Assortative
mating on blood type: Evidence from one million Chinese pregnancies.
Yao Hou et al. Proceedings of the National Academy of Sciences,
December 14, 2022, Vol. 119 (no. 51) e2209643119.
https://doi.org/10.1073/pnas.2209643119
Significance:
In the human population, spousal pairs have been found to share
phenotypes, which demonstrates the highly nonrandom nature of human
mate choice. However, assortative mating on blood type—one of the
most fundamental phenotypes in biological, medical, and psychological
studies—has not been investigated. Using a unique dataset from
China, we provide statistical analysis to test whether matching on
blood type is nonrandom and find a set of strong evidence for
assortative mating on blood type. The findings are robust after we
control for the effect of other possible mechanisms, and show that
the spousal concordance on blood type we observe is attributable to
not only an individual’s mate opportunity but also their mate
choice.
Abstract:
Blood type is one of the most fundamental phenotypes in biological,
medical, and psychological studies. Using a unique dataset of one
million Chinese pregnancies, we find strong evidence from a group of
statistical tests for assortative mating on blood type. After
controlling for anthropometric and socioeconomic confounders,
assortative mating remains robust.
Possible Reasons for Assortative Mating on Blood Type
Having shown robust evidence for assortative mating on blood type, we investigate potential reasons. One possible explanation is that blood type may act as a proxy for other phenotypes. As previously stated, many studies have validated assortative mating on a group of phenotypes, such as BMI, weight, height, and IQ (
10,
16–
21). That is, individuals tend to choose a partner who shares similarities along these dimensions when making mate choices. If blood type is associated with these phenotypes, spousal concordance on blood type will be observed because of assortative mating. Using personal information provided by the dataset, we examine bivariate correlation between blood type and other phenotypes (
Fig. 2). There appear to be some associations between blood type and the phenotypes we examine: education, job type, height, weight, pressure, and drinking habits. However, most associations have a relatively small correlation coefficient lying between −0.03 and 0.03.
[...]
To further explore to what extent assortative mating on blood type can be explained by its correlation with other phenotypes, we performed mediation analysis. Specifically, we first regressed the individual’s blood type on her partner’s using a logistic regression model, then incorporated a mediator—i.e., one of the partner’s phenotypes that might be associated with his blood type—to see whether and to what degree the effect of the partner’s blood type on the individual’s blood type is weakened after the mediator is included in the regression. We report the results of mediation analysis in
Tables 7 and
8. As can be seen, the coefficients of the partner’s blood type decline after we included different mediators in the regression models, which shows that the associations between blood type and other phenotypes can explain assortative mating on blood type to a certain degree. We see from columns 2 to 9 in
Table 7 and column 1 in
Table 8 that the proportion of the coefficients of the partner’s blood type absorbed by mediators varies with blood type. For individuals with type B blood, when all mediators are included, the coefficients of the partner’s blood type are reduced by around 6 to 7%; for those with type A blood, the incorporation of mediators has little effect on estimation results for coefficients of the partner’s blood type, as shown in column 1 in
Table 8. As for those with type AB blood or type O blood, the scale of mediator absorption is about 3 to 4%. However, a large fraction of assortativity remains unexplained. When we further included a group of control variables to isolate our measure of assortative mating from confounding factors—such as population stratification, province-level fixed effects, or even the individual’s phenotypes—in the regression, as indicated by the statistical significance of the coefficients of the partner’s blood type in columns 2 to 4 in
Table 8, we still found strong evidence for assortative mating on blood type. These findings suggest that there could be other potential mechanisms for this pattern we observe in the data. Further investigation into this is left for future research.