DISCRIMINANT ANALYSIS AS A TOOL FOR UNDER FIVE ANTHROPOMETRIC CLASSIFICATION IN GHANA

Abstract

Under five years malnutrition remains an important public health and development problem in developing countries. In this dissertation, anthropometric indicators (heightfor-age, weight-for-height and weight-for-age) of children(less than 5 years) are used as indicators for classifying the nutritional status of children. The study captured 2992 children under five in Ghana from the 2008 Ghana Demographic and Health Survey (GDHS) data set. The multivariate technique of discriminant function analysis was used to classify the nutritional status using selected variables consisting of measured attributes of children and their mothers. The classification function for the discriminant analysis classified correctly 77.9%, 92.4% and 88.4% of the group cases respectively for stunted, wasted and underweight classifications. Logistic regression was used as an investigative tool. The study further identified that the probability of a child being chronically malnourished is certain (1) if the child has low score of BMI. Results from logistic regression, showed that if repeated samples are taken, we are 95% confident that between 0.991 and 0.996 of the malnourished children will be classified. The study identified the body mass index of the child as the major determining factor in classifying the nutritional status of children under five in Ghana. On the basis of the analyses it was recommended that sufficient controls should be exercised in the discriminant function of the three nutritional categories.