the within-person intake of nutrients. For some nutrients, more within-person (or day-to-day) variation than between-person variation may occur. Vitamin A is a good example of this. Intake can be 5,000 retinol equivalents (RE) on a day when the individual snacked on carrots, and close to 0 RE on another day when few fruits, vegetables, and dairy products were consumed. Thus, for some nutrients, the day-to-day variability in intakes for an individual may be larger than the between-person variability in the group. For vitamin A, the within-person variability in intakes may be as much as six times larger than the between-person variability in intakes in typical North American dietary data (Basiotis et al., 1987). For other dietary components such as energy, the day-to-day variability in intakes is about as large as the between-person variability in intakes in the group (Basiotis et al., 1987; Beaton et al., 1983; Guenther et al., 1997; Liu et al., 1978; Looker et al., 1990; NRC, 1986; Nusser et al., 1996; Sempos et al., 1985). This means that if the aim is to estimate the usual intake distribution of a nutrient in a group and have its spread reflect only the between-person variation in intakes, then statistical methods that help reduce this nuisance variance must be used.

Not only do individual intakes differ from day to day, as discussed above, but also how much they differ varies from one person to another. In addition, this variability is not completely random. Individuals with higher average intakes also tend to have more variable intakes than do individuals with lower average intakes (Nusser et al., 1996).

For most nutrients, the distribution of observed mean intakes (and presumably, the usual intake distribution as well) is skewed to the high end rather than being symmetrical. This is particularly true when intakes from supplements are included in the diet. Consider calcium as an example. Mean intake in a group might be 600 mg/day. Very few people would have intakes 500 mg or more below the mean (and it would be impossible to have an intake more than 600 mg below the mean), but there could easily be people in the group consuming intakes 500, 1,000, or even 1,500 mg above the mean. Therefore, the intake of this nutrient has a skewed, asymmetrical distribution. Because most nutrients have skewed, asymmetrical