Figure D-17 and Figure D-18 summarize the information presented in Figure D-13, Figure D-14, Figure D-15 through Figure D-16. In Figure D-17, the three curves represent the bias of the prevalence estimate relative to the true prevalence for three values of the EAR and when the correlation between intakes and requirements is close to 0. The solid line with dots shows the expected bias when the EAR is 55 units for varying values of the *SD*_{r}*.* The dotted line with stars corresponds to the bias at varying values of *SD*_{r} when the EAR is 70. Finally, the dashed line with squares indicates the expected bias when the EAR is equal to the mean intake and the true prevalence is 50 percent. Notice that when *SD*_{r} is high relative to SD_{i}, the bias in the estimated prevalence can be substantial. Consider for example, the case where the EAR is 55 and the *SD*_{r} is 40. The bias in the estimated prevalence is approximately 11 percent. This might not seem significant until one recalls that for an *SD*_{r} of 30 and an EAR of 55, the true prevalence in the group is approximately 20 percent (see Figure D-l). Thus, the bias in the estimate of prevalence corresponds to a full 50 percent of the true prevalence in the population.

**FIGURE D-17** Effect of the standard deviation of requirement (*SD*_{r}) on bias of the estimated prevalence of inadequate intakes using the Estimated Average Requirement (EAR) cut-point method for 10 values of the *SD*_{r} For all values of the *SD*_{r}, mean intake = 90, *SD* of intake = 30, and correlation between intake and requirement = 0.01. The EAR was set at 55 units (solid line with dots), 70 units (dotted line with stars) and 90 units (dashed line with squares).