methodology overstated the total number of participants because errors in predicting the number of pregnant women and children were larger in absolute value than the errors in predicting postpartum women and infants. For example, in 1996, the number of children predicted to participate was overestimated by 28.6 percent and the number of pregnant women predicted to participate was overestimated by 44.5 percent. This is in comparison to an underestimation of 21.6 percent for infants and 23.1 percent of postpartum women. In recent years, prediction error rates for pregnant women and children have fallen in magnitude, while error rates for postpartum women and infants have risen. For example, in 1999, the numbers of children and pregnant women participants were overestimated by 19.6 and 21.3 percent, respectively, but the numbers of infants and postpartum women were underestimated by 31.8 and 32.9 percent, respectively, that year.

The figures shown in Tables 3–1 and 3–2 exemplify the wide variation in coverage rates and prediction error rates by category of eligibility. In general, coverage rates have been increasing over recent years. However, prediction error rates have been decreasing (in absolute value) for pregnant women and children but increasing for infants and postpartum women.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement