Validity of Teacher-Child Interactions in Primary Grades: Testing Alternative Classroom Assessment Scoring System (CLASS) K–3 Models

• In Event: Roundtable Session 50
  In Roundtable Session: Advancing the Science of Classroom Observation: New Applications and Directions From the Field

Abstract

Objectives: This study examined the internal structure of the Classroom Assessment Scoring System, K-3 with elementary classrooms. Confirmatory factor analysis was used to test the structural validity of the original CLASS model (Pianta, La Paro, & Hamre, 2008) and five alternative structures.

Perspective/Theoretical Framework: Prior research has indicated that interactions between teachers and students may be critical for students’ academic and social-emotional outcomes (e.g., McCormick & O’Connor, 2014; Rudasill et al., 2010). Empirical work on this topic has identified systematic classroom observations as a practical and effective way to assess these interactions (Zaslow et al., 2012). The CLASS K-3 (Pianta et al., 2008) is one such observational instrument that has been widely used in both research and practice to measure teacher-student interaction quality. The primary domains assessed within CLASS K-3 are Emotional Support, Classroom Organization, and Instructional Support. The structural validity of CLASS has been tested across pre-kindergarten and elementary samples, and each study has identified some variations in the best-fitting structural model from those that preceded it. Additionally, researchers have used thresholds for fit that vary in the stringency of criteria to determine best-fitting models. Thus, testing these various models on the same sample with consistent fit criteria would allow for direct comparisons of the previously proposed structures, which was the goal of the current study. The original CLASS model (Hamre et al., 2007; Pianta et al., 2008), modified structures identified by Pakarinen et al. (2010) and Sandilos et al. (2014), and two bifactor structures proposed by Hamre et al. (2014) and Jones et al. (2013) were tested with the current sample.

Methods: This study utilized baseline observation data from the Social Skills Improvement System – Classwide Intervention Project (DiPerna & Lei, 2009). A total of 141 first and second grade classrooms across seven elementary schools in the mid-Atlantic region of the U.S. were examined.

Data Sources/Evidence: The majority of the participating teachers were female (89.4%) and Caucasian (97.2%). The average years of teaching experience was 14.4 (SD = 9). The CLASS K-3 is a structured observation system where trained observers rate teachers on 10 dimensions on a 7-point scale. For the current study, CLASS internal consistency (.81-.92) and percent interrater agreement (91-99%) were high. Descriptive statistics are reported in Table 2.1. The fit of the six structures are reported in Table 2.2. Factor loadings for each model are reported in Figures 2.3-2.5. Although all but one model yielded some fit indices that met a priori thresholds, the best-fitting model was a modified version of CLASS which loaded Regard for Student Perspectives onto both the Emotional Support and Instructional Support domains.

Scholarly Significance: The evidence for structural validity presented in this paper and other studies demonstrates a fair amount of consistency and continues to support a three-domain, 10-dimension structure of CLASS. Across the pre-kindergarten and early elementary samples, some dimensions and domains exhibit less stability than others. Potential reasons and implications for the structural nuances of the CLASS model across grades and samples will be discussed in the presentation.

Authors

• Lia Elaine Sandilos, University of Virginia

• Sarah Wollersheim Shervey, University of South Dakota

• James C. Diperna, The Pennsylvania State University

• Weiyi Cheng, The Pennsylvania State University

• Pui-Wa Lei, The Pennsylvania State University