Malaysian Journal of Mathematical Sciences, May 2015, Vol. 9, No. 2
Hierarchical Multiple Regression Model in Obtaining Body Weight Equation from Anthropometric Measurements
H. J. Zainodin and S. J. Yap
Corresponding Email: zainodin@gmail.com
Received date: -
Accepted date: -
Abstract:
Body weight has long been used as a parameter for calculation in many medical procedures. But when body weight is not available and its visual weight is inaccurate a body weight equation for both genders from anthropometric measurements is developed. This is the main aim of the study besides identifying the significance of interaction variables in hierarchically multiple regression analysis. The interaction variables involved here are up to the fifth-order (product of 6 independent variables). Here, the equation is developed using hierarchically multiple regression analysis. A modified method on the Zainodin-Noraini multicollinearity remedial method is proposed in this work to remedy the multicollinearity problem. Then, coefficient test is carried out on these models to eliminate insignificant variables from models that are free from multicollinearity problem. Ultimately, the equation developed in this work is free from multicollinearity problem and insignificant variables. The proposed modified method is found to be easier, time-saving more accurate and less prone to errors. An important finding in this work is that the interaction variables are significant and is best included in statistical analysis in order to yield a better prediction.
Keywords: Body weight equation, hierarchically multiple regression models, interaction variable, insignificant variable, multicollinearity effect, modified method
Indexing
