Within-Day Variability in the Quality of Classroom Interactions during Third and Fifth Grade

Page 1

The University of Chicago Presshttp://www.jstor.org/stable/10.1086/660682 .

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to TheElementary School Journal.

http://www.jstor.org

Page 2

WITHIN-DAY VARIABILITY IN THE

QUALITY OF CLASSROOM

INTERACTIONS DURING THIRD AND

FIFTH GRADE

Implications for Children’s Experiences and ConductingClassroom Observations

Timothy W. Curby!"#$!" %&'#()(*+"$'*,-

Megan Stuhlman)(*+"$'*,- #. +*$!*(*&

Kevin Grimm)(*+"$'*,- #./&0*.#$(*&1 2&+*'

Andrew MashburnLia Chomat-MooneyJason DownerBridget HamreRobert C. Pianta)(*+"$'*,- #. +*$!*(*&

!"#$%!&$The quality of classroom interactions has typically beenstudied using aggregates of ratings over time. However,within-day ratings may contain important variability.This study investigated within-day variability using theNICHD Study of Early Childcare and Youth Develop-ment’s observational data during grades 3 and 5. The firstquestion examined the stability of classroom interactionquality within a day, and the second question examinedthe extent to which time of day, activity setting, numbersof adults and children, and instructional grouping pre-dicted the quality of classroom interactions within a day.Results indicated that classroom interactions were moststable within emotional support dimensions and leaststable within instructional support dimensions. Resultsalso indicated that the start of the day and transitionswere associated with lower-quality classroom interac-tions. Academic activities, more teachers, fewer stu-dents, and group instruction were associated with higherquality. Implications are discussed in terms of under-standing students’ experiences and conducting class-room observations.

$'( ()(*(+$!%, #&'--) .-/%+!) +#0)%" 334 , ()%5"$ 3© 2011 by The University of Chicago. All rights reserved. 0013-5984/2011/11201-0002 $10.00

Page 3

WI T H I N the classroom setting, there is theoretical and empirical evi-dence that classroom interactions—primarily between teachers andstudents—are the most direct conduits of children’s learning (Bron-fenbrenner & Morris, 2006; Creemers, 1994; Nye, Konstantopoulos, &

Hedges, 2004; Pianta, 1999). There are also substantial differences between class-rooms in the nature and quality of classroom interactions, and these differences arerelated to students’ academic learning gains and social development (Early et al.,2005; Hamre & Pianta, 2005). What has yet to be explored is how stable these class-room interactions are within a day and how specific environmental factors (time ofday, activity setting, numbers of adults and students, and instructional grouping)may be systematically related to within-day classroom interaction quality. Learningmore about factors that influence the quality of classroom interactions informs aricher understanding of the nature of students’ classroom experiences as well asprovides guidance for improving these settings.

History of Classroom Observation Research

A long line of research has indicated the utility of classroom observations (Gage &Needels, 1989). Throughout the 1970s and 1980s, the majority of classroom observa-tions used in educational research studies were implemented as part of the process-product research paradigm. Seeking to better comprehend the components of effec-tive teaching, this research consisted primarily of correlational studies examining theassociations between classroom process variables, like specific teacher behaviors, andstudent products or outcomes, namely, student achievement. For example, Borg(1979) and Good, Grouws, and Beckerman (1978) examined how the number ofpages in a curriculum presented to students by the teacher related to achievement;other work focused on the number of test items taught through lectures (Arehart,1979; Armento, 1977; Cooley & Leinhardt, 1980; Dunkin & Doenau, 1980; Nuthall &Church, 1973; Smith, 1979). Process-product researchers also measured teachers’time allocation to determine how it related to students’ opportunities to learn (Bro-phy & Evertson, 1976; Fisher et al., 1980; Stallings, 1975) and the extent to whichteachers’ classroom management behaviors were correlated with student achieve-ment (Brophy & Evertson, 1976; Coker, Medley, & Soar, 1980; Fisher et al., 1980;Good & Grouws, 1979; Stallings, 1975).

Our study focuses on classrooms in a way that flows out of this research yet issomewhat different. The observational tool used in our study does not employ fre-quency counts of certain behaviors or the amount of time spent in different activities(although these elements may be indirectly present). Rather, the quality of classroominteractions is judged on how effectively they provide students with the emotional,organizational, and instructional supports needed to maximize learning (Hamre &Pianta, 2007; Hamre, Pianta, Mashburn, & Downer, 2007).

Stability of Classroom Interactions

Most elementary classrooms will generally have the same teachers and students overthe course of the year, suggesting that there may be a high degree of similarity, orstability, in the quality of interactions taking place across a school year. Yet, at thesame time, classrooms are dynamic, complex social systems. Even within a day,

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! 37

Page 4

several factors may contribute to change in the quality of interactions taking place.For example, classroom interactions can take place among students, between teach-ers and individual students, and between teachers and groups of students. Teachersalso have multiple objectives when interacting with students in their classes. Theinherent complexity of the classroom as a social setting suggests that the teachers’objectives will be related to changing aspects of classroom interactions. Therefore,interactions in the classroom take place in what at first glance might be considered astable context but may actually demonstrate considerable variability over time as afunction of changing environmental factors.

The degree to which behaviors are stable—in this case teaching behaviors—ishighly influenced by how that behavior is defined. As evidenced by the aboveprocess-product research, teaching behaviors have been defined in various ways andhave therefore obtained varying results. For example, one way to view classrooms hasbeen to examine the frequencies of discrete teaching behaviors (e.g., the number ofquestions asked), the use of particular phrasing, or the use of instructional activities.According to these metrics, classrooms appear to be fairly unstable (e.g., Darling-Hammond, Wise, & Pease, 1983). However, a different view of teaching behaviors,employed by Meyer, Linn, and Hastings (1991), focused mostly on the amount oftime teachers devote to certain classroom activities. Using this definition, rank or-dering of these teacher behaviors over the course of one school year to the next andfrom the morning to the afternoon appeared to be relatively stable.

Some prior work has examined the stability in classroom interactions across dif-ferent days within a school year. Using data from the NICHD Study of Early ChildCare and Youth Development (SECCYD), correlations of scores across days in thequality of classroom interactions were quite high (Chomat-Mooney et al., 2008;NICHD ECCRN, 2005), suggesting overall consistency across days in teachers’ be-havior. Notably, these analyses aggregated several ratings made within a day. In otherwords, ratings were made for several times a day for each classroom, and thesewithin-day ratings were averaged. Thus, there still may be a great deal of within-dayvariability that, up to this point, has been masked by the chosen analytic method.

Studies that have examined associations between more global patterns of teachers’classroom behaviors and students’ outcomes suggest that there are stable between-teacher differences in teaching behaviors (Dolezal, Welsh, Pressley, & Vincent, 2003;Early et al., 2005; Hamre & Pianta, 2005; Matsumura, Patthey-Chavez, Valdes, &Garnier, 2002; Meyer, Wardop, Hastings, & Linn, 1993). In other words, if teachers’behaviors were erratic or random, empirical findings would likely show no relationbetween such behaviors and student outcomes (Meyer et al., 1991). Instead, findingsconsistently indicate that ratings of classroom interaction quality predict students’achievement gains (e.g., Early et al., 2005; Mashburn et al., 2008).

Potential Sources of Instability

Although lack of stability may in part represent random fluctuation stemming fromeither idiopathic sources within a teacher or measurement error, such variation mayalso represent the influence of systematic factors that underlie changes in a giventeacher’s behavior. Thus, we were interested in examining the classroom conditionsthat are likely to function as sources of systematic variation in teachers’ behaviors(see also Chomat-Mooney et al., 2008). We propose four potential areas that could

38 ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 5

contribute to variation in predictable, organized ways: time of day, activity setting,numbers of adults and students in the classroom, and size of instructional groupings.

Time of Day

With regard to time of day, it may be that certain times, such as the very beginningor end of the school day, may make it difficult to have high-quality instructionalinteractions as teachers and students are more oriented toward the physical andcognitive transitions taking place during these times. These times may also include ahigher proportion of interactions focused on managerial instructions and thus in-clude fewer opportunities for facilitating engagement or providing individualizedemotional support. Another perhaps underestimated factor in the dynamic nature ofinteractions is that physiological changes occur throughout the day in teachers andchildren. For example, children and teachers fatigue throughout the day (Finn &Pannozzo, 2004), which may lead to lower-quality classroom interactions.

Activity Setting

It is also possible that there may be changes in the quality of teachers’ interactionswith students when they are involved in different activities. For example, teachers’behaviors during mathematics lessons tend to be more structured and rigid thantheir behaviors during language arts lessons (Wiley, Good, & McCaslin, 2008). Itmay also be that when engaged in comprehension or problem-solving activities, thequality of interactions related to instructional support would be highest, and con-versely instructional support could reasonably be expected to be lowest during tran-sitions or the management of classroom business (e.g., taking attendance, adminis-tering spelling tests). Productivity, defined in the present work as managing routinesto provide maximal opportunities for learning, is also expected to be lower duringtransition times as opposed to during times devoted to formal instruction.

Numbers of Adults and Students in the Classroom

Child-staff ratios have been cited as a salient factor of children’s development inthe child-care literature (Howes, Phillips, & Whitebook, 1992; NICHD ECCRN,1996), but not all research has shown that class size, teacher-child ratios, or presenceof teacher’s assistants are consistently or substantially linked with the quality ofteachers’ interactive behaviors or children’s outcomes in early elementary school(NICHD ECCRN, 2005; Stuhlman & Pianta, 2009). Given the salience of this ques-tion for policy-related decisions, we also include the numbers of teachers and stu-dents in the classroom as a possible factor that may systematically alter the frequencyand quality of classroom interactions.

Size of Instructional Groupings

Finally, the salience of group sizes in predicting quality of classroom interactionsis also an area of interest. There has been much discussion within the educationalcommunity around the issue of groupings. Irrespective of the intent of the group,empirical study of group size often finds improved student outcomes when moretime is spent in small groups as opposed to whole groups or individual work (Lou et

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! 3;

Page 6

al., 1996), although there is some support for the positive effects of large- and whole-group instruction (Marliave & Filby, 1985; Rosenshine, 1976). We examined this areato determine whether interactive quality was associated with the proportion of timestudents spent in different-sized groups. Notably, evaluation of the impact of homo-geneity or heterogeneity of ability level within groupings on quality of classroominteractions is outside the scope of the present work (e.g., Slavin, 1987).

Our Study

We examined classroom interactions according to a developing body of literaturethat documents three broad domains of classroom interactions that influence chil-dren’s learning (Hamre et al., 2007): emotional support, classroom organization,and instructional support. In terms of emotional support, teachers’ ability to createsocially supportive environments and to be emotionally and academically engagedand responsive are highly salient factors that promote students’ feelings of comfortand engagement in the classroom and thus contribute to achievement gains as well(Brophy & Good, 1986; Eccles & Gootman, 2002; Pianta, 1999; Weinstein, 2002).With regard to classroom organization, when teachers organize classrooms to max-imize the time that students are able to be engaged in interesting tasks, organize theirown behavior so that they can adeptly facilitate student engagement, and minimizethe degree to which managing behavior detracts from valuable learning time, stu-dents are able to regulate their own behavior more successfully and learn more fromthe schooling experience (Brophy & Evertson, 1976; Burnett, 2003; Cameron, Con-nor, & Morrison, 2005; McCaslin et al., 2006). For instructional support, previouswork has indicated the potential role of higher-order cognitive activities (e.g., activ-ities involving analysis, reasoning, or conceptual understanding) that are relevantand meaningful for students (Bransford, Brown, & Cocking, 1999; Cameron et al.,2005; Curby et al., 2009; Hamre & Pianta, 2005).

Within-day variability is the focus of our study for three important reasons. First,focusing on within-day variability allows us to align our methods with our theory ofdevelopment. According to the bioecological model of development (Bronfen-brenner & Morris, 2006), the reciprocal interactions between students and theirenvironment (i.e., teacher) would serve as proximal processes—the mechanism ofdevelopment. Proximal processes occur in microtime (Bronfenbrenner & Morris,2006), that is, the moment-to-moment experiences of children. Therefore, we ex-amined proximal processes (i.e., classroom interactions) on the time scale in whichthey occurred. Second, students do not experience average ratings; students experi-ence all the ups and downs of a classroom within a day. These ups and downs (i.e., thenonstable portion of classroom interactions) speak to the consistency of their expe-rience. We examined whether students’ days on average are marked by relative con-sistency or relative variability. Third, classroom interaction ratings are actually madewithin a day. In other words, ratings are made over the course of several observationcycles during an observation day. If the ratings themselves are systematically chang-ing as other changes happen in the classroom (e.g., time of day or doing math insteadof reading), then those interested in ratings of teachers need to know of the potentialinfluence of these factors in order to take them into account when planning obser-vations or analyzing data.

4: ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 7

Research Questions

We examined how stable the quality of classroom interactions are and how theymay systematically change within a day. To accomplish this, two broad researchquestions were addressed: (1) How stable are domains and dimensions of classroominteraction quality within a school day? (2) Does the quality of interactions within aday vary as a function of environmental factors (time of day, activity setting, num-bers of adults and students present, and grouping) in the classroom?

Method

Participants

Data were obtained from classrooms in which one of the students was a partici-pant in the National Institute of Child Health and Human Development’s Study ofEarly Child Care and Youth Development (NICHD SECCYD). The NICHD SEC-CYD was a prospective, longitudinal study of environments that influence children’sdevelopment; participants were recruited at the time of the child’s birth from 10 sitesin the United States. (Methods for the entire NICHD SECCYD are described in detailat http://secc.rti.org, and the interested reader is referred to this resource for a morecomplete review of study methods.) Briefly, the NICHD SECCYD is a longitudinalstudy of key developmental contexts for children from birth through age 15. Familieswere recruited through hospital visits to mothers shortly after the birth of a child in1991. Of the initial pool of eligible mothers, 1,364 completed a home interview whenthe child was 1 month old and were enrolled in the study. In third grade, 827 of thesechildren were still enrolled in the study. In fifth grade, 791 children participated.

Measures and Procedure

Classroom observation system. As part of the NICHD SECCYD data collection,trained research assistants observed third- and fifth-grade classrooms. Ratings weremade of the quality of classroom interactions and classroom characteristics. TheClassroom Observation System for Third Grade and the Classroom ObservationSystem for Fifth Grade (COS-3 and COS-5, respectively) are upward extensions ofthe Classroom Observation System for First Grade (COS-1), described in detail inprevious publications (e.g., NICHD ECCRN, 2002). The COS has repeatedly dem-onstrated predictive validity with regard to students’ academic and social function-ing (Hamre & Pianta, 2005; NICHD ECCRN, 2002, 2003, 2005, 2006).

Observations using the COS-3 and COS-5 occurred between November (thirdgrade) or January (fifth grade) and June and were structured in the same way. Eachclassroom was observed for eight 25-minute observation cycles totaling almost 4hours of observation. During each of the eight cycles, data collectors spent the first 5minutes and the last 10 minutes dedicated to observing the quality of classroominteractions the study children experienced. Ratings were made on a 7-point scaleranging from “uncharacteristic” to “highly characteristic” on several dimensions.These dimensions were then combined into three domains of classroom interaction:Emotional Support, Classroom Organization, and Instructional Support (Hamre etal., 2007). Emotional Support consists of five dimensions: positive climate, negativeclimate (reversed), teacher sensitivity, teacher detachment (reversed), and teacher

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! 43

Page 8

over-control (reversed). Classroom Organization consists of two dimensions: chaosand productivity. Instructional Support consists only of richness of instruction.(Quality of feedback was available for the COS-5 only, but we did not analyze thosedata as there would have been no comparison with third grade.) Brief narrativedescriptions and descriptive statistics for each dimension are provided in Table 1.

In addition, data collectors spent an additional 10 minutes per cycle recordingdiscrete, interval-level codes on 30-second observe/30-second coding intervals, for atotal of 80 minutes of discrete behavior ratings. These behavioral codes includemeasures of activity setting (i.e., class is involved in literacy–word level activities,literacy– comprehension, math computation, problem solving, science, social stud-ies, computers/technology, transitions/management). A summed score was com-puted (ranging from 0 to 10) for each cycle representing the number of minutes thateach activity setting occurred within that cycle. This coding also included the num-ber of adults and children present and the instructional grouping (teacher interactswith whole class, subgroup, or individuals) during each cycle.

Training, reliability, and validity. Identical approaches to training and reliabilitytesting were used for the third- and fifth-grade data collections with the COS-3 andthe COS-5, respectively. Observers from all 10 sites trained on practice videotapesusing a standardized manual that provided extensive descriptions of codes and an-chor points, prior to attending a centralized training workshop. After the centraltraining workshop, coders conducted pilot observations and trained on two morevideotaped cases. Following training, all observers had to pass a videotaped reliability

Table 1. Descriptive Statistics for the Quality of Interactions Children Experience in Third andFifth Grade

Description

ThirdGrade Fifth Grade

M SD M SD

Emotional support:Positive climate Interactions are characterized by emotionally

warm and supportive gestures andverbalization

5.05 1.03 5.11 .98

Negative climate Interactions are characterized by negativity,sarcasm, anger, irritation, and peer conflicts

1.18 .64 1.31 .75

Teacher sensitivity The teacher accurately detects and interpretsstudents’ cues, provides comfort, andresponds to distress

4.90 1.26 4.83 1.17

Detachment The teacher is disconnected from students’activities and emotional and academic needs

1.67 1.05 1.84 1.13

Classroom over-control

Interactions stifle students and learningopportunities are diminished

2.58 1.69 2.13 1.32

Classroom organization:Chaos Teacher ineffectively manages students so that

disruption and chaos predominate1.51 0.96 1.40 .85

Productivity Teachers manage instructional time so thatstudents have maximal opportunities to learn

4.87 1.41 4.87 1.40

Instructional support:Richness of instruction Teachers use a variety of strategies to promote

student thinking and understanding ofmaterial at a deeper and more complex level

2.09 1.26 3.59 1.62

Note.—Observed range for all scales is 1–7.

44 ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 9

test covering six cycles of ratings. Criteria for passing were at least an 80% match(within 1 scale point) with a master coder on the classroom interaction coding and a60% match on the time-sampled codes. All coders met or exceeded these levels on areliability test before being certified to conduct observations in the field.

Beyond their agreement, the consistency of the ratings was also checked. In fifthgrade, the average correlation between the coders’ ratings and the master codes onthe videotaped test was r ! .79 (p " .05). For the time-sampled codes, the correlationwith master codes scores was r ! .85 (p " .05). These consistency estimates werehighly similar to those of third grade. Observers each also conducted a minimum oftwo paired visits scheduled randomly during the data collection window for thepurpose of estimating live reliability. Correlations between observers exceeded r !.84 for all but one of the 38 time-sampled codes, with the lower estimate due to theinfrequency of the observed behavior. Average live reliability across all global ratings,estimated using correlations, was r ! .71.

Analysis and Results

How Stable Are the Classroom Interaction Experiences of Students within aSchool Day?

To address the question of stability in the quality of interactions that childrenexperience in a school day, autoregressive models were fit to the multiple cycles ofdata using a structural equation modeling framework. These autoregressive modelshave the advantage of providing a single stability estimate for each dimension (asopposed to correlations that can only make pairwise comparisons). Each autoregres-sive model predicts the latter time point with the previous time point (see Fig. 1).Because we actually had eight cycles of observation, we constrained the association(b) between any two adjacent time points to be the same while also removing anestimate of measurement error (e), also constrained to be equal over time. Eachmodel incorporated all dimensions in each domain. For example, the domain Class-room Organization is indicated by the dimensions Productivity and Chaos. There-fore, one Classroom Organization model was run (for each grade) that included anautoregressive component for Productivity and an autoregressive component forChaos.

The stability estimate for each dimension represents stability from one cycle to thenext (or approximately 25 minutes later). Stability estimates range from 0 to 1, withhigher values indicating greater stability. Table 2 presents results from autoregressivemodels showing the stability in observable dimensions over the first 4 hours of aschool day. Dimensions are grouped according to domain, and model fit values are atthe domain level. Overall, all models provided adequate model fit (RMSEA ! .10).

Figure 1. Autoregressive model used to make stability estimates for each dimension. Note: errorvariances (e) and unstandardized stability estimates (b) are constrained to be equal over time.

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! 4<

Page 10

Unstandardized estimates ranged from .38 to .68 (standardized estimates range from.35 to .68), indicating moderate to moderately high levels of stability. Some of therange was linked to grade level; autoregressive coefficients consistently were slightlyhigher in third grade than in fifth grade.

Does the Quality of Interactions within a Day Vary as a Function ofEnvironmental Factors in the Classroom?

Next, we tested the degree to which the quality of interactions throughout the dayvaried systematically as a function of (1) the time of day the observation took place(i.e., observation cycle), (2) the activity settings near when the observation tookplace, (3) the number of adults and children present during the observation, and (4)the instructional grouping of students in the classroom. Because multiple observa-tions were made corresponding to the classroom that the study child was in at anygiven observation cycle, these data are no longer independent. Therefore, traditionaltechniques (e.g., regression) that do not take into account the nesting of these data(multiple observations nested within each study child) would be inappropriate.Therefore, multilevel models were run to address this research. Multilevel modelscan account for the fact that multiple observations (n ! 8) of classroom interactionswere nested within each study child. When conducting a multilevel analysis, the firststep is to determine the extent to which the nesting is a source of variance in theoutcomes. These unconditional models only account for the nesting and do notinclude any predictors. After establishing this unconditional model, models were runseparately with the various classes of predictors included.

Time of day. After having established the unconditional models, time was enteredas a predictor. Cycles 2 through 8 were added as dummy-coded predictors in level 1(within-child) models to estimate the extent to which these quality ratings differedfrom the quality rating made at the beginning of the school day (Cycle 1). Results ofmultilevel models are presented in Table 3. The intercepts can be interpreted as theaverage score for that dimension during Cycle 1 (which is the comparison group).The coefficients for Cycles 2– 8 are unstandardized betas, which indicate the magni-

Table 2. Autoregressions Assessing Within-Day Stability inClassroom Interaction Dimensions for Third and Fifth Grade

Third Grade Fifth Grade

b " b "

Emotional Support:Positive climate .51 .45–.51 .46 .42–.46Negative climate .59 .51–.59 .42 .39–.42Teacher sensitivity .53 .47–.53 .45 .41–.45Detachment .48 .44–.48 .38 .35–.38Over-control .68 .56–.68 .61 .52–.61

Model fit (RMSEA) .08 .06Classroom Organization:

Productivity .49 .44–.49 .42 .38–.42Classroom chaos .53 .47–.53 .47 .42–.47

Model fit (RMSEA) .10 .08Instructional Support:

Richness of instruction .45 .41–.45 .48 .43–.48Model fit (RMSEA) .07 .10

4= ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 11

Tab

le3.

Ass

ocia

tion

sbe

twee

nT

ime

ofD

ayan

dth

eQ

ualit

yof

Cla

ssro

omIn

tera

ctio

ns

Posi

tive

Clim

ate

Neg

ativ

eC

limat

eT

each

erSe

nsit

ivit

yD

etac

hmen

tO

ver-

Con

trol

Cha

osPr

oduc

tivi

tyR

ichn

ess

ofIn

stru

ctio

n

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Inte

rcep

t5.

115.

101.

121.

254.

944.

801.

711.

822.

462.

021.

471.

364.

724.

311.

702.

97C

ycle

2(n

!86

6,81

0)-.

01-.

01.0

1.0

4.0

6.0

5-.

04-.

02.1

4**

.17*

**-.

06-.

06.3

4**

*.6

3***

.42*

**.5

8**

*C

ycle

3(n

!86

9,81

5)-.

02-.

01.0

5*.0

9**

-.01

.02

.03

.03

.14

**.1

5***

.02

-.01

.32*

**.6

4**

*.5

5***

.77*

**C

ycle

4(n

!86

7,81

4)-.

02-.

07.0

7**

.11*

**-.

03.0

2.0

6.0

8.0

9.0

9.1

0**

.02

.20

***

.60

***

.49

***

.70

***

Cyc

le5

(n!

868,

814)

-.03

.03

.07*

*.0

7*-.

07.0

4.0

9*

.04

.14

**.0

7.0

9*

.01

.21*

**.5

8**

*.4

1***

.76

***

Cyc

le6

(n!

868,

812)

-.03

.02

.06

**.0

9**

-.09

*.0

6.0

4.0

3.0

8.0

9*

.10

**.0

8*

.25*

**.5

2***

.45*

**.6

2***

Cyc

le7

(n!

856,

815)

-.08

*-.

05.1

1***

.13*

**-.

10*

-.10

*.0

4.0

3.0

7.1

3**

.13*

**.1

0**

.20

***

.44

***

.43*

**.6

4**

*C

ycle

8(n

!83

8,80

5)-.

01.0

3.0

9**

*.1

4**

*-.

09*

-.01

.12*

*.0

0.1

2*.1

1*.1

2***

.15*

**.2

1***

.49

***

.39

***

.74

***

Not

e.—

Ref

eren

cegr

oup

isC

ycle

1.*p

!.0

5.**

p!

.01.

***p

!.0

01.

Page 12

Tab

le4.

Ass

ocia

tion

sbe

twee

nT

ype

ofC

lass

room

Act

ivit

ies

and

the

Qua

lity

ofC

lass

room

Inte

ract

ions

Posi

tive

Clim

ate

Neg

ativ

eC

limat

eT

each

erSe

nsit

ivit

yD

etac

hmen

tO

ver-

Con

trol

Cha

osPr

oduc

tivi

tyR

ichn

ess

ofIn

stru

ctio

n

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Inte

rcep

t5.

265.

151.

171.

325.

034.

721.

741.

962.

412.

061.

491.

494.

764.

471.

873.

42W

ord-

leve

lact

ivit

ies

-.03

***

-.01

.00

.00

-.02

**.0

1*.0

0-.

01*

.04

***

.01*

.00

-.01

**.0

4**

*.0

6**

*-.

02*

-.01

Com

preh

ensi

on-.

02**

*-.

01.0

0-.

01*

-.01

.01*

.00

.00

.01

.01*

-.01

-.02

***

.04

***

.06

***

.05*

**.0

3***

Mat

hco

mpu

tati

on-.

02**

-.02

***

.00

.01

-.01

.01*

.00

-.01

*.0

3***

.02*

**.0

0-.

01**

.04

***

.07*

**-.

01-.

01Pr

oble

mso

lvin

g-.

01.0

1*.0

0-.

01.0

0.0

3***

-.01

-.04

***

.01

-.01

.01

-.01

.06

***

.09

***

.15*

**.1

3***

Scie

nce

.01*

.01*

.00

.01*

.00

.01

-.01

-.01

*-.

01.0

0.0

1*.0

0.0

3***

.04

***

.08

***

.08

***

Soci

alst

udie

s.0

0.0

0.0

0.0

0.0

0.0

1*-.

01-.

02**

*.0

0.0

0.0

1-.

01.0

2*.0

4**

*.0

6**

*.0

5***

Enri

chm

ent

.01

.03*

**.0

0.0

1-.

03**

.02

.02*

-.03

***

-.01

.01

.02*

.01

.00

.03*

*.0

1.0

3**

Com

pute

rs/t

ech

-.02

*-.

01.0

0.0

0-.

03**

*-.

02.0

2*.0

3**

-.03

*-.

02*

.00

-.02

**.0

0.0

1-.

01.0

1T

rans

itio

n/m

anag

emen

t-.

02**

*-.

02.0

0.0

1**

-.02

**-.

02**

.00

.00

.00

-.01

*.0

2***

.01*

-.09

***

-.14

***

-.03

***

-.09

***

*p!

.05.

**p

!.0

1.**

*p!

.001

.

4>

Page 13

Tab

le5.

Ass

ocia

tion

sbe

twee

nN

umbe

rof

Adu

ltsan

dC

hild

ren

Pres

enta

ndth

eQ

ualit

yof

Cla

ssro

omIn

tera

ctio

ns

Posi

tive

Clim

ate

Neg

ativ

eC

limat

eT

each

erSe

nsit

ivit

yD

etac

hmen

tO

ver-

Con

trol

Cha

osPr

oduc

tivi

tyR

ichn

ess

ofIn

stru

ctio

n

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Gra

de3

Gra

de5

Inte

rcep

t4.

784.

831.

261.

414.

824.

921.

992.

112.

421.

931.

561.

314.

954.

551.

462.

59N

o.of

adul

ts.1

0**

*.0

6**

-.03

*-.

03.1

0**

*.0

1-.

05*

-.04

-.09

**-.

06*

.01

.02

.08

*.0

5.1

0**

*.1

3***

No.

ofch

ildre

n.0

0.0

0.0

0*

.00

-.02

***

-.01

***

.01*

**.0

0.0

1***

.01*

**.0

1**

.01*

**-.

01**

.00

.00

.01*

**Pr

op.s

mal

l/la

rge

grou

p.3

4**

*.5

6**

*.0

3.0

8*

.20

***

.33*

**-.

04-.

13*

-.58

***

-.48

***

.17*

**.2

0**

*.1

7**

.36

***

.85*

**1.

26**

*Pr

op.w

hole

grou

p.1

9**

*.2

7***

.00

.06

**.2

2***

.34

***

-.28

***

-.65

***

.18

***

.06

*-.

04-.

01.1

1**

.32*

**.6

4**

*.9

9**

*

Not

e.—

Prop

.sm

all/

larg

egr

oup

and

prop

.who

legr

oup

are

calc

ulat

edas

prop

orti

ons

and

dono

thav

ere

fere

nce

grou

ps.

*p!

.05.

**p

!.0

1.**

*p!

.001

.

47

Page 14

tude and direction of the difference between that cycle and Cycle 1. Results indicatefew systematic changes in Positive Climate, Teacher Sensitivity, Detachment, andOver-Control. However, other dimensions (Negative Climate, Chaos, Productivity,and Richness of Instruction) were often significantly different during Cycles 2– 8than during Cycle 1. Negative Climate and Chaos tended to be higher during Cycles2– 8 than during Cycle 1. Furthermore, Negative Climate and Chaos ratings looked asif they were increasing incrementally across the observation segments. Productivityand Richness of Instruction were significantly higher in Cycles 2– 8 than during thefirst cycle of the day, suggesting that instructional activities are much less likely tooccur at the very beginning of the school day compared to later in the school day.

To follow up on the observation that Negative Climate and Chaos seemed to bebuilding throughout the day, we conducted formal tests to see whether there was infact a linear trend. To conduct these tests, we constructed growth models with thecontinuous cycle number included as a predictor of Negative Climate and Chaosscores (as opposed to dummy-coded cycles) and looked for a significant linear slope.A significant cycle number coefficient would reveal a linear increase in NegativeClimate and Chaos as time passes within a day. For Negative Climate, the estimatedintercepts (Cycle 1) are 1.23 and 1.25, and slopes are .014 (p " .0001) and .016 (p ".0001) for third and fifth grade, respectively. For Chaos, the estimated intercepts(Cycle 1) are 1.44 and 1.30, and slopes are .024 (p " .0001) and 0.024 (p " .0001) forthird and fifth grade, respectively. Thus, results indicate that there were significant,albeit small, positive linear increases in Negative Climate and Chaos across the ob-servation day.

Activity setting. In this model, the number of minutes spent in each activitysetting (e.g., comprehension, math computation) was entered at level 1 instead of thedummy-coded cycle numbers. The associations between the minutes spent in differ-ent activity settings and the dimension ratings are presented in Table 4, in which afew clear patterns emerge across grade levels between the type of activity setting andthe dimension ratings. One way to examine the table is to look across the table to seehow particular activity settings (e.g., word-level activities) are associated with vari-ous dimensions. For example, more time (i.e., higher percentages of time) spent intransition/management tended to be associated with lower levels of quality acrossmany dimensions in both third and fifth grade. In contrast, engaging in moreproblem-solving activities was somewhat predictive of higher ratings. The patternsbecome clearer, however, when looking down Table 4 to see how particular dimen-sions of quality (e.g., Chaos) are related to various activity settings. In general, di-mensions of Classroom Organization and Instructional Support tended to be pro-portionately more sensitive to activity settings than dimensions of EmotionalSupport. For the five Emotional Support dimensions (e.g., Positive Climate), therewere 40 significant associations. For the three other dimensions (two ClassroomOrganization dimensions and one Instructional Support dimension), there were 35significant associations. Furthermore, these latter associations tended to be stronger.However, most of the associations were quite small (even when significant), and thusthere was weak support for systematic links between activity settings and quality ofclassroom experiences.

Number of adults and students present and grouping. Using the same analyticapproach as above, the number of adults, number of children, and the proportion ofthe activity that was whole or small/large group were entered as level 1 predictors.

48 ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 15

These variables were entered together because group size is related to the number ofadults and children present in the room. Furthermore, time the teacher spent work-ing with individuals was not entered as a grouping type because it was highly redun-dant with the information provided about the groups. Table 5 shows that in generalas the number of adults in the classroom increased, ratings of the quality increasedacross all measured dimensions except Chaos. Likewise, as the number of studentsincreased, lower quality was generally evident across the outcomes. With regard toinstructional grouping, more time spent in small/large groups or in whole groupswas generally related to better quality.

Discussion

Two important findings became evident from our study: (1) Both stability and vari-ability were evident in students’ experiences of classroom interactions, and (2)changing characteristics of classrooms were associated with varying levels of class-room interaction quality.

Stability and Variability

Overall, results of observations in over 1,600 third- and fifth-grade classroomsindicate that students experienced much stability in the quality of classroom inter-actions, but there is also substantial variability depending on the domain, at leastacross the first 4 hours of the school day (Curby, Grimm, & Pianta, 2010). Therelatively lower stability estimates indicate that children’s experience of classroomswas least consistent within a day when it came to instruction. This can be thought ofin terms of teachers changing objectives for the day—that is, teachers may confinehigh-quality instructional interactions, as defined by the COS, to instructional activ-ities. Conversely, aspects of classroom organization and emotional support mayremain more constant throughout the school day because they reflect more about acharacteristic response of the teacher than just an overt response to what is happen-ing in the classroom. In other words, teachers’ emotional support can be expressed inalmost any classroom interaction, whereas high-quality instruction may take placeprimarily in the context of instructional activities. Thus, our results are more similarto studies that examined the amount of time teachers devote to certain classroomactivities (e.g., Meyer et al., 1991) than studies that examined the frequency of certainbehaviors (e.g., Darling-Hammond et al., 1983).

Findings also suggest that fifth graders experience less stability in their classroominteractions than third graders. This may be a reflection of the fact that students aremore likely to change teachers throughout the day in fifth grade, so their experienceof less consistent quality in classroom interactions may be due to the fact that they aremore likely to be interacting with multiple teachers over the course of a day than arethird graders. Regardless of whether the lower levels of stability in interactions aredue to inconsistencies between teachers or instability within a given teacher, theimplication is that fifth graders experience more fluctuation in the quality of theirinteractions with teachers than do third graders. The extent to which this variabilitymay matter for students’ learning is unknown, but there are indications that, at leastin prekindergarten, within-day variability in emotional support hinders academicand social development (Curby, Brock, & Hamre, 2010).

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! 4;

Page 16

Characteristics of Classrooms

Time of day. Our analyses revealed several factors systematically associated withvariability in students’ experiences. During the first 25 minutes of the school day,classes were less productive and used fewer instructional strategies to promote stu-dent thinking and understanding in comparison to later in the day. This suggests thatclassrooms have an initial settling-in period at the very beginning of the school day.Students were often just arriving in the classroom during the first observation cycle.Teachers, therefore, may be unwilling to offer much in the way of instruction whensome students are arriving or getting organized. As students arrive they may still betired because of inadequate sleep, be interested in socializing with friends, or wish toconverse with the teacher. These student behaviors may disrupt planned instruc-tional activities or simply place many competing demands on teachers’ attention.The confluence of these reasons may make the first 25 minutes of the day the mostdifficult for teachers to offer high-quality interactions to their classes.

Although the first 25 minutes showed the lowest levels of instruction and produc-tivity, it is noteworthy that as the school day progressed, classroom interactionstended to be more controlling and more negative. The finding that negativity andchaos increased as time progressed is consistent with previous findings that teachers’use of nonverbal criticism increased over the course of the school day (Cheffers &Lombardo, 1979). When coupled with the information about classroom instructionincreasing, an image of the class begins to emerge. As teachers begin to place moreinstructional demands on students, they are also using techniques to demand stu-dent adherence to the activities, and, at least in some cases, negative interactionsoccur between and among students and teachers. Student and teacher fatigue mayplay a role in this as well, particularly as the day wears on (Finn & Pannozzo, 2004).After hours of classroom activities, students and teachers alike are likely to havedepleted at least some of their attentional and emotional reserves, making lower-quality classroom interactions more likely as a result. This is a potentially seriousissue, the mechanism of which should be explored in future research so that appro-priate interventions can be designed and implemented.

Activity setting. We also examined whether instability in students’ experiences ofclassroom interaction quality varied as a function of the proportion of time studentsspent in different activity settings (i.e., science, math computation, transitions, etc.).The type of activity setting demonstrated relatively few systematic relations to ratingsof interactive quality, but one of the most consistent findings was for transitions:greater time spent in transitions was linked with lower-quality classroom interac-tions across dimensions. Thus, classrooms that went through many activities quickly(necessitating more transitions) or those classrooms that tended to take a lot of timeto transition between activities typically had lower levels of interactional quality(Brophy & Evertson, 1976). Given that transitions account for nearly 17% of class-room time (NICHD ECCRN, 2003), they may limit the quality of classroom inter-actions.

Number of adults and children. The numbers of adults and students presentduring an observation and the size of the grouping were shown to be predictors ofclassroom interaction quality. Children in classrooms with fewer children, greaternumbers of adults, and more time spent in whole or group instruction also experi-enced higher-quality emotional, organizational, and instructional supports than did

<: ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 17

students who spent more time with larger numbers of peers and smaller numbers ofteachers and less time in whole-group or subgroup instruction. These findings arelargely consistent with other work, suggesting the value of more teachers and fewerstudents in a classroom (e.g., Howes et al., 1992; NICHD ECCRN, 1996). When moreteachers are present, they are able to better meet the needs of children in the class-room by being more available to respond to individual students.

Instructional grouping. The present study also indicated the value of group in-struction (Lou et al., 1996; Rosenshine, 1976). Small/large- and even whole-groupinstruction was found to be a positive predictor of interaction quality. In under-standing why group work was positively associated with interaction quality, it ishelpful to think about the alternative. If students are not in some sort of group, theyare working individually or working individually with a teacher. It is likely that whenteachers are having students do individual work, the teachers are frequently usingthat time to do other activities, such as taking attendance or setting up for the nextactivity. Conversely, when teachers are working with a group (of any size), they tendto be engaged in the task at hand and offering the supports that students need as bestthey can. In this way, students (on average) are likely to experience higher-qualityinteractions while in a group, even though the feedback they experience when work-ing individually with a teacher may be more appropriate for an individual child(while the other students do not have the benefit of those interactions). Furthermore,it is worth noting that these results do not take into account the type of grouping (i.e.,heterogeneous or homogeneous) and that the effects of group instruction interactwith student characteristics (e.g., ability, socioeconomic status) in producingachievement gains (Marliave & Filby, 1985; Rosenshine, 1976; Slavin, 1987). Thus, thisfinding may speak more to the challenge of being a teacher than the value of groupinstruction.

Implications for Efforts to Support or Improve Teaching Practice

Drawing from this understanding of students’ experiences of stability and insta-bility in the quality of their interactions with their teachers, we can derive severalimplications for best practices in classroom teaching. For example, the provision ofconsistent (and high) levels of instructional support throughout the day stands out asa goal. This study suggests that teachers need better training to see all classroommoments, whether the focus is on management or science, as opportunities forlearning that can either be missed or taken advantage of. Students’ interest in andunderstanding of curricular concepts can be expanded during what might otherwisebe downtimes such as during transitions or when conducting classroom business ifteachers intentionally integrate informative discussions during these times and focuson making them meaningful for students. Furthermore, this study suggests thatteachers need training on how to effectively transition between activities during aclass period to minimize the time spent in transition. As a result, we remind thereader of the good deal of training material available on means for creating smoother,more efficient transitions (e.g., Simonson, Fairbanks, Briesch, Myers, & Sugai, 2008).

This study also suggests that teachers and administrators need to be made awarethat, at least on average, levels of negativity and chaos may increase over the course ofthe day. Awareness is the first step to addressing the problem, and teachers may beable to reverse this trend simply by being aware of it. This can be accomplished by

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! <3

Page 18

focusing on proactive and productive strategies for providing clear, positive, andattainable behavioral expectations, evaluating whether those expectations are beingmet, reinforcing desired behaviors, effectively redirecting unwanted behaviors be-fore they escalate, and managing stress.

Finally, knowing that levels of quality appear to increase when there are moreteachers and fewer students is consistent with other literature (e.g., Howes et al., 1992;NICHD ECCRN, 1996). But why is more time spent in groups related to higher-quality classroom interactions? Perhaps teachers are most engaged in learning tasksduring group work, whereas during alternate classroom arrangements the teacherattempts to do other necessary, non-learning-related activities (e.g., organizing foranother activity). More research is needed to understand why quality tends to behighest during group work. Once a better understanding is reached about why thishappens, strategies at a policy level (i.e., in regulating numbers and students andteachers) or at the classroom level (i.e., in planning for more group work) can bedevised to put this knowledge to work in classrooms.

Implications for Conducting Classroom Observations of Interaction Quality

Obtaining information about the stability and predictors of change in the qualityof classroom interactions has important implications for those who conduct class-room observations (for basic descriptive purposes, professional development effi-cacy evaluation, intervention research, accountability purposes, etc.). Implicationsfor both planning observations and interpreting results are discussed in turn below.

Planning for classroom observations. Given the relatively high degree of stabilityin ratings, our study suggests that relatively few observations within a day will pro-vide good estimates of within-day variability. Indeed, this is evidenced in the obser-vation measure that grew out of COS-3 and COS-5. Although it is not exactly thesame, the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, & Hamre,2008) grew directly out of the COS measures.1 The CLASS recommends that class-rooms be observed for at least four cycles, with the subsequent ratings being aggre-gated. Our study does not speak to the number of days that should be observed to geta good estimate of between-day variability. Although multiple studies have foundconsistency between different days of observation (NICHD ECCRN, 2005), westrongly recommend that any personnel decisions be based on multiple days ofobservation. In addition, for observers particularly interested in dimensions of in-structional support, our results indicate that there is less stability in instructionalsupport and that, therefore, more cycles of observations need to be completed inorder to get a more reliable estimate. Likewise, results indicate that a greater numberof observation cycles should be completed in older versus younger grades due to thelower levels of stability in fifth grade.

Regardless of whether observations are intended to describe classroom condi-tions, associate classroom processes with student outcomes, or evaluate teachers’response to intervention, researchers and school personnel need to be aware of anysystematic fluctuations in ratings of classroom interactions that can be expected tooccur over the school day. Our results suggest that those planning to observe class-room interaction quality may choose to begin all observations at least 25 minutesafter the official start of the school day, because ratings made during the first 25minutes of the day appear to be systematically lower than subsequent observations. If

<4 ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 19

the initial interactions of the school day are of particular interest, it would be wise tomake sure observations uniformly include this time for all classrooms, as startingsome observations earlier and some later may make comparisons between class-rooms less valid.

In terms of activities and size/groupings, researchers and school personnel facedifficult choices. Some teachers may spend more time than others in a given instruc-tional context (e.g., problem solving or transitions). If observers wish to observeteachers’ typical practice, observations should be conducted irrespective of the in-structional context. However, if two teachers spend roughly the same amount of timein transitions and more formal academic activities, but one happens to be observedduring a transition and the other happens to be observed during formal instruction,differences in ratings of classroom interaction quality may not reflect true differencesin students’ experiences of those two classrooms. Similar implications hold for num-ber of adults, number of students, and groupings. Therefore, we recommend as a firststep that observations include documentation of the start time of the observation,the activity settings occurring during the observation, numbers of teachers and stu-dents, and grouping type to allow these factors to be accounted for statistically inanalyses or systematically analyzed as they relate to quality of classroom interactions.Beyond that, researchers and school personnel should decide whether they want tostandardize these aspects of their observations (e.g., observe all classrooms duringsmall-group math lessons) or assume that the environmental factors that happen tobe observed represent valid snapshots of students’ typical experiences and thereforeshould not be standardized. Observing and collecting data on multiple occasionsmay be helpful in resolving these issues, as would observing a block of time thatincludes both planned lessons and other activities such as transitions or individualwork.

Evaluating results of classroom observations. If, in the planning stages, factorssuch as time of day, activity, and social setting are held constant or recorded asvariables that can be entered into analyses, the variance associated with theseenvironmental factors can be separated from variance that represents true, stablecharacteristics of students’ typical experiences. This partitioning of variance maybe particularly important when determining which changes in ratings of class-room quality are true indications of changes in practice and which may be func-tions of fluctuation in ratings expected with changing temporal or environmen-tal factors. For example, if a teacher is observed preintervention at the start of theschool day during transition activities with a large number of students and thenis observed postintervention 1 hour into the school day during a social studieslesson with a smaller group of students, differences in the quality of instructionalinteractions may not reflect changes due to the intervention but may instead bea function of differing temporal and environmental conditions. Therefore, ifthere is a systematic difference between when control and intervention class-rooms are observed, a lack of positive change in quality may not reflect a lack oftreatment effect as much as the conditions surrounding the observation. Beingintentional about scheduling observations and record keeping around the con-ditions during observations will assist users of observational tools in makingaccurate interpretations based on their obtained data.

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! <<

Page 20

Limitations

There are two limitations to the present study that should be noted. First, in termsof generalizability, the classrooms under consideration were limited to the third andfifth grades; early elementary or middle school classrooms may not necessarily followidentical patterns. However, the fact that on many counts third- and fifth-gradeclassrooms follow similar patterns mitigates this concern to a degree, at least across afairly narrow age range.

A second limitation relates to the unit of analysis in the stability analyses—thestudent’s experience as opposed to the behavior of a specific teacher. Many times astudent had only one teacher, so these two levels of analysis were the same, but inother (unknown) cases, students were interacting with different teachers during themorning of observation. Thus, findings represent the stability of the students’ expe-riences and not necessarily the stability of behavior of individual teachers. One of thedifferences between third- and fifth-grade findings was that fifth-grade interactionpatterns appeared to be less stable than they were in third grade. This may representa trend that as students move up in grades, they are more likely to have greaternumbers of teachers and are therefore less likely to experience consistency in theirexperiences in school over time. Clearly, study of a particular teacher should focus onthat teacher’s interactions with students and not the behaviors of other teachers.However, if a goal is to understand the nature of students’ school experiences, un-derstanding the implications of having different teachers over the course of a morn-ing is relevant and informative.

Conclusions

Within the context of the complex social and instructional dynamics of the class-room, we sought to understand the extent to which students experience stability inthe quality of their interactions with teachers and to examine potential factors thatcould be systematically associated with instability in interactive quality. We foundthat, in general, students experience moderate stability in the quality of interactionsthey have with their teachers, particularly in terms of the emotional support teachersprovide. However, in examining sources of instability in students’ experiences, anumber of factors demonstrated associations with higher levels of interactive quality,most notably engaging in more planned learning time (less time spent on transitionsand management), teaching in small/large- or whole-group formats, and havingfewer students and more teachers. These findings have implications for understand-ing students’ classroom experiences, provide practical information about how tomost effectively support teachers in optimizing interaction quality, and offer guid-ance to those collecting and interpreting observationally based data on quality ofclassroom interactions, either for research or professional development purposes.

Notes

The authors would like to thank several granting agencies for their generous support. We wouldlike to thank the William T. Grant Foundation and the Institute of Education Sciences, U.S. De-partment of Education, through grant R305A060021 to the University of Virginia, for their supportof this research. The first author would also like to thank the Institute of Education Sciences, U.S.Department of Education, for its predoctoral fellowship support through the University of Virginia

<= ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 21

(R305B040049). Additionally, we are thankful for the National Institute of Child Health and Hu-man Development for their funding of the Study of Early Child Care and Youth Development. Theopinions expressed are those of the authors and do not represent the views of the U.S. Departmentof Education or any of the granting agencies.

1. In the current published version of the CLASS (Pianta, La Paro, & Hamre, 2008), PositiveClimate, Negative Climate, Teacher Sensitivity, and Productivity are all basically the same as theyappeared in the COS. Detachment was not incorporated into the CLASS. Classroom Over-Controlin the COS morphed into Regard for Student Perspectives in the CLASS—the main distinctionbeing that Over-Control is a negative indicator (reversed in analyses) and Regard for StudentPerspectives is a positive indicator. Chaos in the COS was broadened and reversed into the positiveindicator Behavior Management in the CLASS. The CLASS also added the dimension of Instruc-tional Learning Formats, which captures the degree to which teachers actively engage studentsthrough their efforts and materials. Richness of Instruction was narrowed to focus on ConceptDevelopment (rote vs. conceptual learning). The later version of the COS included Quality ofFeedback, as does the CLASS. The CLASS also grew to incorporate Language Modeling as a mea-sure of the degree to which teachers promote conversation in the class and elaborate on students’language with more advanced language.

References

Arehart, J. (1979). Student opportunity to learn related to student achievement of objectives in aprobability unit. Journal of Educational Research, 72, 253–269.

Armento, B. J. (1977). Teacher behaviors related to student achievement on a social science concepttest. Journal of Teacher Education, 28, 46 –52.

Borg, W. R. (1979). Teacher coverage of academic context and pupil achievement. Journal of Edu-cational Psychology, 71, 635– 645.

Bransford, J. D., Brown, A. L., & Cocking, R. R. (Eds.). (1999). How people learn: Brain, mind,experience, and school. Washington, DC: National Academy Press.

Bronfenbrenner, U., & Morris, P. (2006). The bioecological model of human development. InR. M. Lerner (Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human develop-ment (6th ed., pp. 793– 828). Hoboken, NJ: Wiley.

Brophy, J., & Evertson, C. (1976). Learning from teaching: A developmental perspective. Boston: Allyn& Bacon.

Brophy, J., & Good, T. (1986). Teacher behavior and student achievement. In M. Wittrock (Ed.),Handbook of research on teaching (pp. 328 –375). New York: Macmillan.

Burnett, P. C. (2003). The impact of teacher feedback on student self-talk and self-concept inreading and mathematics. Journal of Classroom Interaction, 38, 11–16.

Cameron, C. E., Connor, C. M., & Morrison, F. J. (2005). Effects of variation in teacher organiza-tion on classroom functioning. Journal of School Psychology, 43, 61– 85.

Cheffers, J. T. F., & Lombardo, B. J.(1979). The observation and description of teaching behavior andinteraction of selected physical education teachers. Paper presented at the National AAHPERConvention, New Orleans.

Chomat-Mooney, L. I., Pianta, R. C., Hamre, B. K., Mashburn, A. J., Luckner, A. E., Grimm, K. J.,. . . Downer, J. T. (2008). A practical guide for conducting classroom observations: A summary ofissues and evidence for researchers. New York: William T. Grant Foundation.

Coker, H., Medley, D. M., & Soar, R. S. (1980). How valid are expert opinions about effectiveteaching? Phi Delta Kappan, 62, 131–134.

Cooley, W. W., & Leinhardt, G. (1980). The instructional dimensions study. Educational Evaluationand Policy Analysis, 2, 7–25.

Creemers, B. P. M. (1994). The effective classroom. London: Cassell.Curby, T. W., Brock, L., & Hamre, B. (2010). Teachers’ emotional consistency predicts children’s

achievement gains and social skills. Manuscript submitted for publication.Curby, T. W., Grimm, K. J., & Pianta, R. C. (2010). Stability and change in early childhood class-

room interactions during the first two hours of a day. Early Childhood Research Quarterly, 25,373–384.

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! <?

Page 22

Curby, T. W., LoCasale-Crouch, J., Konold, T. R., Pianta, R., Howes, C., Burchinal, M., . . . Bar-barin, O. (2009). The relations of observed pre-K classroom quality profiles to children’s aca-demic achievement and social competence. Early Education and Development, 20, 346 –372.

Darling-Hammond, L., Wise, A. E., & Pease, S. R. (1983). Teacher evaluation in the organizationalcontext: A review of the literature. Review of educational research, 53, 285–328.

Dolezal, S. E., Welsh, L. M., Pressley, M., & Vincent, M. M. (2003). How nine third-grade teachersmotivate student academic engagement. Elementary School Journal, 103, 239 –269.

Dunkin, M., & Doenau, S. (1980). A replication study of unique and joint contributions to variancein student achievement. Journal of Educational Psychology, 72, 394 – 403.

Early, D., Barbarin, O., Bryant, D., Burchinal, M., Chang, F., Clifford, R., . . . Weaver, W. (2005).Pre-K in eleven states: NCEDL’s Multi-State Study of Pre-K and Study of State-Wide Early Edu-cation Programs (SWEEP). Retrieved from http://www.fpg.unc.edu/~ncedl/pdfs/SWEEP_MS-_summary_final.pdf

Eccles, J., & Gootman, J. A. (2002). Community programs to promote youth development. Washing-ton, DC: National Academies Press.

Finn, J. D., & Pannozzo, G. M. (2004). Classroom organization and student behavior in kinder-garten. Journal of Educational Research, 98, 79 –92.

Fisher, C. W., Berliner, D. C., Filby, N. N., Marliave, R., Cahen, L. S., & Dishaw, M. M. (1980).Teaching behaviors, academic learning time, and student achievement: An overview. In C.Denham & A. Lieberman (Eds.), Time to learn (pp. 7–32). Washington, DC: Department ofEducation and the National Institute of Education.

Gage, N. L., & Needels, M. C. (1989). Process-product research on teaching: A review of criticisms.Elementary School Journal, 89, 253–300.

Good, T. L., & Grouws, D. A. (1979). The Missouri Mathematics Effectiveness Project. Journal ofEducational Psychology, 71, 355–362.

Good, T. L., Grouws, D. A., & Beckerman, T. (1978). Curriculum pacing: Some empirical data inmathematics. Journal of Curriculum Studies, 10, 75– 81.

Hamre, B. K., & Pianta, R. C. (2005). Can instructional and emotional support in the first grademake a difference for children at risk of school failure? Child Development, 76, 949 –967.

Hamre, B. K., & Pianta, R. C. (2007). Learning opportunities in preschool and early elementaryclassrooms. In R. C. Pianta, M. J. Cox, & K. Snow (Eds.), School readiness and the transition tokindergarten (pp. 49 – 84). Baltimore: Brookes.

Hamre, B. K., Pianta, R. C., Mashburn, A., & Downer, J. (2007). Building and validating a theoreticalmodel of classroom effects in over 4,000 early childhood and elementary classrooms. Retrieved fromhttp://www.fcd-us.org/resources/resources_show.htm?doc_id!507559

Howes, C., Phillips, D. A., & Whitebook, M. (1992). Thresholds of quality: Implications for thesocial development of children in center based child care. Child Development, 63, 449 – 460.

Lou, Y., Abrimi, P. C., Spence, J. C., Poulsen, C., Chambers, B., & d’Apollon, S. (1996). Within-classgroupings: A meta-analysis. Review of Educational Research, 66, 423– 458.

Marliave, R., & Filby, N. K. (1985). Success rate: A measure of task appropriateness. In C. W. Fisher& D. C. Berliner (Eds.), Perspectives on instructional time (pp. 217–235). New York: Longman.

Mashburn, A. J., Pianta, R. C., Hamre, B. K., Downer, J. T., Barbarin, O., Bryant, D., . . . Howes, C.(2008). Measures of classroom quality in pre-kindergarten and children’s development of ac-ademic, language and social skills. Child Development, 79, 732–749.

Matsumura, L. C., Patthey-Chavez, G. G., Valdes, R., & Garnier, H. (2002). Teacher feedback,writing assignment quality, and third-grade students’ revision in higher and lower achievingschools. Elementary School Journal, 103, 3–25.

McCaslin, M., Good, T., Nichols, S., Zhang, J., Wiley, C., Bozack, A. R., . . . Cuizon-Garcia, R.(2006). Comprehensive school reform: An observational study of teaching in grades 3 through5. Elementary School Journal, 106, 313–332.

Meyer, L. A., Linn, R. L., & Hastings, C. N. (1991). Teacher stability from morning to afternoon andfrom year to year. American Educational Research Journal, 28, 825– 847.

Meyer, L. A., Wardrop, J. L., Hastings, C. N., & Linn, R. L. (1993). Effects of ability and settings onkindergarteners’ reading performance. Journal of Educational Research, 86, 142–160.

<> ! $'( ()(*(+$!%, #&'--) .-/%+!) '"9,"%5"$ 4:33

Page 23

National Institute of Child Health and Human Development Early Child Care Research Network.(1996). Characteristics of infant childcare: Factors contributing to positive care giving. EarlyChildhood Research Quarterly, 11, 269 –306.

NICHD Early Child Care Research Network. (2002). The relation of global first grade classroomenvironment to structural classroom features, teacher, and student behaviors. ElementarySchool Journal, 102, 367–387.

NICHD Early Child Care Research Network. (2003). Social functioning in first grade: Associationswith earlier home and child care predictors and with current classroom experiences. ChildDevelopment, 74, 1639 –1662.

NICHD Early Child Care Research Network. (2005). A day in third grade: A large-scale study ofclassroom quality, teacher, and student behaviors. Elementary School Journal, 105, 305–323.

NICHD Early Child Care Research Network. (2006). The relations of classroom contexts in theearly elementary years to children’s classroom and social behavior. In A. C. Huston & M. N.Ripke (Eds.), Contexts of development in middle childhood: Bridges to adolescence and adulthood(pp. 217–236). New York: Cambridge University Press.

Nuthall, G., & Church, J. (1973). Experimental studies of teaching behavior. In G. Chanan (Ed.),Towards a science of teaching (pp. 9 –25). Slough, Bucks, England: National Foundation forEducational Research in England and Wales.

Nye, B., Konstantopoulos, S., & Hedges, L.V. (2004). How large are teacher effects? EducationalEvaluation and Policy Analysis, 26, 237–257.

Pianta, R. C. (1999). Enhancing relationships between children and teachers. Washington, DC: Amer-ican Psychological Association.

Pianta, R. C., La Paro, K. M., & Hamre, B. K. (2008). Classroom assessment scoring system—K–3.Baltimore: Brookes.

Rosenshine, B. (1976). Recent research on teaching behaviors and student achievement. Journal ofTeacher Education, 27, 61– 64.

Simonson, B., Fairbanks, S., Briesch, A., Myers, D., & Sugai, G (2008). Evidence-based practices inclassroom management: Considerations for research to practice. Education and the Treatmentof Children, 31, 351–380.

Slavin, R. (1987). Ability grouping and student achievement in elementary schools: A best evidencesynthesis. Review of Educational Research, 57, 293–336.

Smith, M. S. (1979). Strategies, language transfer and the simulation of the second language learn-er’s mental operations. Language Learning, 29, 345–361.

Stallings, J. (1975). Implementation and child effects of teaching practices in Follow-Through class-rooms. Monographs of the Society for Research in Child Development, 40, 1–133.

Stuhlman, M. W., & Pianta, R. C. (2009). Profiles of educational quality in first grade. ElementarySchool Journal, 109, 323–342.

Weinstein, R. (2002). Reaching higher: The power of expectations in schooling. Cambridge, MA:Harvard University Press.

Wiley, C. H., Good, T. L., & McCaslin, M. (2008). Comprehensive school reform instructionalpractices throughout a school year: The role of subject matter, grade level, and time of year.Teachers College Record, 110, 2361–2388.

,6*$2- &(2 .*.,6-!$&2" +&$*&5*0*,- ! <7