Associations between child and teacher characteristics and quality of teacher–child relationships:...

25
This article was downloaded by: [Brian Collins] On: 13 March 2013, At: 16:41 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK European Early Childhood Education Research Journal Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/recr20 Associations between child and teacher characteristics and quality of teacher–child relationships: the case of Hungary Bernadett Koles a , Erin E. O'Connor b & Brian A. Collins c a Central European University, Frankel Leo ut 30-34, Budapest, H-1023 b New York University, 239 Greene Street, 6th Floor, New York, NY, 10012 c Hunter College, CUN, 695 Park Avenue 1032, New York, NY, 10065 Version of record first published: 11 Mar 2013. To cite this article: Bernadett Koles , Erin E. O'Connor & Brian A. Collins (2013): Associations between child and teacher characteristics and quality of teacher–child relationships: the case of Hungary, European Early Childhood Education Research Journal, 21:1, 53-76 To link to this article: http://dx.doi.org/10.1080/1350293X.2012.760337 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and- conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Transcript of Associations between child and teacher characteristics and quality of teacher–child relationships:...

This article was downloaded by: [Brian Collins]On: 13 March 2013, At: 16:41Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

European Early Childhood EducationResearch JournalPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/recr20

Associations between child andteacher characteristics and quality ofteacher–child relationships: the case ofHungaryBernadett Koles a , Erin E. O'Connor b & Brian A. Collins ca Central European University, Frankel Leo ut 30-34, Budapest,H-1023b New York University, 239 Greene Street, 6th Floor, New York, NY,10012c Hunter College, CUN, 695 Park Avenue 1032, New York, NY,10065Version of record first published: 11 Mar 2013.

To cite this article: Bernadett Koles , Erin E. O'Connor & Brian A. Collins (2013): Associationsbetween child and teacher characteristics and quality of teacher–child relationships: the case ofHungary, European Early Childhood Education Research Journal, 21:1, 53-76

To link to this article: http://dx.doi.org/10.1080/1350293X.2012.760337

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representationthat the contents will be complete or accurate or up to date. The accuracy of anyinstructions, formulae, and drug doses should be independently verified with primarysources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand, or costs or damages whatsoever or howsoever caused arising directly orindirectly in connection with or arising out of the use of this material.

Associations between child and teacher characteristics and qualityof teacher–child relationships: the case of Hungary

Bernadett Kolesa*, Erin E. O’Connorb and Brian A. Collinsc

aCentral European University, Frankel Leo ut 30-34, Budapest, H-1023; bNew YorkUniversity, 239 Greene Street, 6th Floor, New York, NY 10012; cHunter College, CUNY, 695Park Avenue, 1032 New York, NY 10065

ABSTRACT: The purpose of the current study was to investigate variations inteacher–child relationships in childcare classrooms in Budapest, Hungary (N =172 children in 43 classrooms), and to examine whether variations wereassociated with child and/or teacher characteristics. In addition, cultural variationwas examined with reference to an American comparison group (N = 36 childrenin nine classrooms). Teacher–child relationships were found to vary in the inlevels of closeness, conflict and over-dependence. There was more variationwithin as opposed to between classrooms, indicating that child attributes play animportant role in teacher–child relationships. Girls had better relationships withtheir teachers than boys, characterised by higher levels of closeness and lowerlevels of conflict. Higher levels of shyness were associated with more conflictedteacher–child relationships for boys, and less conflicted ones for girls. Teacherswith higher levels of neuroticism and depression tended to report more conflictedrelationships with children. Hungarian teachers reported more closeness in theirrelationships with younger children, whereas American teachers reported higherlevels of relationship closeness with older children. Hungarian teachers reportedhigher levels of over-dependence of the children in their classrooms than didAmerican teachers. Educational implications as well as limitations andrecommendations for future research are discussed.

FRENCH: Cette étude a pour objectif d’analyser des variations des relationsenseignant-enfant dans les sections de crèche à Budapest, en Hongrie (N = 172enfants dans 43 classes), et d’examiner si les variations sont liées auxcaractéristiques de l’enfant et/ou celles de l’enseignant. De plus, la diversitéculturelle a été examinée par rapport à un groupe de référence américain (N = 36enfants dans 9 classes). L’analyse montre que les relations enseignant-enfantvarient suivant le degré de proximité, de conflit et de sur-dépendance. Il y adavantage de variations au sein des sections que dans les différentes sections dela crèche, indiquant que les profils d’enfants jouent un rôle important dans lesrelations enseignant-enfant. Comparées aux garco̧ns, les filles ont de meilleuresrelations avec leurs enseignants, caractérisées par un niveau plus élevé deproximité et par un niveau inférieur de conflit. Les niveaux plus élevés detimidité ont été liés aux relations enseignant-enfant plus conflictuelles chez lesgarco̧ns et moins conflictuelles chez les filles. Les enseignants d’un niveau plusélevé de névrose et de dépression ont tendance à montrer des signes de relationsplus conflictuelles avec des enfants. Les enseignants hongrois montrent unniveau de proximité plus élevés dans leurs relations avec les enfants les plusjeunes, alors que les enseignants américains le montrent avec les enfants plusâgés. Les enseignants hongrois ont remarqué des niveaux plus élevés de sur-dépendance chez les enfants dans leurs classes que les enseignants américains.

© 2013 EECERA

*Corresponding author. Email: [email protected]

European Early Childhood Education Research Journal, 2013Vol. 21, No. 1, 53–76, http://dx.doi.org/10.1080/1350293X.2012.760337

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Les implications éducatives et leurs limites, ainsi que des suggestions pour defutures recherches, sont discutées.

GERMAN: Zweck dieser Studie war die Untersuchung eventueller Unterschiede inden Personal-Kind-Interaktionen in frühpädagogischen Tageseinrichtungen inBudapest, Ungarn (N = 172 Kinder in 43 Kita-Gruppen) sowie herauszufinden,ob die Unterschiede mit Merkmalen der Kinder und/oder der Fachkräftezusammenhingen. Zudem wurden mit Bezug auf eine amerikanischeReferenzgruppe (N = 36 Kinder in 9 Kita-Gruppen) kulturelle Variationenuntersucht. Die Personal-Kind-Interaktionen zeigten Unterschiede vor allemhinsichtlich gegenseitiger Nähe, Konfliktbelastung und Grad der Abhängigkeit.Die Unterschiede zwischen den einzelnen Kita-Gruppen waren größer alserwartet. Dies deutet darauf hin, dass die Attribute der Kinder eine wichtigeRolle in ihrer Beziehung zu den Fachkräften spielen. Die Mädchen hatten einbesseres Verhältnis zu ihren Bezugspersonen als die Jungen, was sich vor allemdurch eine engere Beziehung und wenige Konflikte auszeichnete. Beischüchternen Kindernt wurde bei Jungen eine konfliktreichere Beziehung, beiden Mädchen eine eher konfliktarme Beziehung zu den Fachkräften verbunden.Fachkräfte mit einer Neigung zu höherer emotionaler Instabilität und Depressionbeurteilten ihre Beziehungen zu den Kindern des Öfteren als konfliktgeprägt. Dieungarischen Fachkräfte beurteilten ihr Verhältnis eher zu jüngeren Kindern alsenger, die amerikanischen Fachkräfte hingegen vor allem zu älteren Kindern. Dieungarischen Fachkräfte berichteten über einen höheren Grad übermäßigerAbhängigkeit bei den Kindern in ihren Gruppen als ihre amerikanischenKollegen. Eventuelle Implikationen für den Bildungsbereich sowie Hinweise aufBegrenzungen der Studie und Empfehlungen für weitere Untersuchungen werdendiskutiert.

SPANISH: En la presente investigación hemos analizado qué diferencias hay en larelación maestra-niño en las guarderías de Budapest (N = 172, en total 43 grupos) ysi estas diferencias se explican con las características de las maestras o con las de losniños. Además, hemos hecho una comparación incluyendo un gruponorteamericano (N = 36, en total 9 clases). Según nuestros resultados lasrelaciones maestra-niño muestran diferencias en las dimensiones de la cercanía,conflictos y dependencia. Se han presentado mayores diferencias dentro de unaula que entre los diferentes grupos, es decir, las características de los niñostienen un papel muy importante en la relación maestra-niño. En su totalidad lasniñas tenían una relación más cercana con las maestras y con menos conflictosque los niños. En caso de los niños la mayor timidez ha estado en coherenciacon la relación conflictiva maestra-niño, mientras en caso de las niñas con larelación menos conflictiva. Asimismo, las maestras que tienen un nivel más altode neurotismo y depresión que sus colegas han referido más relacionesconflictivas maestra-niño. Las maestras húngaras han desarrollado relaciones máscercanas con los niños más jóvenes mientras las norteamericanas con losmayores. Las maestras húngaras han relatado más dependencia en cuanto a larelación maestra-niño que sus colegas norteamericanas. Al final se hacen conocertanto las implicaciones relacionadas con la enseñanza como las limitacionesrelacionadas con la investigación y las futuras posibilidades.

Keywords: teacher–child relationships; child development; early childhoodeducation; relationship quality; Hungary

Introduction

The majority of children in Hungary and the United States attend childcare prior tokindergarten (Bachu 1995). The relationships that children form with teachers duringthese early years have been demonstrated to have long-lasting effects on children’s

54 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

emotional/behavioral adjustment, social development, and academic achievement (e.g.Pianta 1999). Understanding factors associated with the quality of children’s earlyrelationships with teachers is therefore essential. Little empirical work, however, hasbeen conducted to examine variation within and between classrooms with respect toteacher–child relationship quality in childcare. Such an examination would allow foran investigation of the relative effects of teacher, classroom and child characteristicson relationship quality. Furthermore, no studies have been conducted to examinecross-cultural variation in relationship quality. Such a study would provide a valuableassessment of the influence of social context on relationship quality (Melhuish andMoss 1991); and provide alternative perspectives and policy implications regardingteacher–child relationships (Lamb 1998). The purpose of the current study was toexamine within- and between-classroom variation in relationship quality in childcaresin Hungary and the United States.

Teacher–child relationships

Young children develop attachment or attachment-like relationships with teachers (e.g.Anderson et al. 1981; Howes and Hamilton 1992). Attachment relationships represent aspecific type of relationship within which children seek feelings of safety and security.Early attachment relationships are particularly meaningful, because children developexpectations, or internal working models, about interpersonal interactions throughthese relationships (Bowlby 1988). Initial internal working models of attachmentrelationships ‘become abstracted from schemas of specific interactions into generalbeliefs about’ caregiving relationships (Howes and Matheson 1992, 26). Thesemodels of early relationships develop into the child’s sense of him- or herself in relationto others, and often become templates for future relationships.

The teacher–child relationship is characterised by three relatively independentdimensions: closeness, conflict and over-dependence. Closeness describes the amountof warmth and positive affect in the relationship (Pianta and Nimetz 1991; Spilt andKoomen 2009), and relates to whether the child can use the relationship as a secure-base from which to explore. Conflict, on the other hand, describes the amount of discor-dance and anger in the relationship (Pianta and Nimetz 1991; Spilt and Koomen 2009).Lastly, overdependence describes the amount of overly clingy and immature behaviordemonstrated in the relationship (Pianta 1999). In close relationships, teachers providechildrenwith positive behavioral support andmodel appropriate coping skills that encou-rage self-regulation (Doll 1996; Howes and Hamilton 1993; Howes, Hamilton, andMatheson 1994; Pianta 1997). Non-close relationships, on the other hand, appear toevoke feelings of ineffectiveness in teachers, and to discourage them from puttingtime into the child (Spilt and Koomen 2009). Furthermore, in conflictual relationships,teachers tend to focus more on controlling children’s behavior than providing a suppor-tive environment for them (Hamre and Pianta 2001; Pianta, Steinberg, andRollins 1995).Within what might be described as an overdependent relationship, that is, one in whicha child seems to depend extensively on the teacher, a child is unlikely to explore theschool environment or engage in relationships with peers.

Associations between teacher–child relationships and individual characteristics

An extensive body of research indicates that there is variation in relationship qualityacross teachers and children, and that child and teacher characteristics are responsible

European Early Childhood Education Research Journal 55

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

for this variation. Specifically, child gender has been shown to influence the quality ofthe teacher–child relationship, such that teachers generally report lower quality relation-ships with boys than girls (Pianta 1999; Rudasill and Rimm-Kaufman 2009). Forexample, in a three-year longitudinal study of teacher–child relationships in early child-hood, researchers found that teachers reported more closeness and less conflict in theirrelationships with girls than with boys (Howes 2000).

Previous research has demonstrated that child temperament correlates with variationin relationship quality as well. More specifically, shyness, i.e. the extent to which achild evidences a slow or inhibited approach in situations involving novelty or uncer-tainty, is associated with the quality of the teacher–child relationship (Rudasill et al.2006). Uninhibited children tend to interact with teachers more than their shy peers.These children seem to enjoy closer relationships with their teachers (e.g. Lerner,Lerner, and Zabski 1985; Patrick, Yoon, and Murphy 1995; Rimm-Kaufman et al.2002; Rimm-Kaufman and Kagan 2005; Skarpness and Carson 1986). On the otherhand, shyer children have been shown to initiate fewer interactions with their teachers,which in turn are associated with lower levels of teacher–child relationship closeness(Rudasill and Rimm-Kaufman 2009). Furthermore, in a study of preschool children,shy children had both less close and less conflictual relationships with teachers thantheir less shy peers (Rydell, Bohlin, and Thorell 2005).

In a recent study, Rudasill and Rimm-Kaufman (2009) found that children withlower levels of effortful control tended to have more conflicted relationships withtheir teachers, whereas children with higher levels tended to have closer relationshipswith their teachers. Furthermore, the authors found that both shyness and effortfulcontrol contributed to the frequency of interactions between children and their tea-chers. More specifically, children who were characterized as less shy appeared toinitiate more interactions with their teachers, and children who were characterizedwith lower levels of effortful control appeared to receive more interactions initiatedby their teachers.

Child anger also appears to correlate with relationship quality. Children who exhibithigher levels of anger tend to have relationships with teachers characterised by lowlevels of closeness and high levels of conflict (Howes 2000). Research on temperamentand parent–child relationships suggests that the effects of temperament on relationshipquality vary as a function of child gender. For example, results from one study demon-strated that the effects of child anger on relationship quality had stronger negativeimplications for boys than girls (Kawaguchi et al. 1998). No studies, however, haveexamined potential differential associations between temperament and the quality ofthe teacher–child relationship among male and female children.

The birth order of the child has been cited in some studies as a factor influencingchildren’s environment and experiences (e.g. Kowalski et al. 2004; Sulloway 1996).Although the effect of birth order is not as widely studied as the effect of some otherconstructs, such as gender, some researchers have shown that birth order is importantfor children’s development within the family. As Zajonc (2001) summarises: ‘each suc-cessive child enters into a different environment and begins a particular cycle ofgrowth. At the same time, each successive child changes and keeps on changing thefamily environment’ (p. 490). In addition, as subsequent children arrive to thefamily, they may encourage or sometimes necessitate older children’s emerging rolesas an ‘aid,’ a ‘tutor,’ or even a ‘surrogate parent’ (Zajonc 2001). These experiencesin turn may influence children’s behaviors, and perhaps even the relationships theyform with others, such as teachers.

56 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Characteristics of the teacher’s personality may influence the quality of the teacher–child relationship as well. The majority of work on adult personality characteristics andrelationship quality has focused on mother–child dyads. Parallels, however, may bedrawn between the mother–child and teacher–child relationship. Research investigatingmother–child dyads demonstrates that children have more positive interactions withmothers who report low levels of neuroticism and high levels of agreeableness(Belsky and Barends 2002). Furthermore, researchers have found that maternal neuroti-cism is the most consistent predictor of a negative mother–child relationship (Belsky,Crnic, andWoodworth 1996). Based on these studies, one may hypothesise that charac-teristics of the teacher’s personality would be associated with relationship quality aswell.

The comparison of childcare education in Hungary and in the United States

Historically, there had been marked differences between American and Hungarianapproaches to childcare. These differences had to do with culture, governmental regu-lation or lack thereof, societal beliefs about childcare, education and licensure of child-care providers, and status and remuneration of childcare professionals. Thesedifferences manifested themselves in greater emphasis on an individualist approachto child rearing in the US, as opposed to a collectivist approach to child rearing inHungary; little childcare regulation and limited support tied to low socioeconomicstatus in the US, as opposed to universal and government supported childcare inHungary; and greater emphasis on childcare as a caring in the US, as opposed to a con-ception of childcare as an educational tool in Hungary (Lamb 1998). A summary ofthese and other dimensions along which Hungary and the US childcare policies andpractices show historical and current variation is found in Table 1.

Because there is no universal childcare system in the United States for families, theuse as well as the type of non-maternal care is an individual choice (Senninger 1992).Limited governmental childcare support tends to be accessible for low-income families

Table 1. Summary of different historical and current dimensions between childcare practicesin Hungary and in the US.

United States Hungary

Approach to childrearing Individualist CollectivistUse of childcare Up to individual

familiesPartially mandatory

Childcare support Limited Government supportedBased on SES Universal

Teacher education No regulation Rigorous requirements andcontinuous education

Teacher salary when compared withelementary school teachers

Lower wages Comparable wages

Consistency of care providerthroughout childcare exposure

Greaterinconsistency

Greater consistency

Main purpose of childcare Caring tool Educational tool

Sources: (Fodor et al. 2002; Lamb 1998; Senninger 1992; Adamik 1991; Vari 1989; Csapo 1984).

European Early Childhood Education Research Journal 57

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

participating in the Program Personal Responsibility and Work Opportunity Reconci-liation Act (PRWORA) (Kahn and Kamerman, 1998). On the other hand, for Hungar-ian families, there is a centralised and government-sponsored partially mandatorychildcare system available to all childcare-aged children (Fodor et al. 2002).

Furthermore, prior to kindergarten, childcare arrangements tend to be of varyingtype and quality in the United States, with the primary aim of providing care for thechild while the parent is at work (Lamb 1998). In contrast, the primary aim of the Hun-garian childcare system is to provide a high-quality educational institution for all chil-dren to aid their later school transition and academic success (Adamik 1991; Lamb1998).

A related point concerns the necessary amount of education in order to becomea childcare provider. In the US, where young children are exposed to a variety ofchildcare arrangements prior to their elementary school entry, there are no universaleducational requirements (Howes 1991). In Hungary, on the other hand, where chil-dren attend universal childcare throughout the entire three-year period prior toelementary school entry, rigorous criteria shape the preparation of childcare pro-fessionals (Csapo 1984). They must have a specific college degree from theCollege of Child Care Education, complete an internship, and participate infollow-up training programs at least once every seven years (Vari 1989). In theUS, there is no universal or mandatory teacher preparation for childcare providers(Lamb 1998).

In addition to the lack of educational requirements, childcare teachers in the US tendto earn low wages and most receive no benefits (Olmstead and Senninger, 1995). Incontrast, childcare teachers in Hungary receive a salary similar to that of elementaryschool teachers, and are provided with various governmental benefits, including 13months of pay and two-months of summer vacation (Adamik 1991; Vari 1989). Fur-thermore, Hungarian teachers (on both the childcare and primary school level) enjoyrather high social prestige, with these professions ranked at seventh and sixth, respect-ively, on a prestige scale out of 40 jobs (Nagy 1998). These data are in line with reportsof high job satisfaction and low turn-over rates among teachers in Hungarian childcareand suggest that being a childcare teacher in Hungary is more likely than in the US to beone’s permanent profession (Nagy 1998). These differences in job satisfaction and turn-over rates between childcare teachers in the US and Hungary may have importantimplications for the teacher–child relationship. For example, in a study investigatingcorrelates of positive and negative behaviors of childcare educators, researchersfound a positive relationship between the work environment (including teachers’ atti-tudes towards work and their job satisfaction) and caregiver’s affectionate behavior,and a negative one between teachers’ job perception and the expression of anger inthe classroom (Mill and Romano-White 1999).

Furthermore, due to the low turn-over rates and greater job permanence in Hungar-ian childcare, the same childcare teacher is likely to stay with the same group of chil-dren for the entire duration (i.e. three years) of childcare (Adamik 1991). In contrast,childcare aged children in the US are often exposed to more than one educationalprogram prior to their entering elementary school (Howes 1991). This consistencyversus inconsistency between childcare professionals throughout a program is a par-ticularly important cultural distinction between Hungary and the US given the roleof consistency in the teacher–child relationship.

While the generic policies and regulations in Hungary remained practicallyunchanged, except for updates and slight improvements, major advancements in

58 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

early childhood education had been implemented in the US. Many states haveimproved their quality standards, increased availability and access for three- andfour-year-olds to state pre-K program, increased total funding as well as spendingper child for state programs, and progressed in numerous ways with respect to provisionof preschool education for children (Barnett et al. 2009). Nevertheless, due to the rela-tive novelty of the changes implemented in the US, the marked disparities among statesprimarily in meeting quality benchmarks with respect to programs and teacher qualifi-cations, as well as the fact that Hungary maintains its universal childcare available to allchildren, cross-cultural comparisons may provide meaningful insights into variousaspects of early childhood education.

Background for the current study

This study builds on work conducted in the US in Cambridge, Massachusetts (Koles2002, Koles and O’Connor 2003), and previously reported in the Journal of EarlyChildhood Teacher Education (JECTE Koles, O’Connor and McCartney 2009). Inthe American sample, variation in teacher–child relationship quality within the class-rooms was predicted by child-level variables (e.g. gender, anger and shyness) andtheir interactions. The variation in teacher–child relationship quality between class-rooms was predicted by teacher-level variables (i.e. teacher personality traits). Thecurrent project extended this work to a different cultural context.

The current study and its contributions

The purpose of this study is to add substantively to the literature by identifying potentialpredictors of teacher–child relationship quality. This study employs a varied methodo-logical approach that includes a simultaneous within- and between-teacher design, eval-uating teacher–child dyadic relationships of the same teacher with different children intheir classroom and including information from a number of different teachers. Thestatistical application used for analysis allows the examination of data on individualsnested within classrooms. Through such design, it was possible to examine potentialfactors associated with teachers’ differential treatment towards different children inthe same classroom.

Prior research indicates that variation in teacher–child relationship quality wouldbe expected to exist both across classrooms due to differences in teacher character-istics, and within classrooms due to variations in child and teacher characteristics(Pianta 1999; Pianta, Nimetz, and Bennet 1997). Only one earlier study has been con-ducted to examine variation between versus within classrooms in teacher–childrelationship quality: Hamre et al. (2007) investigated variation in relational conflictbetween teachers and children in pre-kindergarten classrooms in the US. Theyfound variation in average levels of relational conflict between classrooms andbetween teachers and individual students within classrooms, indicating that teachersform different quality relationships with children in their classrooms based on childcharacteristics and the goodness-of-fit between children and teachers. Additionalresearch to examine factors associated with variation in relational conflict and toinvestigate variation in relational closeness and overdependence is necessary.Another noteworthy characteristic of the design in the current study was the cross-cul-tural sample. This is the first empirical study to compare teacher–child relationshipsacross two different cultures.

European Early Childhood Education Research Journal 59

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Research questions

The following four research questions served to guide the current analyses.

(1) How much variation exists in closeness, conflict and overdependence inteacher–child relationships within and across classrooms?

(2) What child-level factors predict variation in closeness, conflict and over-depen-dence in teacher–child relationships?

(3) What teacher-level factors predict variation in closeness, conflict and overde-pendence in teacher–child relationships?

(4) Do cultural factors and practices account for differences in teacher–childrelationships?

Methods

Participants

Budapest, Hungary

Nine districts were sampled randomly from the total of 23 districts in Budapest. Foreach of these nine districts, where there was an average of 25 government sponsoredcenters, five childcare centers were chosen using random selection. The directors ofthese centers were contacted by telephone, and details regarding the study wereexplained. Of the 45 childcare center directors, all agreed to participate. As a result,45 centers participated in the current study. Because the majority (94%) of Hungarianchildcare centers are government-sponsored (Lannert and Halasz 2003), as opposed toprivate or church-affiliated, only government-sponsored centers were considered forrecruitment.

Ten letters were distributed to the parents of three- to four-year-olds in each class-room in each center, thus approximately 50% of the average class of 20 children werecontacted. Two of the classrooms were entirely excluded from the analysis due to insuf-ficient teacher-reports as well as missing information on the teacher questionnaires. Asa result, the final sample consisted of 43 teachers working with a total of 172 children.Of the families who responded to the letter of invitation, four children (two females andtwo males) ranging between the ages of 37 and 65 months (mean = 51 months) wererandomly selected from each classroom for observation. Children spent an averageof 35.5 hours per week in childcare.

Cambridge, Massachusetts

Except for the size of the sample, the data collection method and the measures usedwere identical for the American and Hungarian samples (see Koles, O’Connor, andMcCartney 2009 for a detailed description of the American sample). Ten teachersfrom 10 classrooms were recruited from childcare centers located in Cambridge, Mas-sachusetts. One classroom was excluded from the analysis due to insufficient teacherreports as well as missing information on the teacher questionnaires. As a result, thefinal sample consisted of nine teachers and 36 children.

Four children (two females and two males) ranging between the ages of 41 and 62months (mean = 49 months) were randomly selected from each classroom for obser-vation. The sample comprised 75% European-American, 11% African American, 8%Latino-American and 6% Asian-American participants. All the teachers in the studywere European-American.

60 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Overview of data collection

The study utilised a multi-reporter approach, including teacher and mother reports.Teacher–child relationships were assessed via a questionnaire administered to eachteacher (Student Teacher Relationship Scale, Pianta 1992). At the same time, teachers’personality and psychological adjustment were also assessed using self-report scales(the NEO Personality Inventory, (Costa and McCrae 1985); the My Feelings scale[adapted from the Center for Epidemiological Studies Depression Scale (CES-D)],(Radloff 1997)). A measure of children’s temperament was obtained through a ques-tionnaire given to the mothers (Children’s Behavior Questionnaire, Rothbart, Ahadi,and Hershey 1994). Finally, mothers were asked to fill out a questionnaire that provideddemographic variables, such as the child’s age and birth order.

Measures

Teacher–Child Relationship Scale (STRS, Pianta 1992). This is a self-report measureobtained from teachers, assessing their feelings and beliefs about three features oftheir relationship with the target child. The Conflict subscale comprises 12 items thattap the extent to which the teacher–child relationship is characterised by disharmo-nious, antagonistic interactions (e.g. ‘This child and I always seem to be strugglingwith each other.’; α = 0.94).1 The Closeness subscale is an 11-item index of thedegree of warmth and open communication present in the teacher–child relationship(e.g. ‘I share an affectionate, warm relationship with this child.’; αs ranged from0.88 to 0.91). Finally, the Overdependency subscale contains four items that assessthe degree to which the child seems overly dependent on the teacher (e.g. ‘This childasks for my help when he/she does not really need help’; αs ranged from 0.72 to0.80). Teachers rate items in terms of how applicable each statement is to theircurrent relationship with a particular child. Responses range from 1 = definitely doesnot apply, to 5 = definitely applies. STRS scores are associated with Attachment Q-Set ratings of teachers and students such that higher STRS scores correlate withmore secure relationships (Howes and Ritchie 1999).

Children’s Behavior Questionnaire (Rothbart et al. 1994). This measure obtainsparental reports of temperament and social behavior in childhood. Aspects of tempera-ment are assessed using an 80-item scale (a subset of the original measure). Items arerated on a 7-point scale ranging from 1 = extremely untrue, to 7 = extremely true. Thefollowing five subscales of temperament are obtained: (1) Anger/Frustration (e.g.‘Rarely gets irritated when s/he makes a mistake.’; α = 0.86);2 (2) Child’s AttentionalFocusing (e.g. ‘Has trouble sitting still when s/he is told to.’; α = 0.80); (3) Child’sInhibitory Control (e.g. ‘Acts very friendly and outgoing with new children.’; α =0.85); (4) Child’s Sadness (e.g. ‘Her/His feelings are easily hurt by what parentssay.’; α = 0.69); and (5) Child’s Shyness (e.g. ‘Seems to be at ease with almost anyperson.’; α = 0.90). The five dimensions of temperament were associated with parentreports of social behavior.

Basic Demographics Questionnaire. This measure obtains information regardingbasic background characteristics of the child (age, ethnicity, birth order), as well asinformation regarding the child’s childcare exposure.

The NEO Personality Inventory (Costa and McCrae 1985). This measure is a 36-item questionnaire using a 5-point Likert-type scale, assessing teacher personalityand psychological adjustment using three subscales. The Neuroticism subscale includes

European Early Childhood Education Research Journal 61

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

12 items that assess the extent to which the teacher is anxious, depressed or hostile (e.g.‘I often feel inferior to others’; α = 0.84).3 The Extraversion subscale is a 12-item indexof the extent to which the teacher is sociable, optimistic and fun loving (e.g. ‘I reallyenjoy talking to people.’; α = 0.75). Finally, the Agreeableness subscale contains 12items that assess the extent to which the teacher is trusting, forgiving and helpful(e.g. ‘I try to be courteous to everyone I meet.’; α = 0.74). Responses range from 1 =strongly disagree, to 5 = strongly agree. Teachers rate items in terms of how descriptiveeach statement is of themselves.

My Feelings (adapted from the Center for Epidemiological Studies DepressionScale [CES-D], Radloff 1997). This measure is a 20-item questionnaire assessingteachers’ level of distress and depression. The scale is rated on a 4-point scale offrequency from 1 (less than once a week) to 4 (5–7 days a week). A total score iscalculated for each participant, including all 20 items. Sample statements include ‘Ifelt sad.’; ‘My sleep was restless.’; and ‘I feel hopeful about the future.’ Scores of16 or higher on this scale are considered to be in the range for clinical depression.The internal consistency (alpha) for this measure was 0.85 in the general population(Radloff 1997).

Please note that all questionnaires and surveys used in the current study have beentranslated from English to Hungarian, and then back translated by a third party, in orderto assess and ensure the consistency in meaning and maintain the validity of thequestions.

Data analysis plan

To examine whether child and/or teacher characteristics were associated with variationin closeness, conflict and over-dependence between, as well as within classrooms, wefit a series of classic two-level school effects models (see Singer [1998] for detaileddescription on the method). This technique is useful for situations incorporating dataat two levels of an organisation (Singer 1998), in this case amongst students withinclassrooms. In particular, this model allows for the examination of a level-1 (student-level) outcome as a function of both level-1 (student-level) and level-2 (teacher-level) predictors. As prior research had examined the effects of child and teachercharacteristics on the teacher–child relationship in the American sample, only theHungarian sample was included in the first set of analyses, for research questions 1through 3, examining within and between classroom variation and teacher and childcharacteristics associated with this variation. The combined American and Hungariansample was used to investigate research question 4.

Initially, unconditional means models for closeness, conflict and over-dependencewere fit (Singer 1998). Unconditional means models serve as a baseline for latermodel building. Unconditional means models have two variance components: onerepresents the variation between classroom means (intercept), and the other representsthe variation among students within classrooms (residual). These two variance com-ponents can be used to estimate intraclass correlations through which one can deter-mine the percentage of the explainable variation in a given outcome variable betweenclassrooms by using the formula: intercept/(intercept + residual); as well as the percen-tage of explainable variation within classrooms by using the formula: residual/(intercept + residual). Finally, interactions were tested between gender and anger,gender and shyness, and gender and sadness, to examine whether the effects ofchild temperament varied as a function of child gender. All student-level predictors

62 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

were added separately to each of the three baseline models for closeness, conflict andover-dependence to examine how much variation was attributable to each individualpredictor. Teacher-level predictors followed the same procedure as did child-levelpredictors, except no interactions were tested.

Once the models including each predictor were fit, it was possible to investigatethe proportion of between- as well as within-classroom variation explained by eachindividual predictor or by a combination of predictors. In order to examine theamount of between- or within-classroom variation that is explained by a particularpredictor, one subtracts the appropriate variance component of the model includingthe predictor from the same variance component of the baseline model, and thendivides by the variance component of the baseline model. To examine between-class-room variation that is explained by a predictor, the appropriate variance component isthe intercept. To investigate within-classroom variation that is explained by a predic-tor, the appropriate variance component is the residual. The scores for closeness, con-flict and overdependence were regressed on child characteristic variables (anger,shyness, gender and birth order), as well as on interactions between the temperamentvariables and gender, in a series of models. Each of the teacher-level predictors (neur-oticism, extraversion, agreeableness and depression) was then regressed separately oneach of the outcome variables (closeness, conflict and over-dependence) in a series ofmodels. Lastly, the country level variable was regressed on each of the outcomevariables.

Results

Descriptive statistics and correlations

Descriptive statistics for the outcome and predictor variables for the Hungarian sampleare presented in Table 2.

On average, teachers reported levels of closeness in the top quartile of the possiblerange with a fairly small standard deviation that was one-seventh of the mean. Bothaverages for levels of teacher-reported conflict and over-dependence were in thelower half of the possible range with the standard deviations slightly over one-thirdof the mean. These results indicate that teachers in the Hungarian classrooms tendedto report relationships with their students that were higher on positive and lower onnegative attributes. The means and the standard deviations of the three STRS subscalesare similar to those in previous work using the STRS (e.g. Kesner 2000; Pianta 1994;Saft 1995).

With respect to the child temperament measures, mothers generally reportedaverage levels for their children on each of the five behavioral constructs of anger,control, shyness, sadness and focus. With respect to the teacher characteristic indi-cators, teachers, on average, reported levels of neuroticism and depression that wereon the lower half of the possible range, and levels of agreeableness and extraversionthat were on the upper half of the possible range. There was little variation in agreeable-ness and extraversion and also relatively low variation in neuroticism. However, thevariation in depression scores was relatively high: two-thirds of the mean. Scores onthe child temperament measure and the teacher neuroticism and depression scaleswere not significantly different between the Hungarian and American samples (seeKoles et al. 2009). The majority of children in the Hungarian and American sampleswere the first born.

European Early Childhood Education Research Journal 63

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Variations in teacher–child relationships within and between classrooms –research Question 1

In the three baseline models for closeness, conflict and over-dependence (Table 3) inthe teacher–child relationship, the intercept and residual variance components repre-senting the variations between and within classrooms in each outcome variable, allshowed values different from zero. These results indicated that there was variation tobe explained between as well as within classrooms in all cases. Through the calculationof the intraclass correlations of the baseline models, we found that for most outcome

Table 2. Descriptive statistics including mean values and standard deviations for the outcomevariables (the three STRC subscales and the two observational variables), and each of thestudent and teacher-level predictors.

Mean Standard Deviation Sample Range Possible Range

Outcome variablesSTRS subscales (teacher ratings)Relationship conflict 23.5 9.0 12–53 12–60Relationship closeness 44.6 6.9 16–55 11–55Child overdependence 8.7 3.2 4–19 4–20PredictorsStudent-level predictorsChild temperament (parent ratings)Child anger 3.9 0.6 2–5 1–7Child control 4.1 0.6 3–6 1–7Child shyness 4.0 0.6 2–6 1–7Child sadness 3.6 0.6 2–5 1–7Child focus 3.4 0.7 2–5 1–7Age 51.1 5.8 37–65Birth order 0.5 0.5 0–3Teacher-level predictors (teacher ratings)Neuroticism 30.0 6.5 20–48 12–60Extraversion 42.0 5.4 28–52 12–60Agreeableness 46.7 5.0 33–56 12–60Depression 9.0 6.3 0–35 0–60

Table 3. Estimates of variance in teacher–child relationships.

Variance components

Teacher–child relationship variable

Conflict Close Over-dependence

Between classrooms (teacher level)Variance estimate 14.4 7.6 4.7Percent of total 17.6% 15.8% 44.8%

Within classrooms (student level)Variance estimate 67.2 40.47 5.8Percent of total 82.4% 84.2% 55.2%

64 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

variables, the majority of variability occurred within classrooms, as opposed to betweenclassrooms. The values of within- and between-classroom variations were as follows:conflict 82 and 18%; closeness 84 and 16%; overdependence 55 and 45% respectively.In other words, the only outcome variable indicating a relatively even distribution of thepercentage of within- and between-classroom variation was child over-dependence onthe teacher.

Effect of child-level predictors – Question 2

In the models examining predictors of the levels of conflict in teacher–child relation-ships (Table 4), gender showed a significant main effect. In particular, as seen inModel 1, teachers, on average, reported higher scores for levels of relationship conflictwith respect to boys, and the difference between the average scores for girls and boysdiffered by 3.01 points. Approximately 9% of the explainable within-classroom vari-ation in teacher–child relationship conflict can be attributable to gender.

A marginally significant interaction effect was found between child gender andchild shyness on levels of conflict in the relationship between teachers and children.This data is presented in Figure 1.

As apparent from Figure 1, virtually no difference in conflict existed among boysand girls who could be characterised as less shy. However, there was a greater discre-pancy in conflict among those boys and girls who could be characterised as more shy,and the effect of gender increased gradually as the levels of child shyness increased. Ingeneral, there was a positive trend for boys, such that higher levels of shyness in boystended to be associated with higher levels of conflict in the teacher–child relationship.For girls, on the contrary, there was a negative trend, such that higher levels of shynessin girls tended to be associated with lower levels of conflict in the relationship withteachers. The reduction in the residual from the baseline model to Model 6 indicatedthat approximately 8% of the explainable variation in levels of teacher–child relation-ship conflict could be explained by child gender, levels of child shyness, and theirinteractions.

Figure 1. The effect of the interaction between child shyness and child gender on teacherratings of teacher–child relationship conflict.

European Early Childhood Education Research Journal 65

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Table 4. Examining the effect of child-level predictors on teacher–child relationship conflict (CONFLICT).

Baseline Model 1 Gender Model 2 Order Model 3 Age Model 4 Anger Model 5 Shy Model 6 Gender* Shy

Intercept 23.50*** 25.00*** 23.52*** 18.64** 21.70*** 25.68*** 20.25***Child level predictorsGender −3.01* 10.73Order 1.28Age 0.09Anger 0.46Shy −0.57 1.22SadInteraction effectsGender* Shy −3.45∼

Random effectsIntercept 14.36* 35.80** 11.70∼ 140.28 0.00 68.71∼ 35.95**Residual 67.22*** 61.06*** 67.27*** 66.77*** 66.74*** 67.46*** 61.78***

∼p < 0.10.

*p < 0.05.

**p < 0.01.

***p < 0.001.

66B.Koles

etal.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

In the models examining predictors of teacher–child relationship closeness(Table 5), child gender and age showed significant effects.

In particular, we found from Model 1 that teachers, on average, tended to reporthigher scores for relationship closeness with respect to girls, and the differencebetween the average scores for girls and boys differed by 2.77 points. Approximately11% of the explainable within-classroom variation in teacher–child relationship close-ness can be attributable to gender. As presented in Model 3, the effect of age onrelationship closeness was negative, indicating that teachers tended to report scoresfor relationship closeness that were, on average, 0.21 points higher for younger thanfor older age groups. The residual value from the baseline remained virtuallyunchanged in this model, suggesting that age alone did not seem to have a main con-tribution. For the last STRS outcome variable representing children’s overdependenceon their teachers (Table 6), no significant main effects were found to predict any of thevariation within classrooms.

Effect of teacher-level predictors – Question 3

Tables 7–9 summarise the results for the main effects of the teacher-level predictors oneach of the outcome variables. Before summarising the results, it should be noted thatinformation regarding teacher personality traits as well as qualities of teacher–childrelationships were both obtained from the teacher, and thus we have to be cautiousabout shared reporter variance. In other words, as teachers provided multiple ratingsfor the study, these ratings may reflect their mood and other situational variables,which in turn may influence the findings.

Significant main effects were found when regressing teacher neuroticism or teacherdepression on levels of teacher–child relationship conflict. In particular, as apparent inTable 7, those teachers who scored higher on the neuroticism scale of the NEO also

Table 5. Examining the effect of child-level predictors on teacher–child relationship closeness(CLOSE).

BaselineModel 1Gender

Model 2Order

Model 3Age

Model 4Anger

Model 5Shy

Intercept 44.58*** 43.20*** 45.17*** 55.27*** 42.33*** 40.81***Child level predictorsGender 2.77**Order −1.28Age −0.21*Anger 0.58Shy 0.95Random effectsIntercept 7.60* 14.59* 5.04 0.00 21.79 65.67Residual 40.47*** 35.83*** 39.49*** 40.61*** 40.90*** 5.20***

*p < 0.05.

**p < 0.01.

***p < 0.001.

European Early Childhood Education Research Journal 67

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

tended to report higher levels of conflict with the children in their classroom. Forexample, when comparing two teachers, one scoring 30 and the other 40 points on theneuroticism subscale, one can predict that their relationships with their students, onaverage, will differ by positive 4.5 points with respect to levels of conflict. Based onthe reduction in the intercept variance from the baseline model to Model 1, one can con-clude that approximately 58% of the explainable variation in levels of teacher–childrelationship conflict between classrooms was attributable to levels of teacher neuroticism.

Table 6. Examining the effect of child-level predictors on child’s overdependence on theteacher (OVERDEP).

BaselineModel 1Gender

Model 2Order

Model 3Age

Model 4Anger

Model 5Shy

Intercept 8.74*** 8.52*** 8.63*** 8.54*** 7.43***Child level predictorsGender 0.44Order 0.25Age N/AAnger 0.05Shy 0.33Random effectsIntercept 4.67* 3.62** 4.50** 0 2.55 39.20*Residual 5.76*** 5.50*** 5.77*** 0 5.79*** 5.06***

*p < 0.05.

**p < 0.01.

***p < 0.001.

Table 7. Examining the effect of teacher-level predictors on teacher–child relationship conflict(CONFLICT).

BaselineModel 1

NeuroticismModel 2

ExtraversionModel 3

AgreeablenessModel 4

Depression

Intercept 23.50*** 9.91** 27.94*** 38.40*** 20.15***Teacher-level predictorsNeuroticism 0.45***Extraversion −0.11Agreeableness 0.13Depression 0.37**Random effectsIntercept 14.36* 5.96 14.78* 7.57* 9.44∼

Residual 67.22*** 67.22*** 67.22*** 40.47*** 67.22***

*p < 0.05.

**p < 0.01.

***p < 0.001.

68 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

The results from the regression analyses regressing the other significant teacher-level predictor, teacher depression, on levels of conflict, are listed in Model 4 of thistable. The results of this model show that those teachers who reported higher levelsof depression from the self-scale, also tended to have relationships with the childrenin their classrooms characterised by higher levels of conflict. For example, when com-paring two teachers, one scoring 10 and the other scoring 20 points on the depressionscale, one can predict that their relationships with their students, on average, will differby positive 3.7 points with respect to levels of conflict. The reduction in the intercept

Table 8. Examining the effect of teacher-level predictors on teacher–child relationshipcloseness (CLOSE).

BaselineModel 1

NeuroticismModel 2

ExtraversionModel 3

AgreeablenessModel 4

Depression

Intercept 44.58*** 48.84*** 40.82*** 90.46*** 45.34***Teacher-level predictorsNeuroticism −0.14Extraversion 0.09Agreeableness 0.39Depression −0.08Random effectsIntercept 7.60* 7.13* 7.78* 24.83* 7.73*Residual 40.47*** 40.47*** 40.47*** 166.35*** 40.47***

*p < 0.05.

**p < 0.01.

***p < 0.001.

Table 9. Examining the effect of teacher-level predictors on child’s overdependence on theteacher (OVERDEP).

BaselineModel 1

NeuroticismModel 2

ExtraversionModel 3

AgreeablenessModel 4

Depression

Intercept 8.74*** 9.49*** 6.90* 8.46***Teacher-level predictorsNeuroticism −0.02Extraversion 0.04Agreeableness N/ADepression 0.03Random effectsIntercept 4.67*** 4.79*** 4.76*** 0.0 4.78***Residual 5.76*** 5.76*** 5.76*** 0.0 5.76***

*p < 0.05.

**p < 0.01.

***p < 0.001.

European Early Childhood Education Research Journal 69

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

variance from the baseline model to Model 4 indicated that approximately 34% of theexplainable variation in teacher–child relationship conflict between classrooms couldbe attributable to levels of depression in the teacher. No significant teacher effectswere found for relational closeness and overdependence (Tables 8 and 9, respectively).

Effect of cultural membership – Question 4

After having regressed each of the outcome variables on the variable indicating culturalmembership, the only significant main effect suggesting a difference between the Hungar-ian and theAmerican sampleswaswith respect to levels of child over-dependence. Resultsindicated thatHungarian teachers, on average, reported higher levels of overdependence ofthe children in their classroom than did American teachers (t = 2.01, p < 0.05).

Despite the fact that no further significant main effects were found, there was a mar-ginally significant interaction effect between age and country (β = 5.06, p = 0.06).Namely, there was an apparent positive trend in the American sample such that olderchildren showed higher levels of relationship closeness than did younger children. Incontrast, a negative trend was found in the Hungarian sample such that younger chil-dren showed higher levels of relationship closeness than did older children. Theresults of this model are presented in Figure 2.

As apparent from Figure 2, although the levels of relationship closeness appears to besimilar in both samples for four-year-olds, there is a reverse effect of the variable culturalmembership with respect to the three-year-old and the five-year-old groups. More specifi-cally, the average levels of relationship closeness for three-year-olds is approximately threepoints lower than that of five-year-olds in the American sample, and is approximately fourpoints higher than that of five-year-olds in the Hungarian sample.

Discussion

In line with our expectations, teacher–child relationships varied within classrooms, aswell as between classrooms. Interestingly, the findings demonstrated greater within- as

Figure 2. The effect of cultural membership and age on teacher ratings of teacher–childrelationship closeness.

70 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

opposed to between-classroom variation in closeness, conflict and over-dependence inteacher–child relationships indicating that differences in the teacher–child relationshiptend to be more child–related than teacher–related. In other words, variations inrelationships appear to be better explained by child attributes and the ‘fit’ between chil-dren and teachers rather than teacher attributes.

Several child characteristics were associated with variation in relationship quality.More specifically, girls’ relationships were characterised by lower levels of conflictand higher levels of closeness with their teachers than were boys’ relationships.This finding is consistent with previous work primarily concerning Americansamples (e.g. Howes, Philipsen, and Peisner-Feinberg. 2000), and thus indicatesthat the differential treatment of children based on gender is not specific to oneculture, but rather is a more universal phenomenon. Numerous research studiesclaim that parents as well as teachers tend to treat children differently based ontheir gender. In particular, boys tend to receive more negative and more non-academicfeedback from teachers than girls (Irvine 1986). On the other hand, girls tend to beviewed as more engaged with the school environment (Birch and Ladd 1997), andtend to be characterised in teacher reports as having more closeness and beingmore dependent than boys (Howes et al. 2000).

Interestingly, mother-reported child shyness, had an additive effect with gender onlevels of teacher–child relationship conflict. Higher levels of shyness appeared to bemore ‘beneficial’ for girls than for boys, as for girls higher levels of shyness wereassociated with lower levels of conflict. This finding in the Hungarian sample replicatesthat in the American sample. These findings are interesting, as they imply that teachersmay have different expectations of girls than of boys with respect to their behaviors.Likewise, shy boys may develop more conflictual relationships with teachers thanshy girls due to potential gender-based coping mechanisms (Fagot and Kavanagh2008; Rudasill and Rimm-Kaufman 2009). In addition, acting out behaviors tend tobe expected more from boys than from girls, while withdrawn behaviors tend to beexpected more from girls than from boys. In particular, girls tend to be viewed asshy and introverted more often than do boys (Lamb 1998). As children characterisedas more shy tend to be more socially withdrawn and thus tend to compete less forthe teacher’s attention when compared with children characterised as less shy, the tea-chers’ expectations and initiations become more critical for them. As shyness may be amore expected attribute of girls than of boys, shy girls may not seize the teacher’s atten-tion as readily, and thus the teacher may not try to initiate more interactions with them.

Of the child-level predictors, the child’s age showed a main effect on levels ofrelationship closeness. In particular, younger children had higher levels of closeness,on average, in their relationships with their teachers, when compared to older children.These findings are consistent with developmental theory (Lamb 1998), as well as withthe curriculum of educationally-oriented childcare programs (Nagy 1998).

With respect to teacher-level predictors, those teachers characterised with higherlevels of neuroticism tended to report more conflicted relationships with children. Inaddition, higher levels of depression in teachers were associated with more conflictedrelationships. These findings generally replicate those in the American sample, andindicate that children, in general, appear to benefit from a psychologically healthyteacher, a finding consistent with the reported effects of psychologically healthymothers on children’s well-being (Belsky and Barends 2002).

Interestingly, few differences were found with respect to qualities of teacher–childrelationships between the American and the Hungarian samples. This finding has

European Early Childhood Education Research Journal 71

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

important implications for research on childcare exposure, which has shown negativeeffects of prolonged exposure to children’s social development (Belsky 1988; White1995). Despite the fact that the majority of the children in the Hungarian sampleattended childcare for over 35 hours, which was substantially higher than childrenin the American sample, the distribution of relationship qualities were similar tothose in the American sample (Koles et al. 2009). Therefore, even though the Hun-garian children tended to attend childcare for substantially greater numbers ofhours than the American children, Hungarian children did not appear to experiencemore negativity in the teacher–child relationship than did their Americancounterparts.

Age, however, did play a distinct role with respect to levels of closeness in relation-ships across cultures. More specifically, older American children showed higher levelsof relationship closeness than did younger American children. The opposite was appar-ent for Hungarian children, such that younger Hungarian children showed higher levelsof closeness than did their older counterparts. Due to the small sample size in the Amer-ican study, as well as to the marginal effect of the result, it would be presumptuous todraw major conclusions based on the interaction between child age and country. Never-theless, it is meaningful to consider some potential possibilities. Some of the reasonsmay include the higher likelihood of children’s exposure to multiple centers, and thehigher turnover rate of teachers in American than Hungarian childcare settings(Lamb 1998).

Educational implications

The greater the within- than between-classroom variation in closeness, conflict andover-dependence in the teacher–child relationship has important potential implicationsfor educators as well as for future research. Increasing teachers’ awareness of some ofthe factors influencing their relationships is important for assisting teachers in forminghigh quality relationships with their students. Furthermore, such knowledge base mayaid in protecting children from unnecessary biases and prejudices simply drawn on thebasis of child characteristics. The findings that teachers tended to form differentrelationships with boys and with girls as a function of basic attributes such as childgender or child temperament highlight the need for teachers to be sensitive to theirown biases. Teacher action research projects may be an effective way in which toincrease pre- and in-service teachers’ awareness of biases and interactions that mayharm their relationships with students. More specifically, action research projectsmay support teachers in observing and documenting their own behaviors in responseto child characteristics and in sharing those behaviors with other teachers (Mooreand Gilliard 2008; Rust and Meyers 2006).

Web-based professional development programs may also provide teachers withsupport in developing high quality relationships with students in their classes.Results from one study found that prekindergarten teachers evidenced high levels ofengagement in a web-based program, MyTeachingPartner, designed to provide tea-chers with information and consultation regarding teaching practices and interactionswith students and with a set of techniques designed to build positive and supportiverelationships between teachers and students (Whitaker et al. 2007). Teachers whopartook in the program and who received feedback on their classroom interactionsreported an increase in activities with students in their classes focused on supportinghigh quality relationships (Whitaker et al. 2007).

72 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Limitations

The extensive reliance on teacher reports is a potential shortcoming of the study. In par-ticular, data were obtained through teacher reports assessing both the qualities ofteacher–child relationships and teacher personality traits, potentially leading toshared methods variance. These reports are likely to reflect on the mood of theteacher, and to be further influenced by other situational variables. Future studies,hence, need to assess data from sources other than teacher reports.

Cross-cultural comparisons are important, as they provide valuable policy insights,as well as caution us against potential overgeneralisations of findings (Lamb 1998).However, despite expectations, we failed to find any differences between the Hungarianand the American samples with respect to levels of closeness in teacher–child relation-ships. Further research is recommended utilising a larger sample size for the Americansample. Due to the great discrepancy between the sizes of the two samples, we were notable to draw powerful conclusions, and may also have missed some findings that mayhave been detected with a larger American sample.

One further limitation lies in the fact that although we managed to identify that themajority of variation in all outcome variables was within- as opposed to between-class-rooms, we failed to identify appropriate child-level predictors to explain a sufficientamount of this variation. More specifically, only 4 to 11% of the within-classroomvariation in either of the outcome variables was explained by child-level predictorsgender and age. Hence, future studies are necessary to examine additional student-and teacher-level factors based on substantive reasoning to explain some more of thewithin- as well as between-classroom variation.

Notes1. The alpha coefficients are those reported by Pianta (1992).2. The alpha coefficients are those reported by Rothbart et al. (1994).3. The alpha coefficients are those reported by Costa and McCrae (1985).

ReferencesAdamik, M. 1991. “Hungary-Supporting parenting and Child Rearing. Policy Innovation in

Eastern Europe.” In Child Care, Parental Leave, and the Under 3s-Policy Innovation inEurope, edited by S. B. Kamerman and A. J. Kahn, 115–144. New York: Auburn House.

Anderson, C. W., R. J. Nagle, W. A. Roberts, and J. W. Smith. 1981. “Attachment to SubstituteCaregivers as a Function of Center Quality and Caregiver Involvement.” Child Development52: 53–61.

Bachu, J. 1995. Fertility of American women: June 1994. (U.S. Bureau of Census CurrentPopulation Report pp. 20–482). Washington, DC: US Government Printing Office.

Barnett, W. S., D. J. Epstein, A. H. Friedman, R. A. Sansanelli, and J. T. Hustedt. 2009. TheState of Preschool 2009: State Preschool Yearbook. New Brunswick, NJ: The NationalInstitute for Early Education Research.

Belsky, J. 1988. “The ‘Effects’ of Infant Daycare Reconsidered.” Early Childhood ResearchQuarterly 3: 235–272.

Belsky, J., and N. Barends. 2002. “Personality and Parenting.” In Handbook of Parenting:Vol. 3: Being and Becoming a Parent, edited by M. H. Bornstein, 2nd ed. 415–438.Mahwah, NJ: Lawrence Erlbaum Associates.

Belsky, J., K. Crnic, and S. Woodworth. 1996. “Personality and Parenting: Exploring theMediating role of Transient Mood and Daily Hassles.” Journal of Personality 63 (4):905–929.

Birch, S. H., and G. W. Ladd. 1997. “The Teacher–Child Relationship and Children’s EarlySchool Adjustment.” Journal of School Psychology 35 (1): 61–79.

European Early Childhood Education Research Journal 73

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Bowlby, J. 1988. A Secure Base: Clinical Applications of Attachment Theory. London, UK:Routledge.

Costa, P., and R. McCrae. 1985. The NEO Personality Inventory Manual. Odessa, FL:Psychological Assessment Resources, Inc.

Csapo, M. 1984. “Child Care Special Education in Hungary.” Early Child Development andCare 14: 251–268.

Doll, B. 1996. “Prevalence of Psychiatric Disorders in Children and Youth: An Agenda forAdvocacy by School Psychology.” School Psychology Quarterly 11 (1): 20–46.

Fagot, B. I., and K. Kavanagh. 2008. “The Prediction of Antisocial Behavior from AvoidantAttachment Classifications.” Child Development 61 (3): 864–873.

Fodor, E., C. Glass, J. Kawachi, and L. Popescu. 2002. “Family Policies and Gender in Hungary,Poland and Romania.” Communist and Post Communist Studies 35: 475–490.

Hamre, B. K., and R. C. Pianta. 2001. “Early Teacher–Child Relationships and the Trajectoryof Children’s School Outcomes Through Eighth Grade.” Child Development 72 (2):625–638.

Hamre, B. K., R. C. Pianta, J. T. Downer, and A. J. Mashburn. 2007. “Teacher’s Perceptions ofConflict with Young Students: Looking Beyond Problem Behaviors.” Social Development17 (1): 115–135.

Howes, C. 1991. “Caregiving Environments and their Consequences for Children: TheExperience in the United States.” In Day Care for Young Children: InternationalPerspectives, edited by E. C. Melhuish and P. Moss, 185–197. New York: Routledge.

Howes, C. 2000. “Social–Emotional Classroom Climate in Child Care, Child–TeacherRelationships and Children’s Second Grade Peer Relations.” Social Development 9 (1):191–204. Oxford: Blackwell Publishers.

Howes, C., and C. E. Hamilton. 1992a. “Children’s Relationships with Caregivers: Mothers andChild Care Teachers.” Child Development 63: 859–866.

Howes, C., and C. E. Hamilton. 1993. “The Changing Experience of Child Care: Changes inTeachers and in Teacher–Child Relationships and Children.” Early Childhood ResearchQuarterly 8 (1): 15–32.

Howes, C., C. E. Hamilton, and C. C. Matheson. 1994. “Children’s Relationships with Peers:Differential Associations with Aspects of the Teacher–Child Relationship.” ChildDevelopment 65: 253–263.

Howes, C., and C. C. Matheson. 1992. “Sequences in the Development of Competent Play withPeers: Social and Social Pretend Play.” Developmental Psychology 28 (5): 961–974.

Howes, C., L. C. Philipsen, and E. Peisner-Feinberg. 2000. “The Consistency of PerceivedTeacher–Child Relationships Between Child Care and Kindergarten.” Journal of SchoolPsychology 38 (2): 113–132.

Howes, C., and S. Ritchie. 1999. “Changes in Child – Teacher Relationships in a TherapeuticPreschool Program.” Early Child Education and Development 9: 411–422.

Irvine, J. J. 1986. “Teacher–Student Interactions: Effects of Student Race, Sex, and GradeLevel.” Journal of Educational Psychology 78 (1): 14–21.

Kahn, A. J., and S. B. Kamerman. 1998. Big Cities in the Welfare Transition. NY: ColumbiaUniversity School of Social Work.

Kawaguchi, M. C., D. P. Welsh, S. C. Powers, and S. S. Rostosky. 1998. “Mothers, Fathers,Sons and Daughters: Temperament, Gender, and Adolescent–Parent Relationships.”Merrill-Palmer Quarterly 44 (1): 77–96.

Kesner, J. E. 2000. “Teacher Characteristics and the Quality of Child–Teacher Relationships.”Journal of School Psychology 28 (2): 133–149.

Koles, B. 2002. Teacher–Child Relationships in Child Care: The Influence of Child and TeacherCharacteristics. Cambridge, MA: Harvard University, Library Archives.

Koles, B., and E. O’Connor. 2003. “Teacher–Child relationships and their predictors. Societyfor Research and Child Development.” 70th annual meeting, Tampa, FL.

Koles, B., E. O’Connor, and K. McCartney. 2009. “Teacher–Child Relationships inPrekindergarten: The Influences of Child and Teacher Characteristics.” Journal of EarlyChildhood Teacher Education 30: 3–21.

Kowalski,H. S., S.R.Wyver,G.Masselos, andP.DeLacey. 2004. “Toddlers’EmergingSymbolicPlay: A First-Born Advantage?” Early Child Development & Care 174 (4): 389–400.

74 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Lamb, M. E. 1998. “Nonparental Child Care: Context, Quality, Correlates, and Consequences.”In Handbook of Child Psychology. Vol. 4: Child Psychology and Practice, edited by I. E.Sigel, K. A. Renninger, and W. Damon, 73–133. New York: John Wiley & Sons, Inc.

Lannert, J., and G. Halasz, eds. 2003. Education in Hungary, 2003. Budapest, Hungary:National Institute of Public Education.

Lerner, J. A., R. M. Lerner, and S. Zabski. 1985. “Temperament and Elementary SchoolChildren’s Actual and Rated Academic Performance: A Test of a ‘Goodness-of-Fit’Model.” Journal of Child Psychology and Psychiatry 26 (1): 125–136.

Melhuish, E. C., and P. Moss. 1991. Day Care for Young Children: International Perspectives.New York: Routledge.

Mill, D., and D. Romano-White. 1999. “Correlates of Affectionate and Angry Behavior in ChildCare Educators of Preschool-Aged Children.” Early Childhood Research Quarterly 14 (2):155–178.

Moore, R. A., and J. L. Gilliard. 2008. “Preservice Teachers Conducting Action Research inEarly Education Centers.” Journal of Early Childhood Teacher Education 29: 45–58.

Nagy, M. 1998. “Teacher Training in Central Europe: Recent Developments.” EuropeanJournal of Education 33 (4): 393–406.

Olmstead, P. P. and M. M. Senninger. 1995. A Nation-by-Nation Look at the Findings. In TheIEA Preliminary Study: Early Childhood Care and Education in 11 Countries, edited byP. P. Olmsted and D. P. Weikart, 137–189. Oxford, England: Elsevier Science.

Patrick, H., K. Yoon, and A. M. Murphy. 1995, March. “Personality characteristics, socialcompetence, and early school adjustment: A contextual and developmental perspective.”Poster session presented at the biannual meeting of the Society for Research in ChildDevelopment, Indianapolis, IN.

Pianta, R. C. 1992. The Student–Teacher Relationship Scale. Charlottesville, VA: University ofVirginia.

Pianta, R. C. 1994. “Patterns of Relationships Between Children and Kindergarten Teachers.”Journal of School Psychology 32 (1): 15–31.

Pianta, R. C. 1997. “Adult–Child Relationship Processes and Early Schooling.” Early Education& Development 8 (1): 11–26.

Pianta, R. C. 1999. Enhancing Relationships: Between Children and Teachers. Washington,DC: American Psychological Association.

Pianta, R. C., and S. L. Nimetz. 1991. “Relationships Between Children and Teachers:Associations with Classroom and Home Behavior.” Journal of Applied DevelopmentalPsychology 12 (3): 379–393.

Pianta, R. C., S. L. Nimetz, and E. Bennett. 1997. “Mother-child relationships, teacher-childrelationships, and school outcomes in preschool and kindergarten”. Early ChildhoodResearch Quarterly 12 (3): 263–280.

Pianta, R. C., M. S. Steinberg, and K. B. Rollins. 1995. “The First Two Years of School:Teacher–Child Relationships and Deflections in Children’s Classroom Adjustment.”Development and Psychopathology 7: 295–312.

Radloff, L. S. 1997. “The CES-D Scale: A Self Report Depression Scale for Research in GeneralPopulation.” Applied Psychological Measurement 1: 385–401.

Rimm-Kaufman, S. E., D. M. Early, M. J. Cox, G. Saluja, R. C. Pianta, R. H. Bradley, andC. Payne. 2002. “Early Behavioral Attributes and Teachers’ Sensitivity as Predictors ofCompetent Behavior in the Kindergarten Classroom.” Journal of Applied DevelopmentalPsychology 23 (4): 451–470.

Rimm-Kaufman, S. E., and J. Kagan. 2005. “Infant Predictors of Kindergarten Behavior: TheContribution of Inhibited and Uninhibited Temperament Types.” Behavioral Disorders30 (4): 331–347.

Rothbart, M. K., S. A. Ahadi, and K. L. Hershey. 1994. “Temperament and Behavior inChildhood.” Merril-Palmer Quarterly 40: 21–39.

Rudasill, K. M., and S. E. Rimm-Kaufman. 2009. “Teacher–Child Relationship Quality: TheRoles of Child Temperament and Teacher–Child Interactions.” Early Childhood ResearchQuarterly 24 (2): 107–120.

Rudasill, K. M., S. Rimm-Kafuman, L. Justice, and K. Pence. 2006. “Temperament andLanguage Skills as Predictors of Teacher–Child Relationship Quality in Prekindergarten.”Early Education and Development 17 (2): 271–291.

European Early Childhood Education Research Journal 75

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3

Rust, F., and E. Meyers. 2006. “The Bright Side: Teacher Research in the Context ofEducational Reform and Policy-Making.” Teachers & Teaching: Theory & Practice 12(1): 69–86.

Rydell, A. M., G. Bohlin, and L. B. Thorell. 2005. “Representations of Attachment to Parentsand Shyness as Predictors of Relationships with Teachers and Peer Competence inPreschool.” Attachment and Human Development 72 (2): 187–704.

Saft, E. W. 1995. “A Descriptive Study of the Student–Teacher Relationship Scale used withChild Careers.” Dissertation Abstracts International, 55(10-A).

Skarpness, L. R., and D. K. Carson. 1986. “Temperament, Communicative Competence andthe Psychological Adjustment of Kindergarten Children.” Psychological Reports 59 (3):1299–1306.

Singer, J. D. 1998. “Using SAS PROC MIXED to Fit Multilevel Models, Hierarchical Models,and Individual Growth Models.” Journal of Educational and Behavioral Statistics 24 (4):323–355.

Spilt, J. L., and H. M. Y. Koomen. 2009. “Widening the View on Teacher–Child Relationships:Teachers’ Narratives Concerning Disruptive Versus Nondisruptive Children.” SchoolPsychology Review 38 (1): 86–101.

Sulloway, F. J. 1996. Born to Rebel: Birth Order, Family Dynamics, and Creative Lives. NewYork: Pantheon Books.

Vari, P. 1989. “Care and Education in Hungary for 3- to 6-Year-Olds.” In How Nations ServeYoung Children: Profiles of Child Care and Education in 14 Countries, edited byP. P. Olmsted and D. P. Weikart, 143–149. Michigan: High/Scope Educational ResearchFoundation.

Whitaker, S., M. Kinzie, M. E. Kraft-Sayre, A. Mashburn, and R. C. Pianta. 2007. “Use andEvaluation of Web-Based Professional Development Services Across Participant Levelsof Support.” Early Childhood Education Journal 34 (6): 379–386.

White, B. L. 1995. The New First Three Years of Life. New York: Simon & Schuster.Zajonc, R. B. 2001. “The Family Dynamics of Intellectual Development.” American

Psychologist 56 (6/7): 490–496.

76 B. Koles et al.

Dow

nloa

ded

by [

Bri

an C

ollin

s] a

t 16:

41 1

3 M

arch

201

3