at SciVerse ScienceDirect

Learning and Instruction 23 (2013) 1e9

Contents lists available

Learning and Instruction

journal homepage: www.elsevier .com/locate/ learninstruc

Transactional associations between classroom engagement and relations withteachers from first through fourth grade

I. Archambault*, L.S. Pagani, C. FitzpatrickÉcole de Psychoéducation, Université de Montréal, C.P. 6128, succ. Centre-Ville, Montréal, Québec, Canada H3C3J7

a r t i c l e i n f o

Article history:Received 24 January 2012Received in revised form10 September 2012Accepted 14 September 2012

Keywords:Approaches to learningClassroom engagementTeacherestudent relationsAchievementLearning-related behavior

* Corresponding author. Tel.: þ1 514 343 6111�875E-mail addresses: isabelle.archambault@umon

linda.s.pagani@umontreal.ca (L.S. Pagani), caro(C. Fitzpatrick).

0959-4752/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.learninstruc.2012.09.003

a b s t r a c t

Using structural equation modeling with a population-based cohort of French-speaking children fromQuebec (Canada), prospective associations were made between two previously established factorsunderlying student performance d classroom engagement and the teacherestudent relations. Ourresults show developmental continuity in classroom engagement and teacherestudent relations fromgrades 1 through 4, beyond the influence of confounding child factors (sex, kindergarten cognitive skills,and second grade achievement) and family factors (such as maternal education). Although they wereboth relatively stable over time, closer relations with teachers showed comparatively less stability thanclassroom engagement. That is, classroom engagement showed the most developmental continuity fromone grade to the next. Because intervention programs targeting very young children are among the mostcost-effective (Heckman, 2006), our findings suggest the benefits of investing in evidence-basedprograms, in concert with practitioners, to promote positive teacher relations with students andencourage their active classroom participation and involvement.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

From school entry, students are confronted with importantcognitive and behavioral challenges. While having to adapt toa group-based social environment, they are asked to make an effortduring classroomactivities, listen carefully, follow rules, participate,work autonomously, and invest considerable persistence. Childrenvary in their ability to respond to such expectations (Ladd, Buhs, &Seid, 2000). In turn, the behavioral responses children produce areimportant as they chart the developmental course toward academicsuccess (Johnson, McGue, & Iacono, 2006; Ladd & Dinella, 2009;Skinner, Kindermann, & Furrer, 2009). For youngsters, engagementis amalleable process that evolves over time and is responsive to theinterpersonal transactions in the classroom environment (Simons-Morton & Chen, 2009; Sirin & Rogers-Sirin, 2004). Beyond indi-vidual characteristics (Ladd et al., 2000; Pagani, Fitzpatrick,Archambault, & Janosz, 2010), the bond between teacher andstudent remains a fundamental aspect of the childhood schoolingexperience. Past research has shown that close and supportiveresponses from teachers have important implications for learning-

2; fax: þ1 514 343 6951.treal.ca (I. Archambault),line.fitzpatrick@umontreal.ca

All rights reserved.

related behaviors and achievement in students (Birch & Ladd, 1997;Furrer & Skinner, 2003; Hughes, Luo, Kwok, & Loyd, 2008; Pianta,Steinberg, & Rollins, 1995).

Until recently, there have been few studies that examine thedevelopmental links between student engagement and teacherestudent relations in primary school children. A recent study byHughes et al. (2008) found reciprocal links between teacherestudent relations, engagement, and achievement over threeconsecutive years. Students showing early signs of behavioraldisengagement in class were at higher risk of future academicdifficulties. Considering that engagement and the quality of rela-tions with teachers are important for the academic dimension ofschooling (Hamre & Pianta, 2001; Ladd & Dinella, 2009), researchexamining the developmental continuity of these factors is nowwarranted.

1.1. Defining classroom engagement

The concept of student engagement has inspired a number ofapproaches (Appleton, Christenson, & Furlong, 2008; Finn, 1989;Ladd et al., 2000). Upon conducting an extensive review of theextant literature, Fredricks, Blumenfeld, and Paris (2004) offer threedimensions which help operationalize student engagement. Theseare behavioral, affective, and cognitive. The affective and cognitivedimensions respectively refer to student feelings, interests, and

I. Archambault et al. / Learning and Instruction 23 (2013) 1e92

attitudes toward school (Eccles, Wigfield, Harold, & Blumenfeld,1993; Finn, 1989; Goodenow, 1993) and psychological investmentin learning anduse of self-regulation strategies (Ablard& Lipschultz,1998; Connell &Wellborn,1991; Pintrich&deGroot,1990; Skinner&Belmont, 1993). The behavioral dimension of engagement refers tobehavioral dispositions and conduct when approaching andundertaking school-related tasks (McDermott, Mordell, & Stoltzfus,2001). Such behaviors include compliance to classroom rules andinstructions, attention, efforts, and participation in classroom-related activities (Finn, 1989; Fredricks et al., 2004). Although thecognitive and affective dimensions of engagement remain impor-tant influences on the academic life-course (Archambault, Janosz,Fallu, & Pagani, 2009), student behavioral engagement is oftenseen as a reflection of affective and cognitive involvement(Archambault, Janosz,Morizot, & Pagani, 2009; Eccles, 2004; Ladd &Dinella, 2009). Moreover, classroomparticipation is associatedwithpositive changes in attitudes about school (Ladd & Dinella, 2009).Early signs of classroom effort and participation chart the coursetoward future academic difficulties (Hughes et al., 2008). Finally,learning-related behaviors are more concrete targets for interven-tion (Diamond, Barnett, Thomas, &Munro, 2007). Accordingly,manystudies have draw upon behavioral engagement as a key elementpredicting later academic adjustment.

The contribution of early behavioral engagement on studentacademic attainment outcomes is well established (Alexander,Entwisle, & Dauber, 1993; Ladd & Dinella, 2009; McClelland,Morrison, & Holmes, 2000; Skinner, Wellborn, & Connell, 1990;Valiente, Lemery-Chalfant, Swanson, & Reiser, 2008). For example,Ladd and Dinella (2009) followed a sample of 383 children fromkindergarten to eighth grade and found that children who werecooperatively engaged in the classroom and responded to teacherdemands were most likely to show academic gains over timebeyond important cognitive, family, and demographic characteris-tics. Valiente et al. (2008) further showed that classroom partici-pation partially mediated the link between student self-regulationand GPA even after controlling for important confounding variablessuch as sex, family income, parental education, and prior grades.According toWentzel (1999), childrenwho are actively engaged aremore likely to achieve academically because they seek goals thatare valued bymembers of their school and classroom environment.To the extent that some stability exists in engagement patternsacross the primary grades (Ladd & Dinella, 2009), some childrenmay find themselves on a continuous risk trajectory starting in theearly grades. Such a trajectory may be influenced by both childcharacteristics and features of their learning environment. Teacherslikely have an important influence on this learning environmentgiven their level of responsibility and pedagogical objectives, in theface of heterogeneous group composition (Hijzen, Boekaerts, &Vedder, 2007).

1.2. Classroom engagement and teacherestudent relations

For some authors, classroom engagement can specifically beoperationalized by the quality of student involvement with class-mates and teachers (Finn, 1989; Ladd et al., 2000; Skinner &Belmont, 1993; Skinner et al., 2009, 1990). For others, the relationsstudents build with teachers do not serve as an indicator but rather,as a foundation or consequence of classroom engagement(Battistich, Solomon, Watson, & Schaps, 1997; Birch & Ladd, 1997;Marks, 2000; Rimm-Kaufman, Curby, Grimm, Nathanson, & Brock,2009). In line with attachment theory perspectives (Bowlby, 1980),building a supportive relationship with a significant caregiver iscritical for child development. Parents and teachers are the mostimportant vehicles for socialization and among the most frequentlymentioned sources of support during the school age years (Resnick

et al., 1997). In classrooms that allows teachers to engage in moreindividualized teaching (Blatchford, Bassett, & Brown, 2011), thesupportive reciprocal relationship teachers develop with theirstudents becomes an important predictor of day-to-day engage-ment, which enables children to consolidate the emotional securityneeded to explore their social environment and copewith academicchallenges. An observational study by Ponitz, Rimm-Kaufman,Grimm, and Curby (2009) using a sample of 171 kindergartenersfound that classroom and teacherestudent relationship quality isindirectly associated with children’s reading ability throughimprovements in classroom engagement. According to their study,when classroom organization and teacher emotional and instruc-tional support are high, children successfully adopt learning-relatedbehaviors which contribute to their academic success. Otherresearch shows that when teachers present themselves as warm,supportive, and responsive to individual needs, students respondwith better behavioral engagement in the classroom and achieve-ment (Brophy, 1983; Connell & Wellborn, 1991; Hamre & Pianta,2001; Hughes et al., 2008; McWilliam, Scarborough, & Kim, 2003;Wentzel, 1999). In some cases, a close and supportive relationshipwith teachers may even attenuate risks associated with priorunderachievement and buffer children against the negative effectsof family or sociodemographic adversity (Resnick et al.,1997); whileconversely, negative and controlling teacher behaviors can havea negative influence (Assor, Kaplan, Kanat-Maymon, & Roth, 2005)

Most importantly, the associations between teacherestudentrelationship and student engagement could also be viewed astransactional (Sameroff & Fiese, 1990; Skinner, Furrer, Marchand, &Kindermann, 2008). This idea implies, on one hand, that studentswho perceive their teacher as being supportive might be moreresponsive to academic expectations and demands and tend to bebetter engaged in classroom-related activities. In support of thisview, longitudinal findings indicate that negative changes inTeacherechild relations are associated with a decline in studentachievement, even after controlling for important variables such asIQ (Hamre & Pianta, 2001). On the other hand, students who showclassroom involvement followdirectives and instructions, andworkneatly and cooperatively could reinforce positive teacher attitudesand support. That is, student self-regulation in the classroomgenerates more positive and spontaneous reciprocal responsesfrom teachers, which can ultimately promote their achievement-related outcomes (Koriat, 2012; Rimm-Kaufman et al., 2009).

1.3. Objective and hypotheses

Previous findings suggest that student engagement and rela-tionships with teachers underscore transactional processes whichset the stage for short and long-term educational progress.Consistentwith this body of work, we used a population-based dataset of French-speaking Canadian children from Quebec (Canada) toconduct a more in-depth examination of the links between class-room engagement and teacherestudent relations and address theirdevelopmental course from first grade through fourth grade. Giventhe notion of developmental continuity, our initial hypothesis wasthat first grade classroom engagement would be positively associ-ated with fourth grade engagement (hypothesis 1). Second, weexpected that students who share a close relationship with theirfirst grade teacher would also share a close relationship with theirfourth grade teacher (hypothesis 2). Third, we anticipated thatclassroom engagement and close teacherestudent relations wouldcovary along the same school year, in first and fourth grade(hypothesis 3). Finally, we expect that these two dimensions wouldtransact over time. That is, classroom engagement in first gradeshould predict teacherestudent close relations in fourth grade(hypothesis 4a) and close teacherestudent relations in first grade

2 An attrition analysis comparing retained and non-retained participants ondemographic measures revealed some differences. Children in the retained samplewere more likely to be girls, and to be of Canadian ethnicity. Retained childrenweremore likely to have been exposed to French as their primary home language, lesslikely to come from single-parent families, and generally came from families that

I. Archambault et al. / Learning and Instruction 23 (2013) 1e9 3

should be positively associated with classroom engagement infourth grade (hypothesis 4b).

2. Method

2.1. Participants and procedure

The Quebec Longitudinal Study of Child Development (QLSCD)originates from a randomly selected birth cohort born between1997 and 1998 in the Canadian province of Quebec.1 The Institut dela Statistique du Québec obtained informed consent from parents atall survey points and coordinated the data collection procedures.The QLSCD comprises an infancy/early childhood phase anda school age phase.

For the infancy/early childhood phase, a stratified sample of2694 infants (and their families), representing 94.5% of the targetpopulation, were first assessed at 5 months. These infants wereconsidered eligible for the longitudinal follow-up from 17 monthsto 53 months. From those deemed eligible at the 5 month wave, 8children became ineligible (inconsistencies in documentation) and689 became untraceable. Thus, 1997 infants were eligible for entryinto the longitudinal study from the second wave onward (17months). From this sampling procedure, 1820 cases were retainedwith informed consent, evenly represented of girls and boys. Overa third (39%) of this sample was firstborn. Biological parents werepresent in the household. The average maternal age at birth of thetarget child was 26.93 years (SD ¼ 8.90), and 13.4% of mothers and16.9% of fathers did not have a high school diploma.

Child assessments occurred at 5, 17, 29, 41, and 53 months ofage using computerized parent interviews, parent questionnaires,individual child assessments, and where appropriate, teacherquestionnaires. Mothers were the most common data source forparent questionnaires in 94e99% of cases. For the school-agephase of the QLSCD, children had to be 60-months on September30 to be eligible for kindergarten entry in fall of 2002. Kinder-garten data on individually-assessed school readiness in math andvocabulary knowledge was gathered at home in the spring of 2003for 1155 kindergarten children (607 girls and 538 boys at 65-months). Ten children were derogated into kindergarten theprevious year, resulting in a sample of 1145 kindergartenersavailable for analysis. By the spring of second grade (2005), allparticipants were attending French speaking elementary school.The nature of the random sampling from the birth registry toachieve provincial representativeness resulted in no clustering ofchildren in the same schools or with the same teachers during theschool age phase.

2.2. Measures

The objectives of this paper are achieved through secondaryanalysis of already existing population-based data from establishedinstruments that emanate from the National Longitudinal Study ofChildren and Youth (NLSCY; for details, see: http://www23.statcan.gc.ca:81/imdb/p2SV.pl?Function¼getSurvey&SDDS¼4450&lang¼en&db¼imdb&adm¼8&dis¼2).

2.2.1. Kindergarten control variablesChildren completed the Number Knowledge Test (NKT, abridged

version) and the Peabody Picture Vocabulary Test (PPVT, Frenchadaptation; Dunn, Thériault-Whalen, & Dunn, 1993) with a trainedexaminer. The NKT adjusted for 5-year-olds measure child

1 QLSCD information is available at www.jesuisjeserai.stat.gouv.qc.ca/etude_an.htm.

knowledge of the (1) number sequence (2) ability to mapa sequence to objects using a one to one correspondence (3)cardinal values of numbers; (4) rule relating adjacent cardinalvalues; and (5) the rule that a successive numbers represents a setcomposed of a greater quantity of objects (Okamoto & Case, 1996).The PPVT measured receptive vocabulary knowledge. During thistest, examiners present 175 vocabulary items in increasing level ofdifficulty. A French version of the test was standardized by Dunnet al. (1993) and correlates well with other French languagevocabulary and intelligence tests (Dunn et al., 1993). Mothers alsoprovided reports of their level of educational attainment in terms ofhigh school completion or not.

2.2.2. Second grade academic achievement control variableTeachers rated the child’s math and reading achievement by

selecting from the following options (a) greatly above the class-room mean; (b) slightly above the classroom mean; (c) at theclassroom mean; (d) slightly below the classroom mean; and (e)greatly below the classroom mean. Global Achievement scoreswere calculated for each child by computing themean of their mathand reading scores. Teacher ratings of achievement have beenshown to detect even small changes in academic performance(Pagani, Tremblay, Vitaro, Boulerice, & McDuff, 2001) and todemonstrate as much sensitivity as individual achievement tests(Duncan et al., 2007)

2.2.3. Classroom engagementIn the springof grades1 and4, homeroomteachers reportedupon

7 items of classroom engagement (Pagani et al., 2010). Items wererated from 1 (Never) to 5 (Always) and included: Follows directions;Works independently; Listens Attentively; Works neatly and care-fully; Puts a lot of effort into work; Participates in class; and Followrules. Alphas at grade 1 and grade 4 were .93 and .94, respectively.

2.2.4. Teacherechild relationsTeachers also provided reports of their positive and close

interactions with the child in grades 1 and 4 using four itemsemanating from the StudenteTeacher Relationship Inventory (Pianta,1999) are rated from 1 (Definitely does not apply) to 5 (DefinitelyApplies) and include: I share a close and warm relationship withthis child; This child spontaneously shares information about him/herself; It is easy to be in tune with what this child is feeling; andMy interactions with this child make me feel effective and confi-dent. Alphas were.78 and .76 in grades 1 and 4, respectively.

2.3. Data analytic strategy

2.3.1. Treatment of missing dataSome children (n ¼ 417; 36.4% of the sample) had incomplete

data on one ormore variables in first, second, or fourth grade (18.1%of students had missing value in the first grade, 18.5% in the secondgrade, and 36.2% in the fourth grade).2 We imputed all missing dataand arbitrarily created ten datasets using the NORM multipleimputations program (Schafer, 1999). By drawing values from the

reported higher annual family income. Their parents were less likely to rate theirneighborhoods as dangerous, and reported spending more time engaged in playwith crayons with the child. Mothers were also older at their first child birth, moreeducated, and scored higher on verbal IQ. These variables were included in our dataimputation model.

Table 1Results of the one-factor congeneric measurement model.

c2(df) RMSEA SRMR TLI CFI Cronbachalpha

First grade classroomengagement

4118.694 (21) .059 .029 .953 .975 .903

Fourth grade classroomengagement

4057.364 (21) .054 .022 .968 .981 .894

First gradeteacherestudentrelations

917.435 (6) .061 .017 .950 .992 .779

Fourth gradeteacherestudentrelations

955.293 (6) .066 .010 .969 .995 .762

3 The within factor correlated residuals were between the following items:Follows directions and Follow rules; Follow directions and Listens Attentively.

4 The model fit without these correlated residuals is almost identical to theoriginal model (CFI ¼ .86, TLI ¼ .84, RMSEA ¼ .073, SRMR ¼ .062).

I. Archambault et al. / Learning and Instruction 23 (2013) 1e94

conditional distribution of the variables, NORM uses an iterativemethod based on EM algorithm to impute missing data, dependingon the available and valid observations from the original data set(for technical details, see Schafer, 1999)

2.3.2. Testing our hypothesized modelWe used Mplus version 5 and implemented a two-step

approach to Structural Equation Modeling (SEM, Anderson,Greene, & Loewen, 1988). This approach allowed us to specify andtest the adequacy of the proposed model of classroom engagementand teacherestudent relations measures prior to testing the fullstructural model. We used Confirmatory Factor Analysis (CFAs) tocalculate one-factor congeneric models of classroom engagementand teacherestudent relations at grades 1 and 4. These are modelswithin which a single latent factor is measured by several observedvariables. In doing so, we fixed factor loadings arbitrarily at one unitfor each first variable. The other factor loadings and the variance ofall latent constructs were left to vary. We subsequently examinedthe fit for each measurement model. Once these were established,we used the latent constructs to analyze our hypothesized model.Because the Quebec Longitudinal Study of Child Development isbased on a random sample of a birth cohort, children weregeographically dispersed in classrooms and schools across theprovince, thus precluding the use of multilevel modeling

We first tested a saturated model with all paths betweencovariates (i.e., sex, maternal education, math and language skillsduring kindergarten, and second grade achievement) and latentvariables in first and fourth grade. Next, because some researchfinds that boys tend to be less engaged and take more advantagefrom a positive and warm relationship with teachers (Hamre &Pianta, 2001; Wigfield & Eccles, 2002), yet other researchers findsno sex differences in the influence of controlling teacher behaviorson student engagement (Assor et al., 2005), we tested multiplegroup invariance of the final model for boys and girls to ensure themodel is appropriately specified for both genders (Appleton,Christenson, Kim, & Reschly, 2006).

As recommended by Hu and Bentler (1999), we compareddifferent parameters in order to evaluate the CFA and SEM modelfit: Comparative Fit Index (CFI); Tucker-Lewis Index (TLI); RootMean Square Error of Approximation (RMSEA); Standardized RootSquare Mean Residual (SRMR); and the c2. We chose to rely on theindices that are less sensitive to sample size (TLI, RMSEA, Sharma,Mukherjee, Kumar, & Dillon, 2005). A value of .08 or less for theSRMR (Hu & Bentler, 1999) and of .06 or less for RMSEA areconsidered an adequate fit (Hu & Bentler, 1999; MacCallum,Browne, & Sugawara, 1996). A value of .95 and above is consid-ered an excellent fit for CFI and TLI. Also, the c2 should be nonesignificant (Bentler & Bonett, 1980). To compare CFA models witha different number of factors, we also used Akaike InformationCriterion (AIC; Akaike, 1987), Bayesian Information Criterion (BIC:Schwartz, 1978), and Sample-Size Adjusted Bayesian InformationCriterion (ABIC; Sclove, 1987). A better fit is indicated when abso-lute values of these indices become smaller. However, it is prefer-able to rely on the BIC and ABIC rather than the AIC because theyusually select more parsimonious models. Finally, we examinedadequacy of factor loadings. Although factor loadings exceeding .40are considered acceptable (Hair, Anderson, Tatham, & Black, 1998),we chose to adopt a more conservative standard of .50.

3. Results

3.1. Measurement model

We first used Confirmatory Factor Analysis (CFAs) to calculateone-factor congeneric models of classroom engagement and

teacherestudent relations at grades 1 and 4. Table 1 reports the fitindices for each model. Overall, this table indicates that eachmeasure of classroom engagement and teacherestudent relationsshows good modeling of the data. The RMSEA for teacherestudentrelations in first and fourth grade was slightly above the .06 stan-dard; however, the SRMR, TLI, and CFI reached an ideal fit (Hu &Bentler, 1999; MacCallum et al., 1996). Finally, the c2 was signifi-cant in all models but we decided not to rely on this fit index since itis very sensitive to sample size. We next generated Cronbach’salpha coefficient to assess the reliability of the classroom engage-ment and teacherestudent relations scales at first and fourth grade.As reported in Table 1, all scales presented good reliability, withalpha coefficient ranging from .76 to .90.

3.2. Descriptive statistics and bivariate correlations

We next generated descriptive statistics and correlations for allthe variables and scales. Table 2 presents the results of theseanalyses. As shown, the correlation between classroom engage-ment in first and fourth grade is large (Pearson r ¼ .54); while thecorrelation between teacherestudent relations at the same gradeswas moderate (Pearson r ¼ .24).

3.3. Structural equation model

We then tested our SEM model to verify the prospective linksbetween our measures of classroom engagement and teacherestudent relations. Theoretically significant covariates (i.e., sex,maternal education, math or verbal skills in kindergarten) wereincluded in the model. As recommended by MacCallum et al.(1996), the first step involved testing a saturated model whichincludes all possible paths from the covariates, the observed, andlatent factors. The fit indices of this original model were slightlylower than the adequacy standards (CFI ¼ .86, TLI ¼ .84,RMSEA¼ .073, SRMR ¼ .062). Next, we removed all non-significantpaths between the latent construct and retested the model. Fig. 1illustrates the final model with standardized coefficients for allsignificant paths. In this model, we allowed correlations betweensome of the residuals of classroom engagement and teacherestudent relations items in first and fourth grade. These correla-tions of error terms were suggested by the modification indices;two of themwere within factor (engagement) correlations3 and allthe others correlations (11) were between testeretest items.4 Thec2(319) ¼ 12,186.746 was significant. The CFI of .915 and the TLI of.896were close but did not reach the ideal standard of .95 proposed

Table 2Means, standard deviations, and bivariate correlations.

Variables 1 2 3 4 5 6 7 8 9

1 Maternal education2 Peabody picture vocabulary test (PPVT) .15***

3 Number knowledge test (NKT) .17*** .41***

4 Sex(girl) .01 .01 .015 Classroom engagement at first grade .05 .29 .34*** .24***

6 Teacherestudent relations at first grade .00 .20*** .20*** .13*** .40***

7 Academic achievement at second grade .14*** .33*** .42*** .06* .40*** .17***

8 Classroom engagement at fourth grade .12*** .22*** .23*** .31*** .52*** .17*** .42***

9 Teacherestudent relations at fourth grade �.03 .13*** .06* .13*** .12*** .24*** .18*** .39***

Mean 4.39 80.45 13.25 .53 4.01 4.23 2.63 4.06 4.03SD 3.56 16.80 3.26 .50 .71 .77 1.48 .71 .82

Note. *p < .05; ***p < .001.

I. Archambault et al. / Learning and Instruction 23 (2013) 1e9 5

by Hu and Bentler (1999). However, they were strong enough tosuggest that our model fit the data. This goodness of fit was furtherconfirmed by the RMSEA of .058 and the SRMR of .054. Overall, thismodel explained, 21.9 and 37.1% of classroom engagement at firstand fourth grade, and 8.8 and 11.2% of teacherestudent relationsduring the same time points, respectively.

There are next a few things to say about the contribution of eachcontrol variable above and beyond the others in our structuralmodel. As shown in Table 3, our model shows that girls were moreengaged and experienced better relations with teachers, accordingto first and fourth grade teachers. Moreover, cognitive skills inkindergarten were positively associated with subsequent engage-ment and closer relations with first grade teachers. Kindergartenskills did predict fourth grade teacherestudent relations, but notengagement. The fact that kindergarten skills did not predict fourthgrade engagement is probably due to the fact these skills sharecommon variance with second grade achievement which is a betterand more proximal predictor of fourth grade engagement andteacherestudent relations.

Results of our structural model further indicate that beyond thecontribution of our covariates (i.e., sex, maternal education, mathand language skills during kindergarten, and second gradeachievement), classroom engagement in first and fourth gradeshows considerable stability as do teacherestudent relations acrossthe same school years. Thus, students who were more engaged andhad better relations with their first grade teachers were also moreengaged and had better relations with their fourth grade teachers.However, when both variables are compared, the stability ofteacherestudent relationship was lower than engagement fromgrade to grade. Fig. 1 shows that classroom engagement andteacherestudent relations significantly covaried both in first gradeand fourth grade, which suggests that children who are perceivedas being more engaged by their teachers are also sharing morepositive and warmer relations with them. We finally testedmultiple group invariance of the final model for boys and girls.Results showed that relationships between variables in our modelwere similar for boys and girls (Dc2(19) ¼ 11.571, r > .05;DCFI ¼ .001).5 This last step indicates that the model is appropri-ately specified for both genders.

4. Discussion

Academic success, as an objective, requires more than cognitiveskills. Adequate task-focused classroom behavior has more often

5 According to Cheung and Rensvold (2002), invariance is reached when changeson the CFI are no larger than .01 and when the c2 differences between the models’fit is not significant. However, considering that the c2 is very sensitive to samplesize, we relied on the CFI change.

than not been neglected in the achievement literature. Moreimportantly, how these transact with the school environment hasreceived even less attention. With this in mind, the goal of thestudy was to comprehensively evaluate the early classroom expe-rience as a developmental milestone toward future academicsuccess. We evaluated the prospective developmental stability andrelations between classroom engagement and teacherestudentrelations from first through fourth grade. On the whole, our find-ings provide more insight about the reciprocal and transactionalassociations between early student participation and relations withteachers in the classroom and their importance in setting the stagefor short- and long-term educational progress.

First graders who participated in class, listened carefully andfollowed directions, and persisted and completed their workautonomously weremore likely to remainmore actively engaged infourth grade (hypothesis 1). This developmental continuity,observed in previous work (Hughes et al., 2008), is above andbeyond the influence of pre-existing and concurrent predictors ofachievement. Nevertheless, the stability of engagement patternsover the elementary school grades were relatively moderate. Thiscorroborates previous findings (Ladd & Dinella, 2009). Whilestability is positive for children and particularly for girls who enterschool cognitively skilled with well-adjusted socio-emotional atti-tudes, it is certainly more worrisome for children who are lessprepared and show early signs of disengagement. Especially giventhat engagement and motivational beliefs tend to decline ascurricular demands increase (Jacobs, Lanza, Osgood, Eccles, &Wigfield, 2002). First graders who show less task-orientation andpersistence in classroom life are likely to manifest increasinglydisengaged behavioral patterns, thus exacerbating their academicdifficulties from one grade to the next.

Students who spontaneously talk, trust, and share their feelingswith their first grade teacher were also more prone to develop suchpositive interactions with their teacher in fourth grade (hypothesis2). This description suits girls particularly well because they tend todevelop closer relationswith teachers over time. Eachyear, studentsare educated by homeroom teachers who present distinct person-alities and classroom expectations. This makes the results showingdevelopmental continuity of the quality of teacherestudent rela-tions over time more than interesting, especially because they areteacher-reported. It suggests that different teachers have thepropensity to invest the same students over time, perhaps becausesuch students present underlying characteristics that tend to facil-itate positive interactions with adults directing the classroomenvironment. As such, our results showed that within the sameschool year, children who follow directions, work neatly, and inde-pendently are more prone to share closer relations with teachers,while students who trust and spontaneously discuss with teachertend to be more behaviorally engaged (supporting hypothesis 3).The dynamic set of transactions that occur between teachers and

Fig. 1. Prospective association between classroom engagement and relationships with Teachers. Note. Dotted line represents none significant path.

I. Archambault et al. / Learning and Instruction 23 (2013) 1e96

students seems to positively nourish children’s behavioral andemotional adaptation in the classroom during the same school year.

Awarm teacherestudent relationship in first grade also forecastsfourth grade engagement over and above the contribution ofkindergarten skills and second grade achievement (hypothesis 4a).This finding is certainly among the strongest argument we can giveto highlight the significance of this specific (and malleable) adultechild bond for students. As reported previously, when studentsdevelop a warm and close connection with their teacher, they aremore inclined to subsequently respond positively to curricularexpectations and demands (Brophy, 1983; Furrer & Skinner, 2003).All the same, thenewfindinghere is that all children canbenefit fromthis positive relationship, independently of their cognitive skills orearly engagement behaviors. As many authors highlight the long-term academic impact of beginning school unprepared (Duncanet al., 2007), this finding suggest that positive interactions withteachers can act as a buffer against early childhood vulnerabilities.

Finally, our results showed that children’s engagement patternsdid not forecast Teacherechild relations in fourth grade (hypothesis4b). Thisfindingwas quite surprising. It suggests that although somechildren experience early difficulties to actively participate andrespond to teacher expectations and demands, this situation do notseem to affect their potential to subsequently develop a warm andpositive affiliationwithanother teacher in later years. EachTeachere

Table 3Prospective associations between covariates and latent variables in the structural equati

Variables First grade

Classroom engagement Teacherestuden

Maternal education �.02 �.05Sex (girl) 0.23 0.14PPVT (kindergarten) 0.18 0.16NKT (kindergarten) 0.29 0.17Achievement (2nd grade) e e

Note. Bold characters are significant at p < .05.

child relationship is unique. Fortunately, the nature of this affiliationseems to be quite independent of children previous classroomexperience. This represents good news for some children who, fordifferent reasons, begin their formal schooling with developmentalimmaturity limiting their ability to invest classroom activitiesproperly and strive to meet teacher expectations. Over time, theywill meet different teachers with whom they still can share theirexperience with, which will in turn favor their engagement.

Although this study established the relevance of a positiveteacherestudent relationship for boys as for girls, for intervention,trying to change the relationship a teacher shares with each studentwill be more challenging than attempting to alter student engage-ment. Teacherechild relations are influenced by the dynamic set ofinteractions that occur in the classroom context, some of which areattributable to student traits. Nevertheless, teachers are the adultsand professionals. Their professional demeanor can be modified, ifthey are driven enough to alter specific interactions as though theyare strategic changes in the learning environment. With an under-standing of the impact of their influence and strategic support,teachers can establish a sensitive and responsive environment thatpromotes active classroom participation and involvement bystudents. In line with attachment theory (Bowlby, 1980), encour-aging teachers to take time to exchange with their students, toexpress appreciation of these exchanges, and to be concerned about

on model.

Fourth grade

t relations Classroom engagement Teacherestudent relations

0.06 �.060.20 0.12.02 0.10

�.04 L0.08L0.17 L0.17

I. Archambault et al. / Learning and Instruction 23 (2013) 1e9 7

the way their students feel will provide a more positive and secureclassroomexperience that promotes learning-related behaviors andachievement (Howes, Phillipsen, & Peisner-Feinberg, 2000). As rolemodels, their behavioral responses to emotional arousal, noncom-pliance, and turbulence in the classroom can promote better self-regulation in children (Kochanska, Murray, & Harlan, 2000).

Beyond the promotion of Teacherechild relations, encouragingchildren’s active classroom participation and involvement may bea more productive focus of intervention for boys and girls whopresent early difficulties. Engagement skills may be related toacademic adjustment because they facilitate adaptation to theacademic and social demands of the early school environment(Birch & Ladd,1997). As an individual factor, classroom engagementis also more enduring and less dependent on the changes childrenexperience from one classroom to another. By bringing student tobe attentive, to participate, and work autonomously, teachers andschool practitioners promote their prospective achievement andengagement which in turn, contribute to strengthen the linksstudent will eventually share with other teachers. They are alsolikely to reflect underlying individual differences in self-regulationskills which play an important role in helping individuals adjust tonovel challenges. Given the notion of heterotypic continuity, earlyindividual differences in attention control and temperament arelikely to manifest themselves as greater task-orientation andclassroom engagement by the time children enter school (Putnam,Rothbart, & Gartstein, 2008; Rothbart & Bates, 2006). Furthermore,initial differences in cognitive control can help children develop theability to mobilize attention in the service of emotional regulation,which in turn can be beneficial for sustaining concentration instressful or frustrating situations. Better emotional regulation canalso facilitate positive social interactions (Rothbart & Bates, 2006).Unfortunately, the attention, working memory, and inhibitorycontrol skills that are important to promote child engagement inlearning are rarely taught in primary school (Diamond et al., 2007).Early intervention aimed at improving these skills, by scaffoldinginhibitory control, and facilitating more rewarding academicexperiences in order to stimulate child classroom active partici-pation are thus imperatively needed (Gormley & Gayer, 2008;Lillard & Else-Quest, 2006). An important investment would be totrain kindergarten and primary school teachers in cognitive controlinterventions for children with programs that promote a sensitiveand responsive classroom environment and reinforce guidedexploration, collaborative learning, planning, and problem solvingfor optimal development (Bierman, Nix, Greenberg, Domitrovich, &Blair, 2008; Diamond et al., 2007).

This study is not without limits. First, our measures of behav-ioral engagement and teacherestudent relations were solely basedon teacher reports at different grades which could have introduceda bias in our data, especially for measuring student engagement.Nevertheless, the finding that first grade teacher assessments wereassociated with fourth grade teacher assessments supports thevalidity of our measures. Second, this study only considers theprospective associations between teacherestudent relations andthe behavioral dimension of engagement. In order to better orientspecific practices aimed at promoting student participation in class,future work ought to consider affective and cognitive facets ofengagement. Finally, the measures are based on secondary analysesof already existing longitudinal data from a large scale birth cohort.This makes them less optimal than more precise instrumentationtypically used in studies specifically designed to address a partic-ular objective (i.e., primary data analysis).

Our findings present important conceptual implications for thestudy of classroom engagement. Behavioral engagement bearsinsights into children’s attentive and affective control. Furthermore,it represents an observable facet of classroom adjustment which can

be easily assessed by teachers and practitioners. Promotingengagement promises improvements in child academic adjustmentregardless of previous experience and individual strengths andweaknesses. Nonetheless, assessing the behavioral facet of engage-ment may not be sufficient. It needs to be complemented by infor-mation about children’s emotional and cognitive engagement. Bylooking at the relations students share with teachers, this studyhighlighted the relevance of examining the proximal influence ofchildren’s affective state on their achievement-related behaviors inthe classroom. In any case, it would be important for research toassess how teachers’ practices and relations with students moredirectly influence children long-term affective and cognitiveengagement in the classroom. That is, whether or not it influencestheir value and interest in different subjectmatter, or whether or notit contributes to the development of effective cognitive strategy tomonitor learning. Additionally, it would also be interesting to look athow other sets of interactions students have in their classroomenvironment, such as the relations they share with peers, couldpromote their active participation and cooperative work. Finally,because intervention programs targeting very young children areamong the most cost-effective (Heckman, 2006), our findingssuggest the benefits of investingfinancial resources so researchers, inconcert with practitioners, may develop and validate screening toolsto detect the early signs of less than optimal classroom engagement.

References

Ablard, K. E., & Lipschultz, R. E. (1998). Self-regulated learning in high-achievingstudents: relations to advanced reasoning, achievement goals, and gender.Journal of Educational Psychology, 90(1), 94e101. http://dx.doi.org/10.1037/0022-0663.90.1.94.

Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317e332. http://dx.doi.org/10.1007/BF02294359.

Alexander, K. L., Entwisle, D. R., & Dauber, S. L. (1993). First-grade classroombehavior: its short- and long-term consequences for school performance. ChildDevelopment, 64(3), 801e814, Retrieved from. http://www.jstor.org/.

Anderson, R. N., Greene, M. L., & Loewen, P. S. (1988). Relationships among teachers’and students’ thinking skills, sense of efficacy, and student achievement. TheAlberta Journal of Educational Research, 34(2), 148e165.

Appleton, J. J., Christenson, S. L., & Furlong, M. J. (2008). Student engagement withschool: critical conceptual and methodological issues of the construct.Psychology in the Schools, 45(5), 369e386. http://dx.doi.org/10.1002/pits.20303.

Appleton, J. J., Christenson, S. L., Kim, D., & Reschly, A. L. (2006). Measuring cognitiveand psychological engagement: validation of the student engagement instru-ment. Journal of School Psychology, 44(5), 427e445. http://dx.doi.org/10.1016/j.jsp.2006.04.002.

Archambault, I., Janosz, M., Fallu, J. S., & Pagani, L. S. (2009). Student engagementand its relationship with early high school dropout. Journal of Adolescence,32(3), 651e670. http://dx.doi.org/10.1016/j.adolescence.2008.06.007.

Archambault, I., Janosz, M., Morizot, J., & Pagani, L. S. (2009). Adolescent behavioral,affective, and cognitive engagement in school: relationship to dropout.Journal of School Health, 79, 402e409. http://dx.doi.org/10.1111/j.1746-1561.2009.00428.x.

Assor, A., Kaplan, H., Kanat-Maymon, Y., & Roth, G. (2005). Directly controllingteacher behaviors as predictors of poor motivation and engagement in girls andboys: the role of anger and anxiety. Learning and Instruction, 15(5), 397e413.http://dx.doi.org/10.1016/j.learninstruc.2005.07.008.

Battistich, V., Solomon, D., Watson, M., & Schaps, E. (1997). Caring schoolcommunities. Educational Psychologist, 32, 137e151. http://dx.doi.org/10.1207/s15326985ep3203_1.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness-of-fit in theanalysis of covariance structures. Psychological Bulletin, 88, 588e600. http://dx.doi.org/10.1037/0033-2909.88.3.588.

Bierman, K. L., Nix, R. L., Greenberg, M. T., Domitrovich, C. E., & Blair, C. (2008).Executive functions and school readiness intervention: impact, moderation, andmediation in the Head Start-REDI program. Development and Psychopathology,20, 821e843. http://dx.doi.org/10.1017/S0954579408000394.

Birch, S. H., & Ladd, G. W. (1997). The Teacherechild relationship and children’searly school adjustment. Journal of School Psychology, 35(1), 61e79. http://dx.doi.org/10.1016/S0022-4405(96)00029-5.

Blatchford, P., Bassett, P., & Brown, P. (2011). Examining the effect of class size onclassroom engagement and teacherepupil interaction: differences in relation topupil prior attainment and primary vs. secondary schools. Learning andInstruction, 21(6), 715e730. http://dx.doi.org/10.1016/j.learninstruc.2011.04.001.

Bowlby, J. (1980). Attachment and loss: Loss, sadness and depression, Vol. 3. New York:Basic Books.

I. Archambault et al. / Learning and Instruction 23 (2013) 1e98

Brophy, J. (1983). Conceptualizing students’ motivation. Educational Psychologist, 18,200e215. http://dx.doi.org/10.1080/00461528309529274.

Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes fortesting measurement invariance. Structural Equation Modeling, 9(2), 233e255.http://dx.doi.org/10.1207/S15328007SEM0902_5.

Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and relatedness: Amotivational analysis of self-system processes. Hillsdale, NJ, England: LawrenceErlbaum Associates, Inc.

Diamond, A., Barnett, W., Thomas, J., & Munro, S. (2007). Preschool programimproves cognitive control. Science, 318(5855), 1387e1388. http://dx.doi.org/10.1126/science.1151148.

Duncan, G. J., Dowsett, C. J., Claessens, A., Magnuson, K., Huston, A. C., Klebanov, P.,et al. (2007). School readiness and later achievement. Developmental Psychology,43(6), 1428e1446. http://dx.doi.org/10.1037/0012-1649.43.6.1428.

Dunn, L. M., Thériault-Whalen, C. M., & Dunn, L. M. (1993). Échelle de vocabulaire enimages Peabody. Adaptation française du Peabody Picture Vocabulary test-revised.Toronto: Psycan.

Eccles, J. S. (2004). Schools, academic motivation, and stage-environment fit. Hoboken,NJ: John Wiley & Sons Inc.

Eccles, J., Wigfield, A., Harold, R. D., & Blumenfeld, P. (1993). Age and genderdifferences in children’s self- and task perceptions during elementary school.Child Development, 64(3), 830e847, Retrieved from. http://www.jstor.org.

Finn, J. D. (1989). Withdrawing from school. Review of Educational Research, 59(2),117e142, Retrieved from. http://www.jstor.org.

Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement:potential of the concept, state of the evidence. Review of Educational Research,74(1), 59e109. http://dx.doi.org/10.3102/00346543074001059.

Furrer, C., & Skinner, E. (2003). Sense of relatedness as a factor in children’sacademic engagement and performance. Journal of Educational Psychology,95(1), 148e162. http://dx.doi.org/10.1037/0022-0663.95.1.148.

Goodenow, C. (1993). Classroom belonging among early adolescent students:relationships to motivation and achievement. Journal of Early Adolescence, 13(1),21e43. http://dx.doi.org/10.1177/0272431693013001002.

Gormley, W.,P.,D., & Gayer, T. (2008). Preschool programs can boost school readi-ness. Science, 320, 1723e1724. http://dx.doi.org/10.1126/science.1156019.

Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate dataanalysis (5th ed.). Upper Saddle River, NJ: Prentice Hall.

Hamre, B. K., & Pianta, R. C. (2001). Early Teacherechild relationships and thetrajectory of children’s school outcomes through eighth grade. Child Develop-ment, 72, 625e638, Retrieved from. http://www.jstor.org.

Heckman, J. J. (2006). Skill formationand the economics of investing in disadvantagedchildren. Science, 312, 1900e1902. http://dx.doi.org/10.1126/science.1128898.

Hijzen, D., Boekaerts, M., & Vedder, P. (2007). Exploring the links between students’engagement in cooperative learning, their goal preferences and appraisalsof instructional conditions in the classroom. Learning and Instruction, 17(6),673e687. http://dx.doi.org/10.1016/j.learninstruc.2007.09.020.

Howes, C., Phillipsen, L. C., & Peisner-Feinberg, E. (2000). The consistency ofperceived Teacherechild relationships between preschool and kindergarten.Journal of School Psychology, 38(2), 113e132. http://dx.doi.org/10.1016/S0022-4405(99)00044-8.

Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indices in covariance structureanalysis: conventional criteria versus new alternatives. Structural EquationModeling, 6, 1e55. http://dx.doi.org/10.1080/10705519909540118.

Hughes, J. N., Luo, W., Kwok, O.-M., & Loyd, L. K. (2008). Teacherestudentsupport, effortful engagement, and achievement: a 3-year longitudinal study.Journal of Educational Psychology, 100(1), 1e14. http://dx.doi.org/10.1037/0022-0663.100.1.1.

Jacobs, J. E., Lanza, S., Osgood, D., Eccles, J. S., & Wigfield, A. (2002). Changes inchildren’s self-competence and values: gender and domain differences acrossgrades one though twelve. Child Development, 73(2), 509e527, Retrieved from.http://www.jstor.org.

Johnson, W., McGue, M., & Iacono, W. G. (2006). Genetic and environmental influ-ences on academic achievement trajectories during adolescence. DevelopmentalPsychology, 42(3), 514e532. http://dx.doi.org/10.1037/0012-1649.42.3.514.

Kochanska, G., Murray, K. T., & Harlan, E. T. (2000). Effortful control in earlychildhood: continuity and change, antecedents, and implications for socialdevelopment. Developmental Psychology, 36, 220e232. http://dx.doi.org/10.1037/0012-1649.36.2.220.

Koriat, A. (2012). The relationships between monitoring, regulation and perfor-mance. Learning and Instruction, 22(4), 296e298. http://dx.doi.org/10.1016/j.learninstruc.2012.01.002.

Ladd, G. W., Buhs, E. S., & Seid, M. (2000). Children’s initial sentiments aboutkindergarten: is school liking an antecedent of early classroom participationand achievement? Merrill-Palmer Quarterly, 46, 255e279, Retrieved from.http://search.proquest.com.

Ladd, G. W., & Dinella, L. M. (2009). Continuity and change in early schoolengagement: predictive of children’s achievement trajectories from first toeighth grade? Journal of Educational Psychology, 101(1), 190e206. http://dx.doi.org/10.1037/a0013153.

Lillard, A. S., & Else-Quest, N. (2006). The early years: evaluating montessorieducation. Science, 313, 1893e1894. http://dx.doi.org/10.1126/science.1132362.

MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis anddetermination of sample size for covariance structure modeling. PsychologicalMethods, 1, 130e149. http://dx.doi.org/10.1037/1082-989X.1.2.130.

Marks, H. M. (2000). Student engagement in instructional activity: patterns in theelementary, middle, and high school years. American Educational ResearchJournal, 37, 153e184. http://dx.doi.org/10.3102/00028312037001153.

McClelland, M., Morrison, F., & Holmes, D. (2000). Children at risk for early academicproblems: the role of learning-related social skills. Early Childhood ResearchQuarterly, 15, 307e329. http://dx.doi.org/10.1016/S0885-2006(00)00069-7.

McDermott, P. A., Mordell, M., & Stoltzfus, J. C. (2001). The organization of studentperformance in American schools: discipline, motivation, verbal learning,nonverbal learning. Journal of Educational Psychology, 93(1), 65e76. http://dx.doi.org/10.1037/0022-0663.93.1.65.

McWilliam, R., Scarborough, A. A., & Kim, H. (2003). Adult interactions and childengagement. Early Education and Development, 14(1), 7e27. http://dx.doi.org/10.1207/s15566935eed1401_2.

Okamoto, Y., & Case, R. (1996). Exploring the microstructure of children’s centralconceptual structures in the domain of number. In R. Case, & Y. Okamoto(Eds.), The role of central conceptual structures in the development of children’sthought. Monographs of the society for research in child development, Vol. 60 (pp.27e58).

Pagani, L., Fitzpatrick, C. F., Archambault, I., & Janosz, M. (2010). School readinessand later achievement: a French Canadian replication and extension. Develop-ment Psychology, 46(5), 984e994. http://dx.doi.org/10.1037/a0018881.

Pagani, L., Tremblay, R. E., Vitaro, F., Boulerice, B., & McDuff, P. (2001). Effects ofgrade retention on academic performance and behavioral development.Development and Psychopathology, 13(2), 297e315, Retrieved from. http://journals.cambridge.org/action/displayJournal?jid¼DPP.

Pianta, R. C. (1999). Enhancing relationships between children and teachers. Wash-ington, DC: American Psychological Association.

Pianta, R. C., Steinberg, M. S., & Rollins, K. B. (1995). The first two years of school:teacherechild relationships and deflections in children’s classroom adjustment.Development and Psychopathology, 7(2), 295e312. http://dx.doi.org/10.1017/S0954579400006519.

Pintrich, P. R., & de Groot, E. V. (1990). Motivational and self-regulated learningcomponents of classroom academic performance. Journal of EducationalPsychology, 82(1), 33e40. http://dx.doi.org/10.1037/0022-0663.82.1.33.

Ponitz, C. C., Rimm-Kaufman, S. E., Grimm, K. J., & Curby, T. W. (2009). Kindergartenclassroom quality, behavioral engagement, and reading achievement. SchoolPsychology Review, 38(1), 102e120, Retrieved from. http://search.proquest.com.

Putnam, S., Rothbart,M.,&Gartstein,M. (2008).Homotypicandheterotypic continuityof fine-grained temperament during infancy, toddlerhood, and early childhood.Infant and Child Development, 17, 385e405. http://dx.doi.org/10.1002/icd.582.

Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., Jones, J., et al.(1997). Protecting adolescents from harm: findings from the national longitu-dinal study on adolescent health. Journal of the American Medical Association,278(10), 823e832. http://dx.doi.org/10.1001/jama.1997.03550100049038.

Rimm-Kaufman, S. E., Curby, T. W., Grimm, K. J., Nathanson, L., & Brock, L. L. (2009).The contribution of children’s self-regulation and classroom quality to chil-dren’s adaptive behaviors in the kindergarten classroom. DevelopmentalPsychology, 45(4), 958e972. http://dx.doi.org/10.1037/a0015861.

Rothbart, M., & Bates, J. (2006). Temperament. In N. Eisenberg, W. Damon, &L. M. Richard (Eds.), Handbook of child psychology (6th ed.). Social, emotional, andpersonality development, Vol. 3 (pp. 99e166). Hoboken, NJ US: John Wiley &Sons Inc.

Sameroff, A. J., & Fiese, B. H. (1990). Transactional regulation and early intervention.In S. J. Meisels, & J. P. Shonkoff (Eds.), Early intervention: A handbook of theory,practice, and analysis (pp. 119e149). New York: Cambridge University Press.

Schafer, J. L. (1999). NORM: Multiple imputation of incomplete multivariate data undera normal model, version 2. Software for Windows 95/98/NT.

Schwartz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6,461e464, Retrieved from. http://www.jstor.org.

Sclove, L. S. (1987). Application of model-selection criteria to some problems inmultivariate analysis. Psychometrika, 52, 333e343. http://dx.doi.org/10.1007/BF02294360.

Sharma, S., Mukherjee, S., Kumar, A., & Dillon, W. R. (2005). A simulation study toinvestigate the use of cutoff values for assessing model fit in covariancestructure models. Journal of Business Research, 58(7), 935e943. http://dx.doi.org/10.1016/j.jbusres.2003.10.007.

Simons-Morton, B., & Chen, R. (2009). Peer and parent influences on schoolengagement among early adolescents. Youth & Society, 41(1), 3e25. http://dx.doi.org/10.1177/0044118X09334861.

Sirin, S. R., & Rogers-Sirin, L. (2004). Exploring school engagement of middle-classAfrican American adolescents. Youth & Society, 35(3), 323e340. http://dx.doi.org/10.1177/0044118X03255006.

Skinner, E. A., & Belmont, M. J. (1993). Motivation in the classroom: reciprocaleffects of teacher behavior and student engagement across the school year.Journal of Educational Psychology, 85(4), 571e581. http://dx.doi.org/10.1037/0022-0663.85.4.571.

Skinner, E., Furrer, C., Marchand, G., & Kindermann, T. (2008). Engagement anddisaffection in the classroom: part of a larger motivational dynamic? Journal ofEducational Psychology, 100(4), 765e781. http://dx.doi.org/10.1037/a0012840.

Skinner, E. A., Kindermann, T. A., & Furrer, C. J. (2009). A motivational perspective onengagement and disaffection: conceptualization and assessment of children’sbehavioral and emotional participation in academic activities in the classroom.Educational and Psychological Measurement, 69(3), 493e525. http://dx.doi.org/10.1177/0013164408323233.

I. Archambault et al. / Learning and Instruction 23 (2013) 1e9 9

Skinner, E. A., Wellborn, J. G., & Connell, J. P. (1990). What it takes to do well in schooland whether i’ve got it: the role of perceived control in children’s engagementand school achievement. Journal of Educational Psychology, 82, 22e32. http://dx.doi.org/10.1037/0022-0663.82.1.22.

Valiente, C., Lemery-Chalfant, K., Swanson, J., & Reiser, M. (2008). Prediction ofchildren’s academic competence from their effortful control, relationships, andclassroom participation. Journal of Educational Psychology, 100(1), 67e77. http://dx.doi.org/10.1037/0022-0663.100.1.67.

Wentzel, K. R. (1999). Social-motivational processes and interpersonal relation-ships: implications for understanding motivation at school. Journal ofEducational Psychology, 91(1), 76e97. http://dx.doi.org/10.1037/0022-0663.91.1.76.

Wigfield, A., & Eccles, J. S. (2002). The development of competence beliefs,expectancies for success, and achievement values from childhood throughadolescence. In A. Wigfield, & J. S. Eccles (Eds.), Development of achievementmotivation (pp. 173e195). San Diego, CA: Academic Press.