Poster #18 - Number Line Estimation Patterns in Children at Risk of Mathematics Learning Difficulties

• In Event: Poster Session 18

Abstract

Background: Children with mathematics learning difficulties (MLD) often show less accurate performance on the number line task (Geary et al., 2012; Lafay et al., 2017; Van’t Noordende et al., 2016), and their estimation patterns tend to be less linear compared to their typically developing peers (Geary et al., 2008; Sella et al., 2013). However, limited evidence exists about: i) the evolution of estimation patters in children with MLD that include models arising from the representational shift (Siegler & Opfer, 2003) and proportional judgment accounts (Barth & Palladino, 2011), and ii) how this relates to performance in other mathematical skills.

Aims: The first aim is to better understand the development of number line estimation patterns throughout the first two years of primary school. The second aim is to determine if and how children’s mathematical skills in the first two years of school differ based on the mathematical function that best explains their estimation patterns at entry to school.

Method: We followed 100 Singaporean children (Mage = 83.63 months) for the first two years of primary school. At school entry, the children had been identified as at risk of presenting MLD. Children completed a 0-100 number line task and three mathematical subtests (Math Fluency, Numerical Operations and Math Problem Solving) from the WIAT III (Wechsler, 2009) at four timepoints.

Analyses: To achieve the first aim, we fit children’s estimates an at individual level following Slusser et al. (2013) and consider models arising from the representational shift (i.e., linear and logarithmic) and proportional judgment accounts (i.e., unbounded, one-cycle and two-cycle power models) separately. To address the second aim, we conducted Bayesian ANCOVAs in JASP (JASP Team, 2022). For all analyses, the variable indicating best fitting model at the first timepoint was used as a fixed factor. We studied possible differences in mathematical skills at the first and last timepoint.

Results: Within the representational shift account, we found that the linear model is the most frequent best fitting model across all timepoints and the number of children best fit by this model increases across timepoints. Children whose estimates were best fit by the linear model outperformed those best fit by the logarithmic model in all skills at the first and last timepoint. Within the proportional judgment account, the unbounded power model is the most frequent best fitting model across all timepoints, followed by the one-cycle power model. The number of children whose estimates were best fit by the two-cycle power model increased over time, although there are no clear trends that show a gradual increase in the use of reference points. We found evidence against differences in mathematical skills at the first and last timepoints based on best fitting model (i.e., unbounded, one-cycle, and two-cycle power models) at the first timepoint.

Discussion: Clearer longitudinal trends and stronger evidence for differences in mathematical skills based on estimation patterns were found within the representational shift account. Our findings are discussed in the light of methodological limitations in the study of estimation patterns.

Authors

• Carola Ruiz, Macquarie University
  Presenting Author

• Saskia Kohnen, Macquarie University

• David Muñez

• Rebecca Bull, Macquarie University