Abstract
1. An alternative statistical technique, based on analysis of the 25th percentile, was used to examine teenagers' diets. The technique was developed to identify personal characteristics that affected the prevalence of low energy and nutrient intakes. It provided useful information not obtainable from methods based on average intakes.2. The study sample was a representative sample of 1055 high-school students in Tasmania, Australia, aged 11–16 years.3. Among girls, different characteristics had the greatest influence on the prevalence of low intakes (‘low’ being defined as below the 25th percentile for girls). Fatness had the greatest influence for several dietary components. Fatter girls were more likely to have low intakes of energy, carbohydrate, iron, calcium and niacin-equivalent, and also of ‘empty-energy’ foods. School type had the greatest influence for fat, riboflavin and vitamin A. Exercise had the greatest influence for protein, low intakes being more usual among the least active girls.4. Among boys, height and age generally had the greatest influence on the prevalence of low intakes (‘low’ being defined as below the 25th percentile for boys). Taller and older boys were less likely to have low intakes. This was attributed partly to increased physiological needs, but peer-group influences appeared more important in relation to food choices. Thiamin was unusual, in that recent use of analgesics had the greatest influence on the prevalence of low intakes.