Why Focus on Classrooms?

First, evidence from developmental science strongly indicates that effects of social settings on children’s development are driven by proximal processes (Bronfenbrenner & Morris, 1998), the interactions that children have on a daily basis with adults, peers, organizational structures, and materials. When examining schooling effects, children’s experiences in classrooms constitute the majority of their day and thus constitute the majority of school-based proximal processes. Furthermore, it is classroom level experiences that appear to be most closely associated with student outcomes (Nye, Konstantopoulos, & Hedges, 2004). School-level effects, such as school climate and culture, are important aspects of students’ experiences. However, these effects are more distal from students’ developmental progress and are largely mediated through or moderated by classroom processes (Hamre & Pianta, 2007).

Another reason to consider classrooms, discussed by Pianta (2006), is the recognition that classroom interactions are not simply applications of practices that are only of interest to educators, but rather reflect processes that should be of as much interest to developmental scientists as are interactions children have with parents and peers, if only for the reason that they are intended to produce developmental change. Researchers who have applied developmental theories to classrooms have discovered that classrooms offer an incredibly rich context in which to study basic developmental processes. Unlike in families, in which only a few children are exposed to a given familial context, in classrooms there are many children who are exposed to that environment over time. This allows for modeling aspects of bio-ecological theory (Bronfenbrenner & Morris, 1998), such as the potential for differential responsiveness to different features of a social setting. For example, work by Connor, Morrison, and colleagues (Connor, 2005; Connor, Morrison, & Petrella, 2004; Morrison & Connor, 2002) provides consistent evidence that children coming into classrooms with varying levels of reading ability benefit from very different instructional practices. Their studies have shown that children at risk of reading difficulties benefit from high levels of teacher-directed explicit language instruction but that this type of instruction makes no difference in decoding skills for children with already high skills on this dimension upon school entry. Highly skilled children, in contrast, make the strongest gains in classrooms with more child-led literacy-related activities.

A final reason for a focus on classrooms is that, with increasing focus on the design, implementation, and evaluation of prevention and intervention programs through experimental contrasts, classrooms are very often a location in which these programs are implemented (Greenberg, Domitrovich, & Bumbarger, 2001). It is our contention that such programs will not be entirely successful nor will they be theoretically informative without the full recognition, conceptually and methodologically, that everyday classroom interactions are the medium through which such programs will ultimately produce change. There is consistent evidence that the effects of school-based interventions rise and fall with how they are implemented within schools (Greenberg et al., 2001; Rones & Hoagwood, 2000). Due in part to early failures of some school-based intervention approaches, prevention science has increasingly focused on ways in which to conceptualize and measure classroom level implementation variables as critical to the success of any program of research in this area (Hamre et al., in press; Han & Weiss, 2005; Kam, Greenberg, & Walls, 2003; Valente, Unger, Ritt-Olson, Cen, & Johnson, 2006). These efforts are critical to the successful adaptation of developmental theory into educational interventions.

Current Developmental Theory and Educational Processes—Confluence and Deviations

As more developmental scientists move to study classrooms, there is a need for conceptual models that integrate decades of research on children’s development in home and peer contexts, with educational research. There are two basic approaches to this task, one which highlights the commonalities between developmental processes elucidated through the study of families and peers and their classroom equivalents and one which seeks to understand classrooms as a unique developmental context. Both approaches have made important contributions to our understanding of the role of classroom environments in children’s development.

The first approach views classrooms primarily as extensions of home environments. Thus, developmental processes known to be important causal agents in home settings, such as parent–child interactions, are transferred to the classroom environment. This view draws from ideas that there are underlying mechanisms driving development (Pianta, 2006). Much of the developmentally informed research conducted to date on classroom effects has taken this approach. For example, just as maternal sensitivity is a key mechanism of young children’s development (Ijzendoorn & Bakermans-Kranenburg, 2004; NICHD, ECCRN, 2005), teacher sensitivity has been shown to uniquely contribute to gains in students’ performance (Rimm-Kaufman, Early, & Cox, 2002). Effective classroom teachers have also been shown to share many of the same characteristics of Baumrind’s (1991) description of effective parents (Wentzel, 2002). Our own work (e.g., Hamre & Pianta, 2001; Pianta, 1999) and the work of others (Birch & Ladd, 1996, 1998; Silver, Measelle, Essex, & Armstrong, 2005; Van Ijzendoorn, Sagi, & Lambermon, 1992; Wentzel, chapter 6 this volume) has similarly focused attention on the ways student–teacher relationships mirror child–parent relationships, both in conceptualization and effects.

An approach assuming some underlying mechanism or contextual isomorphism as accounting for classroom effects and experiences is, however, insufficient to explain the entirety of classroom effects. The approach fails to recognize that classrooms and families are often oriented around very different goals and thus similar processes may function in different ways across home and school environments. For example, because teachers and administrators typically view their primary goal as producing academic gains rather than social, personal, or moral development, strategies for managing students’ behavior take on different meanings from the way behavior is managed within families. In classrooms, behavioral control serves primarily as a tool to help get as much instruction accomplished as possible. In contrast, families may use these practices with much broader goals in mind, such as helping children develop important self-regulatory skills. The consequences for children who challenge behavioral regulation or control strategies are also different. Teachers have options such as sending the child out of the classroom or, ultimately, out of the school, that are not available to or desirable for most families. Thus, although the broad concept of behavioral control may be similar across schools and families, the specific interactions, goals, and outcomes associated with this concept are likely to differ. The larger point here is that interactions in classrooms are likely to serve goals that are different from interaction systems in families even when they both draw upon phenotypically similar behavioral domains.

Beyond different goals, the classroom context itself and the interactions that occur within it can be very different from those within families. As just one example, Hughes, Cavell, and Wilson (2001) suggest that classrooms are a unique context in the development of peer relationships, in part, because teacher–student interactions serve as a potent and frequent source of information about other students that can figure prominently in children’s preferences and beliefs about others. Another set of differences between classrooms and families involves constraints on adult–child interaction. Unlike in families, physical contact and comforting is not a feature of teacher–student interactions in third grade or beyond, and although students often voice the importance of supportive relationships with teachers (Resnick et al., 1997), teachers themselves receive very little in the way of training or knowledge about forming such relationships. Another feature that differentiates classrooms and home settings in relation to developmental processes is the absence of genetic mediation in the interactions between people: in families the qualities of a parent’s behavior and a child’s response are linked in part by shared genetic substrates; in classrooms the common genetic mechanisms are not present and so the extent of a teacher’s interactions having an influence on child outcomes is somewhat different in terms of the nature of the mechanism explaining these effects.

A final distinction between home and classroom environments relates to the ways in which these environments change over time. Family environments are much more stable than classrooms. Within the United States, children typically change classrooms, teachers, and peer groups each year and go through major shifts as they move from preschool to elementary school to middle school and high school (Anderman & Anderman, 1999; Eccles & Midgley, 1989; La Paro et al., 2009; Midgley, 2002; Wigfield, Eccles, MacIver, Reuman, & Midgley, 1991). This lack of continuity offers both a challenge and opportunity to developmental scientists and can be an important entry point for links between developmental science and educational practices; Eccles’s work on middle school transition is one example (Eccles, 2004). Another opportunity is the case of “looping,” a practice involving students staying with the same peers and teacher for more than one year, rather than moving from year to year. There is almost no empirical research on looping (see Hampton, Mumford, & Bond, 1997 for an exception), but it offers a unique opportunity to examine the consequences of an educational practice with significant implications for understanding the role of relationships, both with peers and with teachers, in children’s development. Thus again, the changing nature of classroom contexts offers opportunities for developmentally informed research that exploits the natural variation present in these changes (e.g., looping or not; tracked classes and nontracked classes; having fewer or more teachers in a given grade).

In each of the examples cited above, there is a need to move beyond theories and measures developed in home contexts to ensure that the developmental contexts and processes most salient to the classroom environment are adequately captured. It is these adaptations and integrations of developmental theory and educational research that are a focus of this chapter. The examples cited above offer a glimpse of what could be learned through a more systematic exploitation and exploration of this “seam” between home and classroom and the processes that operate to promote children’s gains in and across both settings. The extent to which further research contributes to understanding development in educational settings is in part predicated on further specification of the nature and dimensions of classroom processes as unique in their own right and not a derivative of the literature on families. The next section provides a conceptual framework for understanding classroom environments that integrates years of research from developmental scientists and educational researchers.

The CLASS Framework

In the next section we review the three major domains of classroom environment described in the CLASS Framework, including a discussion of the relevance of the domain to educational practice and the developmental theories on which they are based. Within each broad domain, we review literature on specific classroom dimensions with empirically supported links to child and adolescent outcomes. It is beyond the scope of this chapter to provide an exhaustive review of research in each domain; however, the literature cited provides examples of relevant work from educational and developmental research.

Classroom Observations

There is a long history of classroom observation in educational research, although the extent to which classroom processes can be reliably and validly assessed has been a subject of great debate within educational research community for the past 30 years (Gage & Needels, 1989). Until recently, most standardized observations of classrooms were conducted as a part of research referred to as “process-product” research. Process-product research relied heavily on gathering frequency data on specific types of classroom interactions and occurrences and linking these to student outcomes. Brophy and Good (1986) offer a thorough summary of this research. Within the educational research community, this research was criticized for, among other things, reliance on decontextualized, discrete observations that did not accurately portray the complexity of classroom environments (Gage, 1989; Gage & Needels, 1989). The result of these criticisms was that very few educational researchers used standardized methods of observation during the 1990s, relying instead on qualitative observational methods.

More recently, developmental scientists have used their experiences in observation in home settings to create new observational measures of classroom environments that address some of the early critiques. Much of this work was stimulated by the NICHD Study of Early Childhood and Youth Development (SECCYD; NICHD, ECCRN, 2003a) which followed a cohort of children from birth through early adolescence. As children entered school settings, researchers realized a need to develop new observational tools. The measures used in the NICHD SECCYD proved successful in capturing a wide range of features of the classroom environment that were related to students’ social and academic development across childhood and early adolescence (Gazelle, 2006; Hamre & Pianta, 2005; NICHD, ECCRN, 2002, 2005; Pianta et al., 2006; Rimm-Kaufman et al., 2002). These measures extended previous work by measuring more complex patterns of interactions than can be captured with frequency counts of behavior. Other researchers have developed similar observational measures, some of which are specific to one domain of the classroom environment (e.g., Howes, 2000; Morrison & Conner, 2002; Patrick et al., 1997; Taylor et al., 2003) and others that are more global (Pianta et al., 2007; Pianta, Hamre et al., 2006; Pressley, Gaskins, Solic, & Collins, 2006; Stipek & Byler, 2004).

Researchers interested in conducting classroom observations are faced with a variety of measurement decisions which have important implications for resulting data. One of the issues that arises most frequently is the extent to which measures capture the frequency or quality of behaviors. Examples of behaviors assessed by frequency measures include: time spent on literacy instruction, the number of times teachers ask questions during instructional conversations, and the number of negative comments made by peers to one another. Quality measures may instead rate the degree to which literacy instruction in a classroom matches a description of evidence-based practices, how much instructional conversations stimulate children’s higher-order thinking skills, and the extent to which classroom interactions contain a high degree of negativity between teachers and students and among peers.

Frequency measures typically rely on time sampling methods while quality is assessed using rating systems; however, this distinction can sometimes be blurred. For instance, some measures that rely on assessing the presence of certain behaviors or interactions can focus on behaviors that have a quality component to them (e.g., Morrison & Conner, 2002). Thus, rather than measuring simply the number of questions a teacher asks, researchers interested in the cognitive demands of instructional conversations could, for example, count the number of closed questions (e.g., “What are the three parts of an insect’s body called?”) versus open questions that require higher levels of cognition (e.g., “Tell me two similarities and two differences between a spider and an insect?”). Similarly, many quality measures take into account the frequency of a group of behaviors in making the quality rating. For example, several measures make ratings based on the frequency with which a set of quality indicators are observed across a time period (NICHD, ECCRN, 2002; Pianta et al., 2007). So, for example, a teacher who has one brief interaction with a student constituting a high degree of teacher sensitivity, but who is quite insensitive for the rest of the observation period receives a lower score than does a teacher who is consistently sensitive across time and students.

There are advantages to each type of system. Quality measures assess higher order organizations of behaviors in ways that may be more meaningful that looking at the discrete behaviors in isolation and tend to parse the behavioral stream into more contextually and situationally sensitive “chunks.” For example, teachers’ positive affect and smiling can have different meanings and may be interpreted differently depending on the ways in which this affect is responded to by students in the classroom. In some classrooms teachers are exceptionally cheerful, but their affect appears very disconnected from that of the students. In other classrooms teachers are more subdued in their positive affect but there is a clear match between this affect and those of their students. A measure that simply counted the number of times a teacher smiled at students would miss these more nuanced interpretations. Another indication of the potential importance of observational measures focused on quality is that similar measures were critical to advancement in other areas of human behavior, such as the development of attachment theory (Ainsworth et al., 1978; Bowlby, 1969) and understanding antisocial behavior in youth (Dishion, Spracklen, Brown, & Haas, 1999).

One distinct advantage to using frequency measures, specifically when they are time-coded is the ability to conduct sequential and contingency analyses, thus allowing for studies of the complex interdependence and flow of interactions (Bakeman & Gnisci, 2006). Among the few researchers who have used this methodology in work on classrooms are those focused on the study of children with emotional and behavioral disorders. For example, Sutherland, Wehby, and Yoder (2002) examined the time-sequenced relationship between teacher praise and the provision of opportunities to respond among a group of children with problem behavior. Erickson, Stage, and Nelson (2006) examined the extent to which they could identify reliable antecedent and consequent events related to inappropriate behavior for children with ADHD diagnoses. Future work using these methods may help paint a more vivid picture of the complex interpersonal interactions among teachers and peers in their everyday interactions with one another.

Student and Teacher Questionnaires

The ultimate effect of any observable classroom process on student outcomes is largely mediated by the ways in which individual students make meaning of and respond to it. Therefore, a complete understanding of these environments requires gaining access to the participants’ perceptions of these environments. Questionnaires designed to assess the classroom environment have been used in educational research for over 30 years. Fraser (1998) provides a helpful review of the most commonly used classroom environment instruments. Four major issues related to the use of questionnaire data on classroom environments are: the domains of classroom environment on which they focus; whether the reporter is the teacher or students; the age of student reporters; and the extent to which they measure students’ personal or classroom level perceptions of the environment.

Notable among questionnaires on the classroom environment is that they focus almost exclusively on the emotional and organizational aspects of classrooms, with few if any examples of measures providing data on students’ perceptions of availability and quality of instructional supports. For example, the most commonly used measure, the Classroom Environment Scale (CES: Moos & Trickett, 1987), has scales measuring involvement, affiliation, teacher support, task orientation, order and organization, and rule clarity. One exception comes from work by Midgley and colleagues (1996) using the Patterns of Adaptive Learning Survey. Several items on this survey assess the degree to which teachers report using strategies that encourage understanding, rather than rote knowledge (Deemer, 2004).

Issues related to the reporter on questionnaires include differences between teacher and student report and the age of students. Most classroom environment questionnaires include both teacher and student report versions; however, research examining relations between teacher and student report have found low to moderate associations (Feldlaufer, Midgley, & Eccles, 1988; Fraser & O’Brien, 1985). In general, teachers report more positive environments than do students (Fisher & Fraser, 1983; Fraser & O’Brien, 1985), so capturing student report data allows for researchers to document the range of students’ experiences within the classroom.

Questionnaire methods are much more common in research on middle and secondary classrooms than on elementary classrooms. This is not surprising given the challenges of reliably assessing young children’s perceptions; however, there is growing evidence of effective practices for obtaining self-reports from children as young as 5 (Daniels, Kalkman, & McCombs, 2001; Measelle, John, Ablow, Cowan, & Cowan, 2005). One of the few examples of work on young children’s perceptions of classroom environments comes from Daniels, Kalkman, and McCombs (2001) who used an interview technique with cartoonlike depictions of classroom situations to elicit kindergarten through second grade students’ feelings about their classroom. This measure was validated against external judgments about the classroom with evidence of some agreement, but clearly there is a great need for more work in this area.

One final distinction of relevance for those selecting questionnaires to assess the classroom environment is the extent to which they assess students’ personal or classroom experiences (Fisher & Fraser, 1983; Fraser, Giddings, & McRobbie, 1995; Fraser & Tobin, 1991). Many classroom environment measures have both types of versions. Personal versions ask students questions specifically about their experiences in the classroom (e.g., “My teacher treats me fairly”), while classroom versions of measures ask questions about students’ perceptions of classroom level processes (e.g., “The teacher tends to treat students fairly”). A direct comparison of these two types of measures offers several important distinctions (Fraser et al., 1995). For example, there are greater gender differences when using the personal versions of the measures and students report more favorable environments using the classroom as compared to the personal version. However, each version accounts for independent variance in student outcomes, suggesting they both may be important aspects of the classroom environment to assess (Fraser & McRobbie, 1995).