Positive/negative

Equivocal situation knowledge .874 33.59nnn

Negative/negative

Emotion regulation strategies

Maternal report: venting � .438 � 2.27n .52

Observed venting � .926 � 22.75nnn

Social competence

SCBE oppositional � .832 � 1.66n .46

SCBE isolated � .437 � 1.43+

SCBE sensitive-cooperative .805 1.66n

Sociometric likability .563 1.69n

Sex�Emotional Expressiveness interactions

Sex�Happiness � .972 � 2.54nn .58

Sex� Sadness .673 1.91n

Sex�Anger .589 3.26nnn

Sex�Emotion Knowledge interactions

Sex�Unequivocal Situation Knowledge .824 12.04nnn .70

Sex�Equivocal Situation Knowledge .797 7.12nnn

Positive/negative

Sex�Equivocal Situation Knowledge .882 27.74nnn

Negative/negative

Sex�Emotion Regulation Strategies

Sex�Maternal Report: Venting � .670 � 2.54nn .49

Sex�Observed Venting � .731 � 2.22n

Age�Emotion Knowledge interactions

Age�Unequivocal Situation Knowledge � .761 � 1.53+ .75

Age�Equivocal Situation Knowledge � .845 � 1.54+

Positive/negative

Age�Equivocal Situation Knowledge � .973 � 1.76n

Negative/negative

Age�Emotion Regulation Strategies

Age�Observed Venting 1.00 .00 F

Emotional Expressiveness�Emotion Regulation Strategies

Anger�Maternal Report: Venting 1.00 .00 F

Kindergarten

Social competence

SCBE oppositional � .777 � 12.21nnn .46

SCBE isolated � .414 � 1.84n

SCBE sensitive-cooperative .819 22.31nnn

Sociometric likability .627 8.81nnn

npo.05.nnpo.01.nnnpo.001.+po.10. (one-tailed)

248 Denham et al.

Gong, 1983) was performed. In this procedure, thepartial least squares parameters of a series ofrandom subsamples of the total sample wereiteratively tested, until significance could be esti-mated based on their convergent findings.

Given these procedures, the final model in Figure2, in which nonsignificant hypothesized paths weredeleted, can be summarized by noting that emotionregulation predicted age 3 to 4 social competence.Emotional expressiveness, emotion knowledge,Age�Emotion Regulation, and age 3 to 4 socialcompetence predicted kindergarten social compe-tence. Emotional Expressiveness�Emotion Regula-tion and Age�Emotion Regulation predicted socialcompetence at both ages. Finally, emotional expres-siveness predicted both emotion knowledge andemotion regulationFmore positive children wereboth more knowledgeable about emotions and moreable to regulate them.

The overall model can also be evaluated. Twoindices are available to specify whether the modelaccounts for meaningful variance in the latentvariables. First is the coefficient root mean squareof the covariance (RMS COV[E,U]) between themanifest variable residuals and the latent variableresidualsFthe proportion unaccounted for by themodel (Falk & Miller, 1992). This coefficient is 0 in aperfect model, and values over .20 indicate poor fit.The path model in our study yielded a RMSCOV(E,U) value of .10, indicating good to excellentfit between model and data. Second, the significanceof R2s for each of the endogenous latent variables(i.e., emotion knowledge, emotion regulation, andsocial competence at both ages) were tested andfound to be significant at the po.01 level or better.

Finally, it is important to specify the nature of thesignificant moderation pathways. To do so, weconstructed sets of separate regression equationsfor boys and girls, younger and older children (splitat the median age at study inception, 46 months),and low and high negatively expressive children foreach pertinent predictor. Thus, emotion regulationpredicted age 3 to 4 social competence more stronglyfor girls than for boys, bs5 .376 and .255, pso.01 and.05, respectively. Emotion regulation also predictedage 3 to 4 social competence more strongly foryounger than for older children, bs5 .335 and .293,pso.01 and .05, respectively. Emotion knowledgepredicted social competence more strongly forchildren who were younger, as opposed to older, atthe study’s inception, bs5 .229 and .085, pso.05 andns, respectively, for age 3 to 4 social competence, andbs5 .251 and .209, pso.05 and .10, respectively, forkindergarten social competence. Finally, emotionregulation predicted age 3 to 4 social competencemore strongly for low-negative children than forhigh-negative children, bs5 .365 and .253, pso.01and .05, respectively. Conversely, emotion regulationpredicted kindergarten social competence morestrongly for high-negative children than for low-negative children, bs5 .220 and .097, pso.05 and ns,respectively.

Alternative model. An alternate model was tested,examining direct paths from age 3 to 4 socialcompetence to concurrent emotional competence,and from age 3 to 4 emotional competence tokindergarten social competence (see Figure 3; be-cause age 3 to 4 social competence was seen as apredictor of emotional competence in this model,links between age 3 to 4 social competence and

Table 4

Correlations Among Latent Variables

Latent variable 1 2 3 4 5 6 7 8 9 10

1. Emotional expressiveness y

2. Emotion knowledge .22nn y

3. Emotion regulation .38nnn .00 y

4. Sex�Emotionality � .94nnn � .22n � .30nn y

5. Sex�Emotion Knowledge .20n .95nnn � .02 � .23nn y

6. Sex�Emotion Regulation .26nn � .02 .90nnn � .23nn � .02 y

7. Emotional Expressiveness�Emotion Regulation � .11 .02 � .07 .07 .02 � .07 y

8. Age�Emotion Regulation .02 .07 � .31nnn � .02 .04 � .32nnn .09 y

9. Age�Emotion Knowledge � .03 .00 .08 .07 � .05 .07 � .02 .14+y

10. Age 3 to 4 social competence .09 .13+ .39nnn � .06 .12 .39nnn .13+ � .16n .06 y

11. Kindergarten social competence .17n .23nn .20n � .16n .23nn .21n � .14+ � .08 .02 .38nnn

npo.05.nnpo.01.nnnpo.001.+po.10. (one-tailed)

Preschool Emotional Competence 249

moderator latent variables were not included in themodel). Overall, the RMS COV(E,U) value was again.10, indicating good to excellent fit between modeland data In this alternative model, social compe-tence predicted all three latent variables for emo-tional competence, but only the R2s for emotionknowledge and emotion regulation were significant,as in the main model, suggesting that their inter-relation was as important, if not more so, as thepathways from age 3 to 4 social competence. Two,instead of three, moderation paths to kindergartensocial competence were significant in this model.Overall, the main difference in the two modelsappeared to be that the total R2 for emotion

regulation was enhanced by its prediction by age 3to 4 social competence.

Discussion

We were encouraged by our effort to identify theaspects of emotional competence that contribute to akey developmental task of the preschool period-successful interaction with peers. Simultaneousconsideration of multiple indicators for three hy-pothesized constituents of emotional competencewas a step forward in specifying which elementscontribute to social competence, both concurrently

B = .398, t = 7.60***

B = .218,t = 2.64**

B = -.053, t = -1.51+

B = -.065, t = -285**

B = .401, t = 3.19***

B = -.216, t = -1.75* B = -.211, t = -2.62**

B = .188, t = 1.64*

B = .152, t = 3.45**

B = .233, t = 2.05*

B = .321, t = 11.50***

B = .313, t = 2.26*

Age × Emotion

Knowledge

Sex × Emotion

Regulation

Emotional Expressiv. ×Emotion Regulation

Emotion Knowledge

Emotion Regulation

Kindergarten Social

Competence

Age × Emotion

Regulation

R2kindergarten social competence = .214, F (9, 94) = 27.13 *** ** p < .01. *** p < .001.

Emotional Expressiv.

Age 3 to 4Social

Competence

R2emotion knowledge = .048, F (1, 141) = 7.11**. R2

emotion regulation = .145, F (1, 141) = 23.90 ***. R2age 3 to 4 social competence = .214, F (9, 133) = 36.21 ***.

Figure 2. Longitudinal model of age 3 to 4 emotional competence predicting age 3 to 4 and kindergarten social competence, and age 3 to 4social competence predicting kindergarten social competence, with only significant pathways shown.

npo.05. nnpo.01. nnnpo.001. +po.10. (one-tailed)

250 Denham et al.

and across time. Emotional competence at ages 3 to 4made contributions to both age 3 to 4 and kinder-garten social competence, suggesting that the abil-ities of emotional competence acquired by ages 3 to 4become stable, with continuing import. The patternsof these latent variables’ contributions differ de-pending on whether each is especially important forpreschoolers in general, boys or girls, or relativelyyounger and older preschoolers. Thus, these find-ings enhance our basic understanding of theaffective foundations of young children’s socialcompetence and could allow for targeted preventionefforts with older and younger preschoolers, boysand girls.

The first of three emotional competence latentvariables involved emotional expressiveness. Thislatent variable indexed predominantly happy, notsad or angry, emotions displayed during interaction.

We have argued elsewhere (e.g., Denham, 1986) that‘‘feeling good’’ in many situations not only ‘‘greasesthe cogs’’ of ongoing social interaction, but alsomakes it easier for a child to enter the peer world inthe first place. Furthermore, the enduring patterns ofclear emotional well-being signaled by a happy childare likely to make a positive impression on apreschool teacher, and happier playmates probablydon’t ‘‘make waves’’ in the sometimes-choppy sea ofpreschool interactions; they are just easier for theother children to like. However, we need to knowmore about how happiness works within interaction.Thus, we should examine whether peer entry, peerprovocation, higher level sociodramatic play, amitywith friends, and other hallmarks of ongoing pre-school social interaction are fostered by children’sobserved happiness (e.g., Gottman & Mettetal, 1986;Marcus, 1987). Furthermore, not all happy displays

B =.132, t = 2.01*

B = .142, t = 5.59***

B = .080, t = 2.24*

B = .123,t = 5.72***

B = .137, t = 2.70**

B = .130, t = 3.45***

B = -.175, t = 2.45*

B = -.209, t = = 2.10*

B = -.263, t = 3.06 ***

Age × Emotion

Regulation

Emotional Expressiv. ×

Emotion Regulation

Emotion Knowledge

Emotion Regulation

Kindergarten Social

Competence

R2kindergarten social competence = .214, F (9, 94) = 28.24 ***. ** p < .01. *** p < .001.

Emotional Expressiv.

Age 3 to 4 Social

Competence

R2emotional expressiveness = .014, F (1, 141) = 2.00, ns. R2

emotion knowledge = .057, F (1, 141) = 8.52 **. R2emotion regulation = .286, F (1, 141) = 56.48***.

Figure 3. Competing model, showing age 3 to 4 social competence predicting age 3 to 4 emotional competence and kindergarten socialcompetence, with only significant pathways shown.

npo.05. nnpo.01. nnnpo.001. +po.10. (one-tailed)

Preschool Emotional Competence 251

are created equal. For example, Arsenio and col-leagues (e.g., Arsenio & Lover, 1997) have made itclear that happiness shown during preschool conflictis likely to be a sign of ‘‘happy victimizing,’’ anddefinitely not well received by others.

As well, anger has been reported elsewhere as adetriment to young children’s social competence(Denham et al., 1990; Denham & McKinley, 1993).Chronically angry children are less likely to meetanother’s positive overtures in kind, but more likelyto respond hostilely. Those who more often showsadness seem to puzzle their peers, who mayultimately shun them. These children’s demeanormakes them more difficult, both by omission andcommission, to play with. In fact, dealing withnegative emotions, either intra- or interpersonally,makes effective interaction much more difficultacross the lifespan; in contrast, more positive indi-viduals are like ‘‘interaction magnets’’ (Carstenson &Turk, 1998; Dua, 1993; Marcus, 1987).

So, it is easy to see how a predominance of happydisplays as opposed to angry or sad displays couldlead to others’ positive evaluations of one’s socialcompetence. Children who showed this pattern ofenduring emotional expressiveness with their play-mates as early as ages 3 to 4 probably built uppositive reputations with teachers and peers alike,and probably encountered more social competence-building experiences than their less happy, sadder,more angry counterparts. It was a bit surprising thatemotional expressiveness did not predict age 3 to 4social competence directly. One explanation could bethat variations in emotional expressiveness do notyet affect others’ evaluationsFteachers may expectsome level of negative emotionality, and peers whoare just beginning to make appraisals of others’likability may not yet focus on their emotionalexpressiveness per se.

Turning to the emotion knowledge latent variable,for the total sample (but particularly for those lessthan 46 months old at the study’s inception),emotion knowledge assessed at ages 3 to 4 predictedsocial competence, both contemporaneously and inkindergarten. Older preschoolers (i.e., older than 46months) typically showed less variation on ourmeasure of emotion understanding; perhaps young-er children’s use of such knowledge more uniquelysupported their socially competent behavior orparticularly stood out to raters. Emotion knowledgeas assessed here could give preschoolers informationin critical peer situations, such as group entry andprovocation, by helping to promote conflict resolu-tion and other positive peer behavior (Dunn, 1995;Dunn & Herrera, 1997; Edwards, Manstead, &

MacDonald, 1984; Goldman, Corsini, & de Urioste,1980; Hughes Dunn, & White, 1998).

Last, the latent variable of emotion regulationmay be seen as a second-order outcome of emotionalexpressiveness, given its link with emotional ex-pressivenessFperhaps a child knows that he or shefeels angry, what made him or her angry, how he orshe is experiencing and expressing it, and figures outwhat to do. Given this interpretation, it may be nosurprise that indices of lapses in emotion regulationpredicted age 3 to 4 social competence even morepotently than emotional expressiveness patterns.How one responds to one’s experience of emotionmay enhance or hinder social relations more thanone’s enduring patterns of expressed emotions. Wefound that children who were less likely to vent theiremotions, especially girls and younger participants,were evaluated as more socially competent at eitherages 3 to 4 or in kindergarten. These childrenshowed restrained emotion in response to proble-matic situationsFthey were relatively unlikely tocry or behave aggressively. Uncontrolled negativitymay have been especially conducive to raters’negative evaluations because it contradicted stereo-typical views of girls’ social competence andreinforced notions of the immaturity of childrenwho are ‘‘young in grade.’’

Girls especially may be expected to adhere to thenuances of ‘‘nice’’ behavior toward others, so thattheir emotional regulation may be an especiallyimportant part of the emotional competence thatpredicts concurrent social competence. For example,Fagot (1984) found that teachers of young preschool-ers (i.e., 2 to 3 years old) reacted particularlynegatively when girls showed behaviors in twocategories that included demanding attention; ex-pressing verbal aggression; taking objects; cryingand whining; and hitting, pushing, or shoving. Thus,when a girl or a preschooler who is young ingrade behaves less positively, this negativity may bemore salient than the negative behavior ofboys or more mature children, because girls areexpected to be nice and older children’s negativebehavior may be less overt (Abramovitch, Corter, &Lando, 1979; Crick, Casas, & Mosher, 1997). Inparticular, this set of findings highlights the con-tinuing need to shed light on girls’ social worldsacross time during the preschool period. Forexample, the precursors and emergence of girls’relational aggression in preschool and kindergartenwarrant study.

As well, abilities to refrain from dysregulatedventing particularly benefited children who weremore negatively emotional; their age 3 to 4 social

252 Denham et al.

competence, but not that of less negatively emo-tional children, was predicted by emotion regula-tion. This pattern of moderation fits well with thatalso found by Eisenberg and colleagues (Eisenberg,Fabes, Guthrie, et al., 1996; Eisenberg, Fabes,Murphy, et al., 1996; Eisenberg, Fabes, et al., 1997).It makes sense that emotion regulation should aidthose who most need to regulate!

However, emotion regulation at ages 3 to 4predicted kindergarten social competence slightlybetter for children low in negative emotions. It couldbe that children who experience fewer intenselynegative emotions, and avoid unrestrained outburstswhen they do feel negative, also are more able, bykindergarten, to use more constructive emotion-regulation strategies, such as active problem solving,cognitive restructuring, and avoidance. This long-itudinal linkage for low-emotional children needsfurther study.

It is also noteworthy that emotional expressive-ness patterns predicted both emotion knowledgeand emotion regulation at ages 3 to 4. Firmestablishment of a linkage between being emotion-ally positive and emotion knowledge, already notedin Denham (1986), suggests that it may be profitableto promote children’s positivity so they feel comfor-table and able to learn about their emotional world.As noted earlier, the association of emotionalexpressiveness and emotion regulation also makesintuitive senseFchildren who experience and dis-play more negative emotion probably have more toregulate, but this undertaking may be overwhelm-ing, leading to venting in less experienced 3- and 4-year-olds.

Finally, we did attempt to assess the adequacy of amodel in which age 3 to 4 social competencepredicted concurrent indices of emotional compe-tence, which then predicted kindergarten socialcompetence. It is perhaps not surprising that, dueto the ubiquitous possibility of bidirectional influ-ences in any concurrent correlations, age 3 to 4 socialcompetence did predict indices of emotional com-petence measured at the same period, and the modelfit well. However, evaluation of the overall modelsuggested that the main information gleaned wasthe prediction of emotion regulation by age 3 to 4social competence. This finding hints at the possibil-ity that the burden of others’ evaluationsFwhetherthose of teachers who consider a child difficult andlet this appraisal leak into their interactions, or thoseof peers who decide that a dysregulated child is toomuch trouble to play withFmay add to the child’sdifficulties in regulation. Better evaluation of alter-native models awaits research in which social

competence and emotional competence are eachevaluated at more than two time points.

Because of the ever-present nature of emotion atthe core of social interaction and well-being, dis-cerning the nature of linkages from emotional tosocial competence is a vital task for developmental-ists (Sroufe et al., 1984). In the present study, lack ofpositive affect, inability to regulate affect and copebehaviorally, and stunted abilities to use emotionunderstanding were clear risk factors for thechildren who participated in our project (see alsoDenham, Blair, Schmidt, & DeMulder, 2002). To fleshout our knowledge to make a difference in children’slives both contemporaneously and predictively, weneed even more details such as those uncoveredhere, and we need to overcome some of thelimitations of our current workFfor example, weneed information of the sort gathered here for othercultural and ethnic groups to generalize our findingsand make them more useful.

However, some specific suggestions can be madefor preschool and daycare teachers and parents toaid their charges in attaining particular aspects ofemotional competence that have long-term utility.For example, teaching about feelings may beespecially helpful for children 4 years old and under.All preschoolers, but especially those who are moreemotional, might benefit from learning means toavoid dysregulated coping, to respond to peers’emotions prosocially instead of antisocially.

In sum, we have focused on the separate and jointprediction of indices of social competence viaemotional expressiveness, emotion knowledge, andemotion regulation. We have begun delineatingboundary conditions for these relations (e.g., age,sex, and level of expressiveness) that ultimately mayhave important applied implications.

References

Abramovitch, R., Corter, C., & Lando, B. (1979). Siblinginteraction in the home. Child Development, 50, 997–1003.

Arsenio, W., & Lover, A. (1997). Emotions, conflicts, andaggression during preschoolers’ freeplay. British Journalof Developmental Psychology, 15, 531–542.

Asher, S. R., Singleton, L. C., Tinsley, B. P., & Hymel, S.(1979). A reliable sociometric measure for preschoolchildren. Developmental Psychology, 15, 443–444.

Birch, S. H., & Ladd, G. W. (1997). The teacher-childrelationship and children’s early school adjustment.Journal of School Psychology, 35, 61–79.

Borke, H. (1971). Interpersonal perception of youngchildren: Egocentrism or empathy? Developmental Psy-chology, 5, 263–269.

Preschool Emotional Competence 253

Bower, B. (1985). Caution: Emotions at play. Science News,127, 226–267.

Bridges, L. J., & Grolnick, W. S. (1994). The development ofemotional self-regulation in infancy and early child-hood. In N. Eisenberg (Ed.), Review of personality andsocial psychology. Vol. 15. Social development (pp. 185–211).Thousand Oaks, CA, Sage.

Campos, J. J., & Barrett, K. C. (1984). Toward a newunderstanding of emotions and their development. In C.E. Izard, J. Kagan, & R. B. Zajonc (Eds.), Emotions,cognition, & behavior (pp. 229–263). New York: Cam-bridge University Press.

Carlton, M. P. (2000). Motivation and school readiness inkindergarten children. Dissertation Abstracts InternationalSection A Human and Social Sciences, 60(11-A),3899.

Carlton, M. P., & Winsler, A. (1999). School readiness: Theneed for a paradigm shift. School Psychology Review, 28,338–352.

Carstenson, L. L., & Turk, C. S. (1998). Emotion in thesecond half of life. Current Directions in PsychologicalScience, 7, 144–149.

Chin, W. W. (2001). PLS Graph. Version 03.00 build 279.Houston, TX: Soft Modeling. Retrieved October 29, 2002,from www.plsgraph.com.

Chin, W. W., Marcolin, B. L., & Newsted, P. R. (1996). Apartial least squares latent variable modeling approachfor measuring interaction effects: Results from a MonteCarlo simulation study and voice mail emotion/adop-tion study. Proceedings of the Seventeenth InternationalConference on Information Systems, 21–41.

Cole, P. M., Michel, M. K., & Teti, L. O. (1994). Thedevelopment of emotion regulation and dysregulation:A clinical perspective. Monographs of the Society forResearch in Child Development, 59 (2-3, Serial No. 240),73–102.

Crick, N. R., Casas, J. F., & Mosher, M. (1997). Relationaland overt aggression in preschool. Developmental Psy-chology, 33, 579–588.

Denham, S. A. (1986). Social cognition, social behavior, andemotion in preschoolers: Contextual validation. ChildDevelopment, 57, 194–201.

Denham, S. A. (1998). Emotional development in youngchildren. New York: Guilford Press.

Denham, S. A., Blair, K. A., Schmidt, M., & DeMulder, E.(2002). Compromised emotional competence: Seeds ofviolence sown early? American Journal of Orthopsychiatry,72, 70–82.

Denham, S. A., Cook, M. C., & Zoller, D. (1992). Baby looksvery sad: Discussions about emotions between motherand preschooler. British Journal of Developmental Psychol-ogy, 10, 301–315.

Denham, S. A., & Couchoud, E. A. (1990a). Youngpreschoolers’ understanding of emotion. Child StudyJournal, 20, 171–192.

Denham, S. A., & Couchoud, E. A. (1990b). Youngpreschoolers’ understanding of equivocal emotionsituations. Child Study Journal, 20, 193–202.

Denham, S. A., & Holt, R. W. (1993). Preschoolers’likability as cause or consequence of their socialbehavior. Developmental Psychology, 29, 271–275.

Denham, S. A., & McKinley, M. J. (1993). Sociometricnominations of preschoolers: A psychometric analysis.Early Education and Development, 4, 109–122.

Denham, S. A., McKinley, M., Couchoud, E. A., & Holt, R.(1990). Emotional and behavioral predictors of peerstatus in young preschoolers. Child Development, 61,1145–1152.

Denham, S. A., Mitchell-Copeland, J., Strandberg, K.,Auerbach, S., & Blair, K. (1997). Parental contributionsto preschoolers’ emotional competence: Direct andindirect effects. Motivation and Emotion, 21, 65–86.

Denham, S. A., Renwick, S., & Holt, R. (1991). Working andplaying together: Prediction of preschool social–emo-tional competence from mother–child interaction. ChildDevelopment, 62, 242–249.

Denham, S. A., Zoller, D., & Couchoud, E. A. (1994).Preschoolers’ causal understanding of emotion and itssocialization. Developmental Psychology, 30, 928–936.

Dua, J. K. (1993). The role of negative affect and positiveaffect in stress, depression, self-esteem, assertiveness,Type A behaviors, psychological health, and physicalhealth. Genetic, Social, & General Psychology Monographs,119, 515–552.

Dunn, J. (1994). Understanding others and the socialworld: Current issues in developmental research andtheir relation to preschool experiences and practice.Journal of Applied Developmental Psychology, 15,571–583.

Dunn, J. (1995). Children as psychologists: The latercorrelates of individual differences in understandingof emotions and other minds. Cognition and Emotion, 9,187–201.

Dunn, J., & Herrera, C. (1997). Conflict resolution withfriends, siblings, and mothers: A developmental per-spective. Aggressive Behavior, 23, 343–357.

Edwards, R., Manstead, A. S. R., & MacDonald, C. J.(1984). The relationship between children’s sociometricstatus and ability to recognize facial expressions ofemotion. European Journal of Social Psychology, 14,235–238.

Efron, B., & Gong, G. (1983). A leisurely look at thebootstrap, the jackknife, and cross-validation. TheAmerican Statistician, 37, 36–48.

Eisenberg, N., Fabes, R. A., Guthrie, I. K., Murphy, B. C.,Maszk, P., & Holmgren, R., et al. (1996). The relations ofregulation and emotionality to problem behavior inelementary school children. Development & Psychopathol-ogy, 8, 141–162.

Eisenberg, N., Fabes, R. A., Murphy, B. C., Karbon, M.,Smith, M., & Maszk, P. (1996). The relations of children’sdispositional empathy-related responding to their emo-tionality, regulation, and social functioning. Develop-mental Psychology, 32, 195–209.

Eisenberg, N., Fabes, R. A., Murphy, B. C., Maszk, P.,Smith, M., & Karbon, M. (1995). The role of emotionality

254 Denham et al.

and regulation in children’s social functioning: Alongitudinal study. Child Development, 66, 109–128.

Eisenberg, N., Fabes, R. A., Nyman, M., Bernzweig, J., &Pinuelas, A. (1994). The relation of emotionality andregulation to preschoolers’ anger-related reactions. ChildDevelopment, 65, 1352–1366.

Eisenberg, N., Fabes, R. A., Shepard, S. A., Murphy, B. C.,Guthrie, I. K., & Jones, S., et al. (1997). Contempora-neous and longitudinal prediction of children’s socialfunctioning from regulation and emotionality. ChildDevelopment, 68, 642–664.

Eisenberg, N., Guthrie, I., Fabes, R. A., Reiser, M., Murphy,B. C., & Holgren, R., et al. (1997). The relations ofregulation and emotionality to resiliency and competentsocial functioning in elementary school children. ChildDevelopment, 68, 295–311.

Fagot, B. I. (1984). Teacher and peer reactions to boys’ andgirls’ play styles. Sex Roles, 11, 691–702.

Falk, R. F., & Miller, N. (1992). A primer for soft modeling.Akron, OH: The University of Akron Press.

Gagnon, C., Craig, W. M., Tremblay, R. E., Zhou, R. M., &Vitaro, F. (1995). Kindergarten predictors of boys’ stablebehavior problems at the end of elementary school.Journal of Abnormal Child Psychology, 23, 751–766.

Goldman, J. A., Corsini, D. A., & de Urioste, R. (1980).Implications of positive and negative sociometric statusfor assessing social competence of young children.Journal of Applied Developmental Psychology, 1, 209–220.

Gottman, J. M., & Mettetal, G. (1986). Speculations aboutsocial and affective development of friendship andacquaintanceship through adolescence. In J. M. Gottman& J. Parker (Eds.), Conversations of friends: Speculations onaffective development (pp. 192–237). New York: Cam-bridge University Press.

Gross, J. J. (1998). The emerging field of emotionregulation: An integrative review. Review of GeneralPsychology, 2, 271–299.

Haapasalo, J., & Tremblay, R. E. (1994). Physicallyaggressive boys from ages 6 to 12: Family background,parenting behavior, and prediction of delinquency.Journal of Consulting and Clinical Psychology, 62, 1044–1052.

Howes, C. (1987). Social competence with peers in youngchildren: Developmental sequences. Developmental Re-view, 7, 252–272.

Howes, C., & Smith, E. W. (1995). Relations among childcare quality, teacher behavior, children’s play activities,emotional security, and cognitive activity in child care.Early Childhood Research Quarterly, 10, 381–404.

Hughes, C., Dunn, J., & White, A. (1998). Trick or treat?Uneven understanding of mind and emotion andexecutive dysfunction in ‘‘hard-to-manage’’ preschoo-lers. Journal of Child Psychology & Psychiatry & AlliedDisciplines, 39, 981–994.

Izard, C. E., Fine, S., Schultz, D., Mostow, A., Ackerman,B., & Youngstrom, E. (2001). Emotions knowledge as apredictor of social behavior and academic competencein children at risk. Psychological Science, 12, 18–23.

Jacobsen, T., & Hofmann, V. (1997). Children’s attachmentrepresentations: Longitudinal relations to school beha-vior and academic competency in middle childhood andadolescence. Developmental Psychology, 33, 703–710.

Jenkins, J. M., & Oatley, K. (2000). Psychopathology andshort-term emotion: The balance of affects. Journal ofChild Psychology and Psychiatry, 41, 463–472.

Kochenderfer, B. J., & Ladd, G. W. (1996). Peer victimiza-tion: Cause or consequence of school maladjustment?Child Development, 67, 1305–1317.

Kopp, C. B. (1989). Regulation of distress and negativeemotions: A developmental view. Developmental Psychol-ogy, 25, 343–354.

Ladd, G. (1990). Having friends, keeping friends, makingfriends, and being liked by peers in the classroom:Predictors of children’s early school adjustment? ChildDevelopment, 61, 1081–1100.

Ladd, G. W., Birch, S. H., & Buhs, E. S. (1999). Children’ssocial and scholastic lives in kindergarten: Relatedspheres of influence? Child Development, 70, 1373–1400.

Ladd, G. W., Kochenderfer, B. J., & Coleman, C. C. (1996).Friendship quality as a predictor of young children’searly school adjustment. Child Development, 67, 1103–1118.

LaFreniere, P. J., & Dumas, J. E. (1996). Social competenceand behavior evaluation in children ages 3 to 6 years:The short form (SCBE-30). Psychological Assessment, 8,369–377.

Lewis, M., Sullivan, M., & Vasen, A. (1987). Making faces:Age and emotion differences in the posing of emotionalexpression. Developmental Psychology, 23, 690–697.

Lohmoller, J. B. (1989). Latent variable path modeling withpartial least squares. New York: Springer-Verlag.

Losoya, S., Eisenberg, N., & Fabes, R. A. (1998). Develop-mental issues in the study of coping. International Journalof Behavioral Development, 22, 287–314.

Marcus, R. F. (1987). The role of affect in children’scooperation. Child Study Journal, 17, 153–168.

Murphy, B. C., Shepard, S. A., Eisenberg, N., Fabes, R. A.,& Guthrie, I. K. (1999). Contemporaneous and long-itudinal relations of dispositional sympathy to emotion-ality, regulation, and social functioning. Journal of EarlyAdolescence, 19, 66–97.

Nunnally, J. (1967). Psychometrics. New York: McGraw-Hill.O’Neil, R., Welsh, M., Parke, R. D., Wang, S., & Strand, C.

(1997). A longitudinal assessment of the academiccorrelates of early peer acceptance and rejection. Journalof Clinical Child Psychology, 26, 290–303.

Parke, R. D. (1994). Progress, paradigms, and unresolvedproblems: A commentary on recent advances in ourunderstanding of children’s emotions. Merrill-PalmerQuarterly, 40, 157–169.

Parker, J., & Asher, S. A. (1987). Peer relations and laterpersonal adjustment: Are low accept children ‘‘at risk’’?Psychological Bulletin, 102, 357–389.

Parker, J. G., & Gottman, J. M. (1989). Social and emotionaldevelopment in a relational context: Friendship interac-tion from early childhood to adolescence. In T. J. Berndt

Preschool Emotional Competence 255

& G. W. Ladd (Eds.), Peer relationships in child develop-ment (pp. 95–131). New York: Wiley.

Pianta, R. C. (1997). Adult-child relationship processes andearly schooling. Early Education and Development, 8, 11–26.

Pianta, R. C., Steinberg, M., & Rollins, K. (1995). The firsttwo years of school: Teacher-child relationships anddeflections in children’s classroom adjustment. Develop-ment & Psychopathology, 7, 295–312.

Roberts, W. L. (2002). Software for observing behavior innatural settings. Retrieved October 28, 2002, from http://www.cariboo.bc.ca/ae/psych/roberts/homepage.htm#focal.

Robins, L. N., & Rutter, M. (1990). Straight and deviouspathways from childhood to adulthood. Cambridge, UK:Cambridge University Press.

Rose-Krasnor, L. (1997). The nature of social competence:A theoretical review. Social Development, 6, 111–135.

Rubin, K. D., & Clark, M. L. (1983). Preschool teachers’ratings of behavioral problems: Observational, socio-metric, and social–cognitive correlates. Journal of Abnor-mal Child Psychology, 11, 273–286.

Rubin, K. D., & Daniels–Byrness, T. (1983). Concurrent andpredictive correlates of sociometric status in kindergar-ten and grade 1 children. Merrill–Palmer Quarterly, 29,337–352.

Rubin, K. H., & Ross, H. S. (1988). Toward the study ofsocial competence, social status, and social relations.Monographs of the Society for Research in Child Develop-ment, 53 (1, Serial No. 217), 79–87.

Saarni, C. (1990). Emotional competence. In Ross Thomp-son (Ed.), Nebraska symposium: Socioemotional development(pp. 115–161). Lincoln, University of Nebraska Press.

Shields, A., Dickstein, S., Seifer, R., Guisti, L., Magee, K. D.,& Spritz, B. (2001). Emotional competence and earlyschool adjustment: A study of preschoolers at risk. EarlyEducation and Development, 12, 73–96.

Sroufe, L. A., Schork, E., Motti, F., Lawroski, N., &LaFreniere, P. (1984). The role of affect in socialcompetence. In C. E. Izard, J. Kagan, & R. B. Zajonc(Eds.), Emotions, cognition, & behavior (pp. 289–319).Cambridge, UK: Cambridge University Press.

Strayer, J. (1980). A naturalistic study of empathicbehaviors and their relation to affective states andperspective-taking skills in preschool children. ChildDevelopment, 51, 815–822.

Thompson, R. A. (1994). Emotional regulation: A themein search of definition. In N. A. Fox (Ed.), Thedevelopment of emotion regulation: Biological andbehavioral considerations Monographs of the Society forResearch in Child Development, 59(2-3, Serial No. 240),25–52.

Tremblay, R. E., Pagani-Kurtz, L., Masse, L. C., Vitaro, F., &Pihl, R. O. (1995). A bimodal preventive intervention fordisruptive kindergarten boys: Its impact through mid-adolescence. Journal of Consulting and Clinical Psychology,63, 560–568.

Werts, C. E., Linn, R. L., & Joreskog, K. G. (1974). Intraclassreliability estimates: Testing structural assumptions.Educational and Psychological Measurement, 34, 25–33.

256 Denham et al.