SRCD 2019 Biennial Conference
March 21-23, 2019
Title: Characterizing Working Memory Growth Trajectories throughout Early Schooling (Poster)
Session: Attention, Learning, Memory
Authors: Adrienne Woods, Yanyan Wang, Paul Morgan, George Farkas, Marianne Hillemeier
Abstract: Working memory (WM), one of the foundational subcomponents of executive functions (EF) for young schoolchildren, refers to the ability to briefly maintain and process information without external assistance. WM has been reported as an especially strong predictor of children’s achievement across multiple academic domains. Children with WM deficits may experience difficulties in comprehending text, following multi-step instructions, or effectively using strategies to solve mathematics or science problems. As problem-solving skills become increasingly salient throughout elementary years, children with WM deficits may increasingly struggle in classrooms. Thus, it is vital to develop effective methods for identifying children with WM deficits across the early schooling years. Prior work suggests that the developmental trajectories of children’s WM follow a linear fashion from preschool to middle school before leveling off around age 15. By the age of six, WM is already sufficiently developed to conduct complex WM tests (e.g., backward digital recall). Yet few empirical studies have examined young children’s WM developmental trajectories using large and nationally representative samples with strong statistical controls.
We used data from the ECLS-K:2011 to classify WM growth trajectories among 12,000 students between Kindergarten and spring of 4th grade using latent class growth analysis (LCGA). LCGA is a form of finite mixture modeling in which participants are grouped into maximally-similar clusters according to their growth curves. WM was assessed using the numbers reversed task, in which children were asked to orally repeat increasingly longer strings of numbers in reverse order. Following the identification of trajectories, we conducted a multinomial logistic regression (MLR) to investigate the probability of membership in each trajectory using a number of Kindergarten demographic, achievement, and behavioral variables.
Though analyses are ongoing, preliminary results reveal 5 growth trajectories for WM (Figure 1). Most students follow relatively similar trajectories in which steady growth over time is observed. About half (48.5%) of students display “lower” WM scores in the fall of Kindergarten (classes 2, 4, and 5). Most of these students (44.4%) experienced steeper growth through spring of 4th grade, effectively catching up to their classmates, but 4.1% of students retained a low WM score during this time (class 5). MLR results comparing the probability of membership in classes 1-4 relative to this low-performing class 5 are displayed in Table 1. This group appears to be characterized by students who are older at Kindergarten entry and display poorer cognitive flexibility, mathematics skill, and learning-related behavior during Kindergarten. The students with the consistently best WM scores (class 1) lived in higher-SES households in Kindergarten, and White, Asian, or a race/ethnicity other than Black/Hispanic. There were relatively fewer differences between classes 4 and 5, but these classes were comprised more of Black and/or male students who demonstrated poorer inhibitory control relative to classes 1-3. In the future, we plan to further probe these groups in the hopes of understanding why some students “catch up” to peers and why about 4% of children continue to display poor working memory through 4th grade.
Comments are closed.