Theoretical Framework

dcfif3

chien_et_al-2010-child_development.pdf

Home >Computer Science homework help >Theoretical Framework

Children’s Classroom Engagement and School Readiness Gains

in Prekindergarten

Nina C. Chien and Carollee Howes University of California at Los Angeles

Margaret Burchinal University of California at Irvine

Robert C. Pianta University of Virginia

Sharon Ritchie, Donna M. Bryant, Richard M. Clifford, Diane M. Early

and Oscar A. Barbarin University of North Carolina at Chapel Hill

Child engagement in prekindergarten classrooms was examined using 2,751 children (mean age = 4.62) enrolled in public prekindergarten programs that were part of the Multi-State Study of Pre-Kindergarten and the State-Wide Early Education Programs Study. Latent class analysis was used to classify children into 4 profiles of classroom engagement: free play, individual instruction, group instruction, and scaffolded learning. Free play children exhibited smaller gains across the prekindergarten year on indicators of language ⁄ literacy and mathematics compared to other children. Individual instruction children made greater gains than other children on the Woodcock Johnson Applied Problems. Poor children in the individual instruction profile fared better than nonpoor children in that profile; in all other snapshot profiles, poor children fared worse than nonpoor children.

There is growing concern about children’s lack of readiness for school (Bowman, Donovan, & Burns, 2000). Evidence suggests that children’s school readiness, particularly for children from disadvan- taged backgrounds, is enhanced in prekindergarten programs during the year before kindergarten (e.g., Magnuson, Meyers, Ruhm, & Waldfogel, 2004). About three fourths of the states now offer such programs (Barnett, Hustedt, Friedman, Boyd, & Ainsworth, 2007). Identification of program charac- teristics related to improving skills is important given the huge investment of tax dollars and the importance of having vulnerable children enter

school ready to learn. The current study describes patterns of children’s engagement in prekindergar- ten classrooms. Specifically, we used person-cen- tered analyses to group children into profiles (i.e., groups) of classroom engagement that reflected the dominant literatures on early childhood education. Then, we explored whether profile membership was linked to gains in school readiness during the prekindergarten year. Additionally, we examined whether some profiles were particularly beneficial for poor children.

Classroom Quality: Children’s Classroom Engagement Instead of Classroom Environment

Higher quality prekindergarten programs are associated with more positive child outcomes (e.g., Burchinal et al., 2000). The literature on child-care environmental quality is primarily based on both constructivist theory, in which the adults’ role is to provide children with rich materials that promote child-initiated exploration (Ginsburg & Opper, 1988), and sociocultural theory, in which the adults’

The NCEDL Multi-State and SWEEP Study of Pre-Kindergar- ten were conducted by a team of researchers, including Oscar A. Barbarin, Donna M. Bryant, Margaret Burchinal, Richard M. Clif- ford, Diane M. Early, Carollee Howes, and Robert C. Pianta. This study is supported under the Educational Research and Devel- opment Center Program, PR ⁄ Award R307A60004, as adminis- tered by the Institute of Education Sciences, U.S. Department of Education. However, the contents do not necessarily represent the positions or policies of the U.S. Department of Education, and endorsement by the federal government should not be assumed. NCEDL is grateful for the help of the many children, parents, teachers, administrators, and field staff who part of this study.

Correspondence concerning this article should be addressed to Nina C. Chien, University of California, San Diego, 9500 Gilman Drive, #0927, La Jolla, CA 92093-0927. Electronic mail may be sent to [email protected].

Child Development, September/October 2010, Volume 81, Number 5, Pages 1534–1549

role is to provide frequent sensitive and responsive interactions with children (Vygotsky, 1962). The assessment of quality is typically at the classroom level and involves global assessments across all activities. Many measures focus on the teacher, although some widely used measures examine the relationship between an individual child and teacher.

Examining children’s classroom engagement can provide additional information that is not available in the environmental quality assessments. The importance of focusing on child-level experiences is stressed in Howes’s theoretical model for children’s child-care experiences (Howes, 2000), where chil- dren’s experiences—play activities, peer play, and relationships—are embedded within the context of the classroom. That is, although measures of child- care environmental quality describe the context within which child engagement occurs, a direct examination of child engagement at the child level is needed to understand exactly what children are doing—what activities are occupying children’s time—within those contexts. Although children’s engagement and the classroom social context cer- tainly influence each other (e.g., Howes & Smith, 1995), they remain distinct constructs.

Another reason to look beyond environmental quality to classroom engagement is that common measures of classroom environmental quality might not adequately capture dimensions of the classroom that are linked to child outcomes. A meta-analysis of 20 studies of the relation between measures of child-care quality and child outcomes showed that the relation is weak (r = .12; Burchinal et al., 2008). The types of activities and level of interactions with the caregiver are also known to be important in pre- dicting preschool children’s learning in child-care settings (Howes et al., 2008).

A Person-Centered Approach

Child engagement has thus far been examined using regression, which examines each child engagement activity individually (e.g., Howes et al., 2008). Another method is a person-centered approach, which would consider the entire spec- trum of activity engagement of each child to place children into profiles, or subgroups. As such, chil- dren within the same profile would exhibit more similar patterns of classroom engagement than children in different profiles. A person-centered approach is more holistic than regression because it considers the entire constellation of child engage- ment rather than one child engagement at a time.

One type of person-centered analysis, cluster analysis, was used in an earlier study that clustered children based on their classroom engagement (Tonyan & Howes, 2003). Latent class analysis (LCA) is another person-centered approach that offers several advantages over cluster analysis. One advantage is that LCA provides model fit statistics that allow assessment of the model fit to the data, and the appropriateness of the number of profiles specified. In addition, because LCA is model based, the same results can theoretically be replicated with an independent sample (Muthén & Muthén, 2000). LCA would be an ideal method for identifying profiles of children based on their patterns of class- room engagement.

Patterns of Classroom Engagement

Based on dominant models of early childhood education (e.g., the constructivist or sociocultural models), three profiles might be expected to emerge from a LCA of children’s classroom engagement: a profile that emphasizes free-choice play and explo- ration, a profile that emphasizes teacher instruction, and a profile that emphasizes teacher scaffolding.

One model of early childhood education encour- ages child-directed exploration in activities that the child chooses, positing that many developmental competencies are acquired only through play (John- son, Christie, Yawkey, & Wardle, 1987). For exam- ple, sociodramatic play develops language and problem-solving skills; constructive play (e.g., blocks) helps children learn about symmetry and practice making mental plans. These skills, accord- ing to this model, are not readily learned via tea- cher instruction. Based on this model, we expect to see a profile of children that engages in primarily free-choice play activities.

A second model of early childhood education posits that children learn most from teacher instruc- tional support, defined as large amounts of literacy instruction, high-quality teacher feedback, and tea- cher-led discussions that elicit cognitive skills (Hamre & Pianta, 2005). One study reported that more class time spent on direct and explicit instruc- tion involving teacher feedback was linked to higher levels of student achievement (Meyer, Wardrop, Hastings, & Linn, 1993; Pianta, La Paro, Payne, Cox, & Bradley, 2002). Another study found that increased instructional time in each of four preacademic activities—letter-sound, oral language, being read to, and mathematics—was associated with higher teacher ratings of children’s language and literacy skills (Howes et al., 2008). Based on

Classroom Engagement and School Readiness Gains 1535

this model, we expect to see a profile of children that spends a lot of time in teacher-directed instruc- tional activities.

A third model of learning emphasizes the impor- tance of ‘‘scaffolding’’ by a more knowledgeable other that enables children to think and complete tasks at a higher level than if they were unassisted (Wood, Bruner, & Ross, 1976). An important ele- ment of quality early childhood education is tea- cher scaffolding of children’s learning (Smith, 1996), including scaffolding to develop young chil- dren’s literacy (Henderson, Many, Wellborn, & Ward, 2002). Because whether scaffolding occurs during free play or teacher instruction is less important, we might expect a third profile of children who are not distinguishable from other children by their engagement in free play or instructional time, but rather in their receipt of teacher scaffolding.

Different models of classroom engagement may be more beneficial for different types of child out- comes. Children in classrooms that give children opportunities for free play and exploration might have more developed language and advanced mathematics and spatial skills (Johnson et al., 1987), according to the literature on free-choice play. Children in classrooms utilizing the instruc- tional model might be expected to make greater gains in basic academic skills such as knowing numbers and letters and how to write their names, because these are skills more readily taught via tea- cher instruction and less likely learned through free play. Children in classrooms utilizing scaffolding might have more developed literacy skills (Hender- son et al., 2002) or problem-solving skills (Rogoff, 1990). Including a variety of child outcomes makes it possible to assess whether certain models of early childhood education are best suited for particular domains of learning.

Child Engagement and Poverty Status

Children’s classroom engagement may also be linked to their poverty status. One study found, via teacher interviews, notable differences in the reported beliefs of preschool teachers of lower versus middle socioeconomic status (SES) children (Lee & Ginsburg, 2007). Teachers of lower SES chil- dren reported explicitly focusing on developing children’s literacy and mathematics skills through direct instruction to prepare children for kindergar- ten. Teachers of higher SES children, on the other hand, reported developing skills more indirectly by allowing children to engage in the literacy or math-

ematics activities of their choice in a classroom environment filled with rich learning materials. We therefore expect children’s classroom engagement to differ by poverty status, consistent with teacher beliefs reported in this study.

If preschool teachers of low- versus middle- income children have different beliefs about what constitutes an ideal preschool education, is there any evidence that their beliefs are accurate? That is, do low-income children indeed benefit more from a focus on mathematics and literacy skill develop- ment, whereas middle-income children benefit more from the freedom to choose activities? One study found that children who were at risk of school failure benefited more from classrooms with more direct literacy instruction, evaluative feed- back, and instructional conversations during tea- cher-led discussions, compared to children who were not at risk (Hamre & Pianta, 2005). Therefore, poor children might be expected to benefit more from direct instruction of academic skills than from child-initiated free-play activities.

In the current study, we used LCA to classify children into profiles based on their classroom engagement. We then used profile membership to predict children’s gains from fall to spring of the prekindergarten year, in the areas of language and literacy and mathematics. Next, we examined soci- odemographic differences across the class engage- ment profiles. Finally, we assessed whether different child engagement profiles provided dif- ferent amounts of gains for poor versus nonpoor children.

Method

Participants

Study data come from the National Center for Early Development and Learning Multi-State Study of Pre-Kindergarten, and a follow-up study, the State-Wide Early Education Programs Study (SWEEP). Both studies used a stratified random sampling method to select programs within states, classrooms within programs, and children within classrooms. States with large numbers of children enrolled in public prekindergarten and with pro- grams that have had time to mature were selected. The Multi-State included six states (California, Illinois, Georgia, Kentucky, New York, and Ohio), with 40 programs selected from each state, and began in fall of 2001; the SWEEP included five states (Massachusetts, New Jersey, Texas, Wash- ington, and Wisconsin), with 100 programs

1536 Chien et al.

selected from each state, and began in fall of 2003. Programs were diverse with regard to urbanicity. All programs were state funded, although funding streams differed within and across states; 15% of programs were part of Head Start programs. A total of 701 programs participated, and one class- room was randomly selected from each program. In both the Multi-State and the SWEEP, 94% of classroom teachers agreed to participate. In the Multi-State and SWEEP, 61% and 55% of parents, respectively, gave consent to participate. Of the children with parent consent, four children (two boys and two girls) from each classroom were ran- domly selected to participate, making for a total of 2,966 children.

Sample classrooms had an average class size of 19 children and an adult–child ratio of 1:8.6. A majority of the children (58%) were from families living below the federal poverty line, and about half were boys (49%). The sample included children who were European American (41%), African American (18%), Latino (27%), Asian American (4%), Native American (1%), and of other ethnicities (10%). Mean maternal education was 12.8 years. Children with no classroom engagement observations (n = 215) were dropped from the study, resulting in a final sample of 2,751 children.

Measures

Classroom Observations

The Emerging Academics Snapshot (Ritchie, Howes, Kraft-Sayre, & Weister, 2001) is a measure of children’s classroom engagement that captures children’s moment-to-moment activities. Observa- tions were conducted over 2 days in the Multi- State and over 1 day in the SWEEP, each in the spring. Each child was observed in 20-s interval ‘‘snapshots,’’ followed by a 40-s coding period. The data collector then observed each of the other study children in that classroom before coming back to observe the first child again, continuing in this manner for the entire day. During each 20-s snapshot, children were coded with one of six mutually exclusive activity settings: basics, free choice, individual time, meals, small group, and whole group; one or more preacademic engagements: esthetics, fine motor skills, gross motor skills, letter and sound, mathematics, oral language develop- ment, prereading, read to, science, social studies, and writing; and one or more teacher–child inter- actions: routine, minimal, simple, elaborated, scaf-

folding, and didactic (see Table 1 for a detailed description of each child engagement). Kappas range from .70 to .87.

Global classroom environmental quality was assessed using the Early Childhood Education Rat- ing System–Revised (ECERS–R; Harms, Clifford, & Cryer, 1998) and the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, & Hamre, 2004). The ECERS–R assesses aspects of the preschool classroom, such as room arrangement and furnish- ings and relaxation. Scores from the Multi-State study reflect the mean of a fall and a spring obser- vation, and scores from the SWEEP study reflect a single observation day in the fall. Composite ECERS–R scores range from 1 (inadequate quality) to 7 (excellent quality). Factor analysis of the ECERS–R yielded two factors, one of which was used in the current study: the Teaching and Interactions factor, which consisted of items such as Encouraging Chil- dren to Communicate, Discipline, and Interactions among Children. The CLASS rates the emotional climate, classroom climate, and instructional sup- ports for learning in early childhood classrooms. Classrooms were rated from 1 (low) to 7 (high) on nine dimensions, such as Positive Climate, Teacher Sensitivity, and Concept Development. Data collec- tors rated the classroom and teacher on the nine dimensions about every 30 min throughout an observation day. The CLASS scores for children in the Multi-State study reflect the mean of a fall and a spring observation, and for children in the SWEEP study a single observation day in the spring. Factor analysis of the CLASS yielded two factors, one of which was used in the current study: Instructional Climate, which is a composite of Con- cept Development and Quality of Feedback.

Centralized training of all observers was con- ducted for the three classroom observations. Prior to data collection, observers were certified via reli- ability tests that compared observers’ ratings with experts’ ratings. Interrater agreement was evaluated using weighted Kappas, which adjust for agree- ment due to chance; observers with an overall kappa of at least .60 were certified (see Pianta et al., 2005, for complete details). Kappas ranged from .65 to .81. for the ECERS–R, CLASS, and Snapshot; Kappas of .65 or higher indicate good agreement (Landis & Koch, 1977).

Indicators of School Readiness

Direct child assessments. In the fall and spring, children’s language, preliteracy, and mathematics skills were conducted by a different data collector

Classroom Engagement and School Readiness Gains 1537

from the one who conducted the classroom obser- vations.

Children who spoke a home language other than English were given an English language screener, the Pre-LAS2000 (Duncan & De Avila, 1998). The Pre-LAS was first administered in the fall, and children who did not pass in the fall

were assessed again in the spring (fall: a = .89; spring: a = .89). Children who passed the screener (a score of 31 or higher of 40) were administered the English assessment battery (fall: n = 147; spring: n = 117). Children who did not pass the screener and who spoke Spanish at home were administered the Spanish assessment battery (fall:

Table 1

Definitions for the Emerging Academics Snapshot

Code Description

Activity settings

Basics Napping, toileting, standing in line, cleaning up, or waiting between activities

Free choice Child selects what and where to play or learn, engaging in activities such as individual art projects, blocks,

pretend play, and reading

Individual time Child and the rest of the class each work on a project independently, such as a worksheet or on the computer.

The teacher moves around to help

Meals Eating lunch, breakfast, or snacks

Small group Small-group activities that are teacher organized or teacher led, such as group art projects, writing stories, or

collective building

Whole group Whole-group activities that are teacher initiated, such as stories, songs, calendar, discussions, book reading, and

demonstrations

Preacademic and academic activities

Esthetics Art, drama, or music activities

Fine motor Stringing beads, building with Legos, cutting, or using crayons and markers

Gross motor Running, skipping, jumping, swinging, riding bikes, or playing games such as basketball, catch, run and chase,

dancing, or musical chairs

Letter and sound With guidance from teacher, child identifies letters, sounds out words, talks about letter-sound relationships,

recognizes sounds using rhymes

Mathematics Rote counting, counting with one-to-one correspondence, skip counting, matching numbers to pictures, making

graphs, or playing counting games

Oral language Child interacts with teacher or peers in talking about stories, or telling development stories of their own, or

answering and asking open-ended questions

Prereading Reading stories, identifying words, recognizing symbols and pictures as having meaning, or practicing a class

poem

Read to Teacher reads books and stories to child, engages in talking about the author, showing the cover, or asking

questions about the book

Science Child explores natural phenomena in their environment, uses science equipment, and reads books or talks about

animals, body parts, and life cycles

Social studies Talking, reading, or engaging in activities about their world (e.g., their neighborhood, their school, the farm, the

community workers)

Writing Writing (or using a computer keyboard to write) numbers or letters

Teacher–child interactions

Routine Teacher engages in routine caregiving (e.g., wipe child’s nose)

Minimal Teacher answers child’s direct requests for help, or gives simple verbal directives with no reply encouraged, such

as ‘‘okay,’’ ‘‘stop that!’’

Simple Teacher uses some warm or helpful physical contact or verbally answers the child’s verbal bids but does not

elaborate

Elaborated Teacher engages in some physical response (e.g., thumbs up, high fives, frown, glare) or acknowledges the child’s

statements and responds

Scaffold Teacher (or a more capable peer) does one-on-one work with child and builds on child’s initiations, using visuals,

concrete objects, and gestures to help child learn. Teacher elicits responses and helps child expand his or her

thoughts

Didactic Teacher lectures, gives instructions, models, asks close-ended questions, or demonstrates, such as counting or

saying the days of the week

1538 Chien et al.

n = 394; spring: n = 317). Children who did not pass the screener and who spoke a language other than Spanish at home were not assessed in the Multi-State but were assessed in English in the SWEEP (fall: n = 39; spring: n = 11). With the exception of the Peabody Picture Vocabulary Test (PPVT) and the Test de Vocabulario en Imagenes Peabody (TVIP), English and Spanish assessments were collapsed for analysis so that children who were administered the fall assessments in Spanish and the spring assessments in English (i.e., the 117 children who passed the Pre-LAS in the spring) could be retained for analysis.

The PPVT–3rd Edition (PPVT–III; Dunn & Dunn, 1997) is a test of English receptive vocabulary where the child was shown a set of four pictures at a time and asked to select the picture that best represents the word spoken by the examiner (fall: a = .96; spring: a = .96). A standardized score was computed for this scale. The TVIP (Dunn, Lugo, Padilla, & Dunn, 1986) is the Spanish version of the PPVT (fall: a = .92; spring: a = .93).

The Oral and Written Language Scale–Oral Expression Scale (OWLS; Carrow-Woolfolk, 1995) assesses children’s understanding and use of spo- ken language (English only; fall: a = .92; spring: a = .91). After receiving the examiner’s verbal stim- ulus, the child looked at a picture board and responded orally by answering a question, complet- ing a sentence, or generating a sentence. A stan- dardized score was computed for this scale.

Woodcock-Johnson (WJ) III Tests of Achievement (Woodcock, McGrew, & Mather, 2001) are well- established measures of academic achievement, and the Spanish versions are the Baterı́a Woodcock- Muñoz-Revisada: Pruebas de Aprovechamiento (Woodcock & Munoz-Sandoval, 1996). The Applied Problems subtest is a mathematics test that assessed children’s ability to comprehend the nature of a problem, identify relevant information, and perform simple calculations (English fall: a = .84; English spring: a = .83; Spanish fall: a = .81; Span- ish spring: a = .79). The Letter–Word Identification subtest, administered in the SWEEP study only (n = 1805), asks children to identify letters and then words (English fall: a = .89 English spring: a = .83; Spanish fall: a = .65; Spanish spring: a = .89). Most children moved beyond identifying letters to identi- fying words—that is, their ceiling set included word identification.

Children’s ability to identify letters, numbers, and colors, to count, and to write their names were assessed in both the English and Spanish assess- ment batteries. For identifying letters, children were

shown a set of mixed capital and lowercase letters and asked to identify as many as they could; the maximum score was 26. For identifying numbers, children were shown a sheet of numbers from 1 to 10, presented in random order, and asked to iden- tify as many as they could; the maximum score was 10. For naming colors, children were presented with a page of 10 different colored bears and asked to name the colors; the maximum score was 10. Alphas for identifying letters, numbers and colors for the English and Spanish assessments ranged from .81 to .97. For the counting task, children were asked to count and point with one-to-one corre- spondence using a picture card with 20 teddy bears. If the child counted to 20 correctly, another sheet of 20 teddy bears was presented to allow con- tinued counting. Finally, children wrote their names and the percent of their name written legibly was calculated.

Teacher report. Teacher report of children’s lan- guage and literacy skills in the fall and spring was an average of nine items from the teacher questionnaire of the Early Childhood Longitudinal Studies–Kindergarten Cohort (West, Denton, & Germino-Hausken, 2000). The items assessed chil- dren’s proficiency in speaking (e.g., using complex sentence structure), listening, early reading (e.g., predicting what will happen next), and early writ- ing (e.g., using initial consonants to spell words). The items were rated on a scale from 1 (not yet) to 5 (proficient; fall: a = .91; spring: a = .92).

Poverty Status

Information was obtained on whether the chil- dren’s family lived below the federal poverty threshold ($17,960 for a family of four in 2001; $18,660 for the same family in 2003).

Sociodemographic Covariates

The analysis of group differences across school readiness used the following covariates: child gen- der, child ethnicity, household size, poverty status, mothers’ years of education, and child age at the spring assessment.

Results

Descriptives

The proportion of a day a child spent in a partic- ular activity or setting was computed as the ratio of

Classroom Engagement and School Readiness Gains 1539

the number of intervals the activity or setting was observed divided by the total number of intervals the child was observed. Mean proportions of a day children spent in each activity setting, teacher–child interaction, and preacademic activity are summa- rized in Table 2.

In terms of activity settings, children spent the largest amount of time in free-choice (30%) and whole-group activities (27%), and the least amount of time in individual time (7%) and meals (7%). In terms of preacademic activities, children spent the most time on esthetics (15%), social studies (15%), and science (11%); the least amount of time was spent in prereading (3%), letter–sound (4%), and being read to (5%). In terms of teacher–child inter- actions, children spent the largest amount of time in didactic interactions (31%) and the second largest

amount of time in scaffolding (9%); the least amount of time was spent in routine interactions (1%).

Some of the school readiness indicators were cor- related. In language and literacy, the following pairs of outcomes had a correlation coefficient of .45 or higher: naming letters and WJ Letter–Word (r = .66; all results reported as significant are p < .05 or better), naming letters and percent name written legibly (r = .45), naming letters and teacher report of language and literacy (r = .49), and OWLS and PPVT (r = .68). In mathematics, the following indicator pairs had a correlation coefficient of .45 or higher: WJ applied problems and naming numbers (r = .55), WJ applied problems and highest number counted (r = .49), and naming numbers and highest number counted (r = .55).

Table 2

Descriptives of Snapshot Variables by Snapshot Profiles

Overall

sample

Free play

profile

(Profile 1)

Individual

instruction

profile

(Profile 2)

Group

instruction

profile

(Profile 3)

Scaffolded

learning

profile

(Profile 4) Significant differences

(p < .05)aM SD M SD M SD M SD M SD

Activity settings

Basics 0.21 0.10 0.20 0.08 0.25 0.09 0.26 0.10 0.16 0.08 SL < FP < II = GI

Free choice 0.30 0.17 0.41 0.12 0.13 0.11 0.15 0.10 0.29 0.13 II = GI < SL < FP

Individual time 0.04 0.07 0.02 0.04 0.21 0.08 0.03 0.04 0.03 0.04 FP = GI = SL < II

Meals 0.12 0.07 0.14 0.07 0.11 0.07 0.10 0.06 0.12 0.07 GI < SL < FP; II < FP

Small group 0.06 0.09 0.04 0.06 0.04 0.07 0.11 0.12 0.06 0.08 FP = II < SL < GI

Whole group 0.27 0.13 0.20 0.09 0.27 0.11 0.36 0.12 0.34 0.12 FP < II < FP < GI

Preacademic and academic activities

Esthetics 0.15 0.09 0.14 0.09 0.17 0.09 0.15 0.09 0.16 0.10 FP < SL; FP < II; GI < II

Fine motor 0.10 0.08 0.09 0.07 0.17 0.08 0.09 0.07 0.10 0.08 FP = GI = SL < II

Gross motor 0.06 0.06 0.08 0.06 0.03 0.04 0.04 0.05 0.06 0.05 II = GI < SL < FP

Letter and sound 0.04 0.05 0.03 0.03 0.08 0.07 0.05 0.05 0.05 0.05 FP < GI = SL < II

Mathematics 0.08 0.06 0.06 0.05 0.10 0.08 0.10 0.07 0.11 0.07 FP < GI < SL; FP < II

Oral language development 0.06 0.06 0.04 0.04 0.06 0.05 0.05 0.04 0.14 0.08 FP < GI = II < SL

Prereading 0.03 0.04 0.03 0.04 0.04 0.05 0.03 0.04 0.05 0.05 FP = GI < II < SL

Read to 0.05 0.05 0.04 0.04 0.06 0.05 0.06 0.05 0.08 0.06 FP < II = GI < SL

Science 0.11 0.09 0.11 0.09 0.09 0.08 0.09 0.08 0.16 0.12 GI < FP < SL; II < SL

Social studies 0.15 0.11 0.17 0.11 0.08 0.08 0.11 0.08 0.21 0.12 II < GI < FP < SL

Writing 0.01 0.03 0.01 0.02 0.03 0.05 0.01 0.02 0.02 0.03 FP = GI < SL < II

Teacher–child interactions

Routine 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 None

Minimal 0.03 0.03 0.03 0.03 0.02 0.03 0.02 0.03 0.02 0.03 None

Simple 0.05 0.04 0.06 0.05 0.04 0.04 0.04 0.04 0.05 0.05 II = GI < SL < FP

Elaborated 0.04 0.04 0.04 0.04 0.03 0.04 0.03 0.03 0.06 0.05 II = GI < FP < SL

Scaffold 0.09 0.08 0.06 0.06 0.09 0.07 0.08 0.06 0.21 0.08 FP < GI < II < SL

Didactic 0.31 0.16 0.25 0.14 0.39 0.16 0.35 0.14 0.37 0.17 FP < GI < SL < II

Note. FP = free play profile; II = individual instruction profile; GI = group instruction profile; SL = scaffolded learning profile. aBonferroni post hoc contrasts were used.

1540 Chien et al.

Latent Class Analysis

Latent class analysis was conducted in a series of steps, the first of which was fitting a one-profile model to the data to establish a baseline. Models with successively increasing numbers of profiles (up to a model with five profiles, which failed to converge) were then tested. Models were compared using the following model fit indices: the Akaike information criterion (AIC; Akaike, 1987), the Bayesian information criterion (BIC; Schwartz, 1978), and the adjusted BIC (ABIC; Sclove, 1987). Smaller values on these indices indicate better fit, but there is no criterion for ‘‘good’’ fit; therefore, these values are only useful for comparing two or more models. Additionally, an entropy approach- ing 1.0 indicates a clear distinction between profiles (Celeux & Soromenho, 1996). The four-profile model was the best fitting model with the smallest AIC ()186005.275), BIC ()185277.150), and ABIC ()185667.961), and the largest entropy (0.825). The LCA models were tested using Mplus (Version 4.0; Muthén & Muthén, 1998-2006).

Profiles 1, 2, 3, and 4 included 51%, 9%, 27%, and 13% of the children, respectively. Means and standard deviations of classroom engagements for each profile are summarized in Table 2. Using anal- ysis of variance (ANOVA), we identified significant differences in classroom engagement across pro- files. Children in Profile 1 spent by far the most time in free choice (41%), compared to other pro- files, and less time than all other groups on a vari- ety of preacademic engagements. Profile 1 is

therefore labeled the ‘‘free play’’ profile. Children in Profile 1 also spent a little more time engaged in gross motor activities (8%) than any other profile and more time than Profiles 2 and 3 in social stud- ies (17%). Children in Profile 2 spent much more time than any other profile in individual time (21%), fine motor skills (17%), and letter–sound (8%). Thus, Profile 2 is labeled the ‘‘individual instruction’’ profile. Children in Profile 3 spent more time than any other profile in whole group (36%) and small group (11%), and this profile is therefore labeled the ‘‘group instruction’’ profile. Children in Profile 4 spent much more time engaged in scaffolding interactions with teachers (21%) than any other profile, and also the most time in elaborated teacher–child interactions (6%). In addition, Profile 4 spent more time than any other profile on many preacademic activities. Therefore, Profile 4 is labeled the ‘‘scaffolded learning’’ pro- file. Profile 4 spent more time than Profiles 2 and 3 on free-choice activities (29%).

Using ANOVA, we found that the snapshot pro- files differed significantly on some demographic characteristics. An examination of household size, maternal education, poverty status, and ethnicity showed that children in the free play and scaffold- ed learning profiles appeared to be somewhat more privileged than children in the individual instruc- tion and group instruction profiles (see Table 3). Specifically, the free play and scaffolded learning profiles have smaller households and more years of maternal education than children in the individual and group instruction profiles. Children in the indi-

Table 3

Demographic Information by Snapshot Profiles

Free play

profile

Individual

instruction

profile

Group

instruction

profile

Scaffolded

learning profile Significant differences

(p < .05)aM or % SD M or % SD M or % SD M or % SD

Household size 4.35 1.43 4.81 1.56 4.57 1.45 4.38 1.34 FP = SL < II = GI

Child is poor 0.57 — 0.69 — 0.57 — 0.50 — FP = SL = GI < II

Mothers years of education 12.77 2.36 12.06 2.58 12.53 2.37 13.00 2.67 II < GI < SL; II < FP

Child age (spring) 5.04 0.32 5.09 0.30 5.06 0.32 5.06 0.31 None

Latino 22% — 41% — 29% — 28% — FP < GI < II; SL < II

African American 16% — 27% — 24% — 14% — SL = FP < II = GI

Asian American 15% — 13% — 10% — 14% — GI < FP

European American 46% — 19% — 37% — 43% — II < GI < FP; II < SL

Gender (male) 50% — 45% — 49% — 50% — None

Note. FP = free play profile; II = individual instruction profile; GI = group instruction profile; SL = scaffolded learning profile. Percentages are base rates within each profile. aBonferroni post hoc contrasts were used.

Classroom Engagement and School Readiness Gains 1541

vidual and group instruction profiles were more likely to be Latino or African American than chil- dren in the free play or scaffolded learning profiles. Children in the individual instruction profile were also most likely to be poor compared to all other children. No gender differences emerged between profiles.

Predicting Academic Gains Based on Profile Membership

We assessed the predictive validity of the snap- shot profiles by examining fall-to-spring gains in school readiness between the four profiles. We used analysis of covariance (ANCOVA) and covaried for household size, poverty status, maternal education, child race ⁄ ethnicity, child gender, and child age. Change was assessed by including fall scores as covariates (i.e., predictors) of the spring scores,

such that the residual variance represents change from fall to spring.

Table 4 shows the scores for each profile of children and for each outcome. For example, the score of 12.19 on naming letters in the column ‘‘Free play profile’’ indicates that after adjusting for covariates, children in the free play profile were able to name an average of 12 letters. The overall F tests revealed significant differences across profiles for the following indicators of school readiness: naming letters, WJ letter–word identification, percent of name written legibly, teacher report of language and literacy skills, WJ applied problems, naming numbers, and highest number counted. Bonferroni post hoc contrasts were then used to examine pairwise differences between profiles.

The free play profile made the smallest gains across language and literacy and mathematics.

Table 4

Residual Change Scores From Fall to Spring of Prekindergarten, by Snapshot Profiles

Free play

profile

Individual

instruction

profile

Group

instruction

profile

Scaffolded

learning

profile

F-test, significant

differences (p < .05)a Partial

eta2

n = 1398 n = 245 n = 755 n = 353

M SE M SE M SE M SE

Language and literacy

Naming letters 12.19 0.17 14.64 0.41 14.21 0.22 13.73 0.34 F(3, 2391) = 22.76***

FP < SL = GI = II

0.03

Naming colors 8.96 0.05 9.00 0.12 9.03 0.06 9.09 0.10 F(3, 2392) = 0.53 0.00

PPVT (standardized score) 96.81 0.25 96.77 0.65 97.24 0.34 96.83 0.49 F(3, 2279) = 0.39 0.00

TVIP (standardized score) 85.76 1.24 84.41 1.48 84.15 1.18 83.94 2.15 F(3, 235) = 0.37 0.01

OWLS (Oral and Written Language) 94.82 0.24 94.29 0.65 94.37 0.33 94.63 0.49 F(3, 2022) = 0.50 0.00

WJ letter–word identification 342.92 0.74 351.28 1.35 350.21 0.76 346.68 1.14 F(3, 1548) = 18.76***

FP < SL = GI = II

0.04

Percent name legible 83.41 0.63 89.32 1.50 88.85 0.83 85.76 1.24 F(3, 2359) = 10.82***

FP < GI = II

0.01

Language and literacy (teacher report) 2.94 0.02 3.15 0.06 3.06 0.03 3.09 0.05 F (3, 1984) = 6.64***

FP < SL = GI = II

0.01

Mathematics

WJ applied problems 412.35 0.36 416.50 0.85 413.14 0.47 412.63 0.70 F(3, 2372) = 6.73***

FP = SL = GI < II

0.01

Naming numbers 6.40 0.07 6.85 0.17 6.91 0.10 6.85 0.14 F(3, 2391) = 7.42***

FP < SL = GI

0.01

Highest number counted 19.22 0.28 22.58 0.66 21.87 0.37 20.99 0.55 F(3, 2336) = 14.92***

FP < SL = GI = II

0.02

Note. FP = free play profile; GI = group instruction profile; II = individual instruction profile; OWLS = Oral and Written Language Scale–Oral Expression Scale; PPVT = Peabody Picture Vocabulary Test; SL = scaffolded learning profile; TVIP = Test de Vocabulario en Imagenes Peabody; WJ = Woodcock-Johnson. aBonferroni post hoc contrasts were used. ***p < .001.

1542 Chien et al.

Specifically, the free play profile showed less growth than all other profiles in naming letters, WJ letter–word identification, teacher report of lan- guage and literacy skills, and number counting. The free play profile also showed less growth than the individual and group instruction profiles in writing their names, and less growth than the group instruction profile and the scaffolded learn- ing profile on number counting. Finally, the free play profile showed less growth than the individual instruction profile on the WJ applied problems. There was only one significant difference among the individual instruction, group instruction, and scaffolded learning profiles: the individual instruc- tion profile made the greatest gains on the WJ applied problems.

Although effect sizes (partial eta2) were small, the differences in gains between groups are mean- ingful. For example, children in the free play profile scored 8 points lower on the WJ letter–word identi- fication than children in the individual and group instruction profiles; this is large for differences that emerged over the course of less than 1 year. Addi- tionally, measurement error for assessments involv- ing younger children tend to be larger than that for older children, and larger measurement errors result in smaller effect sizes (Burchinal, 2008).

In summary, the free play profile showed less growth across indicators of language ⁄ literacy and mathematics compared to the other three profiles. The individual instruction profile outperformed all other groups on WJ applied problems.

Snapshot Profiles as Distinct from Classroom Environmental Quality

We wanted to show that the snapshot profiles contribute information about children’s classroom experiences that is distinct from and adds to infor- mation conveyed by classroom ECERS-R scores. Ideally, we wanted to find no mean differences in ECERS-R scores across the four snapshot profiles. However, there were indeed significant differences: The free play profile (M = 4.09, SD = 0.73) and the scaffolded learning profile (M = 4.14, SD = 0.75) each had significantly higher scores than the group instruction profile (M = 3.42, SD = 0.74) and the individual instruction profile (M = 3.17, SD = 0.63). In spite of finding significant mean differences, the fairly large standard deviations of ECERS–R scores indicates significant overlap in ECERS–R scores across different snapshot profiles; this in turn sug- gests that the snapshot profiles provide information not already conveyed by the ECERS–R scores.

To further examine whether the Snapshot pro- files contain unique information about children’s experiences, we repeated the ANCOVA by profile membership, adding the ECERS–R Teaching and Interactions Scale and the CLASS Instructional Climate Scale as covariates in two separate sets of analyses. All analyses yielded nearly identical results to the original analyses, suggesting again that the snapshot profiles captures information about children’s experiences unique from that captured by the ECERS–R and CLASS (results are available upon request).

Poverty Status and Children’s Classroom Engagement

To test whether the relationship between gains in school readiness and profile membership varied across poverty status, we conducted a 2 (poverty status) · 4 (snapshot profile) ANCOVA, covarying for gender, ethnicity, household size, poverty sta- tus, maternal education, and child age. Table 5 shows adjusted mean scores by profile membership and by poverty status; interaction effects are also shown.

The interaction between poverty status and snap- shot profile membership was significant for WJ letter–word identification and highest number counted, and there was a marginal effect (p = .055) for WJ applied problems.

For WJ letter–word identification, highest num- ber counted, and WJ applied problems, poor chil- dren in the free play, group instruction, and scaffolded learning profiles made smaller gains than their nonpoor peers. In the individual instruc- tion profile, however, poor children actually made greater gains than nonpoor children across these measures of school readiness.

Discussion

It is useful to think about classroom engagement in terms of number of minutes children spend on each activity during a typical program day. For example, the most frequent activity setting was free play (30%), which occupied an average of 45 min of a part-day, 2.5 hr program—quite a longtime. In contrast, < 8 min ⁄ day was devoted to each of the following literacy activities: prereading (3%), letter– sound (4%), and being read to (5%).

Moving beyond whole-sample averages, there was substantial variation in children’s classroom engagement. Using a person-centered approach, we identified four profiles of children with distinct

Classroom Engagement and School Readiness Gains 1543

patterns of classroom engagement: a free play pro- file for whom class time was dominated by free- choice activities; two instructional profiles that spent a lot of time receiving teacher instruction, whether in an individual or group format; and one scaffolded learning profile that engaged in large amounts of teacher scaffolding.

We had expected a profile of children that received frequent teacher instruction. We did not expect, however, two instructional profiles that were distinct in terms of the format—individual or group—in which they received teacher instruction. Advantages of individual settings include increased attention for teacher directions and greater need for self-regulation while giving children an opportu- nity to practice skills they already learned (Stright & Supplee, 2002). An advantage of the group activ- ity setting is that children have the opportunity to share ideas and learn from one another. These results suggest that among the teachers who prac- tice the instructional model of early childhood edu- cation, some teachers prefer to deliver instruction via individual seatwork while other teachers prefer to deliver instruction via group work.

Next, we found that the free play profile made the smallest fall-to-spring gains across many measures of language and literacy and mathematics compared to the two instructional profiles and the

scaffolded learning profile. Recall that we had included a variety of assessments to explore whether different models of early childhood educa- tion were best suited for different domains of learn- ing; our results did not support this hypothesis.

Given that the free play profile was by far the largest profile, making up 51% of the sample chil- dren, it was discouraging to find that it was also the profile with the smallest gains in child out- comes. Although these results suggest that the instructional and scaffolding models of early child- hood education are more beneficial than the free- choice play model, a caveat exists. Recall that the study sample was at higher demographic risk than the national average: A full 58% of the sample is poor, and the average maternal education is only 12.9 years. Recall also the research showing that instructional support was beneficial for at-risk chil- dren, but not necessarily for children not at risk (Hamre & Pianta, 2005). It is possible, therefore, that our finding that classroom instruction was more beneficial than free-choice play is partially the result of having a sample of predominantly at-risk children.

Our study results revealed two further points of interest regarding the scaffolded learning profile. First, the scaffolded learning profile spent a large amount of time in free-choice activities (29%): less

Table 5

Residual Change Scores From Fall to Spring of Prekindergarten, by Snapshot Profiles and by Poverty Status

Free play profilea Individual

instruction profilea

Group

instruction

profilea

Scaffolded

learning

profilea

Profile

Memership · Poverty Status

Nonpoor Poor Nonpoor Poor Nonpoor Poor Nonpoor Poor

F-testn = 604 n = 794 n = 76 n = 169 n = 321 n = 434 n = 178 n = 175

Language and literacy

Naming letters 12.44 12.00 13.60 15.02 14.72 13.82 13.79 13.72 1.93

Naming colors 9.08 8.87 8.73 9.10 9.05 9.01 9.13 9.05 1.56

PPVT (standardized score) 97.20 96.45 97.52 96.11 98.07 96.44 96.93 96.87 0.71

OWLS 95.21 94.44 95.42 93.32 95.19 93.58 95.85 93.25 1.19

WJ letter–word identification 346.31 340.40 349.05 351.70 350.75 349.90 348.46 345.35 3.19*

Percent name legible 84.45 82.57 87.61 89.98 88.59 89.06 85.57 86.06 0.82

Language and literacy skills 3.00 2.90 3.16 3.13 3.15 2.99 3.12 3.07 0.47

Mathematics

WJ applied problems 412.72 412.09 415.09 416.81 414.56 412.04 414.44 410.98 2.54� Naming numbers 6.58 6.26 6.67 6.89 7.13 6.74 6.79 6.94 1.43

Highest number counted 20.53 18.21 21.54 22.55 23.81 20.33 22.09 20.18 2.69*

Note. Comparisons were made between poor and nonpoor children, within classes. Tests of interactions on the TVIP was omitted due to small cell sizes. OWLS = Oral and Written Language Scale–Oral Expression Scale; PPVT = Peabody Picture Vocabulary Test; TVIP = Test de Vocabulario en Imagenes Peabody; WJ = Woodcock-Johnson. aMean values. �p < .06. *p < .05.

1544 Chien et al.

than the free play profile (41%) but a lot more than either of the instructional profiles (13% and 15%). These results suggest that free play, when accompa- nied by high-quality scaffolding interactions with teachers, remains a model of classroom engagement that may be conducive to children’s learning. Sec- ond, it was disappointing that although the scaf- folded learning profile had more than twice as many scaffolding interactions with teachers as any other profile, children in this profile did not experience the greatest growth compared to other profiles. Subsequent analysis revealed a possible explanation: Children in the scaffolded learning

profile had the highest fall scores across several child outcomes (see Table 6). Recall also that the scaffolded learning profile was the most sociode- mographically advantaged of the four profiles with smaller household sizes, a lower likelihood of being poor, and mothers with higher levels of education. As such, these children may have been better pre- pared for prekindergarten as a result of educational experiences they received in the home. Also, com- ing from more advantaged families may be directly related to the frequency of scaffolding interactions: These children may experience more scaffolding interactions with parents and therefore may be

Table 6

Mean Fall and Spring Scores for All Indicators of School Readiness, by Snapshot Profiles

Free play profile Individual instruction profile

Fall Spring Fall Spring

M SD M SD M SD M SD

Language and literacy

Naming letters 7.30 8.77 11.66 9.57 7.35 8.98 14.19 9.54

Naming colors 8.23 2.98 8.98 2.29 7.59 3.46 8.67 2.47

PPVT (standardized) 94.36 14.75 96.78 14.10 89.99 14.59 91.92 13.71

TVIP (standardized score) 78.76 11.09 82.67 14.86 82.39 13.00 86.49 14.98

OWLS 92.37 12.77 94.53 12.96 87.42 12.15 89.57 12.32

WJ letter–word identification 330.82 27.10 341.34 27.89 339.58 27.08 355.08 32.12

Percent name legible 59.16 37.81 81.89 28.74 66.34 35.94 90.59 20.52

Language and literacy skills 2.32 0.87 2.97 0.96 2.19 0.86 3.03 1.01

Mathematics

WJ applied problems 400.78 20.88 411.66 18.94 398.99 21.23 413.10 17.69

Naming numbers 4.14 3.90 6.19 3.71 4.16 3.99 6.74 3.44

Highest number counted 13.68 9.97 18.52 11.46 13.86 10.20 21.60 12.89

Group instruction profile Scaffolded learning profile

Fall Spring Fall Spring

M SD M SD M SD M SD

Language and literacy

Naming letters 7.42 8.60 13.86 9.51 9.49 9.28 15.05 9.42

Naming colors 7.97 3.14 8.95 2.26 8.27 3.04 9.19 2.04

PPVT (standardized) 92.59 14.78 95.61 14.13 96.49 16.58 97.89 15.53

TVIP (standardized score) 77.77 11.96 83.88 13.89 77.96 15.66 82.17 15.79

OWLS 89.90 13.10 92.29 12.62 93.96 14.21 95.10 13.79

WJ letter–word identification 330.55 24.23 348.25 24.90 335.32 26.12 348.34 25.96

Percent name legible 60.36 36.69 87.67 23.59 68.82 36.25 88.81 22.40

Language and literacy skills 2.16 0.83 2.96 0.95 2.48 1.00 3.21 0.98

Mathematics

WJ applied problems 399.27 20.82 411.73 18.08 404.27 20.60 414.77 20.96

Naming numbers 4.21 3.89 6.83 3.60 5.03 4.03 7.36 3.45

Highest number counted 14.09 9.71 21.32 12.39 14.52 10.11 21.15 11.72

Note. OWLS = Oral and Written Language Scale–Oral Expression Scale; PPVT = Peabody Picture Vocabulary Test; TVIP = Test de Vocabulario en Imagenes Peabody; WJ = Woodcock-Johnson.

Classroom Engagement and School Readiness Gains 1545

more likely to elicit and sustain scaffolding inter- actions with teachers.

Note that for the individual instruction profile, the standard errors of many child outcome mea- sures are larger than for other profiles; this is because the individual instruction profile had fewer children than the other three profiles. Had the stan- dard errors been smaller, more differences between the individual instruction profile and the free play profile may have been statistically significant. Therefore, the small size of the individual instruc- tion profile had the effect of making the results more conservative.

The study examined a variety of child out- comes—from higher order language and problem- solving skills to basic letter and number recognition skills—that are important predictors of later ele- mentary school reading and mathematics. The WJ battery of achievement tests are good predictors of later academic achievement (McGrew, 1986). Less obvious, perhaps, is that basic skills such as naming letters and counting are also important for later aca- demic success. Basic literacy skills developed in preschool such as naming letters and writing one’s own name partially determine reading develop- ment in early elementary school (Lonigan, Burgess, & Anthony, 2000; Storch & Whitehurst, 2002). Like- wise, the lack of basic mathematics skills in kinder- garten, such as the ability to name numbers, predicted having a mathematics learning disability in third grade (Mazzocco & Thompson, 2005). Fur- ther evidence from an early mathematics interven- tion program suggests that teaching basic number sense to children at risk for having difficulties in mathematics improves later mathematics perfor- mance (Griffin, Case, & Siegler, 1994). Although middle-class children may have the opportunity to develop these basic skills at home, low-income chil- dren may not; therefore, it is all the more important that low-income children learn these skills at school. Indeed, Delpit (1996) made the case that low-income African American children need to be taught basic skills so that they can translate their oral language fluency and creativity—which they already possess—into a format that is recognized by mainstream society.

The snapshot profiles captured information about children’s classroom experiences not cap- tured by the ECERS–R or the CLASS. ANCOVA of children’s gains across profiles remained unchanged after the ECERS–R Teaching and Inter- actions Scale and the CLASS Instructional Climate Scale were separately added as covariates. In addi- tion, the two snapshot profiles that exhibited

greater gains across the prekindergarten year—the individual and group instruction profiles—actually had lower average composite ECERS–R scores com- pared to the other two profiles. That is, children in lower quality classroom environments (measured by ECERS–R) experienced classroom engagement (measured by the Snapshot) that was associated with greater gains in school readiness. These results suggest that the Snapshot captured aspects of chil- dren’s classroom experiences not captured by the ECERS–R and CLASS.

Looking across poverty status, we found that children in the individual instruction profile were more likely to be poor than children in any other profile. This is consistent with prior research sug- gesting that teachers of low-SES children tend to focus on developing skills in mathematics and liter- acy (Lee & Ginsburg, 2007), which are often prac- ticed in individual instruction settings. That the individual instruction profile had the highest pro- portion of poor children is fortuitously joined by the finding that poor children in the individual instruction profile were the only ones to make greater gains than their nonpoor peers on indica- tors of mathematics and literacy. This finding is consistent with the view some scholars have expressed that minority and low-income children stand to gain more from a curriculum focused on developing academic skills (Delpit, 1996).

Limitations

Free-choice activities teach children complex skills, such as problem-solving, making plans, and comprehension (Johnson & Yawkey, 1988). Unfor- tunately, these types of higher order skills were not fully assessed by the measures included in the Multi-State and SWEEP studies; thus, gains made by children in the free play profile may not have been captured. Nonetheless, assessments, such as the PPVT, OWLS, and WJ do capture more advanced language and cognitive abilities, and the free play profile still made smaller gains on these assessments than other children. Although this study may not have included the most ideal mea- sures for capturing gains made by the free play profile, the results nonetheless suggest that children in the free play profile made smaller gains in many important domains of development.

In all analyses, we treated children’s English and Spanish scores on various assessments as equiva- lent (with the exception of the PPVT and TVIP). Although most children were assessed at both time points in the same language, some children

1546 Chien et al.

(n = 117) were administered the Spanish assess- ments in the fall and, after passing the Pre-LAS in the spring, were administered the English assess- ments in the spring. Those children would have been dropped had we not combined the English and Spanish scores. Although it is not ideal to cal- culate child gains with Spanish assessments at the first time point and English at the second, it is a conservative measure of gains that preserves the number of cases without sacrificing the validity of the results.

Peer interactions were not captured by the snap- shot measure. However, children certainly learn through peer interactions, including through peer scaffolding interactions with same-ability class- mates (King, Staffieri, & Adelgais, 1998). Also, given that a large amount of time was spent in free play, it would have been instructive to capture whether, during free play, children were engaged in peer interaction and, if so, what type of peer interaction (e.g., parallel play, pretend play). Cod- ing an activity setting simply as free play is some- what limited because very different types of peer interactions, with different outcomes for children’s learning, could all occur during free play. Captur- ing children’s peer engagement would have added a rich dimension to further understanding the rela- tionship between classroom engagement and child outcomes.

Several measures of language and literacy and mathematics were moderately or highly correlated. But because many assessments highlight the differ- ent strengths of different snapshot profiles, it is still important to include each measure. For example, although WJ applied problems and naming num- bers were highly correlated (r = .55), the individual instruction profile excelled (compared to other groups) on WJ applied problems only. Further- more, previous studies have often examined these outcome measures together (Howes et al., 2008).

Implications for Policy and Practice

This study described patterns of children’s engagement in prekindergarten classrooms and explores whether some patterns of engagement bring about greater gains for children than others. The environmental quality literature thus far has encouraged preschool teachers to think about the arrangement of physical space and the materials available within classrooms. Results from the cur- rent study might encourage teachers to think about the allocation of children’s time to various class-

room engagements in a way that benefits children’s learning.

Increasingly, policy makers and advocacy orga- nizations (e.g., Trust for Early Education, National Institute for Early Education Research) are calling for ‘‘high-quality’’ prekindergarten for every child. These findings remind us that quality can be assessed in multiple ways that yield different qual- ity ratings for the same classrooms. For example, a class might receive a relatively high ECERS–R score for ample materials but have low levels of teacher– child interactions and may therefore not be opti- mally preparing children for school.

Teachers, educators, and policy makers also need to consider the factors contributing to the lack of high-quality teacher–child interactions. As pres- sures from federal and state mandates increase, and school districts increase demands for school- ready children, teachers are asked to do increas- ingly more. They are often trying to fit into a very short morning (a) circle time that includes calen- dar, weather, counting, good morning greetings, and a story; (b) small-group time to work on a par- ticular skill; (c) one or two meals; (d) trips to the bathroom; and (e) an inside and outside play per- iod. Although full-day programs do offer more hours in the day, many of the extra hours are spent in napping, toileting, and snacks. To our observers, the classrooms often felt pressured, as if teachers were racing to check off the list of things that need to be done. This kind of atmosphere does not give teachers opportunities to interact with children to talk about their lives, their play, or their ideas, nor does it provide opportunities for teachers to be responsive to children’s interests and needs. Results from this study suggest that more quality instructional time spent with teachers, and less free play time spent without teacher guidance or scaf- folding, would better prepare children for entering school.

References

Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332.

Barnett, W. S., Hustedt, J. T., Friedman, A. H., Boyd, J. S., & Ainsworth, P. (2007). The state of preschool 2007: State preschool yearbook. New Brunswick, NJ: Rutgers–The State University of New Jersey, National Institute for Early Education Research.

Bowman, B. T., Donovan, M. S., & Burns, M. S. (2000). Eager to learn: Educating our preschoolers. Washington, DC: National Academy Press.

Classroom Engagement and School Readiness Gains 1547

Burchinal, M. R. (2008). How measurement error affects the interpretation and understanding of effect sizes. Child Development Perspectives, 2, 178–180.

Burchinal, M. R., Kainz, K., Cai, K., Tout, K., Zaslow, M., Martinez-Beck, I., et al. (2008). Child care quality and child outcomes: Multiple-studies analyses. Paper presented at the University of California, Los Angeles.

Burchinal, M. R., Roberts, J. E., Riggins, R. Jr, Zeisel, S. A., Neebe, E., & Bryant, D. (2000). Relating quality of center-based child care to early cognitive and language development longitudinally. Child Development, 71, 338–357.

Carrow-Woolfolk, E. (1995). Oral and Written Language Scales (OWLS). Circle Pines, MN: American Guidance Service.

Celeux, G., & Soromenho, G. (1996). An entropy criterion for assessing the number of clusters. Journal of Classifi- cation, 13, 195–212.

Delpit, L. (1996). Other people’s children: Cultural conflict in the classroom. New York: New Press.

Duncan, S. E., & De Avila, E. (1998). PreLAS 2000. Mon- terey, CA: CTB ⁄ McGraw-Hill.

Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test (3rd ed.). Circle Pines, MN: Pearson’s Assessment Group.

Dunn, L. M., Lugo, D. E., Padilla, E. R., & Dunn, L. M. (1986). Test de vocabulario en imagenes peabody. Circle Pines, MN: American Guidance Service.

Ginsburg, H. P., & Opper, S. (1988). Piaget’s theory of intel- lectual development (3rd ed.). Englewood Cliffs, NJ: Prentice Hall.

Griffin, S. A., Case, R., & Siegler, R. S. (1994). Rightstart: Providing the central conceptual prerequisites for first formal learning of arithmetic to students at risk for school failure. In K. McGilly (Ed.), Classroom lessons: Integrating cognitive theory and classroom practice (pp. 25–49). Cambridge, MA: MIT Press.

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

Harms, T., Clifford, R., & Cryer, D. (1998). The Early Childhood Environment Rating Scale (Rev. ed.). New York: Teachers College Press.

Henderson, S. D., Many, J. E., Wellborn, H. P., & Ward, J. (2002). How scaffolding nurtures the development of young children’s literacy repertoire: Insiders’ and out- siders’ collaborative understandings. Reading Research and Instruction, 41, 309–330.

Howes, C. (2000). Social-emotional classroom climate in child care, child-teacher relationships and children’s second grade peer relations. Social Development, 9, 191– 204.

Howes, C., Burchinal, M., Pianta, R., Bryant, D., Early, D., Clifford, R., et al. (2008). Ready to learn? Children’s pre-academic achievement in pre-kindergarten pro- grams Early Childhood Research Quarterly, 23, 27–50.

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

Johnson, J. E., Christie, J. F., Yawkey, T. D., & Wardle, F. (1987). Play and early childhood development (Vol. 269). Glenview, IL: Scott, Foresman.

Johnson, J. E., & Yawkey, T. D. (1988). Play and integra- tion. In T. D. Yawkey (Ed.), Integrative processes and socialization: Early to middle childhood (pp. 97–117). Hills- dale, NJ: Erlbaum.

King, A., Staffieri, A., & Adelgais, A. (1998). Mutual peer tutoring: Effects of structuring tutorial interaction to scaffold peer learning. Journal of Educational Psychology, 90, 134–152.

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33, 159–174.

Lee, J. S., & Ginsburg, H. P. (2007). Preschool teachers’ beliefs about appropriate early literacy and mathe- matics education for low- and middle-socioeconomic status children. Early Education and Development, 18, 111–143.

Lonigan, C. J., Burgess, S. R., & Anthony, J. L. (2000). Development of emergent literacy and early reading skills in preschool children: Evidence from a latent-var- iable longitudinal study. Developmental Psychology, 36, 596–613.

Magnuson, K. A., Meyers, M. K., Ruhm, C. J., & Waldfo- gel, J. (2004). Inequality in preschool education and school readiness. American Educational Research Journal, 41, 115–157.

Mazzocco, M. l. M., & Thompson, R. E. (2005). Kindergar- ten predictors of math learning disability. Learning Dis- abilities Research & Practice, 20, 142–155.

McGrew, K. S. (1986). A review of the differential predic- tive validity of the Woodcock-Johnson Scholastic Apti- tude clusters. Journal of Psychoeducational Assessment, 4, 307–317.

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

Muthén, L. K., & Muthén, B. O. (1998-2006). Mplus user’s guide (4th ed.). Los Angeles, CA: Author.

Muthén, B., & Muthén, L. K. (2000). Integrating person- centered and variable-centered analyses: Growth mix- ture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882–891.

Pianta, R., Howes, C., Burchinal, M., Bryant, D., Clifford, R., Early, D., et al. (2005). Features of pre-kindergarten programs, classrooms, and teachers: Do they predict observed classroom quality and child-teacher inter- actions? Applied Developmental Science, 9, 144–159.

Pianta, R., La Paro, K. M., & Hamre, B. K. (2004). Classroom Assessment Scoring System (CLASS). Char- lottesville: University of Virginia.

1548 Chien et al.

Pianta, R., La Paro, K. M., Payne, C., Cox, M. J., & Bradley, R. (2002). The relation of kindergarten classroom envi- ronment to teacher, family, and school characteristics and child outcomes. Elementary School Journal, 102, 225– 238.

Ritchie, S., Howes, C., Kraft-Sayre, M., & Weister, B. (2001). Emerging academics snapshot. Los Angeles: University of California at Los Angeles.

Rogoff, B. (1990). Apprenticeship in thinking: Cognitive development in social context. New York: Oxford Univer- sity Press.

Schwartz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6, 461–464.

Sclove, S. L. (1987). Application of model-selection crite- ria to some problems in multivariate analysis. Psycho- metrika, 52, 333–343.

Smith, A. B. (1996). The early childhood curriculum from a sociocultural perspective. Early Child Development and Care, 115, 51–64.

Storch, S. A., & Whitehurst, G. J. (2002). Oral language and code-related precursors to reading: Evidence from a longitudinal structural model. Developmental Psychol- ogy, 38, 934–947.

Stright, A. D., & Supplee, L. H. (2002). Children’s self-reg- ulatory behaviors during teacher-directed seat-work, and small-group instructional contexts. Journal of Educa- tional Research, 95, 235–244.

Tonyan, H. A., & Howes, C. (2003). Exploring patterns in time children spend in a variety of child care activities: Associations with environmental quality, ethnicity, and gender. Early Childhood Research Quar- terly, 18, 121–142.

Vygotsky, L. S. (1962). Thought and language (E. Hanfmann & G. Vakar, Trans.). New York: MIT Press and Wiley.

West, J., Denton, K., & Germino-Hausken, E. (2000). America’s kindergartners: NCES 2000-070. Washington, DC: National Center for Educational Statistics.

Wood, D., Bruner, J. S., & Ross, G. (1976). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17, 89–100.

Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock Johnson III: Tests of Achievement. Itasca, IL: Riverside.

Woodcock, R. W., & Munoz-Sandoval, A. F. (1996). Bater- ı́a Woodcock Muñoz Revisada: Pruebas de Aprovechamiento. Itasca, IL: Riverside.

Classroom Engagement and School Readiness Gains 1549