Could Differentiated Instruction within Classrooms Explain Pre-K Convergence?

Background: Pre-k convergence refers to the tendency for the positive effects of pre-k on children's academic skills to diminish over time, observed across decades of observational and experimental studies (Ansari et al., 2020). Researchers have tested multiple mechanisms to explain convergence, with mixed results; none of the proposed mechanisms (e.g., sustaining environments, kindergarten readiness, pre-k quality) fully explain the phenomenon (Bailey et al., 2017). That research has largely examined school or classroom effects on convergence, but teachers may intentionally expose children to different experiences "within" classrooms, based on their behavior or skills (Deunk et al., 2018). One key way that teachers differentiate instruction is through small groups and intervention time (individual time with a student) (Campbell, 2014). In this analysis we hypothesize that, in a diverse, low-income population, pre-k attendance may be associated with smaller academic gains in kindergarten because teachers allocate their time based on students' skills (achievement and positive task orientation), leading them to spend the more time with children who had no pre-k prior to school entry and who therefore start kindergarten with lower school readiness (see the conceptual model, Figure 1). Prior analyses with this sample have established that pre-k attenders entered kindergarten with higher academic skills and higher task orientation and experienced substantial convergence during kindergarten (Ansari et al., 2020). Research Questions: (1) Does school readiness predict time spent with teachers in small group or individual settings? (2) Does the proposed pathway (pre-k attendance [right arrow] school readiness [right arrow] time with teacher [right arrow] academic gains) account for lower kindergarten growth among pre-k attenders compared to non-attenders? Setting: The study sample was drawn from a large, culturally and linguistically diverse school district in the mid-Atlantic serving over 150,000 students from pre-K through 12th grade. The district serves a rapidly growing percentage of vulnerable families and children, including a large immigrant population. Although the county overall is affluent, one third of children in the school system qualify as low-income and 25% receive public assistance. The division works with schools and community-based preschools to provide need-based public pre-k to nearly 2,000 children annually. Participants: Pre-k attenders (51.7%) were recruited from public preschools at the start of their 4-year-old pre-k year. Non-attenders (48.3%) were demographically similar students who attended kindergarten in the schools into which our pre-k sample transitioned, and were recruited at the start of kindergarten. The sample includes 2,631 children from 482 kindergarten classrooms. Children were Hispanic/Latine (60.8%), Black (12.3%), White (10.8%) and other races/ethnicities (14.8%). Parents reported speaking Spanish (58.5%), English (23.7%), and other languages (23.7%) at home. Program/Practice: We examined naturally occurring variation in teachers' use of small group and individual time with students. Working more intensively with children who are struggling (while leaving other children to work independently) is one of the key ways that teachers differentiate instruction (Deunk et al., 2018). Research design: Data for this analysis is drawn from a longitudinal observational study, following a cohort of children from pre-k or kindergarten through 5th grade. This analysis uses data from the kindergarten year. Data Collection and Analysis: Children were directly assessed and rated by teachers in the fall and spring and observed multiple times throughout the kindergarten year. Direct assessments included the Woodcock-Johnson III Tests of Achievement (Schrank et al., 2001; letter-word identification, picture vocabulary, applied problems, and quantitative concepts). Teacher ratings of task orientation were drawn from the Teacher-Child Rating Scale (Hightower et al., 1986). Both measures have well-established reliability and validity in studies of young children. Individual child observations were conducted using the Behavioral Coding System (BCS), adapted from the NICHD Study of Early Child Care and Youth Development Classroom Observation System (McCartney et al., 2007). The BCS captures group setting and teacher presence in 30-second intervals across the school day. For this analysis, we tallied the proportion of intervals each child spent in teacher-led small group and teacher-led individual work, separately. Analyses used MPlus to test multilevel path models (children within classrooms) and indirect effects, with separate models for each academic outcome. Analyses controlled for child demographic characteristics and special education services. Models used FIML to address missing data, and all models' fit was acceptable. Results: We present results from the model predicting letter-word identification (LWID); all models followed similar patterns. As we knew from previous analyses, pre-k attendance predicted higher fall LWID (B = 0.202, SE = 0.021, p < 0.001) and task orientation (B = 0.101, SE = 0.025, p < 0.001). Children with higher LWID spent significantly less time in small group and in individual time with their teachers (B = -0.116, SE = 0.034, p = 0.001 and B = -0.071, SE = 0.03, p = 0.016, respectively). Children with higher task orientation spent less time in individual work with their teachers (B -0.206, SE = 0.038, p < 0.001). These three indirect effects from pre-k to small group and individual time were significant (see Table 1). Individual time with teachers significantly predicted gains in LWID, but the effect was negative (B = -0.05, SE = 0.019, p = 0.008) suggesting that more time with teachers was associated with slightly smaller gains from fall to spring. Small group was not associated with gains. The full indirect effect (pre-k to task orientation to individual time to spring scores) was very small but significant (B = 0.001, SE = 0.001, p = 0.046) and suggested that pre-k attenders benefitted academically in that they spent "less" individual time with teachers. Conclusions: These results affirm that teachers allocate time based on children's academic and behavioral skills, and that pre-k students, having stronger school readiness at school entry, receive slightly less time with teachers across the year. Unexpectedly, this "benefitted" children because spending more individual time with teachers was associated with lower achievement gains. It may be that we need more specific, descriptive data on how teachers use their time with students in small group and individual work. We will discuss implications of this work for future research on pre-k convergence.