What Applying Growth Mixture Modeling Can Tell Us About Predictors of Number Line Estimation

Jeffrey M. DeVries, Jörg-Tobias Kuhn, Markus Gebhardt


Number line estimation tasks have been considered a good indicator of mathematical competency for many years and are traditionally analyzed by fitting individual regression curves to individual responders. We innovate on this technique by applying growth mixture modeling and compare it to traditional regression using a sample of 2nd graders (n = 325) who completed both 0–20 and 0–100 number line tasks. We explore the effects of gender, special education needs, and migration background. Using growth mixture modeling, more children were identified as logarithmic responders than were identified using regressions. Growth mixture modeling was able to identify the significant effects of gender on class membership for both tasks, and of special education needs for the 0–20 task. Overall, growth mixture modeling provided a more complete picture of individual response patterns than traditional regression techniques. We discuss the implications of these findings and provide recommendations for future researchers to use growth mixture modeling with future number line task analyses.


number line estimation; growth mixture modeling; mixture modeling; latent growth modeling; special education needs; migration background

Full Text:

Downloads: 84

Copyright (c) 2020 DeVries; Kuhn; Gebhardt