The health consequences of race mixing
Posted by J Richards on Monday, June 6, 2005 at 05:38 PM
The health consequences of race mixing among living organisms could be neutral, beneficial or harmful. This issue has been difficult to address in humans, but data from random, population-based studies have started coming in, and here we consider the first such dataset.
The data in question come from J. Richard Udry’s National Longitudinal Study of Adolescent Health, which sampled a random, nationally representative school-based sample of U.S. adolescents in grades 7 through 12, during 1994-1995. 
Paper-based questionnaires were completed by 83,135 adolescents; a random sub-sample of these individuals plus some individuals in the school roster that had not completed the paper-based questionnaire, totaling 18,924 adolescents, were interviewed at home. The data are reported for the home-interviewed sample.
Racial classifications are based on self-report; the participants were allowed to pick multiple racial categories to describe themselves. Table 1 shows participant characteristics. The participants were asked if they were Hispanic/Latino, but this was not considered to be a racial category. 86% of those who only chose “other” race also described themselves as Hispanic, and 46% of those who described themselves as Hispanic only chose “other” race. 72% of those identifying as American Indian also picked another racial category, usually white.
The data are presented in terms of odds ratios and statistical significance, both of which are briefly explained in the next paragraph, which knowledgeable individuals can skip.
The odds ratio (OR) is the number of times something is more prevalent in one group compared to another. Thus, OR = 2.0 means a two-fold greater prevalence, OR = 1.3 means a 1.3-fold greater prevalence, OR = 1.0 means no difference, OR = 0.5 means half as prevalent, and so on. A statistically significant difference is a difference that most likely represents a genuine difference between the groups compared rather than a difference due to chance factors alone. If the probability that the difference obtained is due to chance is less than 5%, then the difference is typically considered statistically significant, which is denoted as p < 0.05; similarly, if the probability that the difference obtained is due to chance is less than 1%, then this is designated as p < 0.01. All statistically significant differences in Tables 2-4 (p < 0.05) are marked by an asterisk. On the other hand, due to inadequate sample size, genuine differences may not reach statistical significance in a study, although trends toward differences may be clear.
Table 2 shows that mixed-race individuals had an across-the-board higher incidence of health and behavior problems; the specific health problems addressed were assessed in terms of whether they occurred frequently or every day of the previous month.
Given the overall picture in Table 2, sample size limitations, and the fact that the sample is random and representative, all values in Table 2 that are clearly greater than 1.00 but not statistically significant can be considered as genuinely reflecting increased likelihood of health/behavior problems among mixed-race adolescents, and this also applies to Table 3.
Now, it may be that adverse outcomes vary by race combinations. To examine this issue, Table 3 compares various mixed-race combos with their single-race counterparts on health/behavior issues. The general pattern seen in Table 2 is also seen in Table 3, i.e., irrespective of which races are combined, there is an overall increase in health/behavior problems among mixed-race adolescents.
Let us address one confound in interpreting the data in Tables 2 and 3. It is commonly observed that a disproportionate number of white women who end up with black men are obese. Obese women tend to have elevated testosterone levels. [2, 3] Some portion of testosterone is converted to dihydrotestosterone (DHT), which amplifies the effect of testosterone at certain targets.  DHT is also one of the culprits in the genesis of acne.  Therefore, a higher incidence of, say, acne in black/white-mixed offspring with a white mother could simply reflect inheritance of the genetics of elevated androgens from the obese white mother rather than an effect of race mixing.
The point above needs to be considered in a more general scenario. When the races occupying the same geographic region differ in status, as in present-day Western societies, members of the high-status race who marry those of a lower-status race tend to have low status within their group. There are, of course, many examples to the contrary: Carmen Electra (mostly white) with Dennis Rodman (black), Heidi Klum (white) with Seal (black), and David Bowie (white) with Iman (Somali), among others. However, the aforementioned trend is clear, which raises the possibility that the negative correlates of race mixing are largely related to the unhealthiness of one or both parents rather than the deleterious effects of race mixing. This potential confound can be addressed by statistically controlling for demographic variables.
The authors controlled for age, sex, verbal IQ, grade point average, family structure (living with one or both parents), and family education; education can act as a proxy for socioeconomic status, and given an inverse relation between socioeconomic status and obesity in white women,  education can also act as a proxy for obesity in the white mothers of the mixed-race adolescents. Controlling for all these factors left the general picture seen in Tables 2 and 3 unchanged; some statistically significant odds ratios lost statistical significance, but the ones greater than 1.00 can still be considered significant for the reasons addressed above, and there were very few instances of greater-than-1.00 odds ratios diminishing to less than 1.00. Therefore, one can conclude that the negative health/behavior consequences of race mixing are related to race mixing per se rather than the variables controlled for.
An across-the-board increase in health problems should be evaluated in light of the basic architecture of physiological control, wherein the brain and the autonomic nervous system (ANS) are especially relevant. As is seen in the figure below, the ANS is involved in the autonomic (involuntary; automatic) regulation of organs, and its abnormalities could easily be behind Udry’s data.
ANS abnormalities are implicated in behavior problems such as excessive aggression, [7-11] atopy (a group of diseases such as asthma, skin problems such as eczema and psoriasis, allergies, runny nose, sinusitis, migraine), [12-16] headaches, [17-22] and a variety of health problems (see any textbook of medical physiology). Further, damaging some portions of the brain in rats increases the likelihood of cocaine or stimulant dependence, [23, 24] and something similar undoubtedly applies to humans, too. Besides, a number of brain abnormalities are common to both depression and drug dependence,  and substance use is more prevalent among mentally ill individuals. 
Therefore, to the extent that race mixing may increase the likelihood of non-optimal genetic correlation structures, it may be expected to adversely affect organism-level physiological control, which may very well account for Udry’s data.
Alternatively, since part of the ANS deals with stressors, chronic social stressors could themselves cause a number of health problems. In this regard, the data in Tables 2 and 3 do not vary as a function of whether the adolescent is living with one or two parents or as a function of family education, which can act as a proxy for socioeconomic status. Being teased/ostracized for being mixed race also requires consideration, but it cannot be argued that during the time period of the childhood years of the Americans examined, a substantial number of the mixed-race children would have experienced ostracism/discrimination due to being mixed race, especially since a number of them would have grown up in large metropolitan areas, which are known to be more diverse and tolerant. Moreover, the general preference in the African-American community is to prefer African-Americans with lighter skin, [27, 28] and lighter-skinned African-Americans also tend to have higher IQs than their darker counterparts (Table 4, ), which should reduce academic stress. In other words, black/white-mixed individuals, who would typically be assigned and raised as African-Americans, are not necessarily really worse off compared to their darker counterparts with respect to being ostracized/marginalized. Additionally, given white preference for Caucasian features and to the extent that Asian-Americans are envious of some of the facial features of whites, Asian/white-mixed individuals are less apt to be teased for their facial features than the less mixed Asians-Americans. And once again, there is no proof that struggle with identity formation explains Udry’s data. Also, the typical mixed-race adolescent with a white parent should have no doubts as to whether he is white; he wouldn’t look anything like a white person. Further, the data in Tables 2 and 3 do not vary as a function of age, which is relevant because the mixed-race individuals should have resolved their racial identity by their late teens.
On the other hand, there exist several examples in the animal literature where matings between more genetically distant individuals within the same species/different races result in offspring that are less healthy than the parents, on average, [30-37] and this cannot be blamed upon struggle with identity formation. There also exist examples of hybrid vigor, but nothing remotely close to hybrid vigor is seen in Udry’s data.
Consider also the fact that the genetic correlation structures manifesting as different races are so distinct that when a computer was asked to classify DNA data (326 microsatellite markers) from 3,636 individuals self-identifying as either white, East Asian, African-American, or Hispanic, into clusters, without being told which DNA sample came from which racial group, the computer clustered the DNA data into 4 groups, classifying all but 5 individuals into the correct self-identified racial group. 
Therefore, a parsimonious explanation of the across-the-board negative health correlates of race mixing in Udry’s data is that race mixing involving notably different races such as human races, by increasing the likelihood of non-optimal genetic correlation structures, increases the likelihood of deleterious effects.
Udry’s data are compatible with the likelihood of race mixing improving one or more parental traits in some mixed-race offspring, who may be better off than both parents on multiple counts, provided that a greater number of mixed-race offspring are overall worse off than both parents. The former possibility is surely not implausible given that the tremendous racial and species diversity out there implies that nature does not rule out equally-well functioning/better functioning novel genetic correlation structures, which could be brought about by race mixing, though the chances of improvement would typically be slim if more distant races are involved.
Further, those identifying as mixed-race in this study would generally have greater genetic admixture than those identifying as single-race. It is seen in Table 2 that those identifying as mixed race have worse health than even populations known to be highly admixed (American black, Native American, Hispanic). This could be accounted for if one assumes that first-generation hybrids who have the worst health/behavior problems would disproportionately not be very successful in reproducing, i.e., the healthier mixed offspring could, within a few generations, set up a mixed-race population that is healthier, on average, compared to the first-generation hybrids, but for this mixed-race population to approach or exceed the overall health of the original single-race populations, it would take many generations of [naturally] weeding out the unhealthy and settling toward a novel population-typical genetic correlation structure that corresponds to good health (more on this and on hybrid vigor in a subsequent post).
To conclude, it is irresponsible for any scientific organization to pretend that race mixing has no adverse health effects and it is obviously inappropriate to portray race mixing as desirable or virtuous.
Extra stuff (for those interested in additional details):
Table 4 compares mixed-race adolescents with their single-race counterparts on several measures. The GPA (grade point average) and the PVT (picture vocabulary test) categories feature the percentage of individuals in the 75th percentile of the entire sample.
Some of the responses to self-identified race differed between the school-based and home-based assessments (16% of entire sample). Over 90% of the respondents self-identifying as white only, black only, or Asian only at school, self-identified as the same race at home; the changes involved adding/substituting “other” race. Somewhat less than half the respondents picking multiple races at school picked the same categories at the home interview, and to the extent that those identifying as multiracial in school identified as single race at home, given the picture in Table 2, these individuals are likely making the single-race individuals look worse. 34% of those identifying as American Indian only at school identified as white only at home, and 77% of those identifying as white/American Indian at school identified as white only at home, but American Indians were a miniscule percentage of the study sample. The inconsistent responses among the mixed-race adolescents may have resulted from context (school vs. home), assessment technique (paper-based vs. interview), ambiguity of question asking about race, or lack of a fully developed racial concept at the time of the assessment. 
1. Udry JR, Li RM, Hendrickson-Smith J: Health and behavior risks of adolescents with mixed-race identity. Am J Public Health 2003, 93:1865-1870.
Commenting is not available in this channel entry.