156 / 126
156
E
UR
A
MERICA
The results also show the gender differences in the association
between educational attainment and mortality compression in the
United States. Consistent with the findings of Brown et al. (2012),
within educational groups, women have significantly higher life
expectancies, higher modal ages of death, more compressed
mortality, and more rectangular survival curves than do men. The
results of the study replicate the similarities found in Brown et al.
(2012) between the least-educated women and men with some
college and above. However, men who have completed college (16+
years) have a better mortality profile than that of least-educated
women in terms of life expectancy and modal age of death.
Furthermore, although the significant gender gaps exist at any given
level of education, the gaps shrink with increased education. As
pointed out in Brown et al. (2012), complex biological,
socioenvironmental, and sociobehavioral factors may be deeply
involved in the gender differences, and education and gender
differences in smoking histories could play an important role.
This study is not without its limitations. First, while education
plays an important role as a “fundamental cause” of mortality
disparities (Link & Phelan, 1995; Phelan & Link, 2005),
educational attainment per se is subject to multiple social factors,
which include early life background. A life-course perspective to
understand the socioeconomic origins of mortality compression,
known as the “long arm of childhood,” is another important
research topic, and future research needs to address this issue
(Blackwell, Hayward, & Crimmins, 2001; Haas, 2007, 2008;
Hayward & Gorman, 2004; Kuh & New Dynamics of Ageing
Preparatory Network, 2007). Secondly, even though the HRS
respondents are followed into institutions, meaning that some
institutional respondents are included in the current data, the
institutional population was not initially included in the HRS sample
at baseline interview. Therefore, I observe slightly lower estimated
mortality rates than those that are found in the national life tables
(results not shown). Thirdly, the study does not distinguish the