error from within-person variation, but the size of this error has not been quantified.
It is clear that estimating an individual's usual intake for a nutrient from the individual's observed intake alone may lead to an under- or overestimation of that individual's usual intake of the nutrient. However, it is still possible to evaluate the potential error if something is known about the magnitude of the within-person variation in intakes for that nutrient. The individual's observed mean intake is the best estimate available of the individual's usual intake of the nutrient. A pooled estimate of the within-person variability in intakes has been computed for a number of nutrients from nationwide food consumption surveys (see Appendix Table B-2, Table B-3, Table B-4 through Table B-5). The magnitude of the day-to-day variation in intakes of a nutrient will indicate whether the observed mean intake calculated from a few daily records or recalls is a more or less precise estimator of the individual's usual intake of that nutrient. The observed mean intake and the pooled estimate of day-to-day variability in intakes will be used subsequently to guide individual dietary assessments.
The second step in individual assessment is to choose the appropriate DRI to use as a reference standard. In assessing the apparent adequacy of an individual's intake, interest is in whether the individual's nutrient requirement is met. Unfortunately, information on an individual's requirement is seldom, if ever, available. Therefore, the best estimate for an individual's unobservable requirement is the Estimated Average Requirement (EAR), defined as the median requirement of a nutrient for a given life stage and gender group. Obviously there is variation in requirements among individuals, and assumptions have been made about the shape of the requirement distribution. A coefficient of variation (CV) (standard deviation of the requirement divided by the mean requirement × 100) of 10 percent has been assumed for most of the nutrients for which EARs have been established (IOM, 1997, 1998b, 2000). If requirements are normally distributed, a CV of 10 percent means that about 95 percent of individuals would have requirements between 80 and 120 percent of the EAR (± 2 standard deviations). With a CV of 15 percent, as has been estimated for niacin (IOM, 1998b), the corresponding range would be between 70 and 130 percent of the EAR. For some nutrients the CV of the requirement distribution may be even higher, and for other nutrients (e.g., iron requirements of