Stability of Children’s Classroom Behaviors Across Time and Settings

• In Event: Within Classroom Variation is not Noise: Predictors and Impacts of Varied ECE Experiences

Abstract

Within the same classrooms, children can have very different experiences. In fact, prior work has demonstrated that there can be more variability between individual children in a classroom than between classrooms themselves with regard to children’s involvement levels, social-learning interactions, and unoccupied– disruptive behaviors, with 52.5%, 56.6%, and 84.2% of the variance occurring within classrooms, respectively (Nesbitt, Farran, & Fuhs, 2015). Some behaviors appear to be more a function of within-child variation while others vary as a function of the type of pedagogical setting provided (e.g., whole group vs. centers). Knowing better which types of behaviors are which will aid policy makers as well as practitioners.
Generalizability Theory (Cronbach, Gleser, Nanda, & Rajarathnam, 1972), allows for the examination of multiple sources of variance simultaneously to understand the reliability and generalizability of a measurements protocol and has the potential to identify potential sources of variability in measurement. Moreover, it allows researchers to identify approaches to improve the dependability of the data rather than just indicating whether a threshold of acceptable error has been met. The goal of this paper is to apply Generalizability Theory to examine the impact of potential sources of measurement error/variance on 1108 prekindergarten children’s classroom experiences. Using a snapshot coding system that captures children’s and teachers’ behaviors across the entire school day, this study aims to estimate the degree to which measurement of children’s engagement levels, social learning interactions, and off-task behaviors are impacted by the facets of occasion (fall, winter, spring) and setting (teacher-directed, child-directed instruction).
As summarized in Table 1, the greatest variance for the percentage of sweeps children spent speaking, in social learning interactions (associative and cooperative interactions), and off-task behavior (unoccupied, disruptive, or timeout) was attributed to the interaction between child specific differences and settings (16.95% to 25.76% of variance). As such, collapsed across occasion, there was great variability between individual children in the degree to which these classroom behaviors differed across setting (teacher vs child-directed) with some children showing similar levels across settings while others varied across setting. Interestingly, the degree to which talking, social learning, and off-task behavior occurred was relatively consistent across the three occasions, including no difference in setting effects or child rankings across fall, winter and spring. For level of involvement, on the other hand, the occasion of the observation did have an effect when considered in relation to child-level difference and differences associated with setting. As illustrated in Figure 1, across all occasions, children were observed as having higher levels of involvement in child-directed settings compared to teacher-directed; however, there was a significant decrease in involvement in child-directed instruction across the course of the year. These results highlight the importance of capturing children’s classrooms behaviors across instructional settings and occasions of the school year.
For the presentation, implications of these findings for early childhood education practice and policy will be discussed with a particular focus on the utility of Generalizability Theory has for the assessment of children’s classroom experiences.

Authors

• Kimberly Turner Nesbitt, University of New Hampshire
  Presenting Author

• Dale C Farran, Vanderbilt University
  Non-Presenting Author