Data Analysis Strategy for Nutritional Survey of 642 New York City School Children

Publisher Website
Google Scholar
Abstract
A special questionnaire served as a source for 155 categories of nutritional, clinical, dietary, sociologic, and laboratory information on 642 New York City school children. These data were processed by means of various tabulating and computer procedures. The laboratory values for thiamine, nicotinic acid, riboflavin, biotin, pantothenic acid, folic acid, vitamin B12, vitamin B6, vitamin A, β-carotene, vitamin C, vitamin E, triglycerides, and cholesterol were grouped in histograms and plotted on equal and logarithmic-interval probability paper. Only vitamin A, β-carotene, and vitamin C were Gaussian in distribution; the other vitamins and circulating lipids studied were found to be lognormal in distribution. Means, standard deviations, 95% confidence limits, t tests, P values, and correlation coefficients were determined; a correlation matrix was constructed. Cluster analysis helped reveal certain interrelationships in this multivariate problem. Welfare assistance, low reading levels, decreased skin-fold thickness, low circulating vitamin C, nicotinic acid, and B12 were seen in one cluster of Puerto Rican children. Another cluster of Puerto Rican children, showing below average size and weight, had poor diets, and came from families in which the mother was the principal wage earner. Chinese children with above average reading scores had good dietary intakes, high circulating thiamine and B6, and low levels of fat-soluble vitamins and cholesterol. Within a cluster the pattern of relationships between variables was often quite different than that for the total population.