Related Experiment Videos
Using a nonlinear mixed model to evaluate three models of human stature
E P Susman1, J R Murphy, G O Zerbe
1Department of Psychology, Metropolitan State College of Denver, CO 80217-3362, USA.
Growth, Development, and Aging : GDA
|April 29, 1999
Summary
A mixed model approach enhances human stature growth models by incorporating individual variability. This statistical method improves model fit and provides a parsimonious way to estimate population average growth parameters.
Area of Science:
- Biometrics
- Human Growth and Development
- Statistical Modeling
Background:
- Human stature development is complex, with significant inter-individual variability.
- Existing nonlinear growth models often do not adequately account for this variability.
- Mixed models offer a statistical framework to integrate fixed and random effects.
Purpose of the Study:
- To evaluate and compare three nonlinear human stature development models using a modern mixed model approach.
- To assess the impact of incorporating random effects on model fit and parameter estimation.
- To determine the best-fitting model for male and female stature data.
Main Methods:
- Application of the mixed model approach using the NLINMIX Macro in SAS.
- Evaluation of three established nonlinear growth models: Preece and Baines (1978), Jolicoeur et al. (1988, 1991, 1992), and Kanefuji and Shohoji (1990).
- Analysis of height data from 28 males and 25 females, incorporating two random components alongside fixed mean parameters.
Main Results:
- The inclusion of random parameters consistently improved the fit of all evaluated growth models.
- The Jolicoeur et al. model demonstrated superior performance for male stature data.
- The Kanefuji and Shohoji model provided the best fit for female stature data.
Conclusions:
- The mixed model approach offers a more parsimonious and statistically robust method for analyzing human stature growth.
- This approach effectively accounts for individual variation, leading to improved population average growth models.
- Model performance varied by sex, highlighting the need for sex-specific growth modeling.