The predictive value of early mathematical skills on subsequent mathematical performance

Introduction

Identifying students with mathematical difficulties at an early stage is important to be able to start early with intervention. Little is known about what specific mathematical skills can predict later mathematical performance. Although earlier meta- analyses show that later mathematical performance can be predicted, in these studies only 16 to 23% of the variance was explained by differences on the baseline measurement. Former meta-analyses are outdated and these studies did not examine the predictive value of different specific mathematical skills. Therefore meta-analysis on the predictive value of different early mathematical skills has been carried out. Early mathematical skills are skills on general understanding of the meaning and functions of numbers in combination with skills that rely on spatial orientation on subsequent mathematical performance. De research question in this study is: “To what extent are there differences in the predictive value of different early mathematical skills on later mathematical performance?”

Method

The study is a meta- analysis. The early mathematical skills have been divided into three domains after the classification of the Stichting Leerplanontwikkeling Nederland (SLO) (Foundation Learningplan Development): number sense, measurements and geometry, to be able to compare differences. Studies published between the 1^st of March 1998 and 21^st of August 2016 were included. The weighted correlations between mathematical skills in grade one and two and the mathematical skills in grade three or higher. Studies were excluded when participants had impairments other than language-, mathematical-, or learning disabilities or if the study was an intervention study. 334 correlations from 32 articles have been included in this study.

Results

The mean difference in months between the baseline and post- measurement was 17 months (standard deviation was 13). When comparing the predictive values of the different domains, the 5 number sense and the geometry domain seem to predict the subsequent mathematical performance equally well. The domain number sense explains about 13.7% of the variance and the domain geometry explains about 13% of the variance. For the measurement domain only one correlation has been found with a value of .54. The predictive value for this domain is strong (explained variability of 29.2%).

Conclusion

Some caution is needed in the interpretation of this value because it is based on 1 study of 43 children, which may cause an unreliable presentation of the reality. Only 13.1 up to 13.5% of the variance in subsequent mathematical performance is explained by differences in early mathematical skills for the number sense and the geometry domain. This suggests that these early mathematical skills are moderate predictors of subsequent mathematical performance and caution is required while interpreting the test scores of young children. A possible explanation is the large age difference between baseline and post- measurement. A follow-up study on the predictive value of the measuring domain on later mathematical performance is recommended.

Keywords: (vroegtijdige) rekenvaardigheid, rekenprestatie, voorspellen, getalbegrip, meetkunde, meten

By: Andrea Sijp

Supervised by: N. Frans, MSc, dr. W.J. Post en dr.A.M.N. Huyghen