Modeling growth in arm circumference of infants in Jimma Town, south west Ethiopia
AbstractThe paper models the growth in arm circumference of 450 infants during their first year of life. The model is based on longitudinal data obtained from Jimma, a town in southeast Ethiopia. A linear mixed model was employed to see how arm circumference of these infants change over their first year of life. A polynomial model with third degree term is found to be significant. The Loess smoothing technique is applied to check whether our model fit the data set well. The fitted arm circumference is very close to the smoothed curves. Three dropout models are fitted and compared using the likelihood ratio statistic. The results favour the Missing at Random (MAR) process. This makes the likelihood-based analysis justifiable. Our final model is dominated by the linear term early in life, with contribution of the quadratic and cubic terms as the child ages a few months. This indicates that arm circumference growth stops for several months, and possibly there are losses in arm circumferences. This phenomenon can be attributed to malnutrition.
Keywords/phrases: Loess, MAR, MCAR, MNAR
SINET: Ethiop. J. Sci Vol.26(1) 2003: 1-10