ological critique. The reader is referred to National Research Council (2004) where some specific factors and studies are treated in more detail.
Any analysis of racial and ethnic differences in health must take into account the effects of selection processes. Groups being compared will always be selected in some sense: for instance, they will necessarily be made up only of those who survive to the time of the study. This selectivity affects comparisons of group health, and it should therefore be explicitly attended to so that health differences are properly interpreted. (Selection, in the sense we use it here, is not the same thing as Darwinian selection and does not involve the evolution of genetic endowments).
Selection processes appear under various names in the literature, sometimes being referred to as unmeasured heterogeneity or drift and sometimes being labeled with reference to specific selection factors, as in “healthy migrant bias” (Abraido-Lanza et al., 1999). Related processes known as endogeneity or reciprocal causation are basically similar. The underlying processes, which can be represented within the same framework (Palloni and Ewbank, 2004), are potentially ubiquitous and need to be dealt with explicitly in interpreting health differences.
Selection processes operate for any group—racial or ethnic, socioeconomic, religious, residential, or any other—if joining the group is dependent on factors that also affect health and mortality. Membership in the group may subsequently affect health, but initial selection into the group will have a separate, prior, and possibly large effect.
It is indeed possible to join a racial or ethnic group, by self-identification, for instance. This is perhaps most relevant for those of mixed ancestry who can choose their group. If healthier individuals self-identify with one group rather than another, that might not affect their health, but it will affect group health statistics.
A more important instance of accession to a group is immigration, through which one acquires, in effect, a new identity in a new society. The case of immigration provides a good example of a selection process, resulting in the so-called “Hispanic paradox.” As noted in Chapter 1, Hispanic immigrants in the United States are generally in better health than whites despite lower socioeconomic status. This paradox may be partly explained by the fact that immigrants who succeed in entering and staying in the United States seem to be drawn disproportionately from the ranks of those with better health in their countries of origin. This phenomenon is generally