used to assess the prevalence of inadequate dietary intakes. It is far preferable to use the EAR cut-point method and the adjusted distribution of usual intakes to estimate the proportion of a group with inadequate intakes.

In the preceding discussion, the unit of observation implicitly assumed in the dietary assessment is the individual. What if the unit of observation is either the household or the population? Consumption data are frequently gathered for households rather than for individuals. Disappearance data (or food balance sheets) may be collected for a group or an entire population such as a country. However, published requirement estimates usually are related to individuals. For dietary assessment applications, however, estimates of nutrient requirements and nutrient intakes must be at the same level of aggregation: individual, household, or population. Appendix E suggests approaches for evaluating dietary adequacy when the unit of observation is not the individual.

Assessing the proportion of a group or population that is at risk of nutrient inadequacy is an important public health and policy concern. The Dietary Reference Intake (DRI) that is relevant to this type of assessment is the Estimated Average Requirement (EAR). The probability approach, described by the National Research Council (NRC) in 1986, permits an estimation of the prevalence of inadequacy within a group by comparing intakes with the distribution of requirements. This method assumes that the correlation between intake and requirement is low and that the distribution of requirements is known. A shortcut to the probability approach—the EAR cut-point method—allows determination of the prevalence of inadequacy in a group by determining the number of individuals with intakes below the EAR. Like the probability approach, the cut-point method assumes that the correlation between intake and requirement is low and that the variability in intakes is greater than the variability of requirements. However, unlike the probability approach, the cut-point method does not require that the actual shape of the requirement distribution be known, but does require that the distribution be symmetrical. Examples demonstrated the biases that occur when the assumptions of the cut-point method are violated. Assessing the prevalence of inadequacy of iron intake in