Elaina Rose*
Department of Economics
Mail Code 353330
University of Washington
Seattle, WA 98195
(206) 543-5237
erose@u.washington.edu
Abstract
Hypergamy is the tendency for women to marry up with respect to education or other
characteristics associated with economic well-being. For a given level of hypergamy, an
increase in the education of women relative to men will tend to increase the success gap (i.e.,
the disadvantage faced by successful women in the marriage market).
I track the success gap with U.S. Census data and find that the success gap declined
between 1980 and 2000 when womens education increased with respect to mens. This is
because hypergamy was not constant it also declined.
Similarly, we would expect marriage rates to fall for men at the bottom of the distribution.
This was consistent with the data. The decline in hypergamy was concentrated at the top of the
distribution. Over the period, hypergamy increased at the bottom of the distribution.
This research was supported by NIH Grant R03HD41611. A version of this paper was
circulated in July 2003 under the title, Does Education Really Disadvantage Women in the
Marriage Market? I am grateful to Janet Currie, Hank Farber, Shoshana Grossbard-Schechtman,
Levis Kochin and Tom MaCurdy for helpful comments and suggestions. Kisa Watanabe
provided excellent research assistance.
I. Introduction
Marriage has changed substantially in the last several decades. The most notable change
is the overall decline. At any given age, individuals are less likely to be, or have ever been,
married. According to Beckers [1977] work, the decline can be explained by the increase in
womens labor supply and market human capital which has reduced the gains from specialization
and exchange in marriage. Other explanations include the improvement in birth control
technology (Akerloff et al [1995] and Goldin and Katz [2002]) and the increase in welfare
generosity (Murray, 1984) 1. Grossbard-Schechtman [1993] relates the decline in womens
propensity to marry to the marriage squeeze, which, given womens tendency to marry older
men, disadvantaged women born during the post-World War II baby boom. Wilson [1987]
emphasizes the role of the deteriorating labor market for less-skilled men as a key factor in the
decline in marriage within the black community. Changes in family policy such as the
liberalization of divorce laws, as well as shifts in social norms, have reinforced these trends.
Patterns in education have changed considerably as well. Overall, the population of both
men and women in the U.S. has become more educated, and women have become more educated
relative to men. The increase in womens education in the last century is well-documented [e.g.,
Goldin, 1990].
The changes, however, were not monotonic. For instance, Card and Lemieux
[2000] documented the decline in high school completion beginning in the 1970s, particularly
for men.
Hypergamy is the tendency of women to marry-up with respect to factors associated
with socioeconomic status, such as education, income, or caste. It is an empirical regularity
across a number of societies and over time. While there has been little focus on hypergamy in
economic analyses of the marriage market, the topic is closely related to that of assortative mating,
on which there has been theoretical (Becker, 1981; Lam, 1988) and empirical (e.g., Pencavel,
1
Although there is some question about the empirical significance of the incentive effects of
transfer programs (Moffitt, 1992).
1998) work. Both hypergamy and assortative mating are approaches to characterizing marriage
matching patterns. Positive assortative mating refers to the degree of similarity of spouses with
respect to an outcome, negative asortive mating refers to the degree of dissimilarity, and
hypergamy refers to the degree of asymmetry.
There has been considerable work outside of economics which focusing on the role of
social norms in generating hypergamy. For instance, anthropologist Barbara Miller [1981]
studied areas of rural north India and found that strong pressures for hypergamy implied a lack of
suitable husbands for high caste girls. This created a disequilibrium that was resolved through
female infanticide.
In another context, the Talmud advises men to go down a step to take a wife, "
(Yevamot, 63a) , and states that a woman from a more distinguished family than her husband
may consider herself superior and act haughtily toward him (Rashi).2
The notion that social norms generate this empirical regularity remains. For instance, in a
2002 New York Times column, Maureen Dowd stated: Men veer away from challenging
women because they have an atavistic desire to be the superior force in a relationship.
In contrast to the explanations which rely on social norms, economic theory can explain
hypergamy as the outcome of a model of specialization and exchange of the traditional form i.e.,
one in which men specialize in the labor market and women specialize in home production.
Gains from marriage will be greater for couples who are hypergamous with respect to labor
market productivity, or characteristics associated with productivity. Lams [1988] work implies
that as the gains from specialization and exchange decline, positive assortative mating will
increase. When specialization and exchange takes the traditional form, a decline in hypergamy
would go hand-in-hand with the increase in positive assortative mating.
I am grateful to Levis Kochin and David Twersky for helping to find the references from the
Talmud.
There is some empirical evidence that marriage matching patterns have shifted in recent
years. Mare [1991] and Pencavel [1998] have found that positive assortative mating on education
has increased. 3
The essence of this paper is this: Regardless of whether it is attributable to comparative
advantage, or to social norms, hypergamy will be associated with different marriage rates for men
and for women. If women tend to marry up with respect to, say, education, and if education is
distributed similarly by sex, women at the top of the distribution will have more limited options,
and a negative relationship between education and marriage will emerge. I measure the success
gap as to the difference in the propensity to marry for women at the top of the education relative
to those at the median.
Furthermore, all other things being equal, an increase in the concentration of women at
the upper end of the education distribution will tend to increase the success gap. But all other
things have not been equal over the last several decades. For instance, findings in a number of
recent papers suggest that the role of specialization and exchange in marriage has declined.4 As
the source of gains from marriage shifts from specialization and exchange to production and
consumption of public goods, hypergamy, and the associated success gap, would be expected to
decline. Moreover, transformation of social norms from those that encourage hypergamy towards
those favoring more symmetric matchings will tend to reduce hypergamy, and the success gap, as
well.5 Whether or not the success gap has increased over time is an empirical question that is the
subject of this paper.
Using data from the Panel Study of Income Dynamics (PSID), Rose [2001] finds evidence of a
decline in assortative mating and hypergamy with respect to college completion, and parents
education between 1970 and 1990. However, Behrman, Rosenzweig and Taubman [1994] find
negative assortative mating on endowments associated with earnings.
4
For example, Lundberg and Rose [1999] and Gray [1997]. Blau [1998] reports that women in
1988 spend significantly less, and men spend somewhat more, time on housework than in 1978.
5
Goldstein and Kenney report that women with college education are more relatively more likely
to be married in 1980 than in 1960.
The discussion so far has focused on the upper end of the education distribution. An
analogous effect might operate at the lower end of the distribution. The decline in high school
graduation rates since the 1970s, and the deteriorating market for less-skilled male labor
documented by Juhn [1992], combine to reduce the returns to market work for less-educated men.
This hinders their ability to contribute to a traditional specialization-and-exchange marriage and
adds to their marriage market disadvantage. However, as argued above with respect to the upper
portion of the success gap at the upper portion of the distribution, the disadvantage may be
countered by a shift in marriage matching patterns.
In this paper, I use data from the U.S. Census of Population to track the success gap, as
well as education-marriage profiles, education-motherhood profile, and marriage-matching
patterns, for individuals age 40-44, over the period 1980-2000. Following much of the literature
on assortative mating, I focus on the characteristic education, as it is less likely to be
endogenous with respect to marriage outcomes than, say, income or wages. The Censuss large
sample sizes and fine breakdowns of education allow for precise estimates of the effect of each
additional year of education (for the most part) in order to test for non-linearities and nonmonotonicities.
For women, the relationship between education and the likelihood of marriage is an
inverted-U peaking at about twelve years of education. The difference of nearly 15 percentage
points between the likelihood of having ever been married for women with 19 relative to 12 years
of education in 1980 is consistent with a considerable success gap. However, that difference
declined to less than 5 percentage points by 2000. The results for the outcome currently
married are similar, although this profile exhibits sheepskin effects at 12- and 16- years of
education.
Overall, the Census data indicate a tendency towards hypergamy: Husbands are more
likely to be educated than their wives than vice versa. Over the period, spouses education
became more similar and hypergamy declined. However, the decline in hypergamy was confined
to the upper portion of the education distribution.
Section II of this paper describes the variables used in the analysis and documents trends
in key variables over the period. Section III contains the results relating to the success gap. I
measure the gap as the difference in the likelihood of marriage at the highest level of education
relative to that at the median. I present plots of the relationship between education and marriage
and test for shifts in the success gap over time. I also plot education-motherhood profiles and test
for shifts in the motherhood success gap. Section IV contains the results relating to hypergamy.
I develop empirical measures of hypergamy, and test for differences in hypergamy across the
education distribution over time. Section V concludes.
II. Data
The data are from the United States Census of Population Public Use Microdata Sample
(PUMS) (5% sample). Unless otherwise specified, analyses pertain to individuals age 40-44.
Table 1 reports characteristics of the sample in each year for men and women.
Education
One complication is that the coding of education changed between 1980 and 1990. In
1980, each respondent reported the number of years of school attended and whether the final year
was completed. The questions in 1990 and 2000 focused more on degrees attained. For 1980,
some of the lower levels of education were grouped together because of small cell counts. The
resulting variable is Edu-1. To obtain a measure that is comparable across years, some
categories were further collapsed. The resulting comparable measure is Edu-2. The
correspondence between the education measures is outlined in Appendix Table A.I-1.
The means by year in Table 1 indicate that womens education increased more than
mens over the period. On average, women age 40-44 had 12.50 years of education in 1980,
which increased to 13.35 in 2000. The education distributions plotted in Figure 1 indicate that the
increase was driven by an increase in post-secondary education at several levels.
The education of men age 40-44 increased from 1980 to 1990, and declined in the
subsequent decade. The distributions plotted in Figure 2 indicate that the spike is attributable to
increased college attendance by men who would have been draft age in the peak years of the
Vietnam War draft. This is consistent with Card and Lemieuxs [2000] findings indicating that
draft avoidance in the 1960s led to a surge in college education. Interestingly, there was a small
increase in college attendance by women of the comparable cohort which receded for the
subsequent cohort.
Figure 3 plots the differences in the education distributions for men and women. For all
levels of education above high school graduation, the difference between the percentage of
women in the category and the percentage of men in the category increased over the twenty year
period, and for virtually every level from high school completion and below, the differences
between the percentages declined. Clearly, there was a shift in the distribution of education
across the population, with relatively more women with greater than high school education, and
relatively more men with high school education or less.
Marriage
For most of the analyses, the outcome is marriage. Two measures of marital status are
used: whether the individual is currently married (Currently Married, or Current for short),
and whether the individual has ever been married (Ever Married, or Ever). Current is a
dummy variable which equals one if the individual is currently married whether living with
spouse or separated. Ever equals one if Current equals one or if the individual is a widow or
is divorced.
While there has clearly been a decline in marriage, the vast majority of both men and
women have been married at some time in their lives by age 40-44. Even in 2000, 89 percent of
all women, and 85 percent of all men had been married at some point. Due to the possibility of
divorce (and to a minor extent, widowhood), fewer individuals report being currently married
than having ever been married. The percentage of women currently married fell from 81 percent
in 1980 to 72 percent in 2000; the comparable numbers for men are 85 and 72 percent,
respectively.
Differences by Race
The second two panels of Table 1 report statistics for whites and for blacks. As whites
dominate the sample, it is not surprising that the patterns for whites are similar to those for the
sample as a whole, with marriage rates and education being somewhat higher.
Education has increased more markedly for black men relative to white men over the
period, but the increase for black women is similar to that of white women. Marriage rates for
blacks, however, are substantially lower than those for whites, and their decline over the period
has been more precipitous. For instance, in 1980, 66 percent of black women in the sample were
currently married, the percentage fell by 16 percentage points, to 50 percent by 2000.6
Motherhood
One ancillary analysis tracks the outcome motherhood with respect to education.
Unfortunately, only an imperfect measure of motherhood, Mother can be constructed
consistently over the period. Mother is based on individuals residing within the household.
Mothers of children residing elsewhere may be misclassified. This is further complicated by the
fact that in some years it is not possible to distinguish step-children from biological children. In
order to maintain comparability across years, the measure classifies step-children as biological
children. Appendix Table A.I-2 details the method used to develop Mother.
Data are available on children ever born, for 1980 and 1990 only. This includes
children residing elsewhere as well as co-resident children. The variable Mom is based on this
measure.
6
The remainder of the sample consists of individuals classified as Asian or Other. As this is a
heterogeneous group, I didnt do any disaggregated analyses with respect to the remainder.
The statistics in Table 1 indicate that motherhood, as well as marriage, declined over the
period. In 1980, 80 percent of women age 40-44 had a child co-residing, but the percentage fell
to 66 percent by 2000. As women in this age group may have had children in their teens or early
twenties that are no longer co-resident, I compute the proportions for women age 35-49 and 30-34.
For each age category, the proportion of women who were mothers fell by about 10 percentage
points over the twenty-year period.
Women are more likely to report having children ever born than having children coresiding. This is as expected, because the former measure includes children living with another
family member and those who have moved out of the household, while the latter does not. Not
surprisingly, the difference between the two measures is larger for older women, as they are more
likely to have adult children who are no longer co-resident.
Cohabitation
Another ancillary analysis tracks the outcome married or cohabiting. For 1980,
cohabiting was defined according to Casper et als [2000] measure of Persons of Opposite Sex
Sharing Living Quarters (POSSLQ). For 1990 and 2000 cohabiters were identified by the
Census as unmarried partners. While cohabitation overall has increased, it is still relatively
uncommon among individuals in their early 40s. For instance, in 2000, only 3 percent of women
in the sample were cohabiting, while 72 percent were married.7
III.
Results
Although the qualitative research of Manning and Smock (2003) indicates that even the 2000
measure may undercount cohabiters.
education changed between 1980 and 1990. Edu-2 is constructed to be comparable among all the
years, although Edu-1, which is available only for 1980 is more precise.
For the figures, I use the most precise measure of education available for each year,
which is Edu-1 for 1980 and Edu-2 for 1990 and 2000. The Tables report both Edu-1 and Edu-2
for 1980 (Columns 1 and 2, respectively), and Edu-2 for 1990 and 2000 (Columns 3 and 5,
respectively). I also report the differences between 1990 and 1980 results (Column 4) and 2000
and 1990 results (Column 6).
Figure 4 indicates that the percentage currently married is (weakly) increasing with each
year of education up to twelve years, at which point there is a spike. There is a decline for each
of the following levels of education, and then another spike at sixteen years of education, after
which the slope of the profile becomes strikingly negative.
The profile shifts down at lower levels of education, in each of the two subsequent
decades.
For 1990, there are still spikes in the profile at twelve and sixteen years of education;
otherwise the profile is flatter in the latter decades. In 2000, other than the two spikes, the profile
appears to be essentially flat or increasing from high school graduation forward. The three
profiles actually cross at 19 years of education.
The spikes in the Current profile at twelve and sixteen years of education are
reminiscent of Hungerford and Solons [1987] sheepskin effects in earnings which are found
when estimating the relationship between education and earnings. Sheepskin effects in earnings
are the significantly greater estimated increases in earnings at the twelfth and sixteen year of
education relative to other years of education indicating a premium for degree completion.
Hypothesis Tests
In order to perform significance tests, I estimate the following logit model using the three
years of pooled data and the comparable education measure, Edu-2. The model, where M* is the
latent variable associated with the outcome marriage, is:
10
M* =
EEdu 2
E ,1980 DE ,i ,1980 +
EEdu 2
E ,1990 DE ,i ,1990 +
EEdu 2
DE ,i , y is a dummy variable which equals one if individual i in year y had at least E years
of education; where Edu2 = {8, 9, 10, 11, 12, 14, 16 ,18, 19}.
Key findings from the logit models are summarized in Table 2. Asterisks in even
columns indicate that the incremental effect of an additional year of education relative to the prior
level of education was significant: i.e., that the hypothesis E , y = 0 can be rejected. Asterisks in
columns 3, 5, 9 and 11 indicate that the coefficient changed significantly between year y-10
and year y; i.e., that the hypothesis E , y E , y 10 = 0 can be rejected. The full set of logit
results, from which the significance tests are drawn,
are reported in Appendix 2.
The Success Gap
The success gap is measured as the difference in the likelihood of marriage at the median
level of education, or at Edu-2 = 12 (usually they are the same), and at the likelihood at the
highest level of education (Edu-2 = 19). The coefficients are defined in terms of incremental
effects of additional years of education. Therefore, there the success gap is zero, i.e., there is no
19
i, y
= 0.
13
19
i, y
13
19
change in the success gap from year y-10 to year y when the hypothesis
i, y
i , y 10 ] =0
13
is rejected. These measures of the success gap, and the p-values associated with the associated
hypothesis tests, are reported in the bottom panel of Tables 2.
These results, taken together, indicate that there was a significant success gap in 1980 and
1990, which declined significantly in each decade, and disappeared by 2000. The incremental
11
effects are consistent with this finding. Going from 16 to 17, and from 17 to 18 years of
education was associated witih a lower likelihood of marriage in 1980 and 1990, but only the
latter difference was significant in 2000. Again, these differences fell significantly in each of the
decades. The logit results also suggest that the sheepskin effects were significant in each of the
three years i.e., propensity to marry was significantly lower for women with both 12 and 16
years of education relative to those with 14 years of education.
Education-Ever Married Profiles, Hypothesis Tests, and The Success Gap
The profiles for Ever are plotted in Figure 5 and, the associated data are reported in
Table 3. The Ever profiles are similar to those for Current, but they are smoother there are
no spikes at twelve and sixteen years of education. For 1980, the likelihood of having ever been
married is substantially lower at 19 years of education (82.6 percent) relative to that at the median
of 12 years of education (96.1 percent). The difference of 13.5 percentage points reflects a
success gap consistent with Dowd and Hewletts statements. However, this difference fell in
each of the two subsequent decades. By 2000, the difference fell to 4.9 percentage points (90.5
85.6 percent). The compression in the profiles at high levels of education indicates that the widely
noted decline in marriage, at least for women in this age group, has been driven mainly by
women at lower levels of education.
Notably, there are no sheepskin effects in the Ever profile. The difference between the
two profiles is that Ever includes divorced and widowed women, and Current does not. As
widowhood in this age group is rare, the difference between the two profiles reflects divorced
women and suggests that women who tend to drop out from college are more likely to drop out
from marriage.8
8
Another possibility is that women who are divorced are more likely to be attending college at
the date of the interview. I examined this using the 1980, which asks whether the individual has
completed the respective year of education, or is still attending or dropped out. The percentages
currently attending women (men) in the sample were: 3.8 (2.9) percent of married, 4.7 (3.1)
percent of widowed, 6.7 (3.3) percent of divorced, 4.9 (3.2) percent of separated and 6.0 (4.1)
percent of never married. To the extent that interviews were conducted over the summer, the
12
In summary, the success gap, as measured as the difference in the likelihood of marriage
for women with the highest level of education relative to the likelihood for women with the
median, or 12 years of education was significant in each of the three years. However, the gap fell
significantly in the 1980s and the 1990s. The sheepskin effects in terms of the outcome
Current were significant in each year, but there were no sheepskin effects in terms of the
outcome Ever.
III.B. Men, Education and Marriage
In this section an identical analysis is reported for men; The education/marriage profiles
for men are plotted in Figures 6 and 7, and associated statistics are reported in Tables 4 and 5.
For 1980, education appears to increase the likelihood a man is currently married for
levels of education below high school completion. The profile is flat beyond that point, perhaps
with some small declines between 12 and 15 years of education. The profiles shifted down and
became steeper in each of the two decades.
The Ever profile is relatively flat from twelve years of education and beyond for each
of the three years. The profiles shifted downward in each of the subsequent decades, particularly
for the lower levels of education. For men, the decline in marriage over the last several decades
reflects primarily a decline at the lower end of the education distribution.9
Education-Marriage Profiles by Race
The issue of the decline in marriage has been particularly salient for blacks. Wilson
[1987] emphasizes the role of the declining pool of marriageable men in the black community
due to the deteriorating labor market for less skilled men in urban areas. As this is mainly an
percentage currently attending do not reflect those still in school but between years in a program.
9
Because cohabitation has become a partial substitute for marriage over the period (Bumpass et
al, 1991), In Appendix II I look at the outcome Cohabiting whether an individual is currently
married or cohabiting. Appendix Figures A.II-1 and A.II-2 plot the proportion of women and
men, respectively, who are currently married or cohabiting, for each of the three years, and the
percentages and associated regression results are reported in Appendix Tables A.II-1 and A.II-2.
In general, as cohabiting is relatively rare for individuals in their early 40s the patterns are very
similar to those when cohabiters are not classified as married.
13
issue for the least educated, we would expect that the positive relationship between education and
marriage would be stronger for blacks than for whites. Figures 8 through 15 plot the EducationMarriage Profiles for Blacks and Whites10 by sex, and the associate statistics are reported in
Tables 6 through 13.
The patterns for whites are very similar to the patterns for the full sample, which is not
surprising as whites dominate the sample. However, the patterns for blacks are very different.
With the exception of the effect of the nineteenth year of education in 1980 and 2000, there is no
evidence of a success gap for black women. The profiles are either flat or increasing over most of
the range for 1980 and 1990, and in 2000 the profile is positively sloped over most of the range.
As expected, the profiles are steeper for black men relative to white men, and became
significantly steeper over the twenty year period. In 1980, the differences in currently married
between the highest and lowest education categories were 6.8 (= 78.3 - 71.5) percentage points
for black men, and 5.0 (= 85.1 - 80.1) percentage points for white men, but the figures were 37.5
(= 79.4 41.9) for blacks and 17.8 (= 82.8 65.0) percentage points for whites in 2000. While
there has been a marked decline in marriage for blacks overall, the proportion of highly educated
black men who are married is similar to that of white men: consistent with Wilsons theory the
difference in black and white marriage rates lies primarily at the lower end of the education
distribution.
III.C. The Motherhood Success Gap
Because much of the popular concern regarding the success gap focuses on the fact that
career success compromises womens opportunities for motherhood, I also track the relationship
between education and motherhood for women age 40-44. Figures 16 through 18, and Tables 1416, pertain to the measure Mother, based on co-resident children, which can be computed for
1980, 1990, and 2000.
10
I didnt report analyses for races other than Black and White, as this is a smaller and
heterogeneous category.
14
Figure 16 and Table 15 indicate that there was indeed a tradeoff between motherhood and
marriage for women age 40-44 with more than a college degree. In 1980, 81.7 percent of women
with exactly 16 years of education were mothers at age 40-44, while only 63.5 percent of women
with a professional degree or doctorate had children, yielding a difference in the likelihood of
motherhood of 18.2 percentage points. However, as with marriage, the difference fell in each of
the subsequent two decades: to 8.2 percentage points by 1990 and 5.0 percentage points by 2000.
The results in Table 16 indicate that this motherhood success gap was statistically significant,
and subsequent declines in the gap were statistically significant as well.
Because the children of women age 40-44 may have already left the home, motherhood is
more likely to be understated for this age group, using this measure, than for younger women. It
is possible (but not likely) that the apparent decline in the gap is due to an increase in the
tendency for more educated women to have their children sufficiently young that they have left
the house by age 40-44. As a check, I look at the relationship between education and motherhood
for women age 30-34 and 35-39 in Figures 17 and 18, and Tables 15 and 16. While not as
marked, there is a success gap for each of these groups, which declines significantly in the 1980s
(and may increase at the highest level in the 1990s). Overall, for women, education is becoming
less of an impediment to motherhood as well as to marriage.
Results using the outcome Mom for 1980 and 1990 are reported in Appendix A-4.
These findings are qualitatively consistent with those using Mother.
15
puzzle, I examine marriage matching patterns for women age 40-44 and test for a shift in these
patterns.
I characterize married couples as Hypogamous if the husband had less education than
the wife, Same if the spouses reported the same level of education, and Hypergamous if the
husband had more education than the wife. Results are reported in the top panel of Table 17, and
graphed in bar charts in Figure 19.
In 1980, the largest category was Hypergamous (37.6 percent), followed by Same
Education (36.1 percent ) and Hypogamous (26.3 percent). The difference of 12 percentage
points between the proportion of couples in which the wife married up relative to the proportion
who married down, indicates hypergamy overall. However, in each of the subsequent two
decades, hypergamy fell, and hypogamy increased. The patterns for husbands are similar.
To compare the extent of asymmetry among various age groups and cohorts, and across
the education distribution, I define Net Hypergamy as the percentage of couples in a particular
group that are hypergamous minus the percentage that are hypogamous. Figure 20 plots this
index along the education distribution.
There is a decline in Net Hypergamy i.e., in the tendency for women to marry up at
the top of the education distribution in each decade. Overall, the likelihood a woman marries up
declines as her education increases. Net Hypergamy is positive at the bottom of the education
distribution and negative at the top of the distribution. Over the 20-year period, hypergamy
became more common at the bottom portion of the distribution and less common at the bottom
portion of the distribution.
The data in the bottom two panels of Table 17 are consistent with the relationships
suggested by the figures. For women with less than twelve years of education, Net Hypergamy is
significantly positive in each year, and the degree of hypergamy increased significantly in the
1980s and the 1990s. However, at the top of the education distribution, Net Hypergamy fell
significantly in each of the two decades.
16
V.
Conclusions
Marriage and education patterns have shifted dramatically in the last several decades.
This paper relates the two by introducing hypergamy into an economic analysis of marriage
markets. When matching patterns remain constant, an increase in the concentration of women at
the top of the education distribution, and men at the bottom of the distribution will disadvantage
more educated women and less educated men in the marriage market. In this paper, I examined
this implication of the theory by tracking education/marriage profiles, and marriage matching
patterns, from the 1980, 1990 and 2000 U.S. Census, for men and women age 40-44 in those
years..
Contrary to popular beliefs, the increased concentration of women at the top of the
education distribution has not resulted in a worsening of the marriage market prospects of more
educated women. The success gap declined substantially in the 1980s and 1990s. The
marriage market accommodated the shift through a decline in hypergamy at the upper end of the
education distribution.
On the other hand, it appears that the declining economic prospects of men at the bottom
of the education distribution have rendered many below the threshold of marriagiability. The
17
likelihood of a 40-44 year old man with 11 years of education being married fell by over 20
percentage points over the 20-year period, a greater decline than that for women of the same
education level. There was no decline in hypergamy at this end of the spectrum; in fact, some
measures indicate an increase in hypergamy for this group, as less educated women have
increasingly been reaching upward in the education distribution for husbands, or opting out of
marriage entirely.
Several caveats regarding causality must be considered in evaluating these results. For
instance, if later cohorts of more educated women are less negatively selected in terms of
unobservables associated with marriage, the decline in the success gap could be attributed to a
change in the pattern of selection into marriage. Alternatively, it may be that couples do not
match on education, but on some characteristic associated with education, and matching on this
characteristic remains more stable over time.
with women in the earlier cohorts being less likely to remain in school while married. The latter
issue can be addressed with a panel data set which tracks the marital and education histories of
respondents. Other approaches for dealing with causality involve instrumental variables
techniques.
There are some important implications of these results. First, for women, higher
education is no longer the hindrance to marriage, and motherhood, that it once was. The
perception that women face a stark choice between career and family is becoming less accurate in
each successive decade.
Second, the decline in marriage is overwhelmingly a phenomenon of the less educated
segments of the population. Mens education-marriage profiles have gone from being relatively
flat in 1980 to strongly steep in 2000. The worsening labor market opportunities for less-skilled
men have severely limited their ability to contribute to marriage. In terms of policy, measures
designed to encourage marriage are more likely to be successful when targeted towards
improving the economic prospects of men at the bottom of the economic spectrum.
18
References
Becker, Gary S. (1973) A Theory of Marriage: Part I Journal of Political Economy 81:4, 81346.
Becker , Gary S. (1974) A Theory of Marriage: Part II Journal of Political Economy 82:2,
S11-S26.
Becker, Gary S. (1981) A Treatise on the Family (Cambridge: Harvard University Press).
Becker, Gary S. (1985) Human Capital, Effort, and the Sexual Division of Labor Journal of
Labor Economics 81:4, 813-46.
Behrman, Jere, Mark R. Rosenzweig and Paul Taubman (1994) Endowments and the Allocation
of Schooling in the Family and in the Marriage Market: The Twins Experiment Journal of
Political Economy 102:6, 1131-74.
Blau, Francine D. (1998) Trends in the Well-Being of American Women, 1970-1995, Journal of
Economic Literature 36 (Section I).
Bound, John and Sarah Turner (2002), Going to War and Going to College: Did the GI Bill
Increase Educational Attainment? Journal of Labor Economics 20:4.
Bumpass, L. L., J.J. Sweet and Andrew Cherlin (1991) The Role of Cohabitation in Declining
Rates of Marriage Journal of Marriage and the Family 53, 913-27.
Card, David and Thomas Lemieux (2000) Dropout and Enrollment Trends in the Post-War
Period: What Went Wrong in the 1970s? In Jonathan Gruber, ed., An Economic Analysis of
Risky Behavior Among Youth. Chicago: University of Chicago Press.
Card, David and Thomas Lemieux (2001) Draft Avoidance and College Attendance: The
Unintended Legacy of the Vietnam War American Economic Review Papers and Proceedings
91 2001, p.97-102.
Casper, Lynne M, Philip N. Cohen and Tavia Simmons (2000) How Does POSSLQ Measure
Up? Historical Estimates of Cohabitation, Demography 37:2, 237-45.
Dowd, Maureen (2002) The Baby Bust, The New York Times, April 10, 2002.
Goldin, Claudia (1990) Understanding the Gender Gap: An Economic History of American
Women New York: Oxford University Press.
Goldin, Claudia (1998) Career and Family: College Women Look to the Past in F. Blau and R.
Ehrenberg eds, Gender and Family Issues in the Workplace, (Russell Sage Press).
Goldin, Claudia and Lawrence F. Katz (2002) The Power of the Pill: Contraceptives and
Womens Career and Marriage Decisions Journal of Political Economy 110, 730-70.
Goldstein, Joshua R. and Catherine T. Kenney (2001) Marriage Delayed or Marriage Forgone?
New Cohort Forecasts of First Marriage for U.S. Women American Sociological Review 66,
19
506-519.
Gray, Jeffrey S. (1997) The Fall in Men's Return to Marriage: Declining Productivity Effects or
Changing Selection? Journal of Human Resources 32:3, 481-504.
Grossbard-Schechtman (1993) On the Economics of Marriage: A Theory of Marriage, Labor and
Divorce (Westview Press, Boulder, CO).
Hewlett, Sylvia (2002) Creating a Life: Professional Women and the Quest for Children, New
York: Hyperion.
Hungerford, Thomas and Gary Solon (1987) Sheepskin Effects in the Returns to Education,
Review of Economics and Statistics 69:1, 175-77.
Juhn, Chinhui (1992) The Decline in Male Labor Force Participation: The Role of Declining
Opportunities, Quarterly Journal of Economics 107, p. 79-102.
Lam, David (1988) Marriage Markets and Assortative Mating with Household Public Goods:
Theoretical Results and Empirical Implications Journal of Human Resources 23(4) 462-87.
Lundberg, Shelly and Elaina Rose (1998) The Determinants of Specialization within Marriage,
mimeo, University of Washington.
Manning, Wendy and Pamela J. Smock (2003) Measuring and Modeling Cohabitation: New
Perspectives from Qualitative Data, mimeo, University of Michigan.
Mare, Robert (1991) Five Decades of Educational Assortative Mating, American Sociological
Review 56 15-32.
Miller, Barbara (1981) The Endangered Sex : Neglect of Female Children in Rural North India,
London: Cornell University.
Pencavel, John (1998) Assortative Mating by Schooling and the Work Behavior of Wives and
Husbands American Economic Review 88:2, 326-29.
Qian , Zhenchao (1998) Changes in Assortative Mating: The Impact of Age and Education,
1970-1990 Demography 35:3, 279-92.
Rose, Elaina (2004) Education, Hypergamy, and the Success Gap (including results by race)
working paper, University of Washington.
http://www.econ.washington.edu/people/detail.asp?uid=erose.
Rose, Elaina (2001) Marriage and Assortative Mating: How Have the Patterns Changed?
mimeo, University of Washington.
http://www.econ.washington.edu/people/detail.asp?uid=erose.
Wilson, William Julius (1987) The Truly Disadvantaged (University of Chicago Press, Chicago
IL).
20
1980
2000
1990
.448901
10
11
12
14
16
18
19
1990
.381512
10
11
12
14
16
18
19
1990
.088751
-.043518
10
11
12
14
16
18
19
21
1980
2000
1990
83.2
64.8
8
10
11
12
13
14
education
15
16
17
18
19
1990
96.2
82.6
8
10
11
12
13
14
education
15
16
17
18
19
22
1980
2000
1990
86.2
63
8
10
11
12
13
14
education
15
16
17
18
19
1990
95.1
78.7
8
10
11
12
13
14
education
15
16
17
18
19
23
1980
2000
1990
1980
2000
85
1990
97.9
66
81.9
8
10
11
12
13
14
educ ation
15
16
17
18
19
10
11
12
13
14
education
15
16
17
18
19
17
18
19
1990
1980
2000
70.4
1990
91.4
45.2
62.5
8
10
11
12
13
14
educ ation
15
16
17
18
19
10
11
12
13
14
education
15
16
24
1980
2000
1990
1980
2000
87.1
1990
95.6
65
76.8
8
10
11
12
13
14
educ ation
15
16
17
18
19
10
11
12
13
14
educ ation
15
16
17
18
19
17
18
19
1990
1980
2000
79.4
1990
91.6
41.9
53
8
10
11
12
13
14
educ ation
15
16
17
18
19
10
11
12
13
14
educ ation
15
16
25
1980
2000
1990
81.8
63.5
8
10
11
12
13
14
educ ation
15
16
17
18
19
1990
86.4
62.3
8
10
11
12
13
14
educ ation
15
16
17
18
19
1990
85.6
43.8
8
10
11
12
13
14
education
15
16
17
18
19
26
hypogamous
hypergamous
same
.417078
90
80
100
1990
.5
-.5
10
11
12
13
14
education
15
16
17
18
19
27
All
1980
Women
1990
2000
12.50
(2.5)
0.81
(0.4)
0.95
(0.2)
298382
13.37
(2.5)
0.75
(0.4)
0.93
(0.3)
451241
12.62
(2.4)
0.83
(0.4)
0.96
(0.2)
250650
11.98
(2.4)
0.66
(0.5)
0.89
(0.3)
33127
Education
(Meaured as Edu-2)
Currently Married
Ever Married
N
White
Education
(Meaured as Edu-2)
Currently Married
Ever Married
Black
N
Education
(Meaured as Edu-2)
Currently Married
Ever Married
All
All
1980
Men
1990
2000
13.35
(2.4)
0.72
(0.5)
0.89
(0.3)
566050
13.01
(3.0)
0.85
(0.4)
0.93
(0.2)
285184
13.74
(2.7)
0.79
(0.4)
0.91
(0.3)
433806
13.24
(2.6)
0.72
(0.5)
0.85
(0.4)
549878
13.53
(2.4)
0.77
(0.4)
0.94
(0.2)
375956
12.78
(2.4)
0.56
(0.5)
0.83
(0.4)
43754
13.54
(2.4)
0.74
(0.4)
0.91
(0.3)
438778
12.94
(2.2)
0.50
(0.5)
0.72
(0.4)
64759
13.16
(2.9)
0.86
(0.3)
0.94
(0.2)
244044
11.89
(2.7)
0.75
(0.4)
0.88
(0.3)
27343
13.93
(2.7)
0.80
(0.4)
0.92
(0.3)
368816
12.64
(2.5)
0.67
(0.5)
0.83
(0.4)
35922
13.43
(2.5)
0.73
(0.4)
0.87
(0.3)
433549
12.61
(2.2)
0.58
(0.5)
0.73
(0.4)
55916
0.80
(0.4)
0.89
(0.3)
0.83
(0.4)
0.87
(0.3)
357751
0.76
(0.4)
0.79
(0.4)
448973
0.73
(0.4)
0.85
(0.4)
0.76
(0.4)
0.81
(0.4)
504186
0.70
(0.5)
0.75
(0.4)
542553
0.70
(0.5)
0.82
(0.4)
298382
0.77
(0.4)
451241
0.86
(0.3)
285184
0.82
(0.4)
433806
0.75
(0.4)
549878
0.73
(0.4)
567280
0.66
(0.5)
496148
0.75
(0.4)
566050
28
Table 2
Percentage Currently Married, by Education Level
All Women, Age 40-44
(Corresponds to Figure 4)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
76.3**
76.3**
69.4**
-6.9**
70.5**
1.1*
9
79.2**
79.2**
73.7**
-5.5
66.9**
-6.8**
10
80.6**
80.6**
73.7
-6.9*
65.9
-7.8
11
80.2
80.2
72.4**
-7.8
64.8*
-7.6
12
83.2**
82.9**
77.7**
-5.2**
72.0
-5.7*
13
80.8**
.
.
.
.
.
14
79.6**
79.2**
74.0**
-5.2
70.3**
-3.7**
15
78.6*
.
.
.
.
.
16
82.1**
80.6**
76.9**
-3.7
75.2**
-1.7**
17
76.7**
.
.
.
.
.
18
74.2**
74.2**
73.1**
-1.1**
72.7**
-0.4**
19
66.4**
66.4**
71.3**
4.9**
72.6
1.3*
Success
-16.8
-16.5
-6.4
10.1
0.6
7.0
Gapa
Pr(Success
0.004
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
-2.7
13.8
3.3
Gapb
Pr(Success
0.000
0.017
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
29
Table 3
Percentage Ever Married, by Education Level
All Women, Age 40-44
(Corresponds to Figure 5)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
90.1**
90.1**
84.6**
-5.5**
83.2**
-1.4**
9
94.8**
94.8**
92.5**
-2.3
86.8**
-5.7**
10
95.8**
95.8**
93.6**
-2.2
87.5
-6.1
11
96.1
96.1
92.3**
-3.8**
84.9**
-7.4
12
96.1
96.1
94.8**
-1.3**
90.5**
-4.3**
13
96.2
.
.
.
.
.
14
95.6**
95.4**
94.2**
-1.2
89.8**
-4.4
15
95.1*
.
.
.
.
.
16
93.7**
92.9**
91.5**
-1.4
88.2**
-3.3**
17
90.5**
.
.
.
.
.
18
88.8**
88.8**
87.7**
-1.1**
85.4**
-2.3**
19
82.6**
82.6**
88.5
5.9**
85.6
-2.9
Success
-13.5
-13.5
-6.3
7.2
-4.9
1.4
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.922
Gap)=0a
Success
-5.7
7.8
0.8
Gapb
Pr(Success
0.000
0.000
0.004
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
30
Table 4
Percentage Currently Married, by Education Level
All Men, Age 40-44
(Corresponds to Figure 6)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
79.6**
79.6**
71.9**
-7.7**
69.3**
-2.6**
9
82.2**
82.2**
74.8**
-7.4
65.4**
-9.4**
10
83.0
83.0
75.0
-8.0
63.0**
-12.0**
11
83.5
83.5
72.8**
-10.7**
63.5
-9.3**
12
86.0**
86.0**
77.9**
-8.1**
69.3**
-8.6
13
85.8
.
.
.
.
.
14
85.5
85.3
79.1**
-6.2**
72.3**
-6.8**
15
84.9
.
.
.
.
.
16
85.9*
85.7
80.9**
-4.8**
76.9**
-4.0**
17
85.3
.
.
.
.
.
18
86.2
86.2
83.1**
-3.1**
80.6**
-2.5**
19
85.4
85.4
84.6**
-0.8**
83.0**
-1.6
Success
-0.6
-0.6
6.7
7.3
13.7
7.0
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
5.5
6.1
8.2
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
31
Table 5
Percentage Ever Married, by Education Level
All Men, Age 40-44
(Corresponds to Figure 7)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
87.8**
87.8**
82.2**
-5.6**
78.7**
-3.5**
9
92.1**
92.1**
89.5**
-2.6*
82.1**
-7.4**
10
93.0**
93.0**
90.0
-3.0
81.7
-8.3
11
93.8**
93.8*
88.9**
-4.9**
80.9*
-8.0
12
94.4**
94.5**
91.9**
-2.6**
85.5**
-6.4
13
95.1**
.
.
.
.
.
14
94.7
94.5
92.4**
-2.1
86.5**
-5.9
15
94.0**
.
.
.
.
.
16
93.7
93.5**
90.7**
-2.8
86.2
-4.5**
17
93.2
.
.
.
.
.
18
93.5
93.5
91.4**
-2.1**
87.7**
-3.7
19
92.8*
92.8*
92.4**
-0.4**
89.8**
-2.6
Success
-1.6
-1.7
0.5
2.2
4.3
3.8
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
0.0
1.7
4.3
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
32
Table 6
Percentage Currently Married, by Education Level
White Women, Age 40-44
(Corresponds to Figure 8)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
78.3**
78.3**
69.1**
-9.2**
67.9**
-1.2*
9
82.4**
82.4**
76.7**
-5.7*
70.0**
-6.7**
10
83.9**
83.9**
78.0
-5.9
70.2
-7.8
11
83.9
83.9
77.5
-6.4
70.0
-7.5
12
85.0**
84.7**
79.8**
-4.9
75.1**
-4.7**
13
82.8**
.
.
.
.
.
14
81.1**
80.9**
75.9**
-5.0
73.2**
-2.7**
15
80.4
.
.
.
.
.
16
83.4**
81.8**
77.7**
-4.1
76.6**
-1.1**
17
77.7**
.
.
.
.
.
18
74.3**
74.3**
73.9**
-0.4**
73.8**
-0.1*
19
66.0**
66.0**
71.3**
5.3**
73.3
2.0**
Success
-19.0
-18.7
-8.5
10.2
-1.8
6.7
Gapa
Pr(Success
0.733
0.000
0.000
0.003
0.000
0.000
Gap)=0a
Success
-4.6
14.1
0.1
4.7
Gapb
Pr(Success
0.000
0.852
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
33
Table 7
Percentage Ever Married, by Education Level
White Women, Age 40-44
(Corresponds to Figure 9)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
90.7**
90.7**
84.6**
-6.1**
81.9**
-2.7**
9
96.7**
96.7**
95.6**
-1.1**
91.3**
-4.3**
10
97.5**
97.5**
96.8**
-0.7
93.1**
-3.7
11
97.9*
97.9*
96.1**
-1.8**
91.5**
-4.6
12
96.8**
96.8**
96.0
-0.8**
93.4**
-2.6**
13
96.9
.
.
.
.
.
14
96.1**
96.0**
95.2**
-0.8
92.0**
-3.2
15
95.7
.
.
.
.
.
16
94.2**
93.3**
92**
-1.3
89.1**
-2.9**
17
90.8**
.
.
.
.
.
18
88.6**
88.6**
88.0**
-0.6**
85.9**
-2.1**
19
81.9**
81.9**
88.8
6.9**
86.1
-2.7
Success
-14.9
-14.9
-7.2
7.7
-7.3
-0.1
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.567
Gap)=0a
Success
-6.4
8.5
-5.9
0.5
Gapb
Pr(Success
0.000
0.000
0.000
0.413
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
34
Table 8
Percentage Currently Married, by Education Level
Black Women, Age 40-44
(Corresponds to Figure 10)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
62.2**
62.2**
50.6**
-11.6**
45.2**
-5.4**
9
64.5
64.5
55.9**
-8.6
45.2
-10.7**
10
67.0
67.0
54.5
-12.5
45.3
-9.2
11
66.8
66.8
53.8
-13
46.3
-7.5
12
67.2
67.2
58.5**
-8.7**
49.9**
-8.6
13
66.9
.
.
.
.
.
14
64.8
64.2**
55.3**
-8.9
51.5**
-3.8**
15
63.1
.
.
.
.
.
16
64.4
64.2
58.9**
-5.3**
55.3**
-3.6
17
63.9
.
.
.
.
.
18
70.4*
70.4
58.3
-12.1
55.9
-2.4
19
60.7*
60.7*
55.5
-5.2
55.2
-0.3
Success
-6.5
-6.5
-3.0
3.5
5.3
8.3
Gapa
Pr(Success
0.008
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
Gapb
Pr(Success
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
35
Table 9
Percentage Ever Married, by Education Level
Black Women, Age 40-44
(Corresponds to Figure 11)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
83.1**
83.1**
70.7**
-12.4**
62.5**
-8.2**
9
86.8**
86.8**
78.1**
-8.7
65.8
-12.3*
10
89.5**
89.5**
80.4
-9.1
64.9
-15.5
11
89.6
89.6
80.4
-9.2
65.9
-14.5
12
89.6
89.8
84.2**
-5.6**
71.6**
-12.6
13
91.4**
.
.
.
.
.
14
91.4
90.9**
85.8**
-5.1
76.5**
-9.3**
15
90.2
.
.
.
.
.
16
89.9
89.3
85.0
-4.3
76.3
-8.7
17
87.7
.
.
.
.
.
18
91.0
91.0
83.2
-7.8
77.3
-5.9*
19
86.2*
86.2
79.9
-6.3
73.4*
-6.5
Success
-3.4
-3.6
-4.3
-0.7
1.8
6.1
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
Gapb
Pr(Success
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
36
Table 10
Percentage Currently Married, by Education Level
White Men, Age 40-44
(Corresponds to Figure 12)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
80.1**
80.1**
70.1**
-10.0**
65.0**
-5.1**
9
83.9**
83.9**
76.8**
-7.1
66.1
-10.7**
10
84.8
84.8
77.3
-7.5
65.2
-12.1
11
85.6
85.6
75.2**
-10.4**
65.2
-10.0**
12
87.1**
87.1**
79.2**
-7.9**
70.8**
-8.4
13
86.9
.
.
.
.
.
14
86.1
86.0**
79.8**
-6.2**
73.4**
-6.4**
15
85.8
.
.
.
.
.
16
86.2
86.1
81.0**
-5.1**
77.3**
-3.7**
17
85.8
.
.
.
.
.
18
86.5
86.5
83.2**
-3.3**
80.7**
-2.5*
19
85.1**
85.1**
84.5**
-0.6**
82.8**
-1.7
Success
-2.0
-2.0
5.3
7.3
12
6.7
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
4.7
6.7
9.4
4.7
Gapb
Pr(Success
0.000
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
37
Table 11
Percentage Ever Married, by Education Level
White Men, Age 40-44
(Corresponds to Figure 13)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
88.2**
88.2**
81.7**
-6.5**
76.8**
-4.9**
9
93.1**
93.1**
92.1**
-1.0**
85.6**
-6.5**
10
94.5**
94.5**
92.6
-1.9
85.8
-6.8
11
95.2*
95.2*
91.5**
-3.7**
84.7**
-6.8
12
95.1
95.1
92.9**
-2.2**
87.3**
-5.6
13
95.6*
.
.
.
.
.
14
95.0*
94.8
92.9
-1.9
87.5
-5.4
15
94.2**
.
.
.
.
.
16
93.9
93.7**
90.9**
-2.8
86.6**
-4.3**
17
93.3
.
.
.
.
.
18
93.8
93.8
91.4**
-2.4**
87.8**
-3.6
19
92.6**
92.6**
92.4**
-0.2**
89.8**
-2.6
Success
-2.5
-2.5
-0.5
2.0
2.5
3.0
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
-0.5
2.0
3.0
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
38
Table 12
Percentage Currently Married, by Education Level
Black Men, Age 40-44
(Corresponds to Figure 14)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
71.5**
71.5**
56.6**
-14.9**
41.9**
-14.7**
9
74.7*
74.7*
62.0**
-12.7
46.5**
-15.5
10
74.7
74.7
61.9
-12.8
49.2
-12.7
11
76.4
76.4
63.4
-13.0
52.5**
-10.9
12
76.4
76.5
66.5**
-10.0*
57.4**
-9.1
13
77.1
.
.
.
.
.
14
75.9
76.0
70.1**
-5.9**
63.7**
-6.4*
15
76.3
.
.
.
.
.
16
75.4
74.9
72.8**
-2.1*
67.5**
-5.3
17
73.5
.
.
.
.
.
18
76.4
76.4
74.6
-1.8
72.0**
-2.6
19
78.3
78.3
76.3
-2.0
79.4**
3.1*
Success
1.9
1.8
9.8
8.0
22.0
12.2
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
Gapb
Pr(Success
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
39
Table 13
Percentage Ever Married, by Education Level
Black Men, Age 40-44
(Corresponds to Figure 15)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
82.3**
82.3**
69.8**
-12.5**
53.0**
-16.8**
9
87.1**
87.1
76.8**
-10.3
60.4**
-16.4
10
86.4
86.4
78.3
-8.1
63.4
-14.9
11
88.7*
88.7*
80.7*
-8.0
67.0**
-13.7
12
89.3
89.7
83.8**
-5.9
72.9**
-10.9
13
91.6**
.
.
.
.
.
14
91.0
91.2**
87.5**
-3.7
79.3**
-8.2
15
91.4
.
.
.
.
.
16
89.2
89.4*
87.1
-2.3
80.4
-6.7
17
90.0
.
.
.
.
.
18
89.4
89.4
87.8
-1.6
83.4**
-4.4
19
91.3
91.3
89.4
-1.9
88.3**
-1.1
Success
2.0
1.6
5.6
4.0
15.4
9.8
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.018
Gap)=0a
Success
Gapb
Pr(Success
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
40
Table 14
Percentage Mothers, by Education Level
All Women, Age 40-44
(Corresponds to Figure 16)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
75.4**
75.4**
70.5**
-4.9**
69.4**
-1.1*
9
77.6**
77.6**
71.0
-6.6**
64.5**
-6.5**
10
77.6
77.6
70.2
-7.4
63.9
-6.3
11
78.9**
78.9**
71.6**
-7.3
65.5**
-6.1
12
81.3**
81.2**
74.6**
-6.6
69.6**
-5.0
13
80.5**
.
.
.
.
.
14
80.1
80.7**
73.0**
-7.7*
71.3**
-1.7**
15
81.8**
.
.
.
.
.
16
81.7
80.2**
73.2
-7.0**
71.9**
-1.3
17
76.4**
.
.
.
.
.
18
73.6**
73.6**
67.0**
-6.6**
67.4**
0.4**
19
63.5**
63.5**
65.0**
1.5**
66.9
1.9
Success
-17.8
-17.7
-9.6
8.1
-2.7
6.9
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
-8.0
9.7
5.3
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
41
Table 15
Percentage Mothers, by Education Level
All Women, Age 35-39
(Corresponds to Figure 17)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
77.9**
77.9**
72.5**
-5.4**
71.4*
-1.1*
9
84.1**
84.1**
78.0**
-6.1**
72.5**
-5.5**
10
85.1*
85.1*
78.8
-6.3
71.4*
-7.4*
11
86.4**
86.4**
78.0
-8.4**
71.5
-6.5
12
86.4
86.3
80.1**
-6.2**
75.8**
-4.3**
13
85.5**
.
.
.
.
.
14
84.2**
83.9**
76.9**
-7.0
75.2**
-1.7**
15
83.2**
.
.
.
.
.
16
81.1**
79.6**
70.8**
-8.8**
70.3**
-0.5**
17
75.7**
.
.
.
.
.
18
72.1**
72.1**
64.0**
-8.1**
64.9**
0.9**
19
62.4**
62.4**
62.3**
-0.1**
64.7*
2.4*
Success
-24.0
-23.9
-17.8
6.1
-11.1
6.7
Gapa
Pr(Success
0.028
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
-14.6
9.3
3.5
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
42
Table 16
Percentage Mothers, by Education Level
All Women, Age 30-34
(Corresponds to Figure 18)
Edu-1
(1)
1980
(2)
1980
(5)
(6)
Education
2000
2000-1990
8
75.5**
75.5**
69.9**
-5.6**
69.4**
-0.5
9
83.9**
83.9**
78.1**
-5.8*
73.6**
-4.5**
10
85.4**
85.4**
78.6
-6.8
72.9
-5.7
11
85.6
85.6
76.0**
-9.6
71.1**
-4.9
12
82.6**
82.3**
76.9**
-5.4**
72.6**
-4.3
13
81.1**
.
.
.
.
.
14
77.1**
76.1**
71.0**
-5.1**
69.0**
-2.0
15
74.4**
.
.
.
.
.
16
67.4**
65.7**
57.4**
-8.3**
56.3**
-1.1**
17
61.3**
.
.
.
.
.
18
54.5**
54.5**
48.5**
-6.0*
48.0**
-0.5
19
43.8**
43.8**
48.4
4.6**
45.2**
-3.2**
Success
-38.8
-38.5
-28.5
10.0
-27.4
1.1
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
Gapb
Pr(Success
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
43
Table 17
Percent of Marriages by Type
Wives Age 40-44
Hypogamous (Husbands Education < Wifes Education)
Same
(Husbands Education = Wifes Education)
Hypergamous (Husbands Education > Wifes Education)
1980
Hypogamous
Same
Hypergamous
Net Hypergamy (All)
(p: Hypergamy=0)
Net Hypergamy (Ed<12)
(p: Hypergamy=0)
Net Hypergamy (Ed>12)
(p: Hypergamy=0)
26.3
36.1
37.6
11.3
(0.000)
27.2
(0.000)
4.5
(0.000)
25.2
38.8
35.9
10.7
(0.000)
40.2
(0.000)
-4.1
(0.000)
-1.1
2.7
-1.7
-0.6
(0.000)
13
(0.000)
-8.6
(0.000)
27.4
41.7
30.9
3.5
(0.000)
45.4
(0.000)
-18.7
(0.000)
20001990
2.2
2.9
-5
-7.2
(0.000)
5.2
(0.000)
-14.6
(0.000)
44
Appendix I
Details of Data Transformations
Table A.I-1
Measuring Education Using U.S. Census Data
1980 Code
(Highest year of school
completed)
Never attended school
Nursery school
Kindergarten
First grade
Second grade
Third grade
Fourth grade
Fifth grade
Sixth grade
Seventh grade
Eighth grade
Ninth grade
Tenth grade
Eleventh grade
Twelfth grade
First year of college
Second year of college
1990 Code:
(Educational attainment)
2000 Code:
(Educational attainment)
Edu1
Edu2
No school completed,
Nursery school,
Kindergarten,
1st, 2nd, 3rd, or 4th grade,
5th, 6th, 7th, or 8th grade
No school completed
Nursery school to 4th grade
5th grade or 6th grade
7th grade or 8th grade
Ninth grade
Tenth grade
Eleventh grade
Twelfth grade, no diploma
High School graduate:
diploma or GED
Ninth grade
Tenth grade
Eleventh grade,
Twelfth grade, no diploma
High School graduate:
diploma or GED
9
10
11
9
10
11
12
12
13
14
14
14
15
16
17
18
19
14
16
16
18
19
Bachelors degree
Bachelors degree
Masters degree
Professional degree
Doctorate
Masters degree
Professional degree
Doctorate
45
Table A.I-2
Measuring Motherhood Using U.S. Census Data
1980
1990
2000
Maybea
Maybea
Maybe a
NA
Motherb
Maybec
Maybec
Step
Mother b,d
Maybe c
Maybe c
Step
Maybe a
Maybe a
Maybe a
NA
Maybe
Maybe
Maybe
Mother
Maybe
Maybe
Maybe
Mother
Maybe
Maybe
Maybe
Maybe
Maybe
Maybe
Child and associated variables do not distinguish step- vs. biological relationships with
respect to head in 1980.
b
Biological and step-children are distinguished in 1990 and 2000.
d
2000 Census data distinguish biological and adopted children; both are treated as children in
here.
c
Cannot distinguish grandchildren from step-grandchildren, and children-in-law from step
children-in-law in 1990 and 2000.
e
Biological and step-relationships are not distinguished for subfamilies in any year.
f
Biological and step-relationships are not distinguished for parents, grandparents, and parents-inlaw of head for any year.
46
Ed-1
1980 1980
(1)
9
10
11
12
13
14
15
16
17
18
19
0.165
(5.59)
0.089
(2.76)
-0.026
(0.94)
0.202
(9.89)
-0.164
(8.35)
-0.074
(3.00)
-0.062
(2.25)
0.222
(7.92)
-0.331
(10.6)
-0.132
(3.56)
-0.378
(9.74)
N
298382
(2)
1990
(3)
19901980
(4)
2000
(5)
Ever Married
Three years Pooled (Ed-2)
Ed-1
20001990
(6)
1980
1980
(7)
(8)
1990
(9)
19901980
(10)
2000 20001990
(11) (12)
-0.227
(17.54)
-0.112
(6.71)
0.149
(7.33)
-0.393
(15.47)
-0.447
(13.33)
298382
-0.012
(0.40)
-0.526
(9.16)
-0.109
(1.60)
-0.021
(0.39)
0.116
(3.62)
-0.079 0.033
(6.78) (1.63)
47
Table A.II-2
Effect of Additional Education on Likelihood of Marriage
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
All Men Age 40-44
Ed-1
9
10
11
12
13
14
15
16
17
18
19
Currently Married
Three years Pooled (Ed-2)
1980
1980
(1)
0.174
(5.63)
0.054
(1.52)
0.036
(1.09)
0.194
(7.64)
-0.021
(0.83)
-0.025
(0.83)
-0.043
(1.38)
0.077
(2.51)
-0.048
(1.52)
0.072
(1.86)
-0.064
(1.82)
(2)
(3)
0.174 0.144
(5.63) (4.97)
0.054 0.011
(1.52) (0.32)
0.036 -0.110
(1.09) (4.13)
0.194 0.273
(7.64) (15.92)
N
285184
1990
19901980
(4)
-0.030
(0.71)
-0.043
(0.89)
-0.146
(3.44)
0.079
(2.58)
2000
20001990
(5)
(6)
-0.176 -0.320
(7.24) (8.46)
-0.106 -0.117
(3.96) (2.74)
0.021 0.131
(1.06) (3.93)
0.262 -0.011
(22.08) (0.53)
0.075
(5.83)
0.037
(1.75)
0.130
(8.38)
-0.013
(0.50)
-0.027
(0.93)
285184
Ever Married
Three years Pooled (Ed-2)
Ed-1
1980
1980
(7)
0.482
(11.36)
0.140
(2.75)
0.125
(2.50)
0.117
(3.00)
0.121
(2.97)
-0.064
(1.35)
-0.141
(2.94)
-0.053
(1.17)
-0.084
(1.87)
0.060
(1.10)
-0.123
(2.53)
(8)
0.482
(11.36)
0.140
(2.75)
0.125
(2.50)
0.117
(3.00)
285184
1990
19901980
(9)
(10)
0.609 0.127
(15.42) (2.19)
0.054 -0.086
(1.14) (1.24)
-0.117 -0.242
(3.07) (3.85)
2000
(11)
0.218
(7.43)
-0.029
(0.89)
-0.051
(2.03)
0.335
(22.82)
20001990
(12)
-0.391
(7.94)
-0.083
(1.44)
0.066
(1.45)
-0.010
(0.34)
0.039
(1.63)
0.345
(14.03)
0.079
(5.17)
0.228
(4.95)
0.040
(1.41)
0.078
(7.42)
-0.001
(0.05)
-0.176
(5.66)
-0.224 -0.048
(13.71) (1.36)
-0.026
(1.94)
0.198
(9.46)
-0.055
(1.51)
-0.093
(2.34)
285184
0.078 0.132
(3.44) (3.10)
0.145 0.238
(4.50) (4.65)
433806
1268868
0.058
0.136
(1.93)
(6.74)
0.067
0.212
(1.53)
(7.22)
549878
48
Table A.II-3
Effect of Additional Education on Likelihood of Marriage
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
White Women Age 40-44
Ed-1
1980
(1)
9
10
11
12
13
14
15
16
17
18
19
N
0.265
(7.22)
0.103
(2.58)
0.005
(0.14)
0.079
(3.15)
-0.161
(7.22)
-0.116
(4.21)
-0.048
(1.57)
0.202
(6.51)
-0.365
(10.77)
-0.187
(4.66)
-0.395
(9.42)
Currently Married
Three years Pooled (Ed-2)
1980 1990 1990- 2000 20001980
1990
(2)
(3)
(4)
(5)
(6)
Ed-1
1980
(7)
Ever Married
Three years Pooled (Ed-2)
1980 1990 19902000
20001980
1990
(8)
(9)
(10)
(11)
(12)
1.089 1.385 0.296
(15.19) (20.60) (3.01)
0.277 0.337 0.060
(3.15) (3.93) (0.49)
0.209 -0.230 -0.438
(2.48) (3.25) (3.99)
-0.435 -0.009 0.426
(6.84) (0.21) (5.60)
0.850
(15.89)
0.246
(3.82)
-0.232
(4.54)
0.271
(9.60)
-0.535
(6.23)
-0.091
(0.85)
-0.002
(0.02)
0.280
(5.55)
-0.034
(0.95)
-0.201
(13.56)
-0.007
(0.26)
-0.336
(20.88)
0.214
(8.12)
-0.298
0.153
(14.43) (4.83)
0.018
-0.062
(0.56)
(1.16)
438778
-0.160
(5.41)
-0.194
(9.68)
49
Table A.II-4
Effect of Additional Education on Likelihood of Marriage
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
Black Women Age 40-44
Currently Married
Three years Pooled (Ed-2)
Ed-1
9
10
11
12
13
14
15
16
17
18
19
1980
1980
1990
(1)
0.097
(1.57)
0.114
(1.76)
-0.010
(0.19)
0.019
(0.48)
-0.014
(0.29)
-0.095
(1.48)
-0.074
(0.99)
0.055
(0.69)
-0.018
(0.19)
0.291
(2.44)
-0.430
(3.40)
(2)
0.097
(1.57)
0.114
(1.76)
-0.010
(0.19)
0.019
(0.48)
N
33127
2000
Ever Married
Three years Pooled (Ed-2)
Ed-1
(3)
0.214
(3.08)
-0.058
(0.83)
-0.030
(0.59)
0.193
(5.87)
19901980
(4)
0.116
(1.25)
-0.172
(1.81)
-0.020
(0.27)
0.173
(3.35)
(5)
0.000
(0.00)
0.004
(0.05)
0.042
(0.85)
0.144
(5.34)
20001990
(6)
-0.214
(2.09)
0.062
(0.62)
0.071
(1.01)
-0.049
(1.15)
-0.088
(2.59)
-0.131
(5.25)
-0.044
(1.03)
0.066
(3.33)
0.197
(6.18)
-0.041
(0.68)
0.149
(3.93)
0.189
(2.67)
0.150
(5.35)
0.002
(0.03)
0.120
(1.51)
-0.277
(2.54)
33127
-0.027
-0.146 0.025 0.052
(0.49)
(1.53) (0.54) (0.73)
-0.115
0.162 -0.026 0.088
(1.12)
(1.08) (0.31) (0.66)
43754
64759
141640
1980
1980
1990
(7)
0.288
(3.38)
0.256
(2.71)
0.019
(0.23)
-0.007
(0.11)
0.208
(2.58)
0.001
(0.01)
-0.142
(1.15)
-0.029
(0.23)
-0.221
(1.52)
0.350
(1.91)
-0.482
(2.53)
(8)
0.288
(3.38)
0.256
(2.71)
0.019
(0.23)
-0.007
(0.11)
33127
(9)
0.392
(4.85)
0.137
(1.60)
0.001
(0.02)
0.261
(6.21)
19901980
(10)
0.104
(0.89)
-0.119
(0.93)
-0.018
(0.17)
0.268
(3.60)
2000
(11)
0.142
(1.82)
-0.040
(0.53)
0.044
(0.85)
0.268
(9.33)
20001990
(12)
-0.250
(2.22)
-0.177
(1.55)
0.043
(0.52)
0.006
(0.12)
0.176
(3.19)
0.127
(3.66)
-0.049
(0.75)
0.253
(11.04)
0.126
(3.02)
-0.138
(1.43)
-0.063
(1.19)
0.076
(0.69)
-0.011
(0.35)
0.051
(0.82)
-0.063
(0.51)
-0.291
(1.84)
33127
-0.137
-0.074
(1.91)
(0.52)
-0.218
0.072
(1.69)
(0.35)
43754
141640
0.059 0.197
(1.07) (2.16)
-0.213 0.006
(2.18) (0.03)
64759
50
Table A.II-5
Effect of Additional Education on Likelihood of Marriage
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
White Men Age 40-44
Currently Married
Three years Pooled (Ed-2)
Ed-1
9
10
11
12
13
14
15
16
17
18
19
1980
1980
(1)
0.254
(6.96)
0.069
(1.65)
0.063
(1.59)
0.130
(4.29)
-0.019
(0.68)
-0.064
(1.93)
-0.031
(0.90)
0.037
(1.12)
-0.038
(1.14)
0.061
(1.49)
-0.113
(3.04)
(2)
0.254
(6.96)
0.069
(1.65)
0.063
(1.59)
0.130
(4.29)
N
244044
-0.073
(4.31)
-0.004
(0.16)
-0.009
(0.32)
-0.081
(2.67)
244044
1990
19902000
1980
(4)
(5)
0.091 0.052
(1.80) (1.67)
-0.045 -0.042
(0.78) (1.30)
-0.179 -0.000
(3.51) (0.01)
0.099 0.261
(2.69) (17.81)
Ed-1
Ever Married
Three years Pooled (Ed-2)
2000
(11)
0.583
(14.76)
0.021
(0.47)
-0.090
(2.71)
0.219
(11.26)
20001990
(12)
-0.373
(5.75)
-0.053
(0.69)
0.058
(0.96)
0.027
(0.71)
0.011 0.004
(0.94) (0.18)
-0.079 0.200
(5.37) (8.70)
0.107 0.044
(4.83) (1.34)
0.202 0.069
(6.28) (1.47)
433549
51
Table A.II-6
Effect of Additional Education on Likelihood of Marriage
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
Black Men Age 40-44
Ed-1
1980
(1)
9
10
11
12
13
14
15
16
17
18
19
0.163
(2.34)
0.001
(0.01)
0.092
(1.37)
-0.000
(0.00)
0.042
(0.64)
-0.071
(0.87)
0.022
(0.25)
-0.048
(0.51)
-0.097
(0.84)
0.155
(1.08)
0.107
(0.76)
N
27343
Currently Married
Three years Pooled (Ed-2)
1980 1990 1990- 2000 20001980
1990
(2)
(3)
(4)
(5)
(6)
0.163
(2.34)
0.001
(0.01)
0.092
(1.37)
-0.000
(0.00)
0.221
(3.14)
-0.004
(0.06)
0.065
(1.17)
0.134
(3.61)
0.058
(0.59)
-0.005
(0.05)
-0.027
(0.31)
0.135
(2.08)
0.185
(2.67)
0.109
(1.60)
0.134
(2.86)
0.195
(7.35)
-0.037
(0.37)
0.113
(1.12)
0.068
(0.94)
0.061
(1.32)
0.001
(0.02)
0.169
(5.59)
0.168
(3.22)
0.267
(11.53)
0.098
(2.56)
-0.056
(0.78)
0.130
(2.90)
0.186
(2.20)
0.166
(4.70)
0.036
(0.62)
-0.024
(0.26)
0.189
(1.60)
27343
Ever Married
Ed-1
Three years Pooled (Ed-2)
1980 1980 1990 1990- 2000 20001980
1990
(7)
(8)
(9)
(10) (11) (12)
0.377
(4.26)
-0.067
(0.67)
0.214
(2.43)
0.066
(0.93)
0.258
(2.65)
-0.069
(0.56)
0.044
(0.32)
-0.250
(1.82)
0.084
(0.50)
-0.059
(0.29)
0.212
(1.06)
27343
0.377
(4.26)
-0.067
(0.67)
0.214
(2.43)
0.066
(0.93)
0.360
(4.51)
0.081
(0.93)
0.150
(2.24)
0.216
(4.68)
-0.017
(0.15)
0.148
(1.12)
-0.065
(0.59)
0.150
(1.76)
0.302
(4.33)
0.125
(1.80)
0.162
(3.32)
0.282
(9.88)
-0.057
(0.54)
0.044
(0.40)
0.012
(0.15)
0.066
(1.22)
0.056
(0.42)
0.182
(1.06)
27343
0.059 0.003
(0.61) (0.02)
0.161 -0.020
(1.05) (0.09)
35922
119181
0.202 0.143
(2.65) (1.16)
0.405 0.244
(3.06) (1.21)
55916
52
Table A.II-7
Effect of Additional Education on Likelihood of Motherhood (Using Mother)
(t-statistics in parentheses)
Women Age 40-44
Ed-1
1980
9
10
11
12
13
14
15
16
17
18
19
(1)
0.126
(4.36)
0.002
(0.05)
0.072
(2.70)
0.155
(7.80)
-0.056
(2.88)
-0.025
(1.00)
0.112
(3.91)
-0.010
(0.35)
-0.317
(10.27)
-0.151
(4.07)
-0.470
(12.28)
N
298382
(2)
0.126
(4.36)
0.002
(0.05)
0.072
(2.70)
0.155
(7.80)
-0.048
(3.69)
-0.084
(9.78)
-0.037
(2.36)
0.083
(10.94)
0.167
(14.57)
0.069
(3.40)
0.009
(0.82)
-0.060
(2.61)
0.026
(2.80)
0.017
(1.21)
-0.389
(15.42)
-0.549
(16.60)
298382
-0.297
0.092
(20.09)
(3.15)
-0.086
0.463
(3.34)
(11.03)
451241
1315673
-0.212
(15.43)
-0.021
(0.94)
0.084
(4.18)
0.065
(1.92)
566050
1980
2000 - 1990
(6)
-0.245
(6.38)
0.010
(0.24)
0.001
(0.03)
0.034
(1.69)
53
Table A.II-8
Effect of Additional Education on Likelihood of Motherhood (Using Mother)
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
All Women
Ed-1
1980
(1)
9
10
11
12
13
14
15
16
17
18
19
0.529
(15.91)
0.111
(3.01)
0.015
(0.48)
-0.223
(9.83)
-0.102
(6.64)
-0.243
(13.65)
-0.143
(7.84)
-0.342
(19.83)
-0.267
(15.63)
-0.279
(13.45)
-0.428
(18.12)
N
448973
30 34
Three years Pooled (Ed-2)
1980 1990 1990- 2000 20001980
1990
(2)
(3)
(4)
(5)
(6)
0.529
(15.91)
0.111
(3.01)
0.015
(0.48)
-0.223
(9.83)
-0.307
(31.77)
-0.522
(44.01)
-0.399
(28.74)
-0.575
(27.40)
448973
0.431
(15.19)
0.026
(0.83)
-0.148
(6.26)
0.051
(3.52)
-0.097
(2.23)
-0.085
(1.77)
-0.164
(4.11)
0.274
(10.19)
0.206
(7.47)
-0.036
(1.13)
-0.091
(3.58)
0.076
(5.34)
-0.225
(5.67)
-0.062
(1.39)
0.057
(1.66)
0.025
(1.24)
Ed-1
1980
(7)
0.411
(12.37)
0.076
(2.10)
0.100
(3.19)
0.006
(0.24)
-0.078
(4.01)
-0.304 0.003 -0.174 0.131 -0.097
(39.15) (0.25) (20.94) (11.50) (4.09)
-0.078
(3.04)
-0.598 -0.075 -0.548 0.050 -0.144
(69.09) (5.14) (60.98) (3.99) (5.80)
-0.317
(12.91)
-0.358 0.041 -0.332 0.026 -0.190
(23.84) (2.00) (24.60) (1.30) (6.52)
-0.004 0.571 -0.111 -0.107 -0.441
(0.18) (17.38) (4.99) (3.17) (14.37)
542553
496148
1487674
357751
35 39
Three years Pooled (Ed-2)
1980 1990 1990- 2000 20001980
1990
(8)
(9)
(10)
(11)
(12)
0.411 0.295
(12.37) (9.70)
0.076 0.050
(2.10) (1.46)
0.100 -0.047
(3.19) (1.77)
0.006 0.126
(0.24) (7.80)
-0.034 0.157
(4.24) (13.23)
-0.406
(20.19)
-0.542
(20.26)
357751
-0.306 0.100
(21.88) (4.07)
-0.075 0.466
(3.10) (12.91)
504186
1429217
0.072
(5.35)
-0.250 0.057
(17.90) (2.86)
-0.006 0.069
(0.27) (2.10)
567280
54
Appendix III
Results for Cohabitors
1980
2000
1990
83.9
67.5
8
10
11
12
13
14
education
15
16
17
18
19
1990
87.2
69
8
10
11
12
13
14
education
15
16
17
18
19
55
Table A.III-1
Percentage Currently Married or Cohabiting, by Education Level
Female Age 40-44
(Corresponds to Figure A.III-1)
Edu-1
(1)
1980
(2)
1980
(5)
2000
(6)
2000-1990
Education
8
77.1**
77.1**
71.9**
-5.2**
73.6**
1.7**
9
80.0**
80.0**
76.8**
-3.2*
72.3*
-4.5**
10
81.7**
81.7**
77.0
-4.7*
71.1
-5.9
11
81.2
81.2
75.5**
-5.7
70.0*
-5.5**
12
83.9**
83.6**
80.0**
-3.6**
76.1**
-3.9**
13
81.7**
.
.
.
.
.
14
80.4**
80.1**
76.3**
-3.8
73.8**
-2.5**
15
79.4*
.
.
.
.
.
16
82.6**
81.2**
78.6**
-2.6
77.4**
-1.2**
17
77.3**
.
.
.
.
.
18
74.9**
74.9**
74.7**
-0.2**
74.7**
0.0
19
67.5**
67.5**
73.3*
5.8**
75.0
1.7**
Success
-16.4
-16.1
-6.7
9.4
-1.1
5.6
Gapa
Pr(Success
0.003
0.000
0.000
0.001
0.000
0.000
Gap)=0a
Success
-3.0
13.1
1.9
Gapb
Pr(Success
0.000
0.447
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
56
Table A.III-2
Percentage Currently Married or Cohabiting, by Education Level
Male Age 40-44
(Corresponds to Figure A.III-2)
Edu-1
(1)
1980
(2)
1980
(5)
2000
(6)
2000-1990
Education
8
80.6**
80.6**
74.8**
-5.8**
73.1**
-1.7**
9
83.6**
83.6**
78.1
-5.5
70.8**
-7.3**
10
84.4
84.4
79.1
-5.3
69.0**
-10.1**
11
85.0
85.0
76.9**
-8.1**
69.3
-7.6**
12
87.1**
87.1**
81.0**
-6.1*
73.9**
-7.1**
13
87.1
.
.
.
.
.
14
86.7
86.5*
82.0**
-4.5**
76.1**
-5.9**
15
86.1
.
.
.
.
.
16
86.9*
86.7
82.8**
-3.9
79.2**
-3.6**
17
86.3
.
.
.
.
.
18
87.2*
87.2
84.8**
-2.4**
82.3**
-2.5
19
86.4*
86.4
86.5**
0.1**
84.9**
-1.6**
Success
-0.7
-0.7
5.5
6.2
11.0
5.5
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
4.5
5.2
6.5
Gapb
Pr(Success
0.000
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
57
Table A.III-3
Effect of Additional Education on Likelihood of Current Marriage or Cohabiting
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
Ed-1
1980
(1)
9
10
11
12
13
14
15
16
17
18
0.171
(5.71)
0.113
(3.44)
-0.036
(1.29)
0.190
(9.17)
-0.156
(7.74)
-0.084
(3.37)
-0.062
(2.21)
0.207
(7.28)
-0.335
(10.66)
-0.130
(3.47)
-0.363
(9.30)
All Women
Three years Pooled (Ed-2)
1980 1990 19902000 20001980
1990
(2)
(3)
(4)
(5)
(6)
0.171
(5.71)
0.113
(3.44)
-0.036
(1.29)
0.190
(9.17)
0.258
0.087
-0.068
(8.71)
(2.08)
(2.45)
0.010
-0.103 -0.057
(0.31) (2.21) (1.83)
-0.081 -0.045 -0.053
(1.17)
(3.16)
(2.25)
0.261
0.070
0.307
(16.10) (2.66) (22.64)
-0.327
(8.03)
-0.067
(1.48)
0.028
(0.81)
0.046
(2.17)
-0.225 -0.217
(17.09) (23.57)
0.008
(0.51)
-0.119
(15.00)
0.097
(8.00)
0.131
(6.35)
-0.001
(0.04)
0.192
(19.38)
0.062
(4.09)
0.130
(11.23)
-0.396 -0.220
0.176
-0.146 0.074
(15.43) (13.77) (5.83)
(9.85) (3.39)
19
0.018
-0.432 -0.069
0.363
0.087
(0.74) (2.37)
(12.77) (2.49)
(8.27)
N
298382
451241
566050
298382
1315673
Ed-1
1980
(7)
0.205
(6.44)
0.060
(1.65)
0.043
(1.27)
0.179
(6.80)
-0.004
(0.16)
-0.030
(0.95)
-0.050
(1.55)
0.062
(1.98)
-0.050
(1.55)
0.082
(2.05)
-0.073
(2.02)
285184
All Men
Three years Pooled (Ed-2)
1980 1990 1990- 2000 20001980
1990
(8)
(9)
(10)
(11)
(12)
0.205 0.185 -0.020 -0.113 -0.298
(6.44) (6.10) (0.45) (4.46) (7.54)
0.060 0.056 -0.004 -0.087 -0.143
(1.65) (1.61) (0.08) (3.11) (3.19)
0.043 -0.125 -0.168 0.016 0.141
(1.27) (4.42) (3.79) (0.75) (3.99)
0.179 0.244 0.065 0.226 -0.018
(6.80) (13.48) (2.04) (18.22) (0.83)
0.115
(13.60)
0.051
(3.78)
0.016
(0.74)
0.060 0.044
(5.00) (1.77)
0.182
(16.68)
0.122
(7.52)
-0.011
(0.41)
-0.031
(1.05)
285184
58
Appendix IV:
Results for Outcome Mom
1980
1990
94
69.2
9
10
11
12
13
14
educ ation
15
16
17
18
19
1990
93.6
63.3
8
10
11
12
13
14
educ ation
15
16
17
18
19
1990
92
44.7
8
10
11
12
13
14
educ ation
15
16
17
18
19
59
Table A.IV-1
Percentage Mothers, by Education Level (Using Mom)
Female Age 40-44
(Corresponds to Figure A.IV-1)
Edu-1
(1)
(2)
(4)
Education
1980
1980
1990
1990- 1980
8
87.2**
87.2**
85.5**
-1.7**
9
92.2**
92.2**
92.1**
-0.1
10
92.9*
92.9*
92.9*
0.0
11
94.0**
94.0**
91.1**
-2.9**
12
90.8**
90.8**
89.0**
-1.8**
13
90.7
.
.
.
14
89.1**
89.1**
85.7**
-3.4**
15
89.1
.
.
.
16
85.6**
84.2**
77.8**
-6.4
17
80.6**
.
.
.
18
77.5**
77.5**
69.9**
-7.6**
19
69.2**
69.2**
70.6
1.4**
Success
-21.6
-21.6
-18.4
3.2
Gapa
Pr(Success
0.000
0.000
0.000
0.000
Gap)=0a
Success
-15.1
6.5
Gapb
Pr(Success
0.000
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
**: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (1%).
Cells in bold: median of education variable for that year.
a: Success Gap defined relative to 12 years of education.
b: Success Gap defined relative to median education (if median is not 12 years of education).
60
Table A.IV-2
Effect of Additional Education on Likelihood of Motherhood (Using Mom)
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
All Women
Age
Education
8
9
10
11
12
13
14
15
16
17
18
19
30 34
Edu-1
35 39
1980
1980
1990
(1)
81.8**
90.9**
91.7*
92.0
85.5**
84.1**
79.5**
77.0**
68.0**
62.0**
55.0**
44.7**
(2)
81.8**
90.9**
91.7*
92.0
85.3**
.
78.7**
.
66.4**
.
55.0**
44.7**
(3)
82.6**
89.9**
90.1
86.4**
82.2**
.
74.5**
.
58.2**
.
49.0**
49.8
Edu-1
19901980
(4)
0.8**
-1.0**
-1.6
-5.6**
-3.1**
.
-4.2**
.
-8.2**
.
-6.0
5.1**
1980
1980
1990
(5)
85.0**
91.6**
92.7**
93.6**
89.6**
88.8**
86.8**
85.7**
81.5**
76.1**
72.0**
63.3**
(6)
85.0**
91.6**
92.7**
93.6**
89.5**
.
86.4**
.
80.0**
.
72.0**
63.3**
(7)
84.0**
91.4**
91.8
89.7**
86.5**
.
81.4**
.
71.7**
.
63.8**
63.5
Success
-40.8
-40.6
-32.4
8.2
-26.3
-26.2
-23.0
Gapa
Pr(Success
0.000
0.000
0.000
0.000
0.000
0.000
0.000
Gap)=0a
Success
-17.9
Gapb
Pr(Success
0.000
Gap)=0b
*: Columns 2, 4, 6: Ho: E , y = 0 rejected; Columns 3, 5: Ho: E , y E , y 10 = 0 rejected (5%).
19901980
(8)
-1.0**
-0.2
-0.9
-3.9**
-3.0**
.
-5.0**
.
-8.3**
.
-8.2**
0.2**
3.2
0.000
8.3
0.000
61
Table A.IV-2
Effect of Additional Education on Likelihood of Motherhood (Using Mom)
(t-statistics in parentheses)
Women Age 40-44
9
10
11
12
13
14
15
16
17
18
19
Ed-1
1980
(1)
0.547
(12.78)
0.101
(2.06)
0.179
(4.03)
-0.456
(13.72)
-0.014
(0.51)
-0.175
(5.38)
-0.005
(0.15)
-0.316
(9.12)
-0.356
(10.63)
-0.187
(4.76)
-0.430
(10.72)
N
298382
1980
(2)
0.547
(12.78)
0.101
(2.06)
0.179
(4.03)
-0.456
(13.72)
-0.131
(7.73)
-0.303
(26.46)
-0.172
(8.40)
-0.379
(16.09)
-0.533
(43.42)
-0.154
(5.80)
-0.445
(16.31)
-0.528
(15.21)
298382
-0.410
(26.72)
0.033
(1.22)
0.035
(1.11)
0.561
(12.76)
451241
749623
62
Table A.IV-3
Effect of Additional Education on Likelihood of Motherhood (Using Mom)
Incremental Effects of Additional Year of Education from Logit Model
(t-statistics in parentheses)
All Women
Ed-1
1980
(1)
9
10
11
12
13
14
15
16
17
18
19
0.800
(19.43)
0.098
(2.06)
0.035
(0.85)
-0.661
(22.76)
-0.112
(6.85)
-0.307
(16.29)
-0.147
(7.75)
-0.454
(25.62)
-0.265
(15.42)
-0.290
(13.91)
-0.413
(17.50)
N
448973
30 34
Three years Pooled (Ed-2)
1980
1990
1990- 1980
(2)
(3)
(4)
0.800
(19.43)
0.098
(2.06)
0.035
(0.85)
-0.661
(22.76)
0.630
(16.76)
0.021
(0.51)
-0.358
(11.32)
-0.321
(18.20)
-0.170
(3.04)
-0.076
(1.20)
-0.393
(7.57)
0.340
(10.00)
-0.366
(35.71)
-0.455
(54.53)
-0.089
(6.71)
-0.655
(53.97)
-0.742
(84.35)
-0.087
(5.83)
-0.402
(28.83)
-0.566
(26.97)
-0.372
(24.74)
0.031
(1.22)
0.030
(1.46)
0.597
(18.18)
448973
542553
991526
Ed-1
1980
(5)
0.656
(15.44)
0.141
(2.91)
0.140
(3.24)
-0.519
(16.37)
-0.086
(3.96)
-0.187
(7.18)
-0.096
(3.50)
-0.302
(11.59)
-0.325
(13.15)
-0.214
(7.33)
-0.401
(13.05)
357751
35 39
Three years Pooled (Ed-2)
1980
1990
1990- 1980
(6)
(7)
(8)
0.656
(15.44)
0.141
(2.91)
0.140
(3.24)
-0.519
(16.37)
0.705
(16.42)
0.053
(1.05)
-0.257
(6.69)
-0.302
(14.08)
0.050
(0.82)
-0.088
(1.26)
-0.398
(6.87)
0.217
(5.66)
-0.236
(17.08)
-0.386
(39.22)
-0.150
(8.82)
-0.435
(24.34)
-0.546
(53.87)
-0.112
(5.44)
-0.426
(21.05)
-0.514
(19.15)
-0.361
(25.74)
-0.013
(0.54)
0.065
(2.64)
0.501
(13.81)
357751
504186
861937
63