(c) add this to the square of the SD of requirements ([26.5 mg/ day]2 = 702 mg/day), resulting in a value of 3,162; and (d) the SDD is then obtained as the square root of 3,162, which is 56.
Therefore, D (55) divided by SDD (56) is just slightly less than 1. As shown in Appendix Table B-1, a value of about 1 implies an 85 percent probability of correctly concluding that this intake is adequate for a woman in this age category. (Details and further explanation are given in Appendix B.)
It is important to note that this woman's intake was exactly equal to the RDA of 320 mg/day, yet since there are only three days of dietary records, there is only 85 percent confidence that this intake is adequate. Only if true long-term intake had been measured for this woman (which is seldom feasible) could there be 97.5 percent confidence that intake at the RDA is adequate. With only three days of dietary recalls, it would be necessary for her magnesium intake to be 377 mg/day (which is well above the RDA) in order to have 97.5 percent confidence that intake was adequate (see Table 3-1).
Note that the SD of daily intake for the woman is not estimated from her own 3-day records. Instead, the estimated SD of daily intake of magnesium obtained from the CSFII is used. This estimate is a pooled (across all sampled individuals of the same life stage and gender group) SD of daily intake.
Why not use the woman's three days of intake records to estimate her SD of daily intake? As discussed earlier in this chapter, daily intakes may vary considerably from one day to the next. Unless the three days of intake recorded for the woman represent her entire range of intakes of magnesium, the SD that is estimated from her own records is likely to be severely biased. Thus, it is recommended that the pooled SD of daily intake obtained from the CSFII (or from other similar large-scale dietary surveys) be used for individual assessment. This has one serious drawback, however, as it is well known that the SD of daily intake also varies from individual to individual. In particular, it has been suggested that the within-person SD of intake is larger in those individuals with higher consumption of the nutrient (Tarasuk and Beaton, 1991a). Nusser et al. (1996) suggested that for some nutrients the association between mean intake and SD of intake for the individual is approximately linear. At this time, however, no extensive studies have been conducted to allow reliable estimation of the within-person SD of intakes from the individual's intake records. Therefore, even though the pooled SD obtained from CSFII (or other large-scale dietary surveys)