older ages only involve those who survive to these ages; thus, if mortality is higher at younger ages for Hispanics, those who survive to older ages would be a select group in some sense (we discuss this argument, which could apply to all groups, in subsequent chapters). Another factor of note is the large proportion of foreign born among Hispanics (and island born among Puerto Ricans), a proportion that varies by age and reaches its peak around age 40.
As for Hispanics, the mortality rates for Asians are biased downward, but corrected rates are still better than those for whites. In contrast to Hispanics, the Asian mortality advantage appears consistently at all ages. Notably, the proportion foreign born is even higher, at all ages, among Asians than Hispanics.
For American Indians and Alaska Natives, the apparent overall advantage over whites in mortality in old age may be entirely illusory, since their rates are estimated to be 21 percent too low (Rosenberg et al., 1999), mainly because decedents are misclassified and are particularly unreliable at the oldest ages. Somewhat more reliable data come from the IHS (1999), which covers the 60 percent of American Indians and Alaska Natives living on federal reservations or in counties near them. These data show mortality levels for those 65 years and older that are 40 percent higher than those for whites. Survey data linked to subsequent mortality information, in contrast to official death rates, also suggest that American Indian and Alaska Native mortality levels are at least similar to those for whites. Notably, American Indian and Alaska Native mortality at younger ages is substantially higher than among whites: 60 percent higher among infants and almost twice as high for those 20-24 years old (Hummer et al., 2004).
At the oldest ages, for those 85 years and older, mortality rates appear to cross over in some cases, with black mortality falling below that of whites and American Indian and Alaska Native mortality falling below that of Hispanics and Asians and Pacific Islanders (see Figures 1-1 and 1-2). The black-white crossover appears in many data sets, though skepticism persists about data quality because of such problems as age misreporting. Recent evidence still suggests a crossover at ages 90-94 for males and 95 and older for females (Hill et al., 2000). However, the crossovers involving American Indians and Alaska Natives are suspect because of data problems. A crossover also appears, as noted above, for Hispanics relative to whites, but below age 65 and therefore not shown in Figures 1-1 and 1-2.
An important factor in any actual crossover is the possibility of selection: higher mortality earlier in life could leave fewer but more robust members in some racial and ethnic groups. This could, in fact, affect comparisons for older adults as a whole, since survival to age 65 is less common among blacks and American Indians and Alaska Natives.