Gender Differences in Remarriage: Marriage Formation and Assortative Mating After Divorce

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Kevin M. Shafer

Graduate Program in Sociology

The Ohio State University

2009

Dissertation Committee:

Zhenchao Qian, Advisor

Chris Knoester

Reanne Frank

Kevin M. Shafer

2009

Abstract

Divorce and subsequent remarriage have become an important part of American family life in recent decades. Divorce has negative consequences for mental health, physical health, overall well-being, and economic well-being. Remarriage can help divorcees overcome many of these problems, especially financial difficulties. However, there are significant gender differences in the likelihood of remarriage. Men are much more likely to remarry than women. This gender gap has important micro- and macro- level social implications. At the individual level, gender differences in remarriage mean that men are more likely to regain the benefits associated with marriage, including economic benefits, at higher rates than women. At the macro-level, some family scholars argue that first marriage has increasingly resembled remarriage in recent decades

(Popenoe 1993; Cherlin 2004). In first marriage, both men's and women's socioeconomic status is positively associated with first marriage formation and both educational and age homogamy--a move away from traditional marriage where men's, but not women's, economic status was important in the marriage market. However, empirical work focusing on the claim that first marriage and remarriage formation are similar is lacking because little is known about remarriage formation patterns.

I focus on two aspects of remarriage formation to understand how remarriage patterns compare to first marriage formation. First, I analyze the individual iii

characteristics associated with the likelihood of remarriage for men and women. I pay particular attention to socioeconomic status (income, labor force status, and educational attainment), first marriage ties (co-residential children) and time since divorce as important factors in the remarriage market. Second, I examine educational and age assortative mating patterns in second marriages to identify whether remarriage follows a more contemporary homogamous pattern or a more traditional form where men marry less-educated and younger women. I use data from the National Longitudinal Survey of

Youth, 1979 cohort (NLSY79) to answer these questions.

The results show remarriage formation patterns consistent with traditional first marriages where current and long-term economic status has a positive effect on remarriage for men, but not for women. For women, family background, race/ethnicity, age and parental status are associated with remarriage. The analysis of assortative mating shows that high-status men tend to marry less-educated, young women and both educational and age homogamy is unlikely for both men and women, regardless of socioeconomic status. These patterns are different from those present in contemporary first marriages where both men and women emphasize on spousal economic status as valuable traits. These findings suggest that remarriage differs from first marriage in its likelihood and assortative mating patterns and provide a foundation for future work to incorporate gender dynamics in research on remarriage.

Acknowledgments

I wish to thank my advisor Zhenchao Qian for his assistance, patience (which I

tested numerous times), and counseling along the way. I chose Zhenchao as an advisor six years ago as a first year MA student, never thinking he would see me through this entire process. I picked him from a number of capable Sociology faculty members because I thought he seemed like a good person to emulate—both professionally and personally. I am grateful to call him my mentor.

I am also grateful to Chris Knoester, Bob Kaufman, Chris Browning, and Doug

Downey for their help in my graduate student career. Chris Knoester was helpful as a dissertation and general exam committee member. Bob, Chris Browning, and Doug helped me immensely as a younger graduate student. I want to make special mention to two additional faculty members. First, Elizabeth Cooksey was tremendously helpful and encouraging throughout my graduate school career. She was originally to be a member of my committee, but her father’s illness prevented her from serving on the committee in the end. Second, Reanne Frank stepped in at the last minute as a pinch-hitter for

Elizabeth. Thank you for your flexibility and kindness. Also, an additional thank you to

Kelly Hopkins who took care of some issues with the graduate school to insure that all would go well for the defense (and she kept me calm in the process).

I wish to acknowledge the help of the Department of Sociology at Ohio State. I have spent the past 10 years as a student in the department--first as an undergraduate and now as a graduate student. I have grown intellectually and personally in the wonderful environment they provide. I also thank the Initiative in Population Research at Ohio

State and the Department of Sociology at Brigham Young University in Provo, Utah.

They provided, like the sociology department at Ohio State, much needed resources for my research. Colleagues in each department provided helpful comments and suggestions along the way.

I thank my parents, Glenn and Patti, for their encouragement and love in my many years as a student. When I told them I wanted to major in sociology as an undergraduate, they did not laugh. Nor did they laugh when I told them I wanted to be a professor.

Instead, they always persuaded me to keep striving for that goal. I am grateful for their continued positive support.

I also want to thank my many friends and other family members. Foremost among these individuals is Benjamin Gibbs. Not only has he been a trusted friend who has intellectually pushed and prodded me, he helped change my life in an inalterable way.

I have become a better person because of his example, advice, and friendship. For this, I will be eternally grateful. I also wish to thank Danielle Gibbs, Peter and Marilee Merrell,

April Randall, Matthew Randall, Jonathan and Allie Randall, and Amy Randall for their friendship, love, and encouragement.

Most importantly, the vast majority of the credit for this project goes to my wife

Melissa and our son Clark. They have been patient, understand, loving, and supportive

throughout the entire process. The sacrifices they made along the way are greater than

anyone else's. Their contribution has been immeasurable and beyond anything anyone

could ever hope for. Melissa, and certainly Clark, did not know what they were in for

when my wife said "yes" on a cold, rainy Salt Lake City day. I know that the last several months have not been easy for them. But, anything good or praiseworthy in this dissertation stands as a testimony to their prominence in this process and in my life more

generally. I love you both more than I could ever put into words. The simplest thing I

can say is thank you.

Vita

July 20, 1980 …………………………….Born – Columbus, Ohio

2002 ……………………………………...B.A. Sociology, The Ohio State University

2005………………………………………M.A. Sociology, The Ohio State University

2008………………………………………Visiting Assistant Professor of Sociology,

Brigham Young University

2004 to present…………………………..Graduate Research Associate, The Ohio State

University

2003 to present …………………………..Graduate Teaching Associate, The Ohio State

University

Publications

Shafer, Kevin and Zhenchao Qian. Forthcoming. “Marriage Timing and Educational Assortative Mating. Journal of Comparative Family Studies.

Fields of Study

Major Field: Sociology

vii

Table of Contents

Abstract...... ii

Acknowledgements...... iv

Vita...... vii

List of Tables ...... x

List of Figures...... xii

Chapter 1: Introduction...... 1

Chapter 2: Theoretical Perspectives...... 9

Chapter 3: Data, Measures, and Analytic Strategy...... 37

Chapter 4: Gender Differences in Remarriage Formation...... 53

Chapter 5: Educational and Age Assortative Mating in Remarriage...... 82

Chapter 6: Conclusion...... 127

References...... 139

Appendix A: Supplemental Results...... 155

viii

Appendix B: Supplemental Results ...... 160

Appendix C: Supplemental Results ...... 165

List of Tables

Table 1: Descriptive Statistics of Key Independent Variables by Remarriage Outcomes for Men and Women ...... 77

Table 2: Discrete-Time Logistic Regression Results for Remarriage, by Sex ...... 79

Table 3: Differences Between Respondent and Spouse on Key Characteristics, by Sex...... 115

Table 4: Assortative Mating on Education and Age Among Remarried Men and Women,

by Educational Attainment ...... 116

Table 5: Descriptive Statistics of Key Independent Variables by Educational Assortative

Mating Outcomes for Men and Women ...... 117

Table 6: Discrete-Time Multinomial Logistic Regression Results for Educational

Assortative Mating Outcomes, by Sex...... 119

Table 7: Descriptive Statistics of Key Independent Variables by Age Assortative Mating

Outcomes for Men and Women...... 121

Table 8: Discrete-Time Multinomial Logistic Regression Results for Age Assortative

Mating Outcomes for Men and Women, Main Effects...... 123

Table 9: Discrete-Time Multinomial Logistic Regression Results for Age Assortative

Mating Outcomes for Men and Women, Interactive Model...... 125

Table 10: Discrete-Time Multinomial Logistic Regression Results for Educational

Assortative Mating, Full Results...... 156

Table 11: Discrete-Time Multinomial Logistic Regression Results for Age Assortative

Mating Outcomes for Men and Women, Full Results--Direct Effects Model...... 161

Table 12: Discrete-Time Multinomial Logistic Regression Results for Age Assortative

Mating Outcomes for Men and Women, Full Results--Interactive Model...... 166

List of Figures

Figure 1: Discrete-Time Logit Model...... 49

Figure 2: Discrete-Time Multinomial Logit Model...... 50

Figure 3: Log Odds of Remarriage for Men and Women by Years Since

Separation/Divorce ...... 81

xii Chapter 1: Introduction

Due to the high prevalence of divorce since the late 1960s, many Americans have experienced remarriage themselves or as the child of a remarried parent. Indeed, more than 20% of Americans over the age of 40 have been married at least twice (U.S. Census

Bureau 2007) and almost half of all marriages in a year involve at least one previously- married person (Ozawa and Yoon 2002). Remarriage not only provides emotional support and affection lost at divorce but also re-establishes many benefits of marriage

(Cherlin 1992; Waite 1995). Mental health, physical health, and overall well-being are all positively associated with remarriage (Coleman et al. 2000). In addition, remarrying often means an increase in economic stability, especially for many women and children who experience a decline in financial well-being and per-capita income after divorce

(Duncan and Hoffman 1985; Holden and Smock 1991; Peterson 1996).

Remarriage has important macro-level implications. Some social commentators argue that a decline in marriage and a rise in both cohabitation and divorce indicate that the institution of marriage and family has weakened (Popenoe 1993). Increases in remarriage, however, suggest that marriage remains valued as an important institution, although serial monogamy has become more normative than lifetime monogamy

(deGraaf and Kalmijn 2003; Cherlin 2004). One neglected aspect of

remarriage is that remarriage is a selective process. Knowing which divorced men and

women select into remarriages helps identify if family decline and family change vary

across socioeconomic status, parental background, age, and other individual

characteristics.

Racial/ethnic minorities, less-educated individuals, and the unemployed are faced

with an increased likelihood of dissolution (Koo 1984; Wu and Balkrishnan 1994;

Sweeney 1997; Teachman 2002). Life-course events, such as pre-marital cohabitation, parental divorce, and non-marital childbearing, are also associated with divorce (Sweet and Bumpass 1987; Schoen 1992; Wolfinger 1999; Teachman 2002). Among the divorced, some are more likely to remarry than the others; the likelihood of remarriage varies across attributes such as age, race/ethnicity, parental status, educational attainment, labor force status, and income (e.g., Koo and Suchindran 1980; Mott and Moore 1983;

Koo et al. 1984; Kalmijn 1987; Bumpass et al. 1990; Chiswick and Lehrer 1990; Klein

1990; Wu 1994; Sweeney 1997; Lampard and Peggs 1999; deGraaf and Kalmijn 2003;

Wu and Schimmele 2005).

One notable difference is the gender gap in remarriage – men are much more likely to remarry than women after divorce (Bumpass et al. 1990; South 1991; Coleman et al. 2000; deGraaf and Kalmijn 2003; Wu and Schimmele 2005). Some estimates suggest that divorced men are up to 25% more likely to remarry than divorced women

(CDC 2002). Although prior research has not explicitly focused on gender differences in remarriage, several factors correlated with remarrying may help explain this gap. Many divorced men are attractive marriage partners because they experience an increase in

economic well-being after divorce. Many women, on the other hand, often experience financial hardship as a divorcee (Peterson 1996). Women are more likely to care for children after divorce, which can increase the difficulty of finding a new marriage partner

(Goldscheider and Sassler 2006). Divorced women are also older when they are in the marriage market, making them unattractive spousal candidates (England and Mclintock

2008). In total, divorced men seem to be better marriage candidates than divorced women and, as a result, have larger partner pools from which to choose a new spouse.

Researching the gender gap in remarriage helps us, as noted above, understand micro-level stratification and macro-level patterns in marriage and the family. However, there are other reasons to study gender differences in remarriage. The gender gap is demographically significant because family composition may differ across stages of the life-course. Divorce is a critical transition for men, women, and children. For example, because women are more likely to remain single after divorce than men, many women may move into a permanently single state and find themselves raising a child as a single parent.

The gender gap in remarriage also provides context for research with the goal of identifying the nature of remarriage. Many social commentators have argued that remarriage is an incomplete institution that is ungoverned by social norms (Cherlin 1978;

Cherlin 1992). Recently, some sociologists and demographers have argued that first marriage has also become a deinstitutionalized family form that mirrors remarriage

(Bumpass 1990; Popenoe 1993; Cherlin 2004). As a result, these researchers argue implicitly, if not explicitly, that first marriage and remarriage are indistinct from one

another. A number of trends provide evidence for this thesis, they argue, including the decline of traditional marriage with gendered expectations and norms (see Cherlin 2004,

849). Yet, any similarities or differences between first marriage and remarriage are empirically untested claims. Exploring gender differences in remarriage helps answer these questions.

Contemporary first marriage tends to be an economic arrangement (Sweeney

2002) – men and women with high socioeconomic status are more likely to marry and to marry one another than low-status individuals (e.g., Lichter et al. 1992; Qian and Preston

1993; Oppenheimer et al. 1995; Qian 1998; Rose 2004; Schwartz and Mare 2005). This represents a departure from the past when men’s, but not women’s, socioeconomic status was important for marriage. Historically, men formed hypergamous unions by marrying less-educated and younger women. These patterns supported marriages where men were actively engaged in the labor market while women pursued domestic production and care work (Becker 1981).

Remarriage may resemble either contemporary or traditional first marriage. If socioeconomic status influences remarriage formation for both men and women, then there is empirical support for the assertion that first and second marriages are similar to one another. There are good reasons to expect that this may be the case. In recent decades, women’s socioeconomic status has become increasingly important in marriage as men’s income growth has slowed, consumption patterns have changed, and the cost of raising a family has increased (see Sweeney 2002). If divorcees are subject to these same trends, then remarriage should mirror contemporary first marriage. Conversely,

remarriage may more closely resemble traditional first marriage. Typically, divorced

men and women are older than Never-married individuals when they marry (Sweeney

1997). They also tend to be more financially established than their Never-married

counterparts (Sweeney 1997). Men, in particular, enjoy a substantial increase in financial

well-being after divorce (Peterson 1996) which can influence their partner preferences. If

these factors support traditional marriage patterns, women should be evaluated on non-

economic criteria and men may seek out hypergamous, not homogamous, matches in the

remarriage market.

I use two strategies to answer the question of whether remarriage resembles either

traditional or contemporary first marriage. The first empirical study focuses on gender

differences in the individual characteristics associated with remarriage. How personal

traits predict remarriage formation reveals what attributes are attractive and unattractive in the remarriage market. Certain patterns are associated with contemporary marriage formation. In such unions, both men and women are expected to economically provide for the family. As a result, both men's and women's socioeconomic status is positively related to the likelihood of marriage (e.g., Mare 1991; Oppenheimer 1997; Sweeney

1997; Qian 1998; deGraaf and Kalmijn 2003). Although other factors, such as race/ethnicity, age, and parental status, matter in contemporary marriage formation, socioeconomic factors are most important (England and Farkas 1986; Oppenheimer

1988). This stands in sharp contrast with traditional marriage formation where men are evaluated on economic criteria, but women are evaluated on non-economic criteria such as age or social background characteristics (Becker 1981). Thus, distinguishing between

attractive and unattractive characteristics can indicate whether traditional or

contemporary patterns prevail in remarriage.

The second empirical study focuses on assortative mating, or who marries whom.

Studying how people sort on personal characteristics, such as educational attainment and age, provides insights into remarriage as an institution. Hypergamous matches where men are better educated and older than their wives would indicate that remarriage may be more traditional than first marriage where men and women normally marry individuals with similar attributes. If remarrying men and women are more likely make traditional matches, gender roles and expectations may be highly specialized (Becker 1981).

Assortative mating in remarriage can also indicate the quality of potential partners

available in the remarriage market. For example, the available partner pool tends to be

small if a number of marriages are hypogamous, where men have lower status than their

wives (Kalmijn 1998). Finally, a study of marital sorting provides additional insight into

which characteristics are attractive in the remarriage market and which are not.

Several studies have focused on the determinants of remarriage (e.g., Koo and

Suchindran 1980; Mott and Moore 1983; Koo et al. 1984; Kalmijn 1987; Bumpass et al.

1990; Chiswick and Lehrer 1990; Klein 1990; Wu 1994; Sweeney 1997; Lampard and

Peggs 1999; deGraaf and Kalmijn 2003; Wu and Schimmele 2005) and assortative

mating in remarriage (e.g., Dean and Gurak 1978; Mueller and Pope 1980; Jacobs and

Furstenberg 1986; Ni Brolchain 1988; White 1990; Gelissen 2004; Ono 2005) but no

studies, to my knowledge, have focused explicitly on gender and the remarriage

formation process (deGraaf and Kalmijn 2003). Instead, previous studies on remarriage

formation have tended to focus exclusively on the demographic or economic

characteristics associated with remarriage. Research on assortative mating has almost

exclusively studied changes in individual matches between first and second marriages

(e.g., Gelissen 2004). This project fills a significant gap in the literature by documenting

gender differences in the factors associated with remarriage entry and patterns of

educational and age assortative mating.

Significance of this Study

The contribution of this study is in the area of family sociology and demography,

with particular attention paid to the marriage formation and assortative mating literatures.

Prior studies in these areas have almost exclusively examined first marriage or prevailing

marriages, while ignoring the significance of remarriage. Furthermore, prior research has failed to focus gender differences in the formation of and assortative mating patterns present in second marriages. This study addresses this gap in the literature. Prior studies conducted on remarriage, in general, have used data from European countries and

Canada. Research in the United States has been somewhat limited by the availability of nationally representative, longitudinal data which provides a complete marital history for both men and women. This study builds on prior contributions while addressing this shortcoming through the use of high quality, nationally representative data from the

United States.

Summary

In summary, this project focuses on the gender gap in remarriage and both educational and age assortative mating patterns in remarriage. By focusing on these

aspects of remarriage following a divorce, this study intends to better understand the

preferences of men and women when marrying a previously-married individual, gender

dynamics in second marriages, and, more generally, the remarriage market. The findings

of this study have implications for understanding how first marriage and remarriage differ

from one another, later life-course stratification by sex, and individual well-being

following a divorce. In addition, the results also address the macro-level issue of family change or decline.

Chapter 2 outlines findings from first marriage formation and potential differences between first marriage and remarriage within the context of marital search theory. This chapter provides a framework for understanding the empirical findings of this project. In Chapter 3, I discuss the data, variables, and analytic strategy for my analyses. Chapter 4 focuses on the gender gap in remarriage. In this chapter I present results from event-history analyses which describe how socioeconomic characteristics, first marriage ties, and market restrictions affect the timing and likelihood of remarriage for men and women differently. Chapter 5 investigates educational and age assortative mating patterns in remarriage. Using event-history methods, this chapter focuses on how the above characteristics affect who marries whom when forming a second marriage.

Finally, Chapter 6 provides a discussion and conclusion of the empirical findings from

Chapters 4 and 5. This chapter outlines how remarriage and first marriage differ and on the preferences of and constraints on divorcees when remarrying.

Chapter 2: Theoretical Perspectives

Divorce has many negative consequences which are sometimes offset by

remarriage. Compared to continuously married individuals, divorcees tend to do worse

on a number of outcomes (Amato 2000). They have lower levels of psychological well- being, lower happiness, and more psychological distress than married men and women

(e.g., Umberson and Williams 1993; Davies et al. 1997; Simon and Marcussen 1999).

Divorcees report more health problems and have higher mortality rates than currently

married individuals (e.g., Lillard and Waite 1995; Murphy et al. 1997; Aldous and Ganey

1999). Divorced individuals, especially divorced women, also have a lower standard of

living, less wealth, and more economic instability than married couples (e.g., Hao 1996;

Marks 1996; Smock et al. 1999). Divorce also makes many families into households headed by single mothers. Divorced women with children are disproportionately in poverty; with recent studies estimating about half of divorced women with children live below the poverty line (Smock 1994; U.S. Census Bureau 2007). This can have strong negative consequences for children including a higher likelihood of dropping out of high school, having children out-of-wedlock, and living in poverty as adult (McLanahan and

Bumpass 1988). In addition to economic consequences, divorced parents report higher

stress and more difficulty than married parents in raising children (e.g., Thomson et al.

1992; Umberson and Williams 1993; Fisher et al. 1998).

Cherlin (1991) suggests that remarriage can help alleviate some, but not all, of the problems associated with divorce. Some evidence shows that remarried individuals experience less distress than divorced men and women (Shapiro 1996). Others report that remarriage slightly improves physical health among divorcees (Richards et al. 1997).

Remarriage substantially improves economic circumstances for divorcees, especially for women (Ozawa and Yoon 2002).

Gender and Remarriage

Women are less likely than men to remarry after a divorce (Bumpass et al. 1990;

South 1991; Coleman et al. 2000; deGraaf and Kalmijn 2003; Wu and Schimmele 2005)

and the gender gap in remarriage appears large. Men are about 25% more likely to

remarry than women following a divorce. Among 45-64 years olds, the rates of

remarriage among divorced men are double those of divorced women. The existence of a

gender gap in remarriage means that any benefits to remarriage will be distributed

unequally among men and women. Accordingly, the negative consequences of divorce

may be long-term for women, but not men. One important implication of this is that

remarriage differentials can solidify economic differences between men and women later

in the life course.

While studies have focused on the implications of remarriage for men and women

(see Coleman et al. 2000 for an overview of research to-date) or on stepfamily dynamics

(e.g., Cherlin 1978; Cherlin and Furstenberg 1994), few studies have researched the

relationship between gender and remarriage. Existing studies on remarriage are based on a small set of variables, focus on one sex, or do not take into account first marriage effects. In addition, they fail to differentiate between remarriage after divorce and remarriage after widowhood (Wu and Schimmele 2005). Still others only focus on one sex, but do not examine the marriage patterns of both sexes (Bumpass et al. 1990; Smock

1990). Therefore, gender differences in remarriage are not well understood.

Gender dynamics in the formation of remarriage may be traditional. Today, the economic prospects of both men and women are important to first marriage formation

(Sweeney 2002). Contemporary first marriage formation differs from the past when marriage patterns reflected specialization in sex roles (Becker 1981). Historically, men were expected to focus on outside employment and the economic well-being of the family and women were expected to engage in domestic labor. As a result, traditional marriage formation meant that men with good socioeconomic prospects and less- educated women were most likely to marry (Goldstein and Kenney 2001). In first marriage, the move away from traditional marriage patterns is the result of stronger income growth for women then men (Wetzel 1995), more egalitarian gender role attitudes (Thornton 1989), the necessity for a second income as consumption patterns have changed (Bumpass 1990) and other rising cost of raising children (England and

Folbre 1999).

In theory, divorcees should be affected by the same macro-level trends as the

Never-married. However, because divorced men and women enter the marriage market at later ages than the never-married, their personal characteristics are different. Men

often experience an increase in economic well-being after divorce while women experience a decline in well-being. These trends, coupled with first marriage experiences, may lead to traditional patterns in remarriage. Men who get an economic boost after divorce may not need to marry a high-status woman. As a result, highly-educated or high earning divorced women may be unlikely to marry if remarriage is traditional.

There are several reasons to expect that divorcees’ attitudes about marriage may differ from never-married individuals’ attitudes. First, because divorcees have been previously-married, they may have more traditional attitudes about marriage than never- married men and women. Married people profess more traditional values about marriage than never-married individuals1 (Bartling and Broussard 1999). Second, many divorced couples have children which may lead to more traditional values about marriage than never-married individuals who are often childless (Morgan and Waite 1987). Third, divorce is selective of less-educated individuals and divorcees are older than never- married men and women. These individuals tend to have more traditional views of marriage than younger, better educated men and women (Wu and Balkrishnan 1994).

Finally, women who drop out of the labor force often have more traditional gender role attitudes than working women (Stokes and Peyton 1986). Because many divorced women dropped out of the labor force while married (deGraaf and Kalmijn 2003), they

1 An alternative perspective suggests that divorce may liberalize attitudes toward more equitable, and less traditional, marriages (Pyke 1994; Pyke and Coltrain 1996). Specifically, divorced women may seek more power in remarriages (Burgoyne and Morison 1997) and bring more resources to remarriages (Crosbie- Burnett and Giles-Simes 1991; Pyke and Coltrain 1996), than first marriages, making remarriages less traditional. However, Amato and Booth (1991) show that any liberalization of attitudes about marriage and family among divorcees is limited to attitudes about divorce, not about gender roles within marriage. These considerations suggest that remarriage matches are likely to be traditional ones.

may have more traditional preferences for remarriages. In total, the effects of post- divorce economic status, first marriage, and individual attitudes about marriage may mean that remarriage will take on a traditional form.

Differences in Marriage Formation between First Marriages and Remarriages

If remarriage is traditional, remarriage formation differs significantly from contemporary first marriage formation. The qualities of and complications facing previously-married individuals may differ from those of the never-married. First, remarried men and women are, on average, older than people marrying for the first time.

As a result, many divorced individuals are more economically established than never- married men and women (deGraaf and Kalmijn 2003). In first marriage, men and women often select partners based on socioeconomic potential, not current economic status

(England and Farkas 1986; Oppenheimer 1988). Educational attainment (Schwartz and

Mare 2005), occupational stability (Oppenheimer 2003), and other economic indicators are important in first marriage because they help people evaluate potential partners.

Unlike first marriage, people who remarry may not need to use proxies for socioeconomic status. At older ages, occupational status and income earned provide clearer information about a potential spouse’s socioeconomic status (Sweeney 1997). For example, Kalmijn (1994) shows that occupational status has a stronger effect on marriage for older individuals than it does for younger men and women. Thus, remarriage formation might be more traditional than first marriage because economic circumstances differ over the life course.

Compositional differences between the first marriage and remarriage markets may

also contribute to distinctions between first marriages and remarriages. The first

marriage market is larger and more homogeneous than the remarriage market (Kalmijn

1998; Gelissen 2004). Thus, there are more well-matched potential partners in the first

marriage market than the remarriage market. In remarriage, divorcees pick partners from

a small, heterogeneous market made up from a select group of divorcees and older never-

married individuals (Gelissen 2004). Because the remarriage market is thin, some men

and women may not remarry (Lewis and Oppenheimer 2000). As a result, partner

searches in remarriage may differ substantially from strategies employed in first

marriage.

Unlike Never-married individuals, divorcees have prior marital experience which

may influence their search for a new spouse. Many divorced women care for children

from their previous marriages which can affect their prospects of remarrying

(Goldscheider and Sassler 2006). Children add complexity to new relationships because

they often cause financial, emotional, and social hardship for the new spouse (Coleman et al. 2000). Although men are rarely a child’s primary caregiver after divorce, parental

status can affect their ability to remarry as well. Many divorced men make child support

payments, pay alimony and have other financial obligations to their ex-wives which can

be problematic in the remarriage market (deGraaf and Kalmijn 2003). Other first

marriage experiences, such as prior cohabitation history (Teachman et al. 2000) and first

marriage duration (Bumpass et al. 1990) can also affect the prospects of remarrying.

Importantly, divorcees are a select group, which means that the characteristics of individuals seeking a remarriage differ from those seeking a first marriage. Divorce rates vary across individual characteristics and groups. Individually, pre-marital cohabitation

(Schoen 1992), parental divorce (Wolfinger 1999; Teachman 2002), non-marital childbearing (Sweet and Bumpass 1987) and early age of marriage (Teachman 2002) are all well-known predictors of marital dissolution. At the group level, divorce rates vary by race/ethnicity. Black divorce rates are much higher than are rates among Whites,

Hispanics, and Asians (Teachman 2002). Less-educated and unemployed men and women have higher divorce rates than college-educated and continuously employed individuals, as well (Koo 1984; Wu and Balkrishnan 1994; Sweeney 1997). As a result, the remarriage market consists of men and women with more complex union histories, parents, and disadvantaged groups than the first marriage market. Ultimately, this may lead to differences between first marriage and remarriage formation.

Assortative Mating Patterns in Remarriage

In addition to remarriage formation, assortative mating patterns are an important aspect of remarriage. How men and women match on individual characteristics can affect individual well-being. Prior studies show that marrying a partner with similar socioeconomic status, educational attainment, and age have better global health (Huijts et al. 2009) higher marital satisfaction (Tynes 1990; White and Rogers 2000; Brennen et al.

2001), and a lower risk of divorce (Kalmijn et al. 2005) than individuals who marry heterogamously by choosing a spouse with dissimilar characteristics. However, not all characteristics have similar effects on such outcomes—age and education have very

strong effects, while traits such as physical attractiveness have little effect. Marital

sorting patterns can magnify inequality between social strata (Kalmijn 1998). For

example, couples with the same educational attainment have higher social standing and

socioeconomic potential than mismatched couples, especially if both partners are college

graduates (Blackwell 1998). Homogamy can also influence stratification

intergenerationally because couples often transmit their high social standing to children

(Kremer 1997; Fernandez and Rogerson 2001). Therefore, assortative mating has an important influence on a number of outcomes.

While people match on a number of characteristics (Kalmijn 1998), educational attainment and age are two important ones. As noted above, the effect of educational and age assortative mating on individual outcomes tends to be strong. Sorting on these characteristics also helps researchers understand what traits are important in marriage formation and how marriage markets are structured. Educational assortative mating is commonly studied (e.g., Mare 1991; Lichter et al. 1992; Qian and Preston 1993; Qian

1998; Schwartz and Mare 2005) because educational attainment is easily measured and highly correlated with overall socioeconomic status (Fernandez et al. 2005). For example, college graduates are almost invariably more economically stable than are people who did not graduate high school (Blackwell 1998; Oppenheimer et al. 1998).

Educational attainment also reveals clear distinctions between social strata (Blau 1977;

Gottschalk 1997; England 2004; Rose 2004) and the prevalence of educational heterogamy shows the willingness people have to cross socioeconomic boundaries when marrying (Kalmijn 1998).

The degree of age homogamy is associated with attitudes toward gender equality

in marriage (Kalmijn 1998). Small age differences between partners indicate less traditional views of marital gender roles (Atkinson and Glass 1985). Indeed, Richman

(1977) finds that gender equality is less likely when men marry younger women. The

prevalence of age homogamy also indicates the quality of the available partners in the

marriage market (Qian 1998). In markets with numerous well-matched potential spouses,

the rate of age homogamy is high. In lower quality markets, age homogamy is less likely

as people match on other, seemingly more important, attributes. Consequently, age

assortative mating patterns not only indicate how traditional remarriage is, but also the

quality of partners from which to choose.

Assortative Mating in First Marriage and Remarriage

Few studies have focused on assortative mating in remarriage (but see Dean and

Gurak 1978; Whyte 1990; Gelissen 2004). However, assortative mating in remarriage probably differs from assortative mating in first marriage. In addition to the reasons cited above, there are several other reasons to expect differences. These are detailed below for both educational and age assortative mating.

Educational Assortative Mating in First Marriage and Remarriage

Educational homogamy is common in first marriage and has increased in the last four decades (Schwartz and Mare 2005). Sociologists and demographers have forwarded three explanations for the trend toward homogamy in first marriage. First, men and women actively seek out highly-educated partners more often than in the past.

Traditionally, less-educated women marry hypergamously by paring with a highly-

educated man (Rockwell 1976; Rose 2004). However, as noted above, contemporary first marriages often consist of men and women who are equally educated, especially if both are college graduates (Qian 1998; Press 2004; England 2004; Schwartz and Mare

2005). This is due, in part, to economic trends which have made it difficult for men to

establish their careers (Oppenheimer 2003; Sweeney and Cancian 2004) and two incomes

necessary to meet a family's financial needs (Bumpass 1990; Sweeney 2002). As a

result, men and women sort on educational attainment as a strategy to avoid economic

uncertainty.

Second, men and women often meet one another in homogeneous settings

(Kalmijn and Flap 2000). Schools are particularly important for marital sorting because

of their composition--they have balanced sex-ratios and are educationally homogeneous

(Mare 1991). The correlation between educational homogamy and college entry is

strong. As the percentage of men and women entering college has increased (U.S.

Census Bureau 2007) and the gender gap in college graduation rates has closed (DiPrete

and Buchmann 2006) the level of educational homogamy has increased. As a result,

many men and women form partnerships in school which can become homogamous

marriages.

Third, the social distance between educational groups has widened in recent

decades (Blau and Schwartz 1997). Today, men and women with different educational

attainment are unlikely to meet one because many settings have become educationally

segregated (Kalmijn and Flap 2000; Fernandez et al. 2005). High inequality also

contributes to strong barriers as the correlation between income and educational

attainment has grown stronger and more positive in recent decades (Goldin and Katz

2000; Schwartz and Mare 2005). These trends imply that educational groups are less integrated today than in the past, resulting in lower levels of educational intermarriage.

The level of educational homogamy in remarriage should be lower than it is in first marriage. Preferences in remarriage may tend to favor heterogamy, not homogamy.

When people marry for the first time, they may sort on educational attainment because it is a proxy for future socioeconomic status. However, by the time most men and women remarry, an individual’s economic well-being is more established, making proxies for

economic status unnecessary (Sweeney 1997). This may help make remarriage more

traditional than first marriage--especially if divorcees hold more traditional values about gender and marriage than never-married individuals (Coleman et al. 2000; Gelissen

2004). As a result, highly-educated men may marry women with lower educational

attainment. Educational homogamy in remarriage should also be lower in remarriage

because divorcees are more likely to use heterogeneous locations, such as social

organizations, neighborhood groups, and churches to meet potential partners (deGraaf and Kalmijn 2003). In particular, educational institutions are less likely to be used as a

meeting location because the vast majority of divorced men and women have completed their education2. Thus, the social distance between educational groups for divorcees may

be lower than it is for the never-married.

2 In the NLSY79 sample used in this chapter, approximately 3% of men and 7% of women are enrolled in school at anytime between divorce/separation and remarriage/censoring. 19

Assortative Mating by Age in First Marriage and Remarriage

Like educational assortative mating, age homogamy is common in first marriage.

The number of age homogamous marriages increased steadily until the 1970s and has

since leveled off (Qian 1998; Van Poppel et al. 2001). There are three major

explanations for why age homogamy is common in first marriage. First, age homogamy

may be a byproduct of matching on other attributes such as education and race/ethnicity

(Kalmijn 1994). Generally, never-married individuals search for a spouse within a

similar age range. For example, as noted above, highly-educated men and women often

find their partners while in school (Mare 1991). Typically, schools are homogeneous with

respect to age, meaning that matches made while enrolled in school likely are

educationally and age homogamous. In addition to schools, men and women often

engage in age-segregated meeting places which prevent them from forming unions with

people in other age groups (Selten 1993; Van Poppel et al. 2001).

Second, the first marriage market has become increasingly age stratified over time. Cultural barriers between age groups may have strengthened in recent decades and the desire for similarly-aged spouses. Therefore, age homogamy may be the result of a preference for a partner who shares similar interests, likes, and dislikes (Modell 1989;

Easterlin and Crimmins 1991; Mitterauer 1992). Indeed, South (1991) and Raley and

Bratter (2004) report that never-married individuals find marrying a partner 10 years

younger or older than themselves a highly undesirable match.

The third perspective explaining age homogamy trends in first marriage focuses

on the leveling off of homogamy in recent years. This perspective asserts that an

increased emphasis on some characteristics, such as educational attainment, may lead to a

trade-off on age (Qian 1998). In other words, high emphasis on educational attainment

may lead individuals to accept a younger or older partner in order to make the preferred

match. In this case, the benefits of a partner’s high socioeconomic status outweigh any

age differences between the two partners. Nevertheless, the level of age homogamy in

first marriage remains high (Kalmijn 1998; Qian 1998; Van Poppel et al. 2001).

The level of age homogamy should be lower in remarriage than in first marriage.

If remarriage is traditional, many divorcees should prefer a hypergamous matches where older men marry younger women (Jacobs and Furstenberg 1986). Indeed, a number of studies have shown that older men prefer to marry women younger than themselves

(Vera et al. 1985; England and Farkas 1986; Oppenheimer 1988; England and

McClintock 2008). As a result, age homogamy should be relatively uncommon among

men. For divorced women, age preferences suggest that non-marriage is the most likely

outcome, especially as they age. Compositional differences also play an important role in

age assortative mating. The remarriage market is more age diverse than the first marriage

market (Bumpass et al. 1990), meaning that any byproduct from matching on other

characteristics may favor age heterogamy. In fact, race/ethnicity, education, and parental

status tend to be more important characteristics in the remarriage market than age (Jacobs

and Furstenberg 1986; Ni Brochlain 1988; Bumpass et al. 1990; South 1991). In

addition, composition and preferences can combine to favor age heterogamy. The fact

that the number of single women outpaces the number of single men at later ages (White

and Rogers 2000) combined with men's preferences for younger women suggests that

women are more likely to remain unmarried after divorce, especially if they are older

when they do dissolve their first marriage.

Marital Searches After Divorce

Marital search theory argues that people look for the best possible match from the

pool of potential partners (England and Farkas 1986; Oppenheimer 1988). The likelihood and quality of a marriage depend on preferences and constraints. In terms of preferences, some men and women are attractive partners because their characteristics are highly sought after, such as being college-educated or earning a good income. Individuals who possess these traits increase their probability of finding a marriage partner (Lewis and Oppenheimer 2000; Goldstein and Kenney 2001; Sweeney and Cancian 2004).

Other men and women, such as people with poor socioeconomic prospects or those who care for children, are unattractive marriage partners and have difficulty finding a spouse

(Michael 1994; Thornton et al. 1995; Oppenheimer et al. 1997; Clarkberg 1999). As a result, the likelihood of marrying varies across individual characteristics.

Choosing whom to marry, however, is constrained by the size and composition of the available partner pool. The size of the available partner pool varies according to personal characteristics. Individuals with attractive attributes choose from large potential partner pools where they can be selective and find a high-quality spouse. Conversely, men and women with less attractive attributes pick from smaller, more limited partner pools where there is little benefit to being selective (Oppenheimer 1988). Notably, partner pools become increasingly restrictive with time, meaning that marital outcomes

likely change as one spends more time searching for a spouse (Shafer and Qian

forthcoming).

Importantly, the proportion of divorcees that remarry (about 75%) is smaller than

the proportion of people who experience a first marriage (approximately 90%; Teachman et al. 2000; CDC 2002). As a result, individual characteristics can take on added importance when remarrying. Contrasted against first marriage, this means that in remarriage individual attributes do not only affect whom you marry, but if you marry at all. For example, deGraaf and Kalmijn (2003) find that the evidence for economic hypotheses of marriage formation (Oppenheimer 1997) are more strongly supported in remarriage than first marriage. Below, I discuss marital search theory in its application to remarriage by highlighting the preferences and constraints in the remarriage market.

Preferences in Remarriage

Clearly, the characteristics of men and women in the remarriage market differ from those of never-married individuals across numerous dimensions, including socioeconomic and parental status. These two factors affect whether divorced men and women will marry for a second time or not. Below, I chronicle how socioeconomic and parental status can influence the likelihood of remarriage and offer hypotheses by gender.

Economic status. Socioeconomic status is an important characteristic in the

marriage market. High socioeconomic status reflects an ability to support a family

(Sweeney 1997) and increases the likelihood of marrying. Important to remarriage,

however, are gender differences in economic standing after divorce (Hoffman 1977;

Duncan and Hoffman 1985; Holden and Smock 1991; Peterson 1996). After divorce,

men often experience an increase in their economic well-being, while women see a decline in their economic stability. The gap, by all estimates, is large with men experiencing a 10 percent increase in the income-to-needs ratio after divorce while women are subject to a 27 percent decline (Peterson 1996). This gap is primarily the result of differences in per-capita income after divorce (Peterson 1996). Men often earn the majority of household income while married and maintain employment through divorce (deGraaf and Vermeulen 1997). As a result, they have high per-capita income after divorce. Women, on the other hand, are often out of the labor force at the time of divorce and care for children after dissolution, reducing their economic well-being.

Men's post-divorce economic status can make them attractive marriage partners.

Men’s economic position has a strong, positive effect on the likelihood of marriage

(Bennett et al. 1989; Klein 1990; Fossett and Kiecolt 1993; Qian and Preston 1993; Wu

1994; McLanahan and Casper 1995; Oppenheimer et al. 1997; Lampard and Peggs 1999; deGraaf and Kalmijn 2003). Divorced men's marriage prospects may be enhanced because they tend to have more established careers and conspicuous economic status than never-married men (Sweeney 1997). High-status men who divorced in the 1980s, 1990s, and early 2000s3, may be particularly attractive marriage partners. During this period, the economic prospects of young (and presumably never-married) men deteriorated, meaning that more established men may have a competitive advantage in the marriage market despite their marital status (Juhn et al. 1993; Wetzel 1995; Oppenheimer et al. 1997).

3 The years when men and women divorced in the NLSY79 data set. 24

If remarriage is traditional, economic status should not increase the likelihood of remarriage for divorced women. Unlike a contemporary marriage market, where men view women's economic status as important (Kalmijn 1998), traditional markets support a conventional division of household labor. High-status women often opt out of their traditional domestic labor roles (Oppenheimer 1997), making it unlikely they will marry, because the gains to marriage for men are low (Becker 1981). Indeed, South (1991) shows that previously-married men have a strong preference for hypergamy, or marrying women with lower economic status. In addition, previously-married men may find marrying a high-status woman unnecessary. As discussed above, in first marriage, men often marry highly-educated women out of economic necessity. However, divorced men, because they are more economically established, may not need a spouse who can financially contribute to the family (Peterson 1996; deGraaf and Kalmijn 2003). As a result, socioeconomic status may not be an important factor for women's remarriage prospects.

Socioeconomic status can be measured in two ways: current status or long-term prospects. Both have validity when considering the impact of economic standing on marriage formation. Although the importance of both measures to marital searches is detailed in Chapter 4, they will be briefly introduced here. Current economic status, such as income or labor force status, is important in remarriage. As discussed above, income, occupational trajectories, and other indicators of economic status are more established as individuals age. As a result, current economic status may strongly affect remarriage formation because divorcees are older. Long-term economic prospects should also affect

remarriage formation. Educational attainment is an excellent measure of long-term

economic stability (England and Farkas 1986; Mare 1991; Qian 1998; Schwartz and

Mare 2005) because it is correlated with occupational prestige, income, and overall

socioeconomic status (Fernandez et al. 2005). Education is also less subject to yearly

fluctuations which are common in individual income or occupational status.

Furthermore, highly-educated individuals recover quickly from a decrease in income and

experience shorter spells of unemployment than less-educated men and women (Goldin and Katz 2000). Therefore, educational attainment, as a measure of long-term economic status, should play an important role in remarriage (Smock 1990; Spanier and Glick

1990; Wu and Balakrishnan 1994; Sweeney 1997; deGraaf and Kalmijn 2003).

In sum, if remarriage is traditional, the following patterns will be observed:

(1) Current (income) and long-term (educational attainment) economic status will be

positively associated with remarriage for men.

(2) Current (labor force status) and long-term (educational attainment) economic

status will have no effect on the remarriage prospects of women.

First marriage ties. A unique and important characteristic to remarriage is that

many divorcees maintain ties to their first marriages after divorce. Such ties are

maintained because of the social, emotional, and psychological investments divorcees

make with their first spouse through jointly-owned goods, children, and relationships

with friends and in-laws (Ono 2006). Often, these ties persist because of shared

obligations and responsibilities with ex-spouses. Commonly, men and women share

parenting duties through co-custodial arrangements or visitation rights. Children

represent some of the strongest ties an individual can have with their previous marriage

partner. As a result, children can affect the overall attractiveness of divorced men and

women in the remarriage market.

Children affect remarriage because they add complexity to new relationships.

Children represent an emotional investment with an ex-spouse which can lead to jealousy

and distrust in new relationships (Ono 2005). Children require constant care and

attention which can strain new relationships and generate conflict because new spouses have difficulty dedicating time to one another (Cooksey and Fondell 1996; Coleman et al.

2000; Yeung et al. 2000; deGraaf and Kalmijn 2003). Stepfamilies are a non- institutionalized family form which lacks formalized or culturally consistent ways of resolving family issues and problems related to stepparenting (Cherlin 1978; Cherlin

1992). These issues sometimes manifest themselves as problematic behaviors in children. Stepchildren routinely have difficulty in school (e.g., Astone and McLanahan

1991; Teachman et al. 1996), numerous mental health issues (e.g., Zill et al. 1993;

Hanson, McLanahan, and Thomson 1996; Lee et al. 1994) and are more likely to engage in risky and delinquent behaviors (e.g., Day 1992; Astone and Washington 1994;

Hoffman and Johnson 1998). Many scholars have linked these issues to marital quality in remarriage--which declines rapidly and tends to be lower than in first marriage (Hobart

1991; Krudek 1991; Koch and Ihinger-Tallman 1993; Brown and Booth 1996; Kurdek

1999. Similarly, remarried individuals report higher levels of conflict than couples in their first marriages (Coleman et al. 2000).

Men and women who choose to marry a parent with a co-residential child are a select group. Becker (1981) argues that because biological/adopted children are a symbol of social status, high-status individuals are unlikely to partner with someone who cares for a child. Indeed, stepparents tend to have less income and education than men and women who partner with non-parents (Anderson 2000; Blackwell and Lichter 2000;

Clarkberg 1999). As a result, men and women with high socioeconomic status are likely to avoid stepparenting, if possible, by marrying a never-married individual or a divorcee without children.

The effect of children on remarriage should differ by gender. Women, because they are overwhelmingly responsible for the care of children after divorce, are more likely to have strong first marriage ties than men (Teachman and Hecker 1985; Koo,

Suchindran and Griffith 1986; Gerstel 1988; Chiswick and Lehrer 1990; Ono 2005).

Women are also more likely to maintain co-parenting arrangements than men. In comparison to women, men often become absent parents after divorce, have little contact with their children and ex-spouses, severing ties to their first marriages (Stewart 1999).

However, even without the presence of their ex-husbands, women are often obligated to maintain ties with friends and family from their first marriages because of children

(Becker et al. 1977; Manning et al. 2003). These relationships may also be maintained through financial and other economic support that divorced women need for themselves and for their children (Ono 2006). Frequently, this economic help comes from their ex- husbands or their ex-husbands’ families, further strengthening first marriage ties (Gerstel

1988). Therefore, children should have a negative effect on repartnering for divorced

women, but not men.

By way of review, the expected relationships for the effect of first marriage ties

on remarriage are:

(1) First marriage ties, as measured by the presence of co-residential children, will

have no effect on the likelihood of remarriage for men.

(2) Having co-residential children will decrease the likelihood of remarriage for

women.

Constraints in the Remarriage Market

While personal characteristics and preferences affect the likelihood of remarriage,

men and women are also constrained in the remarriage market by opportunities to form

matches. I consider two important constraints in the remarriage market. First, divorced

individuals draw from different partner pools, based on their individual characteristics,

when choosing new partners. These pools differ in size and composition and not only

affect the opportunity to remarry, but also marital sorting in second marriages. Second,

time in the remarriage market can affect the opportunity to remarry and the type of

partner an individual marries. These constraints are discussed in detail below.

Differences in partner pools. Divorced men and women draw from different partner pools when searching for a new marriage partner. Because high socioeconomic status is an attractive attribute (e.g., Becker 1981; England and Farkas 1986;

Oppenheimer 1988; Oppenheimer et al. 1997; Wu and Schimmele 2005), high earning, highly-educated men have the largest potential pools available to them. As a result, these

men have the opportunity to form the most preferential matches. High-status divorced

men may be attractive marriage partners because, as noted above, they are older and more

economically established than never-married men (Peterson 1996; Sweeney 1997).

Indeed, it can take several years for men to establish a stable career after school

completion, even for college graduates (Oppenheimer et al. 1997). The socioeconomic

advantages high-status divorced men have over other men can make them attractive

marriage partners to both never-married and previously-married women. As a result,

divorced men have the ability to draw on both pools of potential partners when looking

for a second marriage.

Previously-married men may not be hindered and stigmatized by their previous

marital status because of their ability to support a new family. This pattern is analogous

to intermarriage patterns among high-status racial and ethnic minorities (see Davis 1941

and Merton 1941). In racial and ethnic intermarriage, status exchange takes place when

high-status minorities exchange their socioeconomic status to lower status Whites for

their higher racial/ethnic status (Fu 2001). Similarly, high-status divorced men should be

able to trade their economic standing to never-married, lower status women for their higher status with respect to marital status and age. Thus, the prevalent marriage pattern among high-status, previously-married men should be the preferred traditional matches with women who are less-educated and younger than themselves (South 1991; Raley and

Bratter 2004). These traditional matches are desirable for high-status men.

Hypergamous matches are advantageous for high-status men because the women they marry are more likely to invest in domestic labor than better educated, older women

(Becker 1981), are unlikely to have been previously-married, or have children from a previous relationship (Ono 2005).

Lower status men may have more difficulty finding a spouse or a good marriage partner than their high-status counterparts. Less-educated and poorer men are undesirable marriage partners because they lack economic resources and stability

(Oppenheimer 1988). As a result, low-status divorced men have a much smaller pool of potential partners than high-status men. This should have two effects. First, low-status divorced men should have lower odds of remarriage than high-status divorced men.

Second, among low-status men who do remarry, traditional marriage patterns will not be present. Unlike their higher status counterparts, lower status men do not have the ability to trade their educational attainment or income to younger women. Thus, educational and age homogamy may be a prevelant pattern for this group of men.

Unlike men, there should be little differentiation in marriage patterns across socioeconomic status for women because they are highly constrained in the remarriage market. However, parental status may play an important role in how constrained divorced women are in their marital opportunities and choices (Thornton, 1977; Koo and

Suchindran 1980; Koo et al. 1984; Teachman and Heckert 1985; Smock 1990; Bumpass et al. 1990; Lundberg and Rose 2003). Women’s prior fertility is limiting for several reasons. First, men desire biological/adopted children (Becker 1981) meaning that men are unlikely to seek out women who are already parents (Goldschreider and Sassler

2006). Second, divorced mothers are often economically disadvantaged which can be burdensome for many potential partners (Klute et al. 2002; Raley et al. 2006). Third,

fatherhood can be a daunting challenge to assume for many men who are not a child’s biological or adopted father. This can lead to feelings of anxiety, stress, and conflict between family members, which are more present in stepfamilies than in biologically- related or adopted families (Coleman et al. 2000).

Compared to women with children, women without children may have a larger partner pool from which to choose. However, women without children still draw from a smaller partner pool than do men. Divorced women also tend to be older than Never- married women. Traditionally, men show a preference for and often marry women who are younger than themselves (Vera, et al., 1985; England and Farkas, 1986;

Oppenheimer, 1988), which would negatively impact women. Well-educated, high earning divorced men may be particularly unavailable to divorced women because their economic advantage provides them with a greater opportunity to marry younger women.

Divorced women are also negatively affected by compositional disadvantages such as unbalanced sex ratios at older ages, which reduce the probability of remarriage (Gelissen

2004). However, the effects of market composition on remarriage are outside the confines of this study. Thus, women without children will be more likely than women with children to experience marriage, but their marriages are likely to take place with less attractive men, such as less-educated men. With regard to age, the expectation is that women without children, like women with children will be likely to marry similarly-aged men.

In review, the effects of available partner pools should be, as follows:

(1) High-status men will have a higher likelihood of remarriage than low-status men.

These marriages are likely to be to women with lower educational attainment and

who are younger.

(2) Low-status men have a lower likelihood of remarriage than high-status men. If

they do marry, they will marry women from the same educational background and

age group as themselves.

(3) Women’s socioeconomic status should have no effect on remarriage. However,

women with children should be less likely to remarry than women with children.

Assortative mating patterns will not differ by social status or parental status for

women.

The effect of time. The opportunity to remarry and marry an attractive partner can decline over time. As an individual spends more time in the remarriage market, fewer and fewer marriage partners are available in the partner pool. With time, potential

partners marry out of the partner pool, especially attractive individuals who are highly

sought after in the remarriage market (Lewis and Oppenheimer 2000). The effect of time

on the prospects for remarriage should differ by gender. In first marriage, men are much

more likely to lower their spousal preferences and become less selective searchers than

women (Lichter 1990; Mare 1991; Raley and Brater 2004; Shafer and Qian forthcoming).

Men’s choices allow them to marry, although that marriage may be to a less preferential

mate. What constitutes a less desirable marriage when marrying for the second time,

however, differs considerably from first marriage. In first marriages, the likelihood of

heterogamy goes up with time in the marriage market. Yet, if remarriage is more

traditional (Bartling and Broussard 1999), undesirable matches are either homogamous

with respect to age and education, or marriages where the woman is better educated

and/or older (Raley and Bratter 2004; England and McClintock 2008).

Women are much more likely to remain unmarried than change their partner

preferences. Unlike divorced men, who can marry younger women, divorced women are

disadvantaged by the fact that the majority of similarly-aged men are either currently

married or prefer younger women (Vera et al. 1985; England and Farkas 1986). This

constraint likely becomes stronger as women age (deGraaf and Kalmijn 2003). Thus, the

likelihood of non-marriage increases for women with more time in the remarriage

market.

Thus, the expected effects of time in the remarriage market on remarriage are:

(1) For men, the likelihood of remarriage will decrease with more time in the

remarriage market. Educational and age assortative mating patterns will

become less traditional over time.

(2) For women, the likelihood of remarriage will decrease with more time in the

remarriage market, and do so at a much steeper pace than it does for men.

Summary of Theory and Implications

Returning to the framework of marital search theory, choices in remarriage are a product of preferences and constraints. Preferences are based on individual characteristics. For example, I noted above how men’s higher socioeconomic status benefits them in the remarriage market. This is primarily driven by their higher attractiveness to women as a potential spouse due to their economic stability. In

comparison, women tend to due worse economically after a divorce making them a less attractive marriage partner. Another important factor is who raises children. We know that women are much more likely than men to care for children after a divorce

(Golschreider and Sassler 2006), meaning that women are often forced to maintain ties back to their first marriages which make them less attractive marriage partners. For men, more traditional matches with younger, less-educated women are preferential. On the other hand, women tend to prefer an economically attractive new spouse that can support a new family.

Importantly, individual characteristics can also constrain men and women in the remarriage market. Marital search theory argues that some individuals will have fewer potential spouses available to them than others. For example, highly-educated and less- educated individuals do not have access to the same partner pools. Individual attractiveness in the marriage market can open up or constrain an individual’s opportunity to marry and if they marry, the choices available to them when selecting a new spouse.

Highly-educated men seeking a second marriage have a large pool of potential partners available to them, because they are well-established financially, and are able to trade their socioeconomic status for desirable traits in a woman. Thus, they are highly likely to remarry and make a good match in the process. Less-educated men and women have smaller pools of potential partners to draw from. Similarly, because being a parent of a co-residential child is an undesirable characteristic in the remarriage market, people with

children have an extremely limited pool of potential partners to choose from. As a result, people with less desirable spousal characteristics are either unlikely to remarry or remarry

a partner with less desirable characteristics.

Chapter 3: Data, Measures, and Analytic Strategy

To test the hypotheses discussed in Chapter 2, I use data from the National

Longitudinal Survey of Youth, 1979 cohort (NLSY79), a panel study beginning in 1979 with available data until 2006. Previous studies of remarriage have been limited by the type of data available to researchers. Most studies of remarriage use data from foreign sources (e.g., Wu and Balkrishnan 1994; Lampard and Peggs 1999; deGraaf and Kalmijn

2003; Wu and Schimmele 2005) because some European and Canadian data sets include large, representative samples with questions focused specifically on the remarriage process. Studies in the United States, to this point, have used somewhat problematic data. For example, Sweeney (1997) used the Wisconsin Longitudinal Study data which includes a complete marital history, but is not nationally representative and does not include data on important characteristics like income. Other researchers, such as Ono

(2005), have used vital statistics data to understand remarriage patterns, but this data lacks predictors and control variables. However, NLSY79 has several advantages which allow me to build on prior research by: (1) focusing on the remarriage process in the

United States using nationally representative data; (2) studying remarriage with longitudinal data that accounts for prior marital history; (3) including numerous measures

of personal characteristics which can affect marriage formation; and (4) accounting for

the dynamic nature of repartnering by considering time in the remarriage market.

Data

The NLSY79 is a panel study of 12,686 men and women who were between the

ages of 14 and 22 on January 1, 1979. NLSY79 is unique because it allows researchers to

follow respondents across important life-history events such as marriage, divorce,

educational achievement, and childrearing while also including data on characteristics

such as spouse’s age and educational attainment. The survey was conducted annually

between 1979 and 1993. Beginning in 1994, NLSY79 moved to a bi-annual survey. The

final round of data collection now available is from 2006. Over its 27 year history,

sample attrition has been relatively low, at approximately 25% (Center for Human

Resources Research 2008).

In NLSY79, 7,114 men and women were married for the first time between 1979

and 2006. Over the same time period, 3,006 male and female respondents from this

group divorced. I specifically focus on divorcees because widowhood is rare in the

sample and widowed individuals tend to remarry differently than divorcees (Wu and

Schimmele 2005). Furthermore, only men and women whose first marriages ended in divorce are included in the analytic sample because higher order divorce is selective

(Teachman 2002). Importantly, men and women who divorced prior to 1979 are

excluded from the sample because information on their first marriages and part of their

time in the remarriage market is missing. This restriction excludes 535 individuals who

were married and divorced in their late teens and early 20s. The final sample consists of

2,364 individuals--1,119 men and 1,245 women.

One additional consideration with NLSY79 is that Blacks and military personnel

were originally oversampled. These oversampled cases were dropped from the data set in

1986 (Center for Human Resources Research 2008). If these respondents divorced in or

prior to 1986 and did not remarry, these individuals are censored in that year. Otherwise,

members of the oversample who divorced and remarried between 1979 and 1986 are included in the analyses. Regressions were run with and without these individuals included and the results are substantively similar to those presented in subsequent chapters. Other censored cases include men and women who did not remarry at the time the left the study (n= 52) or by the last available wave of data in 2006 (n= 1,128).

The data are in event-history format where respondents are followed from the time of separation or divorce until they either remarry or are censored. Years in which a respondent self-identifies as separated but is still legally married are included in the sample because many men and women begin searching for a new spouse, in earnest, when they stop living together4. In addition, Sweet and Bumpass (1987) illustrate that

many men and women do not begin legal proceedings for a divorce until one partner

intends to remarry. If a respondent does not report being separated prior to divorce, they

are considered at-risk for remarriage beginning in the year they were legally divorced

from their spouse. Once a respondent remarries they are no longer at-risk for remarriage

4 South and Lloyd (1995) show that these searches may begin even before separation. However, when men and women begin searching for a new spouse cannot be identified in the NLSY79 data. As a result, I use the proxy measure of separation/divorce. 39

and no longer contribute person-years to the data. Respondents who drop out of the study

before they remarry or have not remarried by 2006 contribute person-years until the time of censoring. The final event-history sample consists of 19,278 person-years. By sex, men contribute 8,780 person-years and women contribute 10,498 person-years to the

sample.

One methodological issues is missing data. Missing data for the independent

variables was replaced by substituting a value on a variable from the year prior to the

missing case. Values for the independent variables are also missing in non-survey years

between 1994 and 2006. In these instances, I substitute the values from the prior survey year for each independent measure. For example, respondents do not report educational attainment in 1995. As a result, I use respondent's education from 1994 as the value of

education in 1995. However, some variables, such as income, have reported values for the non-survey years. For these variables, the reported values, not the substituted values from the year prior, are used. This method is preferable to alternative strategies for handling missing data when using event-history models. For example, multiple imputation can be problematic in event-history data because it may lead to sudden increases or decreases in variables from one year to the next. Thus, the substitution strategy maintains the time-ordered nature of the data while accounting for missing data.

Variables and Measurement

Dependent Variables

There are three dependent variables in this analysis, all of which measure timing to remarriage. However, all three have unique outcomes. The first measures whether or

not a respondent remarries. The second measure focuses on educational assortative

mating in remarriage. The final measure captures assortative mating by age. I discuss

these variables in greater detail below.

The likelihood of remarriage. Chapter 4 examines the likelihood of remarriage

using a dependent measure for duration from the time of dissolution (separation/divorce)

to second marriage. This measure is a dichotomous event variable which indicates

whether or not the respondent remarried in any given year. If the respondent does marry,

the variable specifies the year in which the marriage took place. A respondent is coded

as unmarried in each year they have not remarried. Respondents censored through

attrition or who did not remarry by 2006 are coded as unmarried for each year they are in the sample.

Assortative mating in remarriage. Chapter 5 highlights the educational and age assortative mating patterns of divorced men and women who remarry. This is measured through dependent measures which determine duration to remarriage (the event) and the marital sorting outcome. First, the dependent variable for educational assortative mating has four possible outcomes: (1) an educationally homogamous marriage where men and women have the same level of educational attainment, (2) an educationally heterogamous marriage where the husband has more education than his wife, (3) an educationally heterogamous marriage where the wife has more education than her husband, or (4) remaining unmarried. If the respondent does marry, the event variable not only specifies the type of marriage, but also the year in which the marriage took place.

Each assortative mating outcome is defined using a measure of educational

attainment which measures an individual's progression through the education system.

These levels are: less than a high school graduate (11 or fewer years of schooling), at

least a high school graduate (12-15 years of schooling) or at least a college graduate (16

or more years of schooling). As noted above, a marriage is considered homogamous or heterogamous based on the respondent's and spouse's educational achievement at the time of remarriage. Unmarried and censored respondents are coded in the same manner as noted for the dependent variable used in Chapter 4.

The dependent measure for assortative mating by age is constructed in a similar manner. For this variable there are four possible outcomes: (1) an age homogamous marriage where the husband is no more than one year younger to four years older than his wife, (2) an age heterogamous marriage where the husband is at least five years older than his wife, (3) an age heterogamous marriage where the wife is at least two years older than her husband, and (4) remaining unmarried. These definitions of homogamy and heterogamy are derived from previous studies of assortative mating by age (Qian 1998).

Duration and Time

Duration from entry to either remarriage or censoring is measured from the time of dissolution. The construction of this variable is detailed in the event-history data discussion above. A quadratic term is included in the regression models in order to capture curvilinearity in the likelihood of remarriage over time. Previous research has shown that there can be sharp increases in the odds of remarriage shortly after divorcing, followed by a slow decline in the odds over time (Wu and Schimmele 2005).

Key Independent Variables

The independent variables included in my analyses measure socioeconomic status, first marriage ties, age, and controls correlated with divorce and the likelihood of remarriage.

Current Socioeconomic status. In this analysis, there are three measures of socioeconomic status. Two variables measure current socioeconomic status. For men, income is included as a measure of socioeconomic status and economic stability at a given point-in-time (Lewis and Oppenheimer 2000). Because of the dynamic nature of an individual's income, the variable is time-varying. There are two potential issues associated with the measurement of income in NLSY79. First, there is high positive skew in the income variable. Second, income is a top-coded variable in order to insure the confidentiality of respondents in the data set (Center for Human Resources Research

2008). As a result, I transform the variable and use the natural log of income in my analyses.

For women, income can be a problematic measure of current socioeconomic status because it is correlated with several characteristics associated with divorce. For divorced women, income can come from numerous sources including work, TANF and other welfare programs (Lichter and Jayakody 2002), child support, and alimony

(deGraaf and Kalmijn 2003). Because income is an insufficient measure of socioeconomic status for women, I include a time-varying dichotomous measure for labor force status. Labor force participation is measured as full-time (an average of 35 or more hours per week), year-round (50 or more weeks a year) employment in any given year.

This variable was constructed from variables available in the NLSY79 for the total

number of hours and weeks worked by the respondent in a year.

Long-term socioeconomic status. Current socioeconomic status provide an

incomplete picture of an individual's overall socioeconomic status. Income and

employment circumstances often fluctuate from year-to-year. As a result, I include

educational attainment as a measure of an individual's long term socioeconomic

prospects. Education is a reliable measure of economic status because it influences

earnings potential, employment stability (Oppenheimer 2003), and the ability to

economically recover from a divorce. Generally, highly-educated men and women tend

to have higher economic well-being over the life course and more developed career paths

(England and Farkas 1986; Mare 1991; Oppenheimer et al. 1997; Qian 1998; Schwartz and Mare 2005).

Educational attainment is measured through a set of dichotomous variables constructed from a variable for the number of years of schooling a respondent completed because NLSY79 does not include measures for highest degree received. These time- varying dichotomous variables capture individual progression through the education system and are: less than a high school graduate (11 or fewer years of schooling), at least

a high school graduate (12-15 years of schooling) and at least a college graduate (16 or

more years of schooling).

First marriage ties. As noted in Chapter 2, children represent important ties to

first marriages which can affect the opportunity to remarry, especially for women (Wu

and Schimmele 2005; Goldschreider and Sassler 2006). Therefore, I include

respondent’s parental status as a predictor in the models. This is a time-constant

dichotomous measure which indicates if the respondent cares for a co-residential

biological/adopted/or step-child from a previous marriage5. This variable was

constructed using two measures available in the NLSY79: the presence of a co-residential

child and the age of the child6. If the respondent reported having a co-residential child in

the first year of divorce, I then identify the age of the child in that year. From the child’s

age I can derive if the child was born during the respondent’s first marriage7. If so, the

respondent is coded as 1 for the dichotomous variable. If the respondent does not have

children from their previous marriage, the respondent is coded as 0.

Age. Men’s and women’s age is an important characteristic in remarriage,

especially because the availability of potential partners is partially dependent upon an

individual’s age (deGraaf and Kalmijn 2003; Gelissen 2004). Age can also influence

potential marriage prospects, the type of marriage formed, and individual partner

preferences (Gelissen 2004). In my analyses age is a time-varying continuous measure.

The variable is also centered around the mean of 35.17 years of age for ease of

interpretation.

5 Several alternative constructions were tested for this variable. Some studies have shown that number of children and the age of children are important for remarriage (Wu and Schimmele 2005). Both variables were tested in all models and were statistically insignificant. A time-varying measure of co-residential children, as suggested by Goldschreider and Sassler (2006), was also tested. However, the results for this measure were less theoretically grounded in the concept of “first marriage ties” because men and women could have children after divorce. Nevertheless, the results for the time-varying and time-constant measure of children are substantively similar to one another. 6 I use this strategy because respondent’s fertility history is not available for all years in NLSY79. 7 One potential issue with this strategy is for women with pre-marital births. However, in the sample there are only about 50 women who report a pre-marital birth. A variable was constructed to control for non- marital births but was statistically insignificant in all models. 45

Control Measures

I include several control measures for the respondent’s socioedemographic characteristics and potential risk factors for divorce which may bias the results. These variables are:

− Respondent’s race/ethnicity: a set of time-constant dichotomous variables which

indicates if the respondent is non-Hispanic White (reference category), non-

Hispanic Black, or Hispanic.

− Family socioeconomic status at age 14: a set of time-constant dichotomous

variables for the educational attainment of the respondent’s mother. The variables

indicate if the respondent’s mother was less than a high school graduate

(reference category), a high school graduate, or college graduate.

− Family structure at age 14: a set of time-constant dichotomous variables which

indicate if the respondent grew up in a two parent home, a home with a stepparent

present, with a single parent, or in another family structure arrangement. These

variables are included because marital and relationship history can be transmitted

intergenerationally (Teachman 2002).

− Religious background at age 14: religion has been linked to marital behavior and

the likelihood of divorce (Kalmijn 1998; Teachman 2002). Previous research has

shown that Catholics and Conservative Protestants (Baptist, Southern Baptist,

Pentecostal, and Evangelical) are likely to marry individuals from the same

religious background, irregardless of other attributes. Thus, I include

dichotomous variables which indicate if the respondent is Catholic, Conservative

Protestant, or from another religious background.

− Residential status: a time-varying dichotomous variable for respondents who live

in an urban location, which is defined in the NLSY79 as living in a central city or

adjacent area with a combined population of 50,000 (Center for Human Resources

Research 2008). A second time-varying dichotomous variable for Southern

residence is also included.

− Duration of first marriage: I include a time-constant continuous measure of the

respondent’s first marriage duration using the respondent’s reports for the year in

which he/she first married and divorced. This variable is included because men

and women who experience longer first marriages may have strong ties to their

ex-spouses, ex-spouses families, and have a higher level of commitment to the

institution of marriage (Bumpass et al. 1990; Ono 2005).

− Pre-marital cohabitation: a time-constant dichotomous variable indicating if the

respondent cohabited prior to first marriage is also included as a control variable.

Men and women who cohabit prior to marrying may be a selective group because

they are at a higher risk of divorce than individuals who do not cohabit

(Teachman 2002).

− Educationally/age homogamous first marriage: men and women who married

homogamously in first marriage may desire a similarly-educated/similarly-aged

spouse in their second marriages (Gelissen 2004). To measure this I include a

time-constant dichotomous measure which specifies if the respondent was in an 47

educationally homogamous first marriage, using the same definition of

homogamy as noted above, in the model for educational assortative mating. A

similarly constructed variable indicating if the respondent was in an age

homogamous first marriage is included in the model for marital sorting by age.

Analytic Strategy

I use event history methods to estimate models for the transition to a second marriage among divorced men and women. I use event history techniques because I am specifically interested in timing to remarriage and the effect of both time-varying and

time-constant characteristics on the odds of remarriage (Hosmer and Lemeshow 1999).

This strategy also allows me to differentiate between marital sorting outcomes and

between individuals who did and did not experience an event, including individuals who

are censored through attrition or the end of data collection. Event history models are an

improvement over traditional linear regression techniques because event history methods

allow for easy incorporation of time-constant and time-varying predictors and can

account for both sample attrition and censoring (Allison 1982; Allison 1984). Discrete-

time methods are particularly useful because they allow for they measure the risk of an

event at each given point in time without restrictive assumptions (Allison 1982).

In discrete-time modeling each time unit (or year the respondent is in the risk-set)

is considered a separate observation for the individual. For each time point, the outcome

variable, which indicates whether or not the respondent experienced a remarriage, is

coded as 0 if the respondent did not remarry in that year and 1 if they did remarry. The

hazard in these models is a conditional probability of remarriage, given that the

respondent has not remarried prior to that point. The independent variables are assigned

values for each observed year of data and the effect is essentially averaged over time and individuals in order to get the magnitude of the effect, or coefficient, in any given year

(Allison 1982).

Analysis of the Likelihood of Remarriage

In Chapter 4, I use the discrete-time logit model for my analyses. These models are appropriate because my goals are to understand how socioeconomic status, ties to first marriages, and differences in market restrictions impact gender differences in the likelihood of remarriage. The discrete-time logit model can be defined as:

M N ⎛ P ⎞ ⎜ it ⎟ log⎜ ⎟ = αt + ∑ βm X mi +∑ βn X nit ⎝1− Pit ⎠ m=1 n=1

Figure 1: Discrete-Time Logit Model

Where Pit is the probability of marrying for person i at time t, αt is a set of t-1

dichotomous variables for duration from dissolution, with the first year after dissolution

as the reference group. The model includes m time-constant independent variables and n time-varying variables measured at time t.

For ease of interpretation, I convert the results of the logit models from log odds to odds ratios. An odds ratio of 1.0 indicates that the explanatory variable has no effect on the likelihood of marriage. An odds ratio of greater than 1.0 indicates that the explanatory variable has a positive effect on the likelihood of an event, and increases the 49

odds of the event happening in any given year. An odds ratio of less than 1.0 indicates a

negative effect of the variable on the likelihood of marriage.

Analysis of Assortative Mating Patterns

In my analysis of marital sorting by education and age in Chapter 5, I use a

special case of the discrete-time logit model. Discrete-time multinomial logistic

regression allows me to distinguish homogamous matches from heterogamous matches when men and women remarry. The models also allow me to track how assortative mating changes with additional time in the remarriage market—or, how additional years since separation or divorce influence the sorting behaviors of men and women. The discrete-time multinomial logit model is defined as:

⎛ P ⎞ M N log⎜ ijt ⎟ = α + β X + β X ⎜ ⎟ tj ∑ m mij ∑ n nijt ⎝1− Pijt ⎠ m=1 n=1

Figure 2: Discrete-Time Multinomial Logit Model

Pijt is the conditional probability of marriage after divorce, where j=1 for an

educationally or age homogamous marriage, j=2 for a heterogamous marriage where the

husband is better educated or older, j= 3 for a heterogamous marriage where the wife is

better educated or older and j=0 for individuals not experiencing a remarriage, for person

i at year t since divorce. αtj is a set of t-1 dichotomous variables for time in the

remarriage market, with the first year after divorce as the reference group. The model

includes m time-constant independent variables and n time-varying variables measured at 50

time t. The results in Chapter 5 are presented as relative risk ratios—a special case of

odds ratios. The coefficients in these regressions can be interpreted in a similar manner

to odds ratios, as described above.

All of the models in my analysis of gender differences in the likelihood of and

assortative mating in remarriage are separated by the respondent’s sex. I do this for two

reasons. First, men and women are likely to search for spouses differently in the

remarriage market due to differences in their partner preferences (South 1991).

Unobserved differences between men and women in their partner preferences may bias

the results. Second, there are important differences between men and women after

remarriage, including economic well-being (Peterson 1996) and caring for children

(Goldschreider and Sassler 2006). How these characteristics contribute to the gender gap

in remarriage are of explicit interest in the subsequent analyses. However, in the interest

of parsimony and ease of interpretation, the results are presented separately for men and

women. However, this strategy does not allow me to identify statistically significant

differences in each effect across sex; only describe differences in how the predictors

influence remarriage for men and women independently.

Conclusion

This study uses data from the National Longitudinal Survey of Youth, 1979

cohort to analyze gender differences in remarriage. I include men and women who

experienced a first marriage and then divorced between 1979 and 2006. Two different

aspects of remarriage are considered in this study. First, in Chapter 4, I analyze gender differences in both the timing and likelihood of remarriage following a divorce. Second,

in Chapter 5, I investigate how and why assortative mating by educational attainment and

age differs by gender in remarriage. Individual characteristics such as economic

circumstances, ties to first marriages, and market constraints are considered as important

factors in both analyses. Particularly important characteristics are individual income,

labor force status, educational attainment, and parental status. I also consider the effects of time on the remarriage process by including the duration from separation/divorce and age. It is hypothesized that time and age will have deleterious effects on the odds of remarriage and on the likelihood of forming a desirable match in the remarriage market.

I use discrete-time event history methods in my analyses. In Chapter 4, where I focus on whether divorcees remarry or not and the factors associated with these outcomes, I use discrete-time logistic regression where remarriage is considered the event. In Chapter 5, where I analyze assortative mating patterns by education and age, I use the multinomial form of these models where remarriage is the event, but the type of remarriage

(homogamous or heterogamous) can be identified. These models allow me to predict the timing of remarriage and determine the factors associated with the gender gap in remarriage and gender differences in the partner pool.

Chapter 4: Gender Differences in Remarriage Formation

Divorce and subsequent remarriage are common in American family life, but studies on remarriage are mostly on children’s experiences in stepfamilies and the psychological and economic benefits of marrying again (Cherlin 1978; Cherlin and

Furstenberg 1994; Coleman et al. 2000). Although there are many studies on remarriage formation, few studies have focused on gender. It is well known that women are less likely to remarry after divorce than men (Bumpass et al. 1990; South 1991; Coleman et al. 2000; deGraaf and Kalmijn 2003; Wu and Schimmele 2005). By all accounts, the gap is large—the number of men who remarry after divorce is about 25% higher than it is for women (Centers for Disease Control 2002). At older ages, the gap between men's and women's remarriage rates grows-- divorced men over age 40 are at least twice as likely to remarry as divorced women over 40 (U.S. Census Bureau 2008). While gender differences in remarriage are well documented, little is known about why such a large gender gap exists in post-divorce union formation (but see Sweeney 1997; deGraaf and

Kalmijn 2003; Wu and Schimmele 2005).

Past research has suggested that socioeconomic status (e.g., Sweeney 1997; deGraaf and Kalmijn 2003; Wu and Schimmele 2005) and first marriage ties (e.g.,

Bumpass et al. 1990; Wilson and Clarke 1992; Bennett et al. 1995; Goldscheider and

Sassler 2006; Ono 2006) affect the likelihood of remarriage. However, how these factors affect men and women differently in remarriage is unknown. Socioeconomic status may impact men and women differently in the remarriage market. If remarriage is traditional, men should be evaluated on their economic potential while women should be assessed on non-economic criteria (Becker 1981). First marriage ties are also an aspect of marriage formation which makes remarriage distinctive from union formation among the never- married. Children are a particularly important first marriage tie, and tend to be cared for more by women than men after divorce (Goldschreider and Sassler 2006). A third and often overlooked aspect of marriage, time in the marriage market, may also help explain why men remarry at much higher rates than women. Namely, gender differences in potential partner pool size may lead time to have a more negative impact on women, who draw from smaller pools, than it will on men, who draw from larger pools. In this chapter,

I consider how these three factors contribute to the gender gap in remarriage.

The Current Study and Research Hypotheses

Socioeconomic Status

Studies of remarriage have highlighted the importance of socioeconomic attributes for remarriage formation (e.g., Koo and Suchindran 1980; Bumpass et al. 1990;

Smock 1990; Sweeney 1997; Lampard and Peggs 1999; Gelissen 2004). Socioeconomic measures, such as income, labor force status, and educational attainment often predict entry into remarriage because they indicate potential financial security, resources available to children, consumption patterns, and overall standard of living (Lewis and

Oppenheimer 2000; Goldstein and Kenney 2001; Sweeney 2002). As such, individuals

with high socioeconomic status tend to be attractive partners in the remarriage market

(England and Farkas 1986; Oppenheimer 1988). In contemporary first marriage, both men and women are evaluated on their achieved and future socioeconomic status

(Sweeney 2002). This represents a considerable shift from the past where men were evaluated on economic criteria and women were evaluated on non-economic criteria such as religion, family background, and physical attractiveness (Becker 1981). This shift resulted from important economic and cultural shifts in recent decades: greater income growth for women than men (Wetzel 1995), greater uncertainty for men in the labor market (Oppenheimer et al. 1997), higher consumption levels (Bumpass 1990), and a liberalization of attitudes toward gender roles in the family (Thornton 1989; Barich and

Beiley 1996).

The circumstances surrounding first marriage formation, however, are very different from those in remarriage. In first marriage, economic instability and difficulty evaluating long-term labor market position may lead both men and women to prefer partners with the best socioeconomic prospects. However, because men and women are older when they remarry, there is less uncertainty in economic status because occupational and income trajectories are more established (Oppenheimer et al. 1997).

Furthermore, divorce can affect economic well-being. Men often experience an increase in per-capita and discretionary income after divorce while the majority of women experience a decline in economic well-being (Smock 1994; Peterson 1996; Sweeney

1997; Smock et al. 1999). As a result, socioeconomic status may affect the remarriage prospects of men and women differently.

Economic status is typically linked to men’s marriage prospects because they are

often viewed as the primary provider for a family (e.g., Arensberg and Kimball 1968;

Easterlin 1978, 1987; Becker 1981; Watkins 1984). Thus, men are evaluated on their

ability to support a household at a "minimally acceptable standard" (Easterlin 1978;

Oppenheimer et al. 1997). These expectations are just as strong for divorced men who

are trying to remarry as they are for men marrying for the first time (Sweeney 1997).

Therefore, whether traditional or contemporary in nature, men’s socioeconomic status

should be positively associated with the likelihood of remarriage.

An increase in men’s post-divorce economic well-being can affect the remarriage

prospects of women. High-status divorced men may be attractive potential partners for

younger women who place a strong emphasis on a potential partner's economic status

(deGraaf and Vermeulen 1997; Oppenheimer et al. 1997). This will be discussed in

greater detail in Chapter 5, but the overall effect on the remarriage prospects for divorced

women can be negative because divorced men desire younger, never-married women

(South 1991). Men’s higher status after divorce can also mean that women’s economic

status is unimportant for remarriage. In this scenario, men are capable of providing for a

family, net of any contribution women can make. Women, as suggested by Becker

(1981), would be evaluated on more traditional criteria such as age, physical

attractiveness, and race/ethnicity. As a result, socioeconomic status may have no effect

on women's remarriage formation.

Current economic status. Socioeconomic status can refer to both current status and long-term prospects. Current economic status often refers to income (e.g.,

Oppenheimer et al. 1997; Clarkberg 1999; Sweeney 2002) or labor force status

(Goldscheider and Waite 1986; Oppenheimer et al. 1997) observed at the time of marriage. In first marriage, the effect of current economic status is interpreted as the value attached to the ability for a couple to live at an acceptable standard of living at the time of marriage (Easterlin 1978). Current economic status plays a role in first marriage, but the strength of the effect can grow with age (Kalmijn 1994). When men and women are young and marrying for the first time, both partners expect that their standard of living will improve as one or both of them gain career stability and meet their earnings potential (Xie et al. 2003). However, because men and women are older when they remarry, they should have established careers and incomes. Therefore, current economic status may play an important role in the remarriage market. Indeed, Sweeney (1997) suggests that measures of current economic status, such as income, may be the single most important predictors of remarriage.

For men, income is a sufficient measure of current economic status because it taps overall economic well-being (Oppenheimer et al. 1997). For men, income has a positive effect on remarriage in most studies (e.g., Wolf and MacDonald 1979; Glick 1980; Glick and Lin 1987). Yet some studies show no effect of income on men's remarriage formation (see deGraaf and Kalmijn 2003 for a review). These mixed results may be due to methodological problems including measurement issues and significant missing data

(deGraaf and Kalmijn 2003). In addition, the majority of these studies use European and

Canadian data, where marriage and remarriage may significantly differ from union formation in the United States (e.g., Hueveline and Timberlake 2004). For example,

cohabitation rates after divorce, which vary across socioeconomic status, are much higher

in Canada than in the United States (Wu and Schimmele 2005). As a result, income

should have a significant and positive effect on remarriage for divorced men.

Income is a poor measure of current socioeconomic status for women because it includes child support payments, alimony, and welfare benefits which are the direct result of divorce. Instead, labor force participation can measure current economic status for women because it represents a commitment to economically supporting a family

(Oppenheimer et al. 1997). Regular, full-time, year-round employment indicates a higher

level of labor force commitment and should have a stronger association with remarriage formation than part-time employment or remaining out of the labor force (Oppenheimer

1988). Notably, full-time employment is less universal among divorced women than it is among divorced men8. While men tend to be employed while married and maintain their

employment through divorce, but a significant portion of divorced women do not work

full-time prior to divorce9. As a result, marriage dissolution is often a catalyst for women

to either work outside the home for the first time or to reenter the labor force.

Full-time employment may negatively affect remarriage for women because it

makes marriage economically unnecessary (Becker et al. 1977; Mott and Moore 1983;

deGraaf and Kalmijn 2003) or because women who work are less attractive marriage

partners (Becker 1981). However, some studies find that labor force participation has no

effect on the likelihood of women remarrying (Sweeney 1997). Employment status may

8 In the NLSY79 sample of divorced men and women, approximately 85% of men and 78% of women are working full-time in any given year, on average. 9 In the NLSY79, about 70% of men and less than 50% of women were employed full-time in the year prior to divorce

not be associated with remarriage for women if they are evaluated on traditional criteria

(Becker 1981). A final line of research suggests that any influence labor force

participation has on the propensity to remarry is not due to economic concerns. Instead, deGraaf and Kalmijn (2003) show that employed women tend to use the workplace as a local marriage market to meet potential spouses. In review, there are several conflicting arguments about how women's labor force participation may affect remarriage.

However, if remarriages are traditional unions, working should have no affect on the likelihood of remarriage for women.

Long-term economic status. Educational attainment, as a measure of socioeconomic status, is less susceptible to yearly fluctuations than income or labor force status (Oppenheimer et al. 1997). Instead, educational attainment is a measure of long- term economic potential which influences social, cultural, and material lifestyle

(Oppenheimer 2003). In first marriage, educational attainment is positively associated with marriage (Kalmijn 1998) because men and women use education as a proxy for future economic status (England and Farkas 1986; Mare 1991; Qian 1998; Schwartz and

Mare 2005). In remarriage, education may not be a strong predictor of socioeconomic status because men and women are more economically established at later ages. Instead, educational attainment can serve as an assurance of long-term economic stability and recovery in the face of temporary unemployment or economic setbacks (Sweeney 1997).

Education is also salient in remarriage because the correlation between educational attainment and income is high (Fernandez et al. 2005), is the strongest predictor of career stability (Oppenheimer 2003), and indicates high social standing/status (Blau 1977).

Like income, education can lead to more economic stability and higher levels of financial well-being (Jyman et al. 1975; Easterlin 1978; Jencks et al. 1979; Oppenheimer 1997;

Oppenheimer 2003). College-educated men are also less subject to economic fluctuations and labor market changes than less-educated men, providing highly-educated men with a higher and more sustainable standard of living (Ashenfelter and Ham 1979).

As a result, women tend to be highly attracted to well-educated men, particularly college graduates (Schwartz and Mare 2005; Shafer and Qian forthcoming).

What influence educational attainment has on women's remarriage prospects is ambiguous. In contemporary first marriage, highly-educated women are more likely to marry than their less-educated counterparts (e.g., Lichter et al. 1992; Oppenheimer et al.

1995; Oppenheimer and Lew 1995; Rose 2004) because men are attracted to women who can make financial contributions to the household (Kalmijn 1998; Raley and Bratter

2004). However, in remarriage men may have a preference for traditional matches with less-educated women who are committed to domestic labor (South 1991). This may be due to lower economic need among divorced men with high post-divorce status (Peterson

1996), because divorcees have more traditional marriage preferences than never-married men (Bartling and Broussard 1999), the need for a spouse to take care of children

(Goldscheider and Sassler 2006), market availabilities (Gelissen 2004), or a combination of these factors. This perspective suggests that highly-educated women are penalized in the remarriage market. An alternative view suggests that educational attainment neither benefits nor hurts women in the remarriage market because they are evaluated on criteria such as physical attractiveness, age, and parental status (Becker 1981; Bumpass et al.

1990; Wu and Schimmele 2005; Goldschreider and Sassler 2006; England and Mclintock

2008). However, both perspectives argue that the relationship between educational

attainment and remarriage should not be positive for women. An assertion backed by previous work which shows either a negative effect or no effect of education on

remarriage for women (Mott and Moore 1983; Duncan and Hoffman 1985; Teachman

and Heckert 1985; Bumpass et al. 1990; Chiswick and Lehrer 1990; Smock 1990;

Sweeney 1997; Gelissen 2004; Wu and Schimmele 2005).

In summary, the hypotheses for socioeconomic status are as follows:

Hypothesis 1: Both current economic status and long-term economic status are

attractive attributes for men in the remarriage market. As a result, income and

educational attainment shouldd be positively related to remarriage for men.

Hypothesis 2: For women, the relationship between economic status (both current

and long-term) and remarriage is unclear. If full-time, year-round employed and highly-

educated women are economically independent or men are less attracted to them, there

should be a negative association between economic status and remarriage. On the other

hand, there may be no statistically significant relationship between economic status and

remarriage if women are evaluated on non-economic characteristics.

First Marriage Ties

A unique aspect of remarriage is that it takes place within the context of a failed,

highly-committed prior relationship. Often, divorcees maintain links to previous spouses

after divorce. These links often result from shared investments with an ex-spouse such as

caring and providing for the well-being of a child (Ono 2006). Divorced couples often

maintain legally appointed ties through shared custody or visitation rights. More informally, divorced men and women with children share parenting responsibilities after divorce (Coleman et al. 2000). These ties, specifically with regard to children, can add complexity to new relationships for divorcees. Stepfamilies are a non-institutionalized aspect of American family life that lacks clear rules or behavioral norms (Cherlin 1978).

The presence of a stepparent can lead to greater levels of conflict because of parenting issues, behavioral problems in children, and emotional problems related to divorce

(Cooksey and Fondell 1996; Coleman et al. 2000; Yeung et al. 2000). In total, children can have a negative effect on the likelihood of remarriage.

In considering the effect of children on remarriage, it is important to differentiate

between co-residential and non-residential children. Co-residential children can have

stronger effects on repartnering than non-residential children because new partners will

be more engaged in “active parenting.” The problems related to stepparenting should be

stronger in families where a child is a resident, not a visitor. The strength of ties to a

previous relationship may vary according to gender. Mothers with co-residential children

are more likely to maintain ties to a non-custodial father through financial means such as

child support, generating a visitation schedule, and arranging co-parenting situations.

Previous studies show no association between being a father of a non-residential child

and remarriage (Stewart et al. 2001; Goldscheider and Sassler 2006; Bernhardt and

Goldscheider 2002). The presence of co-residential children, however, shows stronger

effects, though the results still remain mixed (Teachman and Heckert 1985; Koo et al.

1986; Chiswick and Lehrer 1990; deGraaf and Kalmijn 2003; Wu and Schimmele 2005;

Goldscheider and Sassler 2006). Thus, the presence of a co-residential child, rather than

whether a divorcee is a parent, tends to influence remarriage.

Previous studies on how co-residential children influence remarriage patterns among divorced fathers have generated mixed results (Clarkberg et al. 1995; Nock 1998;

Stewart et al. 2001; Stewart et al. 2003; Goldscheider and Sassler 2006). Some research shows that co-residential children are less of a liability in the remarriage market for men than women. Others show that children have no effect on the remarriage prospects of

men, while additional studies suggest a positive relationship.. One reason for the

variation in results is that the measurement of co-residential children is measured

differently across studies. Regardless, it appears that the presence of children will impact

women more than men. Women are much more likely than men to become the primary

caregiver to children after divorce (e.g. Teachman and Heckert 1985; Ono 2005;

Goldscheider and Sassler 2006). Co-residential children should have a strong, negative

effect on the propensity to remarry for divorced mothers. The potential consequences of

and stigma attached to becoming a stepparent hurt women more in the remarriage market

than men because they are more likely to care for children. The reason is clear.

Attitudinal data shows that men are reluctant to take on parental responsibilities and,

thus, less willing than women to marry a partner with a child (Goldscheider and Kaufman

2006). Divorced mothers may also have difficulty in forming a marriage because their

time to search for a new partner is limited by their parental responsibilities (Glenn 2002).

Therefore, despite increased desire and need for marriage, divorced mothers should be

less likely to remarry than divorced women without co-residential children.

In summary, the hypotheses regarding first marriage ties are:

Hypothesis 3: First marriage ties, as measured by the presence of co-residential children, play no role in remarriage for men.

Hypothesis 4: The presence of co-residential children should decrease the likelihood of remarriage for divorced women.

Time in the Remarriage Market

As men and women spend more time searching for a partner after divorce, the odds of remarriage should decline because the available partner pool shrinks with time.

The prospects of remarriage are poorer than the prospects of first marriage from the start because the majority of similarly-aged men and women are currently married at the time of divorce (Vera et al. 1985; England and Farkas 1986). This difference, however, can be compounded by time spent searching for a spouse. Therefore, time in the remarriage market should have a strong negative effect on the likelihood of remarriage. With time, potential partners marry out of the partner pool (Lewis and Oppenheimer 2000) and divorcees grow older, making them less attractive potential partners (England and

Mclintock 2008)10. Finally, many men and women, while searching for a new marriage partner may form temporary unions, such as cohabitation which can reduce the chances of remarriage for men and women (Wu and Schimmele 2005).

Though the effects of time in the remarriage market should be negative for both men and women, the odds of remarriage should drop more precipitously for women.

10 An alternative perspective suggests that older divorcees are a select group that does not want to be married. However, South (1993) shows that there is no statistically significant difference between the previously-married and never-married in the desire to marry. In other words, it appears as if the desire to remain unmarried is no greater among divorcees than the never-married.

Men are not constrained to only marrying similarly-aged spouses, like women. It is

common for men to marry women younger than themselves (Raley and Bratter 2004).

For women, who have a more limited partner pool available to them after divorce,

conducting long searches can mean no marriage. In a limited partner pool, the

opportunity to marry can decline quickly because there are few potential partners who

continue to marry out with time (Oppenheimer 1988). Women's age at divorce can have

a strong effect on the likelihood of remarriage (Bumpass et al. 1990). Men's preference

for younger women means that older divorced women are unlikely to remarry.

Therefore, the hypothesis for the effect of time is as follows:

Hypothesis 5: The likelihood of remarriage should decline with additional years

since separation or divorce. However, the odds of remarriage will decline more sharply

for women than for men.

Other Factors Associated with Remarriage

Social background. Several other characteristics are associated with remarriage.

First, divorce may be an incentive to invest in accumulating human capital through

formal schooling, job training, or seeking out a new career--especially for women

(Holden and Smock 1991; Smock et al. 1999). Such investments may reduce the

likelihood of remarriage as divorcees try to attain higher economic status. Second,

racial/ethnic minorities are less likely to remarry than whites because divorced minorities face very limited marriage markets (Bumpass et al. 1990). Black women are particularly disadvantaged by black men’s high incarceration, unemployment, and mortality rates

(Lichter 1990). Third, family background may affect remarriage behavior. Family

structure can be transmitted intergenerationally (Teachman 2002). Parent's marital status

is highly correlated with an individual's marital status. Men and women who have

stepparents may be more likely to remarry than men and women who grew up with a

single parent or with two married parents. These individuals may have more positive

attitudes about remarriage and forming stepfamilies than men and women who grew up

with a single parent or with two biological parents. Fourth, the composition of the partner pool can vary across locations, with some locations being more advantageous for divorcees than others. Men and women living in urban areas may be more likely to remarry because cities tend to be highly populated, have a higher concentration of potential spouses, and have more balanced sex ratios than rural areas (Lichter 1990;

Lichter et al. 1995; Lewis and Oppenheimer 2000). There may also be regional

differences in remarriage because of differences in population, partner pool composition, and economic structure in different parts of the country (U.S. Census Bureau 2007).

First marriage experiences. Divorcees bring with them the experiences of first

marriage which can influence their likelihood of remarrying. The duration of first

marriage has been linked to the propensity to remarry (Bumpass et al. 1990). Those with

longer first marriages are more “marriage oriented” and more likely to remarry (Becker et

al. 1977). In terms of gender differences, duration of first marriage may serve as a proxy

for a more gendered division of labor in the home (Bianchi et al. 2000). Therefore,

duration of first marriage may affect women more strongly, especially if they have

invested more time in their home production skills. These skills can be transferred easily

to a new marriage. Furthermore, men and women who have spent more of their adult

lives married may find it difficult to adapt to single life compared to those who experienced shorter first marriages (Sweet 1973). In addition, cohabitation experience prior to first marriage may matter as well. They are part of a selective group that are associated with greater risks of divorce (Wolfinger 1999). Several hypotheses have been forwarded to explain this phenomenon including the notion that cohabitation changes marriage expectations because cohabitors are less committed to relationships (Smock

2000). If so, they may be a group who are less likely to remarry than those who did not cohabit prior to their first marriage.

Data and Methods

I use data from the National Longitudinal Survey of Youth, 1979 cohort

(NLSY79), a panel survey of 12,686 men and women interviewed annually from 1979 through 1993 and then biannually from 1994 on. My sample is made up of divorced men and women from the 1979-2006 waves of the NLSY79. The sample consists of 2,471 respondents: 1,179 men and 1,292 women. The sample is in event-history format with

19,278 total person-years: 8,780 person-years for men and 10,498 person-years for women.

The dependent variable measures the duration to remarriage and the likelihood of remarriage. There are four key variables in the analysis: (1) the duration variable which measures in the remarriage market; (2) current economic status, measured by income for men and labor force status for women; (3) long-term economic status, measured by educational attainment; and (4) the presence of first marriage ties, measured by the presence of co-residential children. Control variables include: current age, school

enrollment status, race/ethnicity, family socioeconomic status (mother's education), respondent's family structure at age 14, respondent's religious background, respondent's residential location (urban/rural, south/non-south), duration of first marriage, and if the respondent cohabited prior to first marriage.

The analyses are conducted using a discrete-time logistic regression model which allows me to identify the year of remarriage while including both time-varying and time- constant variables. The sample is separated by gender and I run the models separately for men and women. The data, variables, and analytic strategy are presented in more detail in Chapter 3.

Results

Descriptive Statistics

The means and standard deviations of the key independent and control variables are presented in Table 1. The information for the remarried are taken in the year they remarried and the information for the unmarried are at the time of their censoring, either through data attrition or the end of the data in 2006. 657 men and 686 women experienced remarriage, while 522 men and 606 women did not remarry at the time of censoring.

On average, men and women remarry in about 5-6 years, but men remarry about 6 months earlier. Individuals who do not remarry spend about 11 years in the remarriage market, with men spending slightly less time in the remarriage market than women. The average age of remarriage is 33.9 for men and 32.7 for women. Unmarried men and women are censored at about 44 years of age.

I now turn to the findings by current socioeconomic characteristics, measured by log of income for men and full-time, year-round employment for women. A significant and substantial income difference is evident between remarried and unmarried men.

Remarried men have 9.014 logged income units (an unlogged mean of about $46,400 in

2006 dollars) while unmarried men have a mean logged income of 7.911 (an unlogged mean of about $34,225 in 2006 dollars). Unmarried women are more likely to be full- time, year-round employed (85%) than remarried women (77%). This difference may be caused by economic necessity on part of unmarried women rather than a bias against employed women in the remarriage market.

Educational attainment serves as a measure of long-term economic standing.

Highly-educated men are most likely to remarry while men without a high school degree were least likely. This suggests that educational attainment is strongly linked to remarriage for men. For women, however, the differences across outcomes by educational attainment are not statistically significant. Overall, women are much more likely than men to have co-residential children—about 78% of women have a co- residential child from their divorce, while about 19% of men report having a co- residential child. This number may seem abnormally high for men, but it is likely that when custody is shared, men report a co-residential child (Center for Human Resources

Research 2008). While there is no statistical difference in children from divorce between remarried and unmarried men, remarried women are slightly less likely to have co- residential children in their home versus unmarried women (56.5% versus 62.4%, respectively).

Other factors also are associated with remarriage. Whites are much more likely to

remarry than Blacks or Hispanics. The race/ethnicity effect appears stronger for women

than for men. Remarriage differs significantly by family structure at age 14 for both men

and women. Men who grew up with a stepparent are more likely to remarry than to not.

Women growing up with a single-parent are more likely to remain unmarried than

remarry and women from two-parent households have a higher likelihood of remarriage.

Residential location also affects remarriage, with men and women living in urban areas

and living in the South more likely to remarry. Finally, turning to first marriage

characteristics, it appears that men and women who were married for longer durations are less likely to remarry. However this pattern is potentially misleading because age and time in the remarriage market are uncontrolled. Individuals who cohabited prior to divorce are less likely to remarry than those who did not. The results here suggest that

little gender differences in these characteristics.

Multivariate Analysis

I now turn to the discrete-time logistic models predicting the likelihood of

remarriage versus non-marriage following a divorce. Table 2 presents the odds ratios of

the likelihood of remarriage on the independent variables for both men and women,

respectively. Because the likelihood of remarriage increases and then declines with time

after separation or divorce I include years since dissolution and its quadratic form in the

models. The effect of time on remarriage is shown in Figure 3. This effect is curvilinear,

increasing to an apex at about 7 years since separation/divorce and then decreasing over

time. The effect of time is similar for both men and women. However, the decline in the

odds of remarriage is less steep for women than it is for men--suggesting that a longer

tenure in the remarriage market is less detrimental for women, despite their lower overall

odds of remarrying. Additionally, individuals who divorce at older ages are less likely to

remarry. A one year increase in age is associated with 6.3% reduction for men and 5.6%

reduction for women in the odds of remarriage.

Current socioeconomic status. Many scholars have argued that income should be strongly linked to an increased likelihood of remarriage for men (e.g. Sweeney 1997), and the results support this hypothesis. A one-unit increase in the log of income increases the odds of remarriage by 3.2%. For women, full-time, year-round employment was introduced as the measure of current socioeconomic status. The results show that labor force status is not significantly associated with the likelihood of remarriage for women. Clearly, divorced men may be evaluated on socioeconomic criteria, but not divorced women.

Long-term socioeconomic status. The effect of long-term socioeconomic status

on remarriage is similar to the effect of current socioeconomic status. There is a strong

positive effect of educational attainment on the odds of remarriage for men, but no effect

of education for women. Among men, those with less than a high school diploma have

35.3% lower odds of remarriage than men with a high school diploma. Male college

graduates have 32.3% higher odds of remarriage than their high school educated

counterparts.

First marriage ties. First marriage ties, as measured by the presence of co-

residential children from first marriage, affect remarriage for women, but not men. For

men, the effect of children on remarriage is positive, but insignificant. For women, the

effect of children is significantly negative. The presence of a co-residential child or

children from divorce decreases the odds of remarriage by 14.1% compared to women

without children from a divorce.

Other factors. Several other individual characteristics are associated with the

odds of remarriage. School enrollment is associated with a decrease in the likelihood of

remarriage for women, but not for men. Women who are enrolled in school have 26.4%

lower odds of remarriage than women not in school, in any given year. This suggests that

human capital investment on the part of women is associated with a lower likelihood of remarriage. Race/ethnicity also influences remarriage. Both Black men and women are less likely to remarry than White men and women. Black men have 30% lower odds of remarriage and Black women have 47.6% lower odds of remarriage than their respective

White counterparts. There is no statistically significant difference between Hispanics and

Whites. Thus, like first marriage, Blacks are disadvantaged in the marriage market—

especially Black females.

Family background matters for women, but not men. Women whose mother

graduated from high school have 18.5% higher odds of remarriage than women whose

mother did not graduate from high school. In addition, women who grew up in a two-

parent household or a household with a stepparent are more likely to remarry than women

from a single-parent household. Women from two-parent households have 36.7% higher

odds of remarriage and women from a household with a stepparent have 39.1% higher

odds of remarriage, respectively. These findings suggest inter-generational transmissions

in union formation (Teachman 2002).

Finally, residential location also plays a role in remarriage for women. Women living in the south have more than 30% greater odds of remarriage than women living elsewhere. Duration of first marriage is associated with a higher likelihood of remarriage for men. One additional year of first marriage is associated with a 3.8% increase in the odds of remarriage among men. The corresponding figure for women is not statistically significant. Thus, men who are married for longer periods of time may be more marriage-oriented than divorced women (Becker et al. 1977).

Summary. The findings with regard to the hypotheses are as follows: (1) current economic status is associated with a higher likelihood of remarriage among men; (2) there is no relationship between current economic status and remarriage for women, suggesting that women are evaluated on non-economic criteria; (3) long-term economic prospects, as measured by educational attainment, are positively associated with remarriage for men; (4) educational attainment is not associated with remarriage for women; (5) the presence of co-residential children from divorce has no affect on remarriage for men; (6) women who have co-residential children from divorce are less likely to remarry than women without children; (7) the effect of time on remarriage is similar for men and women—the odds of remarriage increase over time, but this effect is curvilinear, and the odds begin to decline after 7 years in the remarriage market for both sexes.

Discussion

The gender gap in remarriage is large and significant. This gap is important given

the negative personal effects of divorce (e.g., Coleman et al. 2000) and stratifying effects

of divorce (e.g., Peterson 1996). Research on remarriage has been lacking. Typically, studies on remarriage have focused on one set of characteristics (e.g., Sweeney 1997),

one specific measure like having children (e.g., Goldscheider and Sassler 2006) or on one

sex (e.g., Bumpass et al. 1990). The few comprehensive studies of remarriage have

primarily used data from Europe (e.g., deGraaf and Kalmijn 2003) or Canada (e.g., Wu

and Schimmele 2005). Thus, a comprehensive study of remarriage patterns by gender in

the United States and potential explanations for the gender gap have been missing from

the study of remarriage. This chapter addresses this gap by focusing on gender

differences in remarriage and offers explanations for the gender gap in remarriage using

nationally representative data from the United States.

Three attributes were addressed as potential sources for gender differences in

remarriage. First, socioeconomic status can be a source of the gender gap, especially if

remarriage is more traditional than first marriage. In contemporary marriage formation,

both men’s and women’s socioeconomic status increases the likelihood of marriage

(Sweeney 2002). Whereas, in traditional marriage formation, economic attributes are

significantly associated with entry into marriage for men, but not women (Becker 1981).

The evidence suggests that remarriage follows a traditional union formation pattern. In

my analyses, both current and long-term socioeconomic status positively affects the

likelihood of remarriage for men, indicating that men are evaluated as potential marriage

partners on the basis of economic status in remarriage. For women, these effects are not

significant. Instead, characteristics such as race/ethnicity and social background were

associated with the likelihood of remarriage for women. These attributes were

highlighted by Becker (1981) as important characteristics in traditional marriage markets.

One non-traditional attribute women are assessed on in the remarriage market is

the strength of first marriage ties. Having co-residential children from first marriage was

used as a proxy for first marriage ties. The presence of co-residential children is a good

measure because ex-spouses are often involved in co-custodial arrangements or visitation

rights and in co-parenting (Coleman et al. 2000; Goldscheider and Sassler 2006). Often,

caring for a child means that the custodial parent receives financial support from their ex-

spouse in the form of child support (deGraaf and Kalmijn 2003). Having a co-residential child from divorce significantly decreases the odds of remarriage for women. In fact, women with children are about 14% less likely to remarry than women without children.

No such effect exists for men. Therefore, the presence of co-residential children may block women from seeking out a remarriage. This is particularly important, given that

divorced women with children are much more likely to be poor following dissolution

than women without kids (Peterson 1996). It appears, then, that the women to whom

marriage may be most important are the least likely to marry after divorce.

A third potential reason for the gender gap in remarriage is that the timing to remarriage differs by sex. However, the results do not support this hypothesis. Instead, the results show that time in the remarriage market has very similar effects on the likelihood of remarriage for men and women. The effect of time on remarriage is

curvilinear, as noted by the quadratic function in the regression models. On average, the

likelihood of remarriage increases for men and women until they are in the remarriage

market for 7 years, and then begins to decrease. Therefore, time does have a significant

impact on the remarriage prospects for men and women—but does not account for gender

differences in remarriage.

This chapter represents a good starting point for understanding the remarriage

patterns. Yet, an analysis of what predicts entry into remarriage provides an incomplete

picture of these patterns. Additional investigation is required into how traditional

marriages are in remarriage—whether patterns of hypergamy, where women marry better educated and older men are present in second marriages. Furthermore, we know little about the quality of matches made in remarriages by men and women. More in-depth analysis into who remarries whom will also provide insight into the restrictions men and women meet in the remarriage market. Thus, an understanding of who remarries provides only one way to understand gender differences in remarriage among divorcees.

Chapter 5 presents additional analysis into the educational and age assortative mating patterns of divorced men and women, helping to provide further insight into remarriage in the United States.

76 Not Remarried Remarried Men Not Remarried Men Remarried Women Women (n= 657) (n= 522) (n= 686) (n= 606) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Time Years since dissolution † 5.189 3.531 10.655 5.884 * 5.682 3.710 11.330 6.038 * Age † 33.869 5.647 43.870 3.528 * 32.745 6.036 43.865 3.350 *

Current Socioeconomic Status Log of income † 9.014 3.841 7.911 5.461 * ------Full-time employed, lagged † ------0.771 --- 0.847 --- *

Long-term Socioeconomic Status Less than high school † 0.137 --- 0.157 --- * 0.099 --- 0.071 --- High school graduate † 0.726 --- 0.715 --- 0.755 --- 0.736 --- College graduate or more † 0.137 --- 0.128 --- * 0.146 --- 0.193 --- 77 First Marriage Ties Any co-residential children from divorce † 0.185 --- 0.189 --- 0.565 --- 0.624 --- *

Control Variables Currently enrolled in school 0.030 --- 0.013 --- 0.057 --- 0.018 ---

Race and ethnicity non-Hispanic white † 0.600 --- 0.458 --- * 0.628 --- 0.449 --- * non-Hispanic black 0.244 --- 0.341 --- * 0.198 --- 0.351 --- * Hispanic † 0.157 --- 0.201 --- 0.173 --- 0.200 ---

Table 1: Descriptive Statistics of Key Independent Variables by Remarriage Outcomes for Men and Women

Table 1 continued

Not Remarried Remarried Men Not Remarried Men Remarried Women Women Mean S.D. Mean S.D. Mean S.D. Mean S.D. Control Variables, continued Socioeconomic background, age 14 Mother less than high school graduate 0.390 --- 0.423 --- 0.397 --- 0.446 --- * Mother high school graduate 0.428 --- 0.402 --- 0.455 --- 0.376 --- * Mother college graduate 0.145 --- 0.136 --- 0.131 --- 0.152 ---

Family structure, age 14 Single parent home 0.158 --- 0.186 --- 0.130 --- 0.206 --- * Two-parent household † 0.661 --- 0.649 --- 0.710 --- 0.622 --- * Stepparent present 0.125 --- 0.084 --- * 0.101 --- 0.094 --- Other family structure 0.056 --- 0.080 --- * 0.060 --- 0.078 ---

78 Religious background, age 14 Catholic 0.314 --- 0.351 --- 0.324 --- 0.342 --- Conservative Protestant 0.359 --- 0.347 --- 0.334 --- 0.361 --- non-Catholic, non-Conservative Protestant 0.327 --- 0.303 --- 0.343 --- 0.297 ---

Urban residence † 0.753 --- 0.720 --- 0.761 --- 0.751 --- * Southern residence † 0.451 --- 0.412 --- 0.443 --- 0.431 ---

First Marriage Characteristics Duration of first marriage † 6.875 4.804 8.584 5.436 6.602 4.763 9.556 6.021 *

Cohabited prior to first marriage 0.062 --- 0.140 --- * 0.070 --- 0.114 --- * Source: National Longitudinal Survey of Youth, 1979 cohort (NLSY 79) Note: † Indicates a statistically significant gender difference at the p<.05 level; * Indicates a statistically significant difference between marital outcomes within gender at the p<.05 level

Men Women Time Duration since divorce 1.233 *** 1.280 *** Duration since divorce-squared 0.986 *** 0.984 *** Age 0.936 *** 0.955 ***

Current socioeconomic status Log of income 1.032 ** Full-time, year-round employed 0.924

Long-term socioeconomic status Less than high school 0.647 ** 0.887 College degree or more 1.323 * 1.026

First marriage ties Co-residential child/children 1.098 0.859 *

Currently enrolled in school 1.059 0.734 * (ref: not currently enrolled)

Race/ethnicity (ref: non-Hispanic White) non-Hispanic Black 0.700 ** 0.524 *** Hispanic 0.817 0.821

Socioeconomic background, age 14 (ref: mother without high school diploma) Mother high school educated 1.019 1.185 * Mother college educated 1.013 0.893

Table 2: Discrete-Time Logistic Regression Results for Remarriage, by Sex

Table 2 continued

Men Women Family structure, age 14 (ref: single-parent present) Two parents present 0.868 1.367 * Stepparent present 1.213 1.391 * Other family structure 0.730 1.182

Religious background, age 14 (ref: other religion) Catholic 0.982 0.915 Conservative Protestant 1.075 0.995

Urban 0.882 0.902

South 1.136 1.306 **

First marriage characteristics Duration of first marriage 1.038 ** 1.005 Cohabited prior to first marriage 0.768 0.778 (ref: did not cohabit prior to first marriage) N, respondents 1,179 1,292 N, person-years 8,780 10,498 -2 Log likelihood 176.17 228.63 Psuedo R-squared 0.037 0.045 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

0.8

0.6 remarriage

of 0.4

odds

Log 0.2

0 Men Women ‐0.2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Figure 3: Log Odds of Remarriage for Men and Women by Years Since

Separation/Divorce

Chapter 5: Educational and Age Assortative Mating in Remarriage

Mate selection patterns reflect and influence social stratification within a society.

Assortative mating, or who marries whom, affects both intra- and inter-generational social standing and socioeconomic mobility (Kremer 1997; Fernandez and Rogerson

2001), which can magnify inequality between individuals and families (Kalmijn 1998).

This may be particularly important in remarriage because divorce can be a stratifying event in the later life course (e.g., Hao 1996; Marks 1996; Peterson 1996; Smock et al.

1999). Furthermore, marital sorting provides insight into the marriage market that men and women face after they divorce. Indeed, Mare (1991) argues that the study of who marries whom is “a fundamental building block in understanding social structure and social life [15].” Similarly, Atkinson and Glass (1985) argued that marital sorting patterns indicates the level of gender, racial/ethnic, and socioeconomic equality within marriage and Qian (1998) contends that assortative mating is a sign of marriage market quality.

Evidence from first marriage suggests that both men and women have a preference for homogamous marriages and seek out partners with similar social, economic, and demographic characteristics as themselves (England and Farkas 1986;

Oppenheimer 1988; Lichter 1990; Bergstron and Bagnoli 1993; Lichter et al. 1995;

Shafer and Qian forthcoming). Two types of homogamy are particularly relevant for

social stratification and understanding the operation of marriage markets: educational and age homogamy. In first marriage, educational homogamy has increased over the past three decades due, in part, to the closing gender gap in educational attainment, higher average at first marriage, and economic trends which often require both men and women to make economic contributions to the household (e.g. Sweeney 2002). Rates of age homogamy in first marriage increased until the 1970s (Van Poppel et al. 2001), but have recently leveled off or even slightly decreased (Qian 1998).

However, whether high levels of homogamy present in first marriage are also present in remarriage is unknown. In fact, there are two reasons to suspect that assortative mating differs significantly between remarriage and first marriage. First, as discussed in Chapter 2, remarriage may be more traditional than first marriage (Morgan and Waite 1987; Wu and Balkrishnan 1994; Bartling and Broussard 1999) meaning that divorcees may not prefer homogamous matches. In traditional matches, men, particularly men with high socioeconomic status, prefer younger and less-educated women. Second, smaller and selective partner pools in remarriage can also affect assortative mating.

Namely, the number and type of partners available in the remarriage market depend on divorcees’ individual characteristics. The pool may be larger for some individuals, such as high-status men, and small for others, such as such as women with children. In total, these preferences and constraints will influence choices about whether and whom to remarry. This chapter focuses on these choices, with respect to educational and age assortative mating.

Assortative Mating by Education and Age

In Chapter 2, I discussed why assortative mating on education and age are two

important aspects of marital sorting. Educational attainment plays a significant role in

determining socioeconomic status (Blau et al. 1982; Kremer 1997; Fernandez and

Rogerson 2001) and shows clear boundaries between social strata (Blau 1977; Smits et al.

1998; Blossfield and Timms 2003). Education also reflects cultural attributes such as ideals, norms, lifestyle, leisure activities, tastes, intellectual interests, and life experiences

(Blau and Duncan 1967; Hyman et al. 1975; Hyman and Wright 1979; Davis 1982;

Kalmijn 1991). Thus, educational assortative mating reveals important socioeconomic and social aspects to remarriage.

Age provides insight into gender attitudes in marriage, partner preferences, and the available partner pool in the marriage market. With regard to gender and marital

equality, smaller age differences are interpreted as representative of more contemporary

views about gender roles and gender equity in marriage (Richman 1977; Atkinson and

Glass 1985). Patterns of assortative mating by age in remarriage can also provide some

insight into how people marry with regard to prior marital status11. The majority of age

heterogamous remarriages are between older men and younger women. These matches

also tend to be heterogamous with respect to prior marital status—with more older men

11 Studies of assortative mating by marital status are limited (but, see Ono 2005 and Ono 2006), primarily because only vital statistics have information on the marital history of spouses. Survey data such as the NLSY79 do not include such information, making an analysis of matching by marital status difficult.

being previously-married and many younger women having never-married (Ono 2005)12.

Age homogamy can also reflect how many well-matched potential partners are available to single men and women (Qian 1998). This perspective suggests that the ability to match on characteristics such as education, religion, or race/ethnicity and still marry a similarly-aged spouse varies according to the size of the potential partner pool in the marriage market. Therefore, age assortative mating outcomes reflect both preferences and constraints.

Gender Differences in Educational and Age Assortative Mating

Assortative mating patterns in remarriage should vary by gender in remarriage, primarily due to differences in partner preferences and partner pools. I outline these differences below.

Partner Preferences

As discussed in Chapter 2, remarriage may be more traditional than contemporary first marriages with respect to gender roles and expectations. In first marriage, research shows that men and women search for each other with similar education and ages in order to produce the highest level of total or pooled resources possible. However, marrying economically attractive women may be a lower priority for high-status men who have more established careers and more financial stability than their never-married counterparts (Sweeney 1997). Thus, in remarriage markets, the economic benefits to homogamy may be lower for divorced and financially stable men, leading to remarriages

12 Data limitations do not allow me to capture matting by marital status. Nevertheless, age remains an important characteristic for divorced men and women to consider in remarriage. Often, age heterogamy is a good indicator of marital status heterogamy.

with less-educated women. With respect to age, divorced men report a preference to

marry women younger than themselves (South 1991; South 1993). This reflects a

traditional pattern where age hypergamy is relatively common (Raley and Bratter 2004).

For women, partner preferences are similar whether marriage matching takes on a more traditional hypergamous form or contemporary where marriage favors homogamy.

Attitudinal studies of spousal preferences show that women typically have a stronger preference for husbands with high education, high income, and good employment (South

1991). Indeed, women marry based on these preferences, both historically (e.g., Landale and Tolnay 1991) and contemporarily (Lichter et al. 1991; South and Lloyd 1991;

Oppenheimer et al. 1997; Lewis and Oppenheimer 2000). These preferences for high- status men should continue into remarriage (deGraaf and Kalmijn 2003; Gelissen 2004).

With regard to age, women's preferences are more varied than men's (Bytheway 1981).

Women report a higher willingness to marry both similarly-aged and older individuals than are men (South 1991). Therefore, women’s assortative mating patterns may be more varied than men’s.

The Available Partner Pool

Choices in remarriage, however, are not solely a function of men's and women's preferences. Men and women draw their spouses from the pool of available partners when remarrying. From marital search theory (England and Farkas 1986; Oppenheimer

1988) we know that the size of an individual’s potential partner pool is a function of one’s own characteristics as well as constraints in the marriage market. Individuals who have attractive attributes such as high income or high educational attainment are

considered good matches and, as a result, have more potential partners available to them.

However, each individual can be constrained in the marriage market by personal characteristics, the composition of the potential partner pool, social barriers between groups, and other factors. When men and women have attractive attributes and few constraints, they have large partner pools available to them. When men and women have less attractive attributes and many constraints their potential partner pool shrinks. As a

result, individuals who have larger partner pools and more choices are more likely to

make their preferred match in the remarriage market. In terms of gender, men are much

more likely to make preferential matches than women, because of their attractiveness in

the marriage market and ability to engage in status exchange with less-educated, younger

women.

Assortative Mating in Remarriage for Men

Despite their previous marital status, high-status divorced men are highly

attractive partners in the marriage market (Becker 1981; England and Farkas 1986;

Oppenheimer 1988; Oppenheimer et al. 1997; Schwartz and Mare 2005; Wu and

Schimmele 2005) because many divorced men increase their per-capita income and

overall economic well-being after divorce (Peterson 1996). Furthermore, divorced men

tend to be more economically established than never-married men (Sweeney 1997).

High-status men who divorced during the time period under consideration (1979-2006)

are advantaged by macro-level economic trends which have led to deterioration in young

men’s economic prospects (Juhn et al. 1993; Wetzel 1995). Therefore, there are very few

constraints on the preferences of high-status divorced men in the remarriage market. As

a result, they should have traditional, hypergamous assortative mating on education and

age.

An important consideration with respect to age, however, is how old high-status

men are when searching for a new marriage partner. Many men who are younger

divorcees may be inclined to marry similarly-aged, not younger women. Empirical

evidence suggests that men in their 20s and early 30s prefer to marry a woman from the

same general age group, while older men prefer to marry women younger than themselves (Vera et al. 1985; England and Farkas 1986; Oppenheimer 1988; Qian 1998;

England and Mclintock 2008). Furthermore, it is unlikely that divorced men in their 20s will marry younger women because people are only seriously engaged in the open marriage market around age 18 (Lewis and Oppenheimer 2000). This naturally increases

the likelihood of age homogamy for younger divorcees. As a result, age homogamy

should be prevalent among younger high-status divorced men and age hypergamy should

be the norm for their older counterparts.

In summary, the hypotheses for the assortative mating patterns of men are as

follows:

Hypothesis 1: In terms of educational assortative mating, divorced men with high socioeconomic status will be the more likely than lower status men to marry traditionally by partnering with less-educated women.

Hypothesis 2: With respect to assortative mating by age, highly-educated men will be more likely than men with less education to marry a woman younger than themselves. However, this effect will vary by age of divorce. Younger highly-educated

men will be more likely to form age homogamous marriages, while older highly-educated men will be more likely to marry younger women.

Assortative Mating in Remarriage for Women

Unlike men, there are several reasons why socioeconomic status should not affect the assortative mating outcomes of women. First, as noted in Chapter 2, some women have lower economic well-being after divorce (Peterson 1996), which can hurt them in the remarriage market (England and Farkas 1986; Oppenheimer 1988). Second, high- status women may be unattractive marriage partners to high-status men if traditional marriage patterns prevail in remarriage. Third, divorced women, regardless of economic status, can be unattractive marriage partners for men who prefer younger, never-married women (Ono 2005; England and Mclintock 2008). Together, these issues suggest that there may be no discernable assortative mating patterns across socioeconomic status for women. Instead, women may be evaluated on other criteria while matching in the remarriage market. I discuss two of these potential characteristics in detail below.

Age. Among women, youth is a preferred characteristic in the remarriage market.

Women who are younger are much more likely to get remarried than older women

(Teachman and Heckert 1985; CDC 2002). Age may also be associated with making quality matches when remarrying. Because of men's age preferences (England and

Mclintock 2008), younger women may be more likely than older women to marry hypergamously. Young divorced women may have a larger pool of potential partners to draw from when they remarry than older women. Thus, younger women should be more

likely to marry because: (1) men prefer to marry younger women and (2) they have larger

partner pools available to them than older women.

Children. Chapter 2 discusses in detail the influence that first marriage ties, such

as children, can have on remarriage. Indeed, Chapter 4 shows that women with children

are much less likely than women without children to remarry following a divorce.

Children also affect the assortative mating patterns of women. Women with co- residential children from previous relationships are less attractive marriage partners in the remarriage market, meaning that they should have difficulty forming good matches when remarrying. Often, men with high socioeconomic status are unwilling to enter marriages with divorced mothers because children are socially and economically costly. Remarried couples with children report lower marital quality (Coleman et al. 2000) and more complicated relationships (Cherlin 1978) than remarried couples without children.

Women with children often maintain ties with their ex-husband through receipt of child support or shared custodial and parental responsibilities, which can harm new marriages

(Coleman et al. 2000; deGraaf and Kalmijn 2003; Ono 2005). Thus, while caring for children reduces the likelihood of remarriage for women, it should have no effect on assortative mating.

In review, the hypotheses for women’s assortative mating patterns in remarriage are as follows:

Hypothesis 3: Because women’s socioeconomic status is a less significant characteristic in the remarriage market, educational attainment and labor force status will have no statistically significant effect on assortative mating among women.

Hypothesis 4: Women, however, will be evaluated on more traditional criteria in

remarriage. Younger women should be more likely to form educational and age

hypergamous marriages than older women. Because the remarriage prospects of women

decline with age, non-marriage should be the most likely outcome for older women.

Hypothesis 5: Parental status will negatively affect the likelihood of remarriage,

but have no effect on assortative mating.

Time in the Remarriage Market

There are several reasons why time can have a deleterious effect on assortative mating. First, fewer and fewer potential spouses are available to men and women as they age. This effect may be particularly strong in remarriage where the pool of available partners is smaller than in remarriage (Gelissen 2004) and the size of the partner pool gets much smaller with age (deGraaf and Kalmijn 2003). Second, young divorcees may have more opportunity to marry according to their preferences because the availability of potential partners should be higher. Conversely, older men and women can face a tight marriage market, making high-quality matches unlikely—especially as they age. Third, divorcees may cohabit and have children as they search for a new spouse. Forming non- marital families reduces the likelihood of marrying and finding a quality marriage partner. Some research suggests that up to half of men and women cohabit after divorce

(Wu and Schimmele 2005). Importantly, these arrangements have been linked to a lower likelihood of marriage and marrying a high quality man or woman (Bumpass et al. 1990;

Qian, Lichter, and Mellott 2005).

Because the opportunity to marry an attractive partner may decline over time,

divorced men and women may use alternative search strategies to find a new spouse.

One such strategy is to alter marriage preferences in order to match the available partner pool. In first marriage, men and women cite preferences for high-status partners at early ages (Raley and Bratter 2004); however, with additional time in the marriage market individuals can change their search strategies. Men are more likely than women to lower their spousal preferences and become less-selective searchers (Lichter 1990; Mare 1991;

Shafer and Qian forthcoming). Men’s choices allow them to marry, although that marriage may be to a less preferential mate. Women, on the other hand, are much more

likely to remain unmarried rather than to change their marriage preferences (Raley and

Bratter 2004). Divorced women are also disadvantaged because the majority of similarly-aged men are either currently married or have a preference for a younger woman (Vera et al. 1985; England and Farkas 1986). Thus, the likelihood of non- marriage increases for women with more time in the remarriage market, but the likelihood of less-desirable matches increases for men.

Therefore, the hypotheses for the effect of time on assortative mating in remarriage are as follows:

Hypothesis 6: Additional time in the remarriage market, as measured by duration from separation/divorce, will increase the likelihood of less traditional matches for men.

With respect to both education and age, the odds of marrying less-educated, younger

women will decrease with time. However, the odds of homogamy and hypogamy will

decline at a much slower pace.

Hypothesis 7: Additional time in the remarriage market for women will increase the likelihood of women not remarrying. The effect of time should not differ across assortative mating outcomes for women.

Data and Methods

I use the National Longitudinal Survey of Youth, 1979 cohort (NLSY79) to test the above hypotheses. NLSY79 is a panel survey of 12,686 men and women aged 14-22 in 1979. The survey was administered annually from 1979 through 1993, and since 1994 biannually. The sample consists of men and women who divorced between the 1979 and

2006 waves of the data. The sample consists of 1,179 men and 1,292 women for a total of 2,471 individuals. I use event-history data, so the analysis is run on 19,278 person- years, with 8,780 person-years for men and 10,498 person-years for women.

There are two dependent variables in these analyses. The first focuses on educational assortative mating and has four possible outcomes: (1) educational homogamy; (2) the husband more educated than the wife; (3) the wife is more educated than the husband; and (4) not remarried. The second dependent measure is for age assortative mating and also has four outcomes: (1) homogamy; (2) the husband is older;

(3) the wife is older; and (4) not remarried. I focus on the following variables in my analyses: the duration variable which measures in the remarriage market, age of respondent, including an interaction between age and education in the models for marital sorting by age, current economic status, measured by income for men and labor force status for women, long-term economic status, measured by educational attainment, and the presence of first marriage ties, measured by the presence of co-residential children.

Control variables include: current age, school enrollment status, race/ethnicity, family socioeconomic status (mother's education), respondent's family structure at age 14, respondent's religious background, respondent's residential location (urban/rural, south/non-south), duration of first marriage, if the first marriage was educationally or age homogamous, and if the respondent cohabited prior to first marriage.

The analyses are conducted using a discrete-time multinomial logistic regression models which allows me to identify the year of remarriage, the assortative mating outcome, and the effects of both time-varying and time-constant indepdendent variables.

The sample is separated by gender and I run the models separately for men and women.

The data, variables, and analytic strategy are detailed in Chapter 3.

Results

Descriptive Results

Table 3 presents differences between the respondent and their spouse on key characteristics at the time of remarriage by sex of the respondent. The results show important differences between men and women in their marriage patterns. First, male respondents, shown in the top panel of Table 3, are, on average, better educated than their spouses. The average remarried male in the sample has 13.7 years of education, while their spouse has 13.2 years of education. In comparison, female respondents, reported in the bottom panel of Table 3, have lower educational attainment than the female spouses of divorced men (12.9 years of education versus 13.2 years). Female respondents also marry men with lower educational attainment than the average remarried male respondent (13.2 years of education versus 13.7 years). Second, men tend to marry

similarly-aged women. On average, men are approximately 2 to 3 years older than their

spouses. Men who marry female respondents are nearly 3 years older than their wives.

However, this does not account for differences by educational attainment or age in

assortative mating. In terms of income, remarried male respondents earn about $20,000

more than their spouses. Female respondents, in comparison, marry men that make only

$6,000 more than themselves. Interestingly, there is a considerable income difference

between male respondents and the male spouses of female respondents. Male

respondents in the sample make approximately $10,500 more than spouses of female

respondents. Finally, very few men have custody or co-custody of their children. 18% of

male respondents report a co-residential child while 24% of their new spouses report a

co-residential child. By comparison, 56% of divorced female respondents report a co-

residential child while just 11% of their new spouses report having a co-residential child.

Thus, female respondents are more likely to have children than women who marry

divorced male respondents.

The most interesting comparisons in Table 3 are the differences between male

respondents and the men women marry. Male respondents are 1.5 years younger when

they remarry, are better educated, and earn more money than the spouses of female

respondents. A comparison between female respondents and the spouses of male

respondents shows similar differences. Men’s spouses are slightly younger on average,

are much less likely to have children, and have lower income on average. Thus, these descriptive results support the notion that divorced men are making more traditional matches when they remarry and divorced women are marrying less-desirable men.

Educational and age paring by educational attainment. Table 4 shows descriptive

results which illustrate assortative mating on both education and age for men and women.

These results are presented by educational attainment for each sex. 76.3% of less than

high school educated men who remarry partner with a better educated woman. Within this category, men are slightly more likely to marry older woman than similarly-aged or younger women. This suggests that less-educated men are in a good position to marry high-status women when remarrying. Conversely, highly-educated men are most likely to form traditional, hypergamous marriages. 48.3% of college-educated men marry a woman with lower educational attainment and about 40% are older than their wives.

When considering both characteristics together, 22.4% of college-educated men marry women who are both younger and less-educated. The patterns among men with at least a high school diploma show a somewhat even distribution across outcomes, with no real pattern to their marital sorting.

Less-educated women are likely to partner with better educated men. 67.2% of women with less than a high school education marry a man with more education. In terms of age, less-educated women are almost equally likely to marry a similarly-aged man or a man who is older than themselves (36.6% and 35%, respectively). For college- educated women, two outcomes are most likely. 21.8% college-educated women marry a similarly-educated, older man. The same percentage of female college graduates marries men who are less-educated and older than themselves. Thus, while there appears to be a preference for educational homogamy among highly-educated women, it also seems that a significant number of highly-educated women are squeezed in the marriage market.

This squeeze makes it more likely that female college graduates will marry a man with

lower overall status than themselves. Nevertheless, the commonality between these two

outcomes is that these women marry older men, meaning that high-status women may

make somewhat traditional matches when remarrying.

Results for Educational Assortative Mating

Descriptive statistics. The means and percentage distributions for the key

independent variables are reported in Table 5. The results are presented by assortative

mating outcome and by sex at the time of remarriage or censoring. Men’s results are

presented in the upper panel of Table 5 and show variation in the covariates across

outcomes. First, duration since divorce/separation varies by marital outcome. Men who

partner with less-educated women remarry more quickly (4.7 years after

separation/divorce) than men who marry homogamously (5.2 years) and men who marry

better educated women (5.7 years). This pattern suggests that men are more likely to

make desirable matches, such as marrying a less-educated woman, if they remarry

quickly. However, as they spend more time in the remarriage market, less desirable

outcomes become more likely. In terms of age at remarriage, there are only small,

insignificant differences across educational assortative mating outcomes. Men who

marry less-educated women are the youngest when they remarry (32.57 years old), while

men who marry homogamously are oldest when they remarry (33.65 years old).

Turning to current socioeconomic status, the results indicate that high-status men

are more likely to form traditional matches on educational attainment. Men who marry less-educated women have the highest average income—9.89 logged units (an unlogged

mean in 2006 dollars of approximately $41,300). In comparison, men who marry homogamously and men who marry better educated women have nearly identical average incomes. Men who marry homogamously have an average income of 9.06 logged units, or an unlogged average in 2006 dollars of $38,200. Men who marry better educated women have a slightly higher income on the logged scale (9.14 logged units), but a lower average unlogged income (approximately $34,225)13.

The results for long-term socioeconomic status, as measured by educational attainment, lend further support for the notion that high-status men make traditional matches when remarrying. Men with less than a high school diploma are more likely to marry better educated women (17%) than they are to marry homogamously (12%).

Homogamy is the most likely assortative mating outcome for men with a high school diploma. 26% of these men marry homogamously, while only 15% marry a less- educated woman and 16% marry a better educated woman. By comparison, men with a male college degree are slightly more likely to marry a woman with less education than to marry a woman with the same education (30% versus 29%).

Women's results are reported in the lower panel of Table 5. The results show markedly different patterns than exist for men. First, duration since separation/divorce varies differently across outcomes for women than it does for men. Duration to remarriage is shortest among women who marry less-educated husbands (5.18 years since separation/divorce), followed by educational homogamy (5.86 years) and marriages where women marry better educated men (6.09 years). These statistics highlight two

13 This is because the logged scale is unadjusted, but varies with time.

important points. First, women have a longer duration to remarriage from separation/divorce than men do. Second, women who remarry most quickly are marrying the least desirable men in the remarriage market. This supports the assertion that women who remarry quickly may do so out of economic necessity, not desire to remarry

(Sweeney 1997).

There are larger age differences across educational assortative mating outcomes for women than men. Women who marry better educated husbands have the lowest average age of remarriage at just over 31 years of age. By comparison, women who marry similarly-educated men are 32.5 years of age, on average, and the mean age of remarriage for women who marry less-educated men is 33. Therefore, the descriptive statistics support the idea that younger women are the most likely to marry hypergamously. However, as women age, they are more likely to marry similarly- educated and less-educated men.

Current socioeconomic status is more strongly associated with not remarrying than any marital sorting outcome. Indeed, 85% of women who are not remarried are full- time, year-round employed at the time of censoring. Anywhere from 74% to 78% of women who do remarry, depending on assortative mating outcome, are in the labor force.

Similarly, the influence of long-term socioeconomic status differs for women when compared to men's patterns. Higher educational attainment among women support less traditional matches, with 27% of female college graduates being better educated than their husband. Educational homogamy was the most likely outcome for female high

school graduates. Women with less than a high school diploma have a high likelihood of

a traditional match, with 47% of this group marrying a better educated man.

It was hypothesized that children would have a substantial effect on women’s

assortative mating patterns. However, the descriptive results do not initially support this

hypothesis. There seems to be little difference among assortative mating outcomes for

women with or without children.

Regression results. The effects of the key independent variables from the

discrete-time multinomial logistic regression of educational assortative mating patterns

are reported in Table 6. The full results of the regression are reported in Appendix A.

The results for men are reported in the top panel of Table 6. Time has a curvilinear effect on the likelihood for all assortative mating outcomes among men. The likelihood of all

three outcomes, compared to not remarrying, increase at a decreasing rate. However,

there are no statistically significant differences between marital sorting outcomes for

either the duration variable or its quadratic form. This suggests that time only has an

impact on the likelihood of remarriage for men, not on assortative mating. Age has

similar effects on educational assortative mating for men. Men who are above the mean

age in the sample are more likely to remain unmarried than to enter into any of the three

union types. Generally, the odds of any assortative mating outcome decrease about 10%

for each additional year above the mean age. However, there are no statistically

significant differences across marital sorting outcomes.

Socioeconomic status has a strong effect on educational assortative mating for

men. Income increases the likelihood of all three assortative mating outcomes compared

100

to not remarrying. In addition, income has a positive and statistically significant effect on the likelihood of men marrying less-educated women compared to marrying homogamously. A one log unit increase in income increases the odds of marrying hypergamously compared to homogamously by 19.8%. Furthermore, high earning men are unlikely to marry homogamously or hypogamously. Indeed, a one log unit increase in income decreases the odds of marrying a better educated woman by 15.5% compared to marrying a less-educated woman. These results show that higher socioeconomic status is associated with traditional marriage patterns for men.

The results for women's educational assortative mating patterns are reported in the lower panel of Table 6. Time has a similar effect for women as it does men—the effect of duration on remarriage increases at a decreasing rate. Like men, there are no statistically significant differences across assortative mating outcomes, only in comparison to not remarrying. However, the effect of age is much stronger for women than men. The odds of homogamy and marrying a better educated husband are 10.9% and 17.5% lower compared to not remarrying for each additional year above the mean age, respectively. Furthermore, as women age, the odds of marrying a less-educated husband increase, compared to both educational hypergamy and homogamy. Thus, the likelihood of marrying a less attractive husband increase as divorced women age in the remarriage market.

Another difference between men's and women's educational assortative mating patterns is that socioeconomic status is not associated with marital sorting for women.

There are no statistically significant differences in the effect of full-time, year-round

101

employment for women. This finding shows that women’s socioeconomic status is relatively unimportant for remarriage formation—which is consistent with traditional marriage patterns.

Finally, it was hypothesized that co-residential children would have an effect on the quality of matches made by women. Generally, however, children have no effect on educational assortative mating for women. This may be because due to the fact that women with children are less likely to remarry than women without children (e.g.,

Bumpass et al. 1990; also shown in Chapter 4). Therefore, the important difference is whether or not divorced mothers remarry, not how they remarry. The only statistically significant effect of children is for women marrying a less-educated man. Women with children from divorce have 32.2% lower odds of marrying a less-educated man than not remarrying. Nevertheless, women with children are unlikely to remarry. The relative risk ratios are below one for each of the outcomes when compared to non-marriage.

Results for Assortative Mating by Age

Descriptive Results. The descriptive results for key independent variables by age assortative mating outcome are reported in Table 7. These results are presented separately for men and women—with men’s results reported in the top panel and women’s results reported in the bottom panel of Table 7. For men, time in the remarriage market appears to be associated with assortative mating outcomes. Men who marry women at least five years younger than themselves have the shortest duration from separation/divorce to remarriage at 4.87 years. By comparison, men who marry homogamously (wife is one year older to four years younger) and men who marry older

102

women (at least two years older) spend an average of 5.18 years and 5.57 years in the

remarriage market, respectively. In terms of age at remarriage, men who marry

homogamously are younger (32.38 years old at remarriage) than men who marry either

younger (34.78 years old) or older women (34.31 years old). These patterns indicate that:

(1) men who repartner quickly are the most likely to make traditional matches and (2)

younger men are more likely to marry similarly-aged spouses than to marry older women.

Current socioeconomic status does not appear to have a strong effect on the age

assortative mating patterns of men. The income differences across outcomes are very

small. Men who marry younger women and men who marry older women have nearly

identical logged incomes at the time of remarriage. The difference in their unlogged,

2006 adjusted incomes, is also very small--only about a $2,600 advantage for men who

marry younger women. Men who marry homogamously have only slightly lower income

on the logged scale, and have an adjusted income only about $2,200 higher than men who marry older women.

The relationship between assortative mating by age and long-term socioeconomic

status appears to be much stronger. Men with less than a high school diploma and high

school graduates have very similar age sorting patterns. However, assortative mating patterns among college-educated men stand in sharp contrast to less-educated men. 26% of college graduates marry younger women, compared with 21% of less than high school graduates and 20% of high school graduates. Men with a college degree are unlikely to marry older women. Only 14% of male college graduates enter this type of union, while

19% of less than high school graduates and 18% of high school graduates do so.

103

Therefore, the descriptive results indicate that highly-educated men are the most likely to form traditional, hypergamous marriages with respect to age.

Overall, women have different patterns of age assortative mating than men. First, the differences across outcomes in time from separation/divorce to remarriage are much smaller for women than they are for men. Women who marry older men remarry, on average, 5.54 years after separation or divorce. Women who marry similarly-aged men remarry in 5.59 years and women who are older than their husbands in 6.03 years. In terms of age, women’s patterns are similar to those present for men. Women who marry homogamously have the youngest average age at remarriage, at 31.29 years of age.

Women who marry older men (33.89 years old) and women who are older than their husbands (32.81 years old) are one-two years older at remarriage than women who marry homogamously.

Socioeconomic status also has different effects on women’s assortative mating by age than it does for men. At the time of remarriage, 78% of women who marry homogamously are engaged in full-time, year-round work. The comparative numbers for women who marry older men and younger men are 74% and 77%, respectively. Thus, there seems to be very small, inconsequential differences in marital sorting by current socioeconomic status for women.

There are differences in women’s age assortative mating across educational attainment. First, remarriage becomes progressively less likely as education increases.

Second, women with less than a high school education are the most likely to marry similarly-aged and older men. 21% of women with less than a high school degree marry

104

homogamously, while 26% marry older men. The comparable numbers for high school

graduates are 17% and 21%. For college graduates, the percentages are 14% and 22%,

respectively. Women with less than high school or high school diploma are the most

likely to marry younger men. Accordingly, the least educated women are the most likely

to marry hypergamously on age. However, marrying older men is the norm across all

educational groups.

Finally, there are no significant differences across marital outcomes for either women with or women without co-residential children. Therefore, it appears that having

children have little or no affect on age assortative mating for women.

Regression results. The results of the discrete-time multinomial regression of age

assortative mating outcomes can be found in Table 8. This table only reports the results

for key independent variables; the full models can be found in Appendix B. For men, as

reported in the upper panel of Table 8, the results show that additional time in the

remarriage market increases, at a decreasing rate, the likelihood for two of the three

outcomes compared to not remarrying—age homogamy and marrying an older wife.

There are statistically significant differences across the outcomes as well. Compared to

homogamy, the likelihood of marrying a younger woman decreases for men over time.

An additional year in the remarriage market decreases the odds of marrying a younger

woman compared to a similarly-aged woman by 18.6%. Importantly, there is no

curvilinearity in this effect, as noted by the insignificant quadratic term. Marrying an

older woman also increases at a decreasing rate, compared to marrying a younger

woman. These patterns show that while men are more likely to form marriages over

105

time, they are also more likely to form non-traditional matches with additional time in the

remarriage market.

There are also important differences in assortative mating patterns by age for

men. Older men, above the mean age of 35.17, are much less likely to marry

homogamously or marry older women than they are to remain unmarried. An additional

year above the mean age decreases the odds of homogamy by 16.1% and decreases the

odds of marrying an older woman by 11.9%, compared to not remarrying. There are also

statistically significant differences across outcomes. Older men are more likely to marry

a younger woman than marry a similarly-aged woman. A one year increase in age above

the mean is associated with a 22.0% increase in the odds of age hypergamy compared to

age homogamy. Furthermore, older men are unlikely to marry older women. There is no

statistically significant difference between marrying an older woman and marrying

homogamously. Together, these results illustrate that older men are more likely to follow traditional marriage patterns, while younger men are more likely to marry in a manner

similar to first marriages.

Although men’s current socioeconomic status, as measured by the log of income

has essentially no effect on assortative mating by age, there are significant effects by

long-term socioeconomic status. Educational attainment is linked to marital sorting by

age. The results show consistently negative effects for men with a less than high school

education across all three outcomes, when compared to not remarying, indicating that

they are less likely to marry than men who completed high school. College-educated

men, however, are more likely than high school graduates to marry homogamously and to

106

marry a younger woman. Male college graduates have 42.9% higher odds of marrying a similarly-aged spouse compared to not remarrying than a man with at least a high school education. Male college graduates have 52.0% higher odds of marrying a younger woman compared to not remarrying than a male high school graduate.

Women’s results are reported in the lower panel of Table 8. The results are largely similar to men’s with three notable exceptions. First, the effects of duration on age assortative mating are smaller for women than men. Unlike men, time in the remarriage market also has a statistically significant effect on the odds of marrying an older man compared to not remarrying. The odds of marrying hypergamously on age increase at a decreasing rate. However, the rate of increase is less steep for this outcome than it is for both age homogamy and age hypogamy. Second, older women are no more likely to marry older men than they are to marry homogamously. Third, educational attainment has no statistically significant effects on assortative mating by age for women.

Children do not have any significant impact on age assortative mating for women.

The only statistically significant effect of co-residential on marital sorting is for men.

Men with children are more likely to marry younger women, compared to not remarrying

(35.3% higher odds than men without children) and homogamy (50.7% higher odds than men without children). Thus, the hypothesized relationship between children and remarriage for women is not supported.

Education-age interactions. The effect of educational attainment on age sorting may differ by the respondent's age. Specifically, younger men may have a preference for marrying similarly-aged women, while older men have a preference for marrying

107

younger women. Importantly, men with high socioeconomic status are the most capable of marrying according to their preferences because they are generally unrestricted in the remarriage market. As a result, I hypothesized that younger men would be more likely to marry a similarly-aged spouse and older men would marry a younger woman. The results of the interactive model are presented in Table 9. This table takes on an abbreviated form and the full results are reported in Appendix C.

The results for men, reported in the upper panel of Table 9, are substantively similar to those in the main effects model. The fit of the interactive model improves from the main effects model, suggesting that the effect of educational attainment on marital sorting varies by age. The interpreted effect for educational attainment is the odds of a given marital outcome at the mean age in the sample. At the mean age, men with a less than high school education have 38.6% lower odds than high school graduates of marrying a younger woman compared to not remarrying. Less-educated men have the same exact lower odds of marrying an older woman compared to not remarrying. College graduates are much more likely to marry younger women at the mean age. Compared to high school graduates, male college graduates have 44.3% higher odds of marrying a younger woman versus not remarrying.

The interaction between age and education shows that older men are more likely to remain unmarried than marry homogamously, as noted by the negative effect for the college graduate-age interaction. Importantly, the comparison between marrying a younger woman and a similarly-aged woman is positive for college graduates. For every year older above the mean, male college graduates have 14.1% greater odds than high

108

school graduates of marrying a younger woman compared to marrying a similarly-aged

woman. Therefore, marrying a younger woman is more likely for highly-educated, older

men, while homogamy is more likely for highly-educated, young men.

For women, the lower panel of Table 9 shows no statistically significant effects across marital sorting outcomes for educational attainment at the mean age of divorce.

However, there is one statistically significant interaction effect. Older college-educated women are more likely to remain unmarried than to marry a younger man. Female college graduates, compared to high school graduates, experience 16.1% lower odds of marrying a younger man compared to not remarrying.

Summary

In sum, the results in regard to the hypotheses are as follows: (1) the relationship between the likelihood of marrying a less-educated woman and income is positive for men; (2) older highly-educated men are more likely to marry women younger than themselves, while younger college-educated men tend to form age homogamous marriages; (3) unlike men, neither current or long-term socioeconomic prospects effects assortative mating by education and age among women; (4) younger women are more likely to marry hypergamously on education and age than older women, while the

likelihood of hypogamy and non-marriage increases with age for women; (5) having a

co-residential child or children from a prior relationship does not affect assortative

mating for women; (6) additional time in the remarriage market increases the likelihood

of marrying an older woman for men, but has no effect on search strategies in educational

assortative mating; (7) additional time in the remarriage market initially increases the

109

likelihood of all assortative mating outcomes for women, but this is accompanied by a

curvilinear decline in the odds of all outcomes, meaning that as women spend more time

in the remarriage market, the odds of remarriage decline.

Discussion

Assortative mating patterns in remarriage differ significantly from those present

in first marriage. While both educational and age homogamy are likely in first marriage,

homogamy is unlikely in remarriage. Instead, heterogamous marriages are normative in remarriage. However, there are important gender differences in assortative mating.

Remarriage tends to follow a traditional, hypergamous pattern for high-status men, who

are more likely to marry less-educated and younger women than they are to marry

homogamously. Lower status men, on the other hand, have a tendency make less traditional matches, including marrying better educated and older women. For women,

there appears to be very little variation in marital sorting across most individual characteristics, except for age. Generally, older women are more likely to marry less-

educated men, but age has no effect on women’s age assortative mating. Below, I discuss

these findings in greater detail.

For men, high social status is associated with more traditional remarriage patterns.

Income has a positive effect on the likelihood of marrying a less-educated woman. The

results suggest that highly-educated, high earning men are the most likely to follow a

more traditional marriage pattern. Highly-educated men are limited to either homogamous or hypergamous marriages. Because highly-educated men cannot marry

up, the results show that educational hypergamy is prevalent among this group. This

110

pattern represents a significant departure from contemporary assortative mating patterns in first marriage where highly-educated men favor educational homogamy (Schwartz and

Mare 2005). In remarriage, men have high economic well-being and are financially established, which means that the benefit to marrying a highly-educated woman is lower than it is in first marriage. This finding suggests that higher levels of educational homogamy in first marriage are more the result of economic need than a fundamental shift in men’s partner preferences. However, these conclusions should be tempered by the fact that a large number of less-educated men do not remarry in the sample.

Men’s age assortative mating patterns are similar to educational patterns.

Highly-educated men have a higher likelihood of marrying a younger woman than less- educated men. The difference is large. College graduates have 50% higher odds of age hypergamy than high school graduates. Results from the main effects model also show that college-educated men are likely to marry homogamously on age. An interaction effect between age and educational attainment helps us understand these effects. Young, highly-educated men have a high likelihood of marrying homogamously. This difference may be caused by young women’s appeal in the marriage market (England and Mclintock

2008)--younger divorced men are about the same age as highly attractive young women.

Furthermore, younger divorced men are constrained by the minimum age of marriage

(typically age 18), suggesting that it is very difficult for them to marry down on age. In contrast, older college-educated men marry younger women, meaning this group is most likely to follow traditional age assortative mating patterns.

111

In contrast to men, women’s assortative mating patterns are not influenced by socioeconomic characteristics. Again, these results support the conclusion that remarriage is more traditional than first marriage. Both labor force status and educational attainment are statistically insignificant predictors of educational and age assortative mating outcomes. These results stand in stark contrast to first marriage patterns, where high economic status leads to educational and age homogamy for women. .

There are several potential explanations for women’s assortative mating patterns in remarriage. First, women tend to have poor economic well-being after divorce

(Peterson 1996). Many women may lack the sufficient financial resources for a long partner search (Oppenheimer 1988) and may marry at the first opportunity. As a result, some women may not be “choosy,” causing the lack of a pattern in their marital sorting.

Second, divorced women may be less desirable marriage partners in general, constraining their choices (Ono 2005). Third, women’s assortative mating patterns are connected to men’s circumstances. Men’s economic well-being after divorce may make the economic contributions of women obsolete for remarriage formation. Thus, highly-educated, working, divorced women are unlikely to form second marriages because men prefer to make a match with a woman who is dedicated to domestic labor and childrearing.

Finally, divorced women may match on another characteristic not considered in these analyses. For example, using the National Study of Families and Households,

Goldschreider and Sassler (2006) show that women with children from previous relationships are much more likely to marry men with children than men without children. Considering about 60% of divorced women in the NLSY79 data report having

112

a child, including 62% of women who remarry, women may be sorting more on parental

status than other characteristics. Data limitations prevent me from exploring this

relationship. However, this pattern may be an important and could help explain why

children had no effect on assortative mating in this analysis.

Lending support to the conclusion that remarriage is traditional, women’s age has

an important effect on educational assortative mating (Becker 1981). Older women are more likely to marry less-educated men than they are to marry homogamously or hypergamously. However, in terms of age matching, older women have an increased likelihood of marrying a man who is older than themselves.

Closely related to age is time spent searching for a spouse after a divorce. The

results show that time in the remarriage market does not favor one assortative mating

outcome over another. Generally, the curvilinear effect of time for women indicates a

decline in the odds of remarriage over time. The decline in the likelihood of remarrying

over time is less sharp for women than it is men, a pattern which reflects women’s longer

duration to remarriage. However, for men, a decline in the odds of marrying highly

sought after spouses, such as younger women, is accompanied by an increase in the

likelihood of other outcomes. As a result, women are more likely to remain unmarried

after divorce than enter a less traditional or less desirable marriage, as men are.

In conclusion, prior research on assortative mating in remarriage has been

lacking. This chapter addresses this gap in the literature by focusing on the quality of

matches made in remarriage, how marital sorting on education and age in remarriage

differs from assortative mating in first marriage, and the gender gap in remarriage

113

formation. The findings here indicate that, in many ways, remarriage is more traditional than first marriage. High-status men tend to be older and better educated than their spouses, while women are evaluated not on socioeconomic status, but more traditional characteristics such as age. In total, the results of this chapter indicate that assortative mating in remarriage is strongly affected by both preferences and constraints—leading to important differences between men and women in whom they marry.

114

Male Respondents Respondents Spouses Difference Mean years of education 13.7 13.2 0.5*

Mean age at remarriage 33.9 32.1 1.8*

Mean income (2006 adjusted) $35,507 $15,102 $20,402*

Percent with co-residential child 0.18 0.24 -0.06*

Female Respondents Respondents Spouses Difference Mean years of education 12.9 13.2 -0.3*

Mean age at remarriage 32.7 35.4 -2.7*

Mean income (2006 adjusted) $18,986 $25,032 -$6,046*

Percent with co-residential child 0.56 0.11 0.45* Source: National Longitudinal Survey of Youth, 1979 cohort Note: * Indicates a statistically significant difference between the respondent and spouse at the p<.05 level

Table 3: Differences Between Respondent and Spouse on Key Characteristics, by Sex

115

Men

Less than high High school College school graduate graduate or more Same education & age 6.6% 16.5% 20.0% Same education & husband older 9.2% 15.6% 17.7% Same education & wife older 7.9% 15.6% 14.1%

Husband more educated & same age --- 6.3% 14.1% Husband more educated & older --- 10.0% 22.4% Husband more educated & wife older --- 6.0% 11.8%

Wife more educated & same age 27.6% 10.9% --- Wife more educated & husband older 18.4% 6.5% --- Wife more educated & older 30.3% 12.5% ---

Women

Less than high High school College

school graduate graduate or more Same education & age 8.3% 15.8% 16.0% Same education & husband older 10.0% 16.0% 21.3% Same education & wife older 10.0% 12.7% 9.6%

Husband more educated & same age 28.3% 11.2% --- Husband more educated & older 25.0% 10.2% --- Husband more educated & wife older 18.3% 4.5% ---

Wife more educated & same age --- 10.2% 14.9% Wife more educated & husband older --- 9.8% 21.3% Wife more educated & older --- 9.4% 17.0% Source: National Longitudinal Survey of Youth, 1979 cohort Note: numbers may not add to or may exceed 100% due to rounding error

Table 4: Assortative Mating on Education and Age Among Remarried Men and Women, by Educational Attainment

116 Men Husband more Wife more Homogamy educated educated Not remarried Row n (n= 252) (n= 211) (n= 194) (n= 522) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Time Duration since divorce 1,179 5.19 3.25 4.70 3.59 5.65 3.81 11.23 5.76 Age at divorce 1,179 28.63 5.72 28.71 5.55 28.07 5.27 33.15 5.75

Current socioeconomic status Log of income, lagged 1,179 9.29 2.74 9.37 2.42 9.53 2.09 9.10 3.38

Long-term socioeconomic statusa Less than high school graduate 172 0.12 --- 0.00 --- 0.17 --- 0.45 --- 117 High school graduate 850 0.26 --- 0.15 --- 0.16 --- 0.42 --- College graduate or more 157 0.29 --- 0.30 --- 0.00 --- 0.41 ---

a First marriage ties Co-residential child present 221 0.24 --- 0.21 --- 0.12 --- 0.42 --- No co-residential child present 958 0.25 --- 0.15 --- 0.18 --- 0.43 ---

Table 5: Descriptive Statistics of Key Independent Variables by Educational Assortative Mating Outcomes for Men and

Women

Table 5 continued

Women Husband more Wife more Homogamy educated educated Not remarried Row n (n= 277) (n= 170) (n= 239) (n= 606) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Time Duration since divorce 1,292 5.86 3.77 6.09 3.99 5.18 3.38 11.91 5.88 Age at divorce 1,292 26.66 5.46 25.25 5.16 27.67 5.62 32.33 5.93

Current socioeconomic status Log of income, lagged 1,292 0.78 --- 0.74 --- 0.77 --- 0.85 ---

118 Long-term socioeconomic statusa Less than high school graduate 111 0.16 --- 0.47 --- 0.00 --- 0.36 --- High school graduate 964 0.23 --- 0.14 --- 0.19 --- 0.44 --- College graduate or more 217 0.21 --- 0.00 --- 0.27 --- 0.53 ---

First marriage tiesa Co-residential child present 766 0.21 --- 0.13 --- 0.12 --- 0.48 --- No co-residential child present 526 0.24 --- 0.15 --- 0.18 --- 0.41 --- Source: National Longitudinal Survey of Youth, 1979 cohort a Note: Percentages are based on row total, not column total for this set of characteristics

Men Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. not educated vs. not educated vs. educated vs. husband more not remarried remarried remarried homogamy homogamy educated Time Duration since divorce 1.336 *** 1.406 ** 1.363 *** 1.052 1.019 0.969 Duration since divorce, squared 0.985 *** 0.977 ** 0.981 ** 0.991 0.996 1.004 Age, centered 0.901 *** 0.895 ** 0.904 *** 0.994 1.004 1.004

Socioeconomic status Log of income 1.038 ** 1.243 * 1.051 * 1.198 * 1.012 0.845 *

First marriage ties Co-residential child/children from divorce 1.052 0.927 0.861 0.881 0.819 0.929 (ref.: no co-residential child)

119 N-individuals 1,179 N-person years 8,706 -2 Log likelihood 252.61 Psuedo R-squared 0.048

Table 6: Discrete-Time Multinomial Logistic Regression Results for Educational Assortative Mating Outcomes, by Sex

Table 6 continued

Women Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. not educated vs. not educated vs. educated vs. husband more not remarried remarried remarried homogamy homogamy educated Time Duration since divorce 1.367 *** 1.587 *** 1.423 ** 1.161 1.041 0.897 Duration since divorce, squared 0.985 *** 0.982 *** 0.976 *** 0.997 0.991 0.994 Age, centered 0.891 *** 0.825 *** 0.984 0.925 * 1.103 ** 1.192 ***

Socioeconomic status Full-time, year-round employed 0.894 0.863 0.896 0.965 1.003 1.039 (ref.: not full-time, year-round employed) 120 First marriage ties Co-residential child/children from divorce 0.852 0.916 0.678 * 1.076 0.797 0.741 (ref.: no co-residential child) N-individuals 1,245 N-person years 10,438 -2 Log likelihood 338.35 Psuedo R-squared 0.058 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

Men Homogamy Husband older Wife older Not remarried Row n (n= 208) (n= 241) (n= 208) (n= 522) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Time Duration since divorce 1,179 5.18 3.73 4.87 3.43 5.57 3.41 11.23 5.76 Age at divorce, centered 1,179 32.38 5.64 34.78 5.65 34.31 5.37 43.87 3.53

Socioeconomic status Log of income, lagged 1,179 9.10 2.22 9.34 2.59 9.32 2.59 9.10 3.38

Long-term socioeconomic statusa Less than high school graduate 172 0.16 --- 0.21 --- 0.19 --- 0.45 --- High school graduate 850 0.19 --- 0.20 --- 0.18 --- 0.42 --- 121 College graduate or more 157 0.19 --- 0.26 --- 0.14 --- 0.41 ---

Parental statusa Co-residential child present 221 0.18 --- 0.27 --- 0.13 --- 0.42 --- No co-residential child present 958 0.18 --- 0.20 --- 0.19 --- 0.43 ---

Table 7: Descriptive Statistics of Key Independent Variables by Age Assortative Mating Outcomes for Men and Women

Table 7 continued

Women Homogamy Husband older Wife older Not remarried Row n (n= 231) (n= 283) (n= 172) (n= 606) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Time Duration since divorce 1,292 5.59 3.68 5.54 3.63 6.03 3.88 11.91 5.88 Age at divorce, centered 1,292 31.29 5.57 33.89 6.63 32.81 5.17 43.86 3.35

Socioeconomic status Log of income, lagged 1,292 0.78 --- 0.74 --- 0.77 --- 0.85 ---

Long-term socioeconomic statusa 122 Less than high school graduate 111 0.21 --- 0.26 --- 0.17 --- 0.36 --- High school graduate 964 0.17 --- 0.21 --- 0.18 --- 0.44 --- College graduate or more 217 0.14 --- 0.22 --- 0.12 --- 0.52 ---

Parental statusa Co-residential child present 766 0.17 --- 0.21 --- 0.14 --- 0.48 --- No co-residential child present 526 0.20 --- 0.25 --- 0.14 --- 0.41 --- Source: National Longitudinal Survey of Youth, 1979 cohort a Note: Percentages are based on row total, not column total for this set of characteristics

Men Homogamy vs. Husband older Wife older vs. not vs. not not Husband older Wife older vs. Wife older vs. remarried remarried remarried vs. homogamy homogamy husband older Time Duration since divorce 1.305 *** 1.062 1.524 *** 0.814 * 1.168 1.435 *** Duration since divorce-squared 0.989 ** 0.989 ** 0.978 *** 0.999 0.988 * 0.989 * Age, centered 0.839 *** 1.024 0.881 *** 1.220 *** 1.050 0.860 ***

Current socioeconomic status Log of income, lagged 1.084 * 1.008 1.034 0.929 0.953 1.026

Long-term socioeconomic status (ref.: high school graduate) Less than high school 0.612 * 0.639 * 0.662 * 1.045 1.081 1.035 College degree or more 1.429 * 1.520 * 1.114 1.063 0.779 0.733

123 First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.897 1.353 * 1.049 1.507 * 1.169 0.775 N-individuals 1,179 N-person years 8,780 -2 Log likelihood 293.05 Psuedo R-squared 0.048

Table 8: Discrete Time Multinomial Logistic Regression Results for Age Assortative Mating Outcomes, by Sex—Main

Effects

Table 8 continued

Women Homogamy vs. Heterogamy vs. Wife older vs. not not not Heterogamy vs. Wife older vs. Wife older vs. remarried remarried remarried Homogamy Homogamy husband older Time Duration since divorce 1.404 *** 1.174 ** 1.369 *** 0.836 0.975 1.166 * Duration since divorce-squared 0.983 *** 0.985 *** 0.984 *** 1.001 1.000 0.999 Age at divorce, centered 0.873 *** 1.037 * 0.922 ** 1.188 1.056 0.889 ***

Current socioeconomic status Full-time, year-round employed (ref.: not full-time, year-round employed) 0.920 1.085 0.687 * 1.179 0.747 0.634 *

Long-term socioeconomic status (ref.: high school graduate) 124 Less than high school 0.774 1.202 0.820 1.552 1.058 0.682 College degree or more 1.001 0.913 1.203 0.912 1.202 1.317

First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.798 0.967 1.003 1.211 1.257 1.038 N-individuals 1,292 N-person years 10,498 -2 Log likelihood 338.35 Psuedo R-squared 0.050 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

Homogamy vs. Husband older Wife older vs. not vs. not Mennot Husband older Wife older vs. Wife older vs. remarried remarried remarried vs. Homogamy Homogamy husband older Time Duration since divorce 1.326 *** 1.058 1.546 *** 0.798 ** 1.165 1.461 *** Duration since divorce-squared 0.989 ** 0.989 ** 0.977 *** 1.001 0.989 * 0.988 * Age, centered 0.841 *** 1.022 0.889 *** 1.215 *** 1.057 0.869 ***

Current socioeconomic status Log of income, lagged 1.086 * 1.007 1.036 0.927 0.953 1.028

Long-term socioeconomic status (ref.: high school graduate) Less than high school 0.703 0.632 * 0.632 * 0.899 0.899 0.999 College degree or more 1.308 1.443 * 1.197 1.103 0.915 0.829 125 Interactions Less than high school * age, centered 1.033 0.996 0.977 0.964 0.946 0.981

College degree or more * age, centered 0.901 ** 1.028 0.894 ** 1.141 ** 0.992 0.830 *

First marriage ties

Co-residential child/children from divorce (ref.: no co-residential child) 0.907 1.345 * 1.069 1.482 * 1.177 0.795 N-individuals 1,179 N-person years 8,780 -2 Log likelihood 307.69 TablePsuedo R-squared9: Discrete-Time Multinomial Logistic 0.047 Regression Results for Age Assortative Mating Outcomes, by Sex—

Interactive Model

Table 9 continued

Table 9 continued Women Homogamy vs. Husband older Wife older vs. not vs. not not Husband older Wife older vs. Wife older vs. remarried remarried remarried vs. Homogamy Homogamy husband older Time Duration since divorce 1.403 *** 1.176 ** 1.370 *** 0.838 * 0.976 1.164 * Duration since divorce-squared 0.983 *** 0.984 *** 0.984 *** 1.001 1.001 0.999 Age, centered 0.874 *** 1.046 ** 0.926 ** 1.197 *** 1.059 * 0.885 ***

Current socioeconomic status

Full-time, year-round employed (ref.: not full-time, year-round employed) 0.919 1.079 0.684 * 1.175 0.744 0.633 *

Long-term socioeconomic status 126 (ref.: high school graduate) Less than high school 0.803 1.089 0.932 1.336 1.161 0.856 College degree or more 0.983 0.971 1.157 0.937 1.177 1.256

Interactions Less than high school * age, centered 1.005 0.971 1.025 0.967 1.020 1.056

College degree or more * age, centered 0.982 0.958 0.939 * 0.976 0.956 1.256

First marriage ties

Co-residential child/children from divorce (ref.: no co-residential child) 0.803 0.974 1.041 1.213 1.297 1.069 N-individuals 1,292 N-person years 10,498 -2 Log likelihood 324.46 Psuedo R-squared 0.050 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

Chapter 6: Conclusion

Divorce and remarriage are an integral part of American family life. Remarriage

is a unique union type. Demographically, remarriage is more selective than first marriage—the proportion of divorcees who remarry is significantly lower than the percentage of people who ever marry (Teachman et al. 2000). Typically, men and women who remarry are older than people marrying for the first time (deGraaf and

Kalmijn 2003). This may influence their preferences, constraints, and choices when marrying for the second time. Divorcees also face a unique marriage market when remarrying. Generally, divorced men and women choose from a pool of other divorcees and older never-married individuals who tend to have lower socioeconomic status, children, and other select attributes (Teachman 2002; Gelissen 2004). Importantly, divorced men and women have prior marital experiences which can influence their repartnering behavior. Notably, divorce has strong and significant negative psychological, social, and economic consequences (Coleman et al. 2000), which can be overcome, in part, by remarrying (Cherlin 1992). These characteristics help make remarriage distinct from first marriage.

Despite remarriage’s uniqueness, some social commentators have argued that first marriage and remarriage have become similar to one another in recent decades (Cherlin

127

2004). Several trends, according to these researchers, highlight that both first marriage

and remarriage are deinstitutionalized family forms, including the move away from

traditional marriages with gendered expectations and norms (see Cherlin 2004, 849).

However, empirical work focusing on this assertion is lacking. Many studies have

focused on the factors associated with marriage formation and assortative mating patterns

to understand gender ideologies in marriage (e.g., Becker 1981; Atkinson and Glass

1985; Kalmijn 1991; Kalmijn 1998). Both types of studies show that first marriage has

become less gendered in recent decades (e.g., Lichter et al. 1992; Qian and Preston 1993;

Oppenheimer et al. 1995; Qian 1998; Rose 2004; Schwartz and Mare 2005). Women are

increasingly evaluated by their socioeconomic attributes when marrying (Sweeney 2002;

Rose 2004) and educational homogamy has become the norm for highly-educated

individuals (Schwartz and Mare 2005). Given these findings, the question becomes

whether or not similar trends are present in remarriage.

This project is motivated by the question of how similar or different first marriage

and remarriage are from one another with respect to marriage formation and assortative

mating. In other words, this study investigates if remarriage is more like traditional first

marriage with gendered expectations and norms or contemporary first marriage where women are evaluated on the same criteria as men when marrying. Given that men are much more likely than women to remarry, the gender gap in remarriage provides an opportunity to empirically investigate this question. I do this by focusing on gender differences in the formation of and assortative mating in remarriage.

128

First, I explore how individual characteristics affect the timing and likelihood of

remarriage. If remarriage is like traditional first marriage, then socioeconomic status

should positively affect men’s remarriage prospects, but not women’s. The likelihood of

remarriage for women should be influenced by age, parental status, and other characteristics (Becker 1981). However, if remarriage is more like contemporary first marriage, socioeconomic status should increase the likelihood of remarriage for both men and women (England and Farkas 1986; Oppenheimer 1988). Second, assortative mating patterns also highlight how traditional or contemporary remarriage is. Traditionally, marriages are hypergamous on education and age—with highly-educated, older men marrying less-educated, younger women. Contemporary first marriages, however, tend to be educationally and age homogamous (e.g., Qian 1998; Schwartz and Mare 2005;

Shafer and Qian forthcoming). In my analyses, I use data from the National Survey of

Youth, 1979 cohort (NLSY79) to undertake these two empirical questions.

In the analysis of gender differences in remarriage, I use marital search theory which hypothesizes that men and women choose spouses in the marriage market based on personal characteristics (preferences) and individual limitations (constraints). I specifically focused on socioeconomic status, first marriage ties, available partner pools, and time as important factors in both the formation of and assortative mating in remarriage. In terms of socioeconomic status, men tend to have higher economic well- being after divorce because they maintain their income and employment while decreasing the number of dependents, even if they make economic contributions to their ex-wife and any children they might have (Peterson 1996). Women, on the other hand, usually

129

experience a decline in economic well-being after divorce because they have lower incomes than men and often care for children (Holden and Smock 1991). Furthermore, educational attainment, as a measure of long-term economic status, can also influence remarriage. For men, education can have positive effects on remarriage because it helps determine social status, labor market success, and economic stability (Oppenheimer et al.

1997). Conversely, women's education may be unimportant in a market where men are economically established and thus, male divorcees can be socioeconomically attractive partners, while women tend not to be.

Women may find further difficulty in the remarriage market because they care for children. Caring for children from a previous marriage often means that women are required to maintain ties to their ex-husband because of financial transfers through child support and co-parenting (Ono 2005). Marrying a parent can also add complexity to new marriages. Increased levels of conflict and problematic behaviors in children are not uncommon for families with stepparents (Coleman et al. 2000). Thus, parents with co- residential children are not attractive potential spouses in the remarriage market.

In addition to preferences, men and women can also be constrained in the remarriage market. Based on personal attributes, individuals either have large or small potential partner pools available to them. Individuals with large partner pools, such as high socioeconomic status men, have many available partners and can often make a match based on their preferences. Other individuals, however, have fewer potential partners available to them and either make less preferential matches or remain unmarried.

Finally, time can have negative effects on remarriage. The number of potential partners

130

dwindles with time (both years since divorce and age). As a result, people either widen

their search for a spouse or remain unmarried.

As a whole, my results show that remarriage is more like traditional, rather than

contemporary, first marriage. In Chapter 4, I addressed how individual characteristics

predict entry into remarriage differently for men and women. The findings are consistent

with traditional marriage patterns. Both current and long-term socioeconomic status was positively related to remarriage for men, but not for women. These patterns suggest that men, like in first marriage, are evaluated as potential husbands based on their ability to provide for a family (England and Farkas 1986; Oppenheimer 1988). This is not surprising, given that this pattern holds true in both traditional and contemporary first

marriages. Women’s remarriage prospects, on the other hand, were strongly influenced

by social background characteristics such as mother’s education and family structure.

This result, coupled with the non-significant effect of socioeconomic status, indicates that

women are evaluated as potential partners on traditional criteria when marrying (Becker

1981).

In Chapter 5, I analyzed educational and age assortative mating patterns in remarriage. The results, again, provide evidence that remarriage is more traditional than first marriage. Men with high social status, measured by income and educational attainment, have the tendency to make traditional matches by marrying less-educated, younger women. Women’s assortative mating patterns, on the other hand, are not influenced by socioeconomic characteristics. Instead, assortative mating patterns for women are effected by age and social background characteristics. Like the results in

131

Chapter 4, these patterns reveal that women are evaluated on traditional criteria when

remarrying (see Rose 2004; Schwartz and Mare 2005).

An important context for this project is the gender gap in remarriage rates.

Although my analyses cannot account for men’s advantage in remarriage, they do show

potential reasons for this difference. Clearly, high-status men are attractive marriage

partners in both first marriage and remarriage. They also have very few constraints in the

remarriage market. High-status men have large partner pools available to them. Indeed,

the partner pools of high-status divorced men might be particularly large considering that older, divorced men are more economically established than younger, never-married men

(Sweeney 1997). As a result, these men may be able to marry never-married women

(Ono 2005), especially as younger men have more difficulty establishing a stable career, have low income growth, and have been hurt by macro-level economic change (Sweeney

2002).

Previously-married women have greater difficulty in the remarriage market than men. Few attributes, according to the results in Chapters 4 and 5, are considered attractive in the remarriage market. As noted above, neither current nor long term economic status is important for divorced women. Instead, age, parental status, and background characteristics are associated with remarriage. When considered in the context of men’s results, divorced women tend to be unattractive marriage partners. If men prefer less-educated, younger spouses, then women have difficulty in repartnering.

Women are also are more constrained than men in the remarriage market. There are fewer available men in the partner pool at older ages (deGraaf and Kalmijn 2003).

132

Previously-married women are also hindered by the stigma associated with divorce (Ono

2005). Finally, many women are constrained in the remarriage market because they care for children from previous relationships (Wu and Schimmele 2004; Goldschreider and

Sassler 2006).

Clearly, remarriage patterns are different from first marriage and tend to be more traditional in nature. There are several potential explanations for these differences. First, previously-married individuals may have different preferences in remarriage than in first marriage. Married men and women tend to have more traditional values regarding marriage than the never-married (Bartling and Broussard 1999). Thus, because divorcees have been married before, their spousal preferences may differ from those of never- married individuals. Second, individuals holding more traditional gender ideologies and values may be the most likely to form second marriages. If remarriage is selective on this attribute, it follows that marriage formation would be more traditional in remarriage than in first marriage. Indeed, Amato and Booth (1991) report that divorced men and women tend to have more liberalized attitudes toward union formation and dissolution than their

Never-married counterparts—suggesting that some previously-married individuals may resist forming a second marriage. Third, divorced men may not seek out high-status women because they have no need to do so. The onset of contemporary first marriage patterns, where women’s potential economic contributions to the family are valued, coincided with macro-level economic change which affected men more than women.

Specifically, since the 1970s men’s income growth has lagged behind women’s (Wetzel

1995), gender attitudes have become increasingly egalitarian (Thornton 1989), and dual-

133

income households became necessary in the face of changing consumption patterns, the

rising cost of raising children, and other family economic issues (Bumpass 1990; England

and Folbre 1999). Therefore, if women’s economic status is important to men primarily

because of financial uncertainty, it stands to reason that remarriage, which takes place at

later ages, would be more traditional as men become more financially established.

Implications

The results here provide one way to view gender dynamics in remarriage. To the extent that marriage formation and assortative mating patterns reveal gender attitudes,

behaviors, and ideologies within marriage, the analyses indicate that remarriage is like

traditional first marriage. Certainly, the results fit the specialization model which

suggests that these patterns are the result of expectations that men engage in the labor

force while women involve themselves in domestic tasks (Becker 1981). However, there are a number of other ways to measure gendered behavior in marriage, including marital power, decision making, time spent in domestic tasks, and differences in leisure time. To this point, research on gender dynamics in remarriage is limited and has mixed results

(Coleman et al. 2000). Some scholars show that remarriage is more traditional than first marriage, while others show no difference or more liberal attitudes in remarriage (see

Coleman et al. 2000 for a review). Thus, this study provides context for future studies on remarriage and differences between first marriage and remarriage.

Importantly, this project represents a foundation for work on remarriage formation. This project leads to a number of questions about the process of forming a second marriage. First, additional consideration should be paid to the effect of time on

134

the remarriage prospects of men and women. The results in both Chapters 4 and 5 show

that time in the remarriage market has a curvilinear effect on the likelihood of remarriage

and any gender differences in the effect of time are small. However, research in first

marriage has shown variability in the effect of time by educational attainment (Shafer and

Qian forthcoming), suggesting there may be important differences in the timing of

remarriage by social status. Second, my analyses do not account for cohabitation after

divorce. Yet, recent estimates have suggested up to half of all divorcees cohabit after

union dissolution (Xu et al. 2006). Research in first marriage suggests that for groups

with low marriage rates, cohabitation rates are very high (Smock 2000). Thus, the gender difference in remarriage may be offset by cohabitation among women. Furthermore,

research should focus on how individual characteristics are associated with either

remarriage or cohabitation following divorce. Third, as noted above, this project cannot

explain the gender gap in remarriage. However, future research should be dedicated to

explaining this effect, mirroring research on the Black-White marriage gap in first

marriage (e.g., Lichter 1991). Clearly, the remarriage gap has important implications for

both men and women. Identifying the factors associated with this gap should be a

priority in future analyses. Finally, the analysis of other group-level differences in

remarriage may help account for stratification, especially if people become increasingly

stratified later in the life-course. The inability for some individuals and groups to

remarry may help account for part of these differences.

Clearly, the results here support the idea that first marriage and remarriage are

distinct from one another. Unfortunately, prior research into marriage dynamics and

135

marriage formation has been almost exclusively centered on first marriage or prevailing marriages. While this literature has significantly contributed to our understanding of who gets married, who marries whom, and marriage, in general; it has largely ignored a union that nearly 1 in 5 American adults enters into. To date, most research on remarriage has centered on how it affects individuals—either the married couple or the children involved in a new marriage (Coleman et al. 2000). Certainly, this is an important aspect of remarriage that should not be ignored. However, demographers and sociologists should investigate the formation of these marriages. Prior research has shown the importance of the marriage formation process on many individual-level outcomes including mental health, physical health, economic outcomes, and parenting practices (e.g. Amato et al.

2003). Thus, studies of marriage formation, beyond their demographic and sociological significance, lay a foundation for other studies into second marriages.

Limitations

The results should be tempered by the limitations of my analyses. The study is limited by a sample which comes from a specific cohort of men and women born between 1957 and 1967. As a result, the findings presented here may not be indicative of remarriage trends in more recent birth cohorts or in the population, at-large. These concerns are amplified by the recent leveling off or even decline in divorce rates

(Goldstein 1999). If this trend has affected remarriage patterns in recent years is unclear.

I am also limited in my analyses by limited information on socioeconomic characteristics, background variables, and time-varying variables in the NLSY79 data. Similarly, I am limited by the relatively few variables which measure spousal characteristics. For

136

example, I cannot identify if new spouses have been previously-married or have children of their own. Although I speculate that one reason there are gender differences in remarriage is because men and women have different partner pools available to them, macro-level data on marriage markets is limited in the public-use NLSY data. Future research should address this limitation. Despite these limitations, the findings still suggest that remarriage is a traditional union-type, distinct from first marriage, and that there are significant gender differences in remarriage formation.

Discussion

This dissertation investigated gender differences in the formation of and assortative mating in remarriage. Using marital search theory as a context for understanding partner preferences and market limitations on remarriage, I focused on socioeconomic characteristics, first marriage ties, and time as potential sources of gender variation. My findings indicate that remarriage follows a traditional marriage pattern where socioeconomic status positively affects the likelihood of remarriage for men, but not women. Age, social background, and parental status are strong predictors of remarriage for women. Furthermore, the analysis of educational and age assortative mating patterns in remarriage reveals further support for the assertion that remarriage is traditional. High-status men are much more likely than low-status men to marry less- educated, young women. In comparison, there are few discernable assortative mating

137

patterns for women. Because divorce and remarriage continue to be an integral part of

American family life, the results suggest the need for additional research into gender dynamics in remarriage and an emphasis on remarriage as a distinct union from first marriage.

138

References

Aldous, Joan and Rodney F. Ganey. 1999. “Family Life and the Pursuit of Happiness” Journal of Family Issues. 20: 155-180

Allison, Paul D. 1982. “Discrete-Time Methods for the Analysis of Event Histories.” Pp. 61-98 in Sociological Methodology 1982, edited by K. Keinhardt. San Francisco: Jossey-Bass.

Allison, Paul D. 1984. Event History Analysis: Regression for Longitudinal Event Data. Sage Methodology Series.

Amato, Paul R. 2000. “The Consequences of Divorce for Adults and Children.” Journal of Marriage and the Family. 62: 1269-1287.

Amato, Paul R. and Alan Booth. 1991. “Consequences of Parental Divorce and Marital Unhappiness for Adult Well-Being.” Social Forces. 69: 895-914.

Amato, Paul R., David R. Johnson, Alan Booth, and Stacy J. Rogers. 2003. “Continuity and Change in Marital Quality Between 1980 and 2000.” Journal of Marriage and the Family. 65: 1-22.

Anderson, Kermyt G. 2000. “The Life Histories of American Stepfathers in Evolutionary Perspective.” Human Nature. 11: 307-333.

Arensberg, Conrad M. and Solon T. Kimball. 1968. Family and Community in Ireland, 2nd edition. Cambridge: Harvard University Press.

Ashenfelter, Orley and John Ham. 1979. “Education, Unemployment, and Earnings.” The Journal of Political Economy. 87: S99-S116.

Astone, Nan Marie and Sara S. McLanahan. 1991. “Family Structure, Parental Practices, and High School Completion.” American Sociological Review. 56: 309-320

139

Astone, Nan Marie and Mary L. Washington. 1994. “The Association Between Grandparental Co-residence and Adolescent Childbearing.” Journal of Family Issues. 15: 574-589.

Atkinson, Maxine P. and Becky L. Glass. 1985. “Marital Age Heterogamy and Homogamy: 1900 to 1980.” Journal of Marriage and the Family. 47: 685-691.

Barich, Rachel R. and Denise D. Bielby. 1996. “Rethinking Marriage: Change and Stability in Expectations, 1967-1994.” Journal of Family Issues. 17: 139-169.

Bartling, Carl A. and Beth Broussard. 1999. “Effects of Experimental Gender, Testing Context, and Marital Status Upon Marital Role Expectations. International Journal of Adolescence and Youth. 7: 317-326.

Becker, Gary S. 1981. A Treatise on the Family. Cambridge: Harvard University Press.

Becker, Gary S., Elisabeth M. Landes, and Robert T. Michael. 1977. “An Economic Analysis of Marital Instability.” The Journal of Political Economy. 85: 1141- 1187.

Bennett, Neil G., David E. Bloom, and Patricia H. Craig. 1989. “The Divergence of Black and White Marriage Patterns.” The American Journal of Sociology. 95: 692-722.

Bennett, Neil G., David E. Bloom, and Cynthia K. Miller. 1995. “The Influence of Nonmarital Childbearing on the Formation of First Marriages.” Demography. 32: 47-62.

Bernhardt, Eva and Frances Goldscheider. 2002. “Children and Union Formation in Sweeden.” European Sociological Review. 18: 289-299.

Blackwell, Debra L. 1998. “Marital Homogamy in the United States: The Influence of Individual and Paternal Education.” Social Science Research. 27:159-188.

Blackwell, Debra L. and Daniel T. Lichter. 2000. “Mate Selection Among Married and Cohabitating Couples.” Journal of Family Issues. 21: 275-302.

Blau, Peter M. 1977. Inequality and Heterogeneity. New York: Free Press.

Blau, Peter M., Terry C. Blum and Joseph E. Schwartz. 1982. “Heterogeneity and Intermarriage.” American Sociological Review. 47: 45-62.

140

Blossfeld, Hans-Peter and Andreas Timms, eds. 2003. Who Marries Whom? Educational Systems as Marriage Markets in Modern Societies. Dordrecht. The Netherlands: Kluwer Academic Publishers.

Bogenscheider, Karen. 1997. “Parental Involvement in Adolescent Schooling: A Proximal Process with Transcontexual Validity.” Journal of Marriage and the Family. 59: 718-733.

Brennan, Robert T., Rosalind Chait Barnett, and Karen C. Gareis. 2001. “When She Earns More than He Does: A Longitudinal Study of Dual-Earner Couples.” Journal of Marriage and the Family. 63: 168-182.

Budig, Michelle J. and Paula England. 2001. “The Wage Penalty for Motherhood.” American Sociological Review. 66: 204-225.

Bumpass, Larry L. 1990. “What’s Happening to the Family? Interactions Between Demographic and Institutional Change.” Demography. 27: 483-498.

Bumpass, Larry L., James Sweet, and Teresa Castro Martin. 1990. “Changing Patterns of Remarriage.” Journal of Marriage and the Family. 52: 747-756.

Bytheway, William R. 1981. “The Variation with Age of Age Differences in Marriage.” Journal of Marriage and the Family. 43: 923-927.

Centers for Disease Control and Prevention/National Center for Health Statistics. 2002. “Cohabitation, Marriage, Divorce, and Remarriage in the United States.” Vital and Health Statistics. Series 23: 22.

Chase-Lansdale, P. Lindsay, Andrew J. Cherlin, and Kathleen E. Kiernan. 1995. “The Long-Term Effects of Parental Divorce on the Mental Health of Young Adults: A Developmental Perspective.” Child Development. 66: 1614-1634.

Cherlin, Andrew J. 1978. “Remarriage as an Incomplete Institution.” The American Journal of Sociology. 84: 634-650.

______. 1992. Marriage, Divorce, and Remarriage. Cambridge: Harvard University Press.

Cherlin, Andrew J. and Frank F. Furstenberg. 1994. “Stepfamilies in the United States: A Reconsideration.” Annual Review of Sociology. 20: 359-381.

Chiswick, Carmel U. and Evelyn L. Lehrer. 1990. “On Marriage-Specific Human Capital.” Journal of Population Economics. 3: 193-213.

141

Clarkberg, Marin. 1999. “The Price of Partnering: The Role of Economic Well-Being in Young Adults' First Union Experiences.” Social Forces. 77: 945-968.

Clarkberg, Martin, Ross M. Stolzenberg, and Linda J. Waite. 1995. “Attitudes, Values, and Entrance into Cohabitation versus Marital Unions.” Social Forces. 74: 609- 634.

Coleman, Marilyn, Lawrence Ganong, and Mark Fine. 2000. “Reinvestigating Remarriage: Another Decade of Progress.” Journal of Marriage and the Family. 52: 1288-1307.

Cooksey, Elizabeth C. and Michelle M. Fondell. 1996. “Spending Time with His Kids: Effects of Family Structure on Fathers’ and Children’s Lives.” Journal of Marriage and the Family. 58: 693-707.

Crosbie-Burnett, Margaret and Jean Giles-Simes. 1991. “Marital Power in Stepfather Families: A Test of Normative-Resource Theory.” Journal of Family Psychology. 4: 484-496.

Davies, Lorraine, William R. Avison, and Donna D. McAlpine. 1997. “Significant Life Experiences and Depression among Single and Married Mothers.” Journal of Marriage and the Family.59: 294-308.

Davis, Kingsley. 1941. “Intermarriage in Caste Societies.” American Anthropologist. 43: 376-395.

Day, Randal D. 1992. “The Transition to First Intercourse Among Racially and Culturally Diverse Youth.” Journal of Marriage and the Family. 54: 749-762.

Dean, Gillian and Douglas T. Gurak. 1978. “Marital Homogamy the Second Time Around.” Journal of Marriage and the Family. 40: 559-570. deGraaf, Paul M. and Matthijs Kalmijn. 2003. “Alternative Routes in the Remarriage Market: Competing-Risk Analyses of Union Formation after Divorce.” Social Forces. 81: 1459-1498. deGraaf, Paul M. and Hedwig Vermeulen. 2007. “Female Labor-Market Participation in the Netherlands: Developments in the Relationship Between Family-Cycle and Employment.” Pp. 191-209 in Between Equalization and Marginalization: Women Working Part-Time in Europe and The United States of America, edited by Hans-Peter Blossfeld and Catherine Hakim. New York: Oxford University Press.

142

Diprete, Thomas A. and Claudia Buchmann. 2006. “Gender-Specific Trends in the Value of Education and the Emerging Gender Gap in College Completion.” Demography. 43: 1-24.

Duncan, Greg and Saul Hoffman. 1985. “A Reconsideration of the Economic Consequences of Divorce.” Demography. 22: 485-497.

Easterlin, Richard A. 1978. “What Will 1984 Be Like? Socioeconomic Implications of Recent Twists in Age Structure.” Demography. 15: 397-432.

______. 1987. Birth and Fortune. Chicago: University of Chicago Press.

Easterlin, Richard A. and Eileen M. Crimmins. 1991. “Private Materialism, Personal Self-Fulfillment, Family Life, and Public Interest: The Nature, Effects, and Causes of Recent Changes in the Values of American Youth.” Public Opinion Quarterly. 55: 499-533.

Edwards, John N. 1969. “Familial Behavior as Social Exchange.” Journal of Marriage and the Family. 31: 518-526.

England, Paula. 2004. “More Mercenary Mate Selection? Comment on Sweeney and Cancian (2004) and Press (2004).” Journal of Marriage and Family. 66: 1034- 1037.

England, Paula and George Farkas. 1986. Households, Employment and Gender. New York: Aldine.

England, Paula and Nancy Folbre. 1999. “The Cost of Caring.” The Annals of the American Academy of Political and Social Science. 561: 39-51.

England, Paula and Elizabeth McClintock. 2008. “The Gendered Double Standard of Aging in Marriage Markets.” Presented at the International Sociological Research Association RC28 Conference: Social Stratification and Mobility Meetings. Palo Alto, California: August 6-9, 2008.

Fernandez, Raquel and Richard Rogerson. 2001. “Sorting and Long-Run Inequality.” Quarterly Journal of Economics. 116: 1305-1341.

Fernandez, Raquel, Nezih Gunner and John Knowles. 2005. “Love and Money: A Theoretical and Empirical Analysis of Household Sorting and Inequality.” Quarterly Journal of Economics. 120: 273-344.

143

Fisher, Philip A., Beverly I. Fagot, and Craig S. Leve. 1998. “Assessment of Family Stress Across Low-, Medium-, and High-Risk Samples Using Family Events Checklist.” Family Relations. 47: 215-219.

Fossett, Mark A. and K. Jill Kiecolt. 1993. “Mate Availability and Family Structure Among African Americans in U.S. Metropolitan Areas.” Journal of Marriage and the Family. 55: 288-302.

Fu, Vincent Kang. 2001. “Racial Intermarriage Parings.” Demography. 38: 147-159.

Fu, Xuanning and Timothy B. Heaton. 2006. “Interracial Marriage, Family Type, and Consequential Family Socioeconomic Well-Being in Hawaii.” Presented at the Mulitethnic Families Conference, University of Hawaii: February 2006.

______. 2008. “Racial and Educational Homogamy: 1980 to 2000. Sociological Perspectives. 51: 735-758.

Gelissen, John. 2004. “Assortative Mating after Divorce: A Test of Two Competing Hypotheses Using Marginal Models.” Social Science Research. 33: 361-384.

Gerstel, Naomi. 1987. “Divorce and Stigma.” Social Problems. 34: 172-186.

______. 1988. “Divorce, Gender, and Social Integration.” Gender and Society. 2: 343- 367.

Glenn, Norval D. 2002. “A Plea for Greater Concern About the Quality of Marital Matching.” Pp. 45-58 in Revitalizing the Institution of Marriage for the Twenty- First Century, edited by Alan J. Hawkins, Lynn D. Wardle, and David Orgon Collidge. Santa Barbara: Praeger.

Goldin, Claudia and Lawrence F. Katz. 2000. “Decreasing (and then Increasing) Inequality in America: A Tale of Two Half-Centuries.” Pp. 37-82 in The Causes and Consequences of Increasing Inequality, edited by F. Welch. Chicago: University of Chicago Press.

Goldscheider, Frances and Sharon Sassler. 2006. “Creating Stepfamilies: Integrating Children into the Study of Union Formation.” Journal of Marriage and Family. 68: 275-291.

Goldscheider, Frances and Linda J. Waite. 1986. “Sex Differences in the Entry Into Marriage.” The American Journal of Sociology. 92: 91-109.

144

Goldstein, Joshua R. 1999. “The Leveling of Divorce in the United States.” Demography. 36: 409-414.

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: 506-519.

Gottschalk, Peter. 1997. “Inequality, Income Growth and Mobility: The Basic Facts.” Journal of Economic Perspectives. 11: 21-40.

Gullickson, Aaron. 2006. “Education and Black-White Intermarriage.” Demography. 43: 673-689.

Hanson, Thomas L., Sara S. McLanahan, and Elizabeth Thomson. 1996. “Double Jepoardy: Parental Conflict and Stepfamily Outcomes for Children.” Journal of Marriage and the Family. 58: 141-154.

Hao, Lingxin. 1996. “Family Structure, Private Transfers, and the Economic Well-Being of Families with Children.” Social Forces. 75: 269-292.

Hirschman, Charles and Judah Matras. 1971. “A New Look at the Marriage Market and Nuptiality Rates, 1915-1958.” Demography. 8: 549-469.

Hoffman, Saul D. 1977. “Marital Instability and the Economic Status of Women.” Demography. 14: 67-76.

Hoffmann, John P. and Robert A. Johnson. 1998. “A National Portrait of Family Structure and Adolescent Drug Use.” Journal of Marriage and the Family. 60: 633-645.

Holden, Karen C. and Pamela J. Smock. 1991. “The Economic Cost of Marital Dissolution: Why Do Women Bear a Disproportionate Cost?” Annual Review of Sociology. 17: 51-78.

Hosmer, David W. and Stanley Lemeshow. 1999. Applied Survival Analysis. New York: Wiley.

Heuveline, Patrick and Jeffrey M. Timberlake. 2004. “The Role of Cohabitation in Family Formation: The United States in Comparative Perspective.” Journal of Marriage and Family. 66: 1214-1230.

145

Huijts, Tim, Christiaan W.S. Monden, and Gerbert Kraaykamp. 2009. “Education, Educational Heterogamy, and Self-Assessed Health in Europe: A Multilevel Study of Spousal Effects in 29 European Countries.” European Sociological Review. 25: 1-16.

Jacobs, J.A. and F.F. Furstenberg. 1986. “Changing Places: Conjugal Careers and Women’s Marital Mobility.” Social Forces. 64: 714-732.

Jencks, Christopher. 1979. Who Gets Ahead. New York: Basic Books.

Kalmijn, Matthijs 1991. “Status Homogamy in the United States.” American Journal of Sociology. 97: 498-523.

______. 1993. “Trends in Black/White Intermarriage.” Social Forces. 72: 119-146.

______. 1994. “Assortative Mating by Cultural and Economic Occupational Status.” American Journal of Sociology. 100: 422-452.

______. 1998. “Intermarriage and Homogamy: Causes, Patterns, Trends.” Annual Review of Sociology. 24: 495-421.

Kalmijn, Matthijs and Henk Flapp. 2001. “Assortative Meeting and Mating: Unintended Consequences of Organized Settings for Partner Choices.” Social Forces. 79: 1289-1312.

Kalmijn, Matthijs, Van Groenou, and Marjolein Broese. 2005. “Differential Effects of Divorce on Social Integration.” Journal of Social and Personal Relationships. 22: 455-476.

Kitson, Gay C. and Leslie A. Morgan. 1990. “The Multiple Consequences of Divorce: A Decade Review.” Journal of Marriage and the Family. 52: 913-924. Klein, Thomas. 1990. “Remarriage after Divorce in Germany.” Kolner Zeitschrift fur Soziologie und Sozialpsychologie. 42: 60-80.

Koo, Helen and C.M. Suchindran. 1980. “Effects of Children on Women’s Remarriage Prospects.” Journal of Family Issues. 1: 497-515.

Koo, Helen, C.M. Suchindran, and Janet D. Griffith. 1984. “The Effects of Children on Divorce and Remarriage: A Mulitivariate Analysis of Life Table Probabilities.” Population Studies. 38: 451-471.

Kremer, Michael. 1997. “How Much Does Sorting Increase Inequality?” Quarterly Journal of Economics. 112: 115-139.

146

Lampard, Richard and Kay Peggs. 1999. “Repartnering: The Relevance of Parenthood and Gender to Cohabitation and Remarriage among the Formerly Married.” British Journal of Sociology. 50: 443-465.

Landale, Nancy S. and Stewart E. Tolnay. 1991. “Group Differences in Economic Opportunity and the Timing of Marriage: Blacks and Whites in the Rural South, 1910.” American Sociological Review. 56: 33-45.

Lee, Valerie E., David T. Burkham, Herbert Zimiles, and Barbara Ladewski. 1994. “Family Structure and Its Effects on Behavioral and Emotional Problems in Young Adolescents.” Journal of Research on Adolescence. 4: 405-437.

Lewis, Susan K. and Valerie K Oppenheimer. 2000. “Educational Assortative Mating Across Marriage Markets: Non-Hispanic Whites in the United States.” Demography. 37: 29-40.

Lichter, Daniel T. 1990. “Delayed Marriage, Marital Homogamy and the Mate Selection Process Among White Women.” Social Science Quarterly. 71: 802-811.

Lichter, Daniel T., Felicia B. LeClere and Diane K. McLaughlin. 1991. “Local Marriage Markets and the Marital Behavior of African American and White Women.” American Journal of Sociology. 96: 843-867.

Lichter, Daniel T., Diane K. McLaughlin, George Kephart and David J. Landry. 1992. “Race and the Retreat From Marriage: A Shortage of Marriageable Men?” American Sociological Review. 57: 781-799.

Lichter, Daniel T., Robert N. Anderson and Mark D. Hayward. 1995. “Marriage Markets and Marital Choice.” Journal of Family Issues. 16: 412-431.

Lichter, Daniel T. and Rukamalie Jayakody. 2002. “Welfare Reform: How Do We Measure Success?” Annual Review of Sociology. 28: 117-141.

Lillard, Lee A., Michael J. Brien, and Linda J. Waite. 1995. “Premarital Cohabitation and Subsequent Marital Dissolution: A Matter of Self-Selection?” Demography. 32: 437-457.

Lundberg, Shelly and Elaina Rose. 2003. “Child Gender and the Transition to Marriage.” Demography. 40: 333-349.

Manning, Wendy D., Susan D. Stewart, and Pamela J. Smock. 2003. “The Complexity of Fathers Parenting Responsibilities and Involvement with Nonresident Children.” Journal of Family Issues. 24: 645-667.

147

Mare, Robert D. 1991. “Five Decades of Educational Assortative Mating.” American Sociological Review. 56:15-32.

Marks, Nadine F. 1996. “Flying Solo at Midlife: Gender, Marital Status, and Psychological Well-Being.” Journal of Marriage and the Family. 58: 917-932.

McLanahan, Sara and Larry Bumpass. 1988. “Intergenerational Consequences of Family Disruption.” The American Journal of Sociology. 94: 130-152.

McLanahan Sara and Lynne Casper. 1995. “Growing Diversity and Inequality in the American Family.” Pp. 1-45 in State of the Union: America in the 1990s, edited by Reynolds Farley. New York: Russell Sage Foundation.

Merton, Robert K. 1941. “Intermarriage and the Social Structure.” Psychiatry. 4: 361- 374.

Milardo, Robert M. 1987. “Changes in Social Networks of Women and Men Following Divorce.” Journal of Family Issues. 8: 78-96.

Mitterauer, Michael. 1992. A History of Youth. Oxford: Blackwell.

Modell, John. 1989. Into One’s Own: From Youth to Adulthood in the United States 1920-1975. Berkeley: University of California Press.

Mott, Frank L. and Sylvia F. Moore. 1983. “The Tempo of Remarriage among Young American Women.” Journal of Marriage and the Family. 41: 427-436.

Mueller, Charles W. and Hallowell Pope. 1980. “Divorce and Female Remarriage Mobility: Data on Marriage Matches After Divorce for White Women.” Social Forces. 58: 726-738.

Murphy, Mike, Karen Glaser, and Emily Grundy. 1997. “Marital Status and Long-Term Illness in Great Britain.” Journal of Marriage and the Family. 59: 156-164.

Neff, James A. and Tracy Schluter. 1993. “Marital Status and Depressive Symptoms: The Race/Ethnicity and Sex.” Journal of Divorce and Remarriage. 20: 137-160.

Ni Brolchain, Maire. 1988. “Changing Partners: A Longitudinal Study of Remarriage.” Population Trends. 54: 27-34.

Nock, Steven L. 1998. “The Consequences of Premarital Fatherhood.” American Sociological Review. 63: 250-263.

148

Ono, Hiromi. 2005. “Marital History Homogamy Between the Divorced and the Never Married Among Non-Hispanic Whites.” Social Science Research. 34: 333-356.

______. 2006. “Homogamy Among the Divorced and the Never-married on Marital History in Recent Decades: Evidence from Vital Statistics Data.” Social Science Research. 35: 356-383.

Oppenheimer, Valerie K. 1988. “A Theory of Marriage Timing.” American Journal of Sociology. 94: 563-591.

______. 1997. “Women's Employment and the Gain to Marriage: the Specialization and Trading Model.” Annual Review of Sociology 23:431-53.

______. 2003. “Cohabiting and Marriage During Young Men’s Career-Development Process.” Demography. 40: 127-149.

Oppenheimer, Valerie K. and Vivian Lew. 1995. “How Important Was Women’s Economic Indepdendence?” Pp. 105-138 in Gender and Family Change in Industrialized Countries, edited by Karen Oppenheim Mason and An-Margritt Jensen. Oxford: Oxford University Press.

Oppenheimer, Valerie K., Matthijs Kalmijn and Nelson Lim. 1997. “Men’s Career Development and Marriage Timing During a Period of Rising Inequality.” Demography.34: 311-330.

Ozawa Martha N. and Hong-Sik Yoon. 2002. “The Economic Benefits of Remarriage: Gender and Income Class.” Journal of Divorce and Remarriage. 36: 21-39.

Peterson, Richard R. 1996. “A Re-evaluation of the Economic Consequences of Divorce.” American Sociological Review. 61: 528-536.

Popenoe, David. 1993. “American Family Decline, 1960-1990: A Review and Appraisal.” Journal of Marriage and the Family. 55: 527-542.

Press, Julie E. 2004. “Cute Butts and Housework: A Gynocentric Theory of Assortative Mating.” Journal of Marriage and Family. 66: 1029-1033.

Pyke, Karen D. 1994. “Women’s Employment as a Gift or Burden?: Marital Power Across Marriage, Divorce, and Remarriage.” Gender and Society. 8: 73-91.

Pyke, Karen D. and Scott Coltrane. 1996. “Entitlement, Obligation, and Gratitude in Family Work.” Journal of Family Issues. 17: 60-82.

149

Qian, Zhenchao. 1997. “Breaking Racial Barriers: Variations in Interracial Marriage Between 1980 and 1990.” Demography. 34: 263-276.

______. 1998. “Changes in Assortative Mating: The Impact of Age and Education, 1970-1990.” Demography. 35: 279-292.

Qian, Zhenchao and Samuel H. Preston. 1993. “Changes in American Marriage, 1972 to 1987: Availability and Forces of Attraction by Age and Education.” American Sociological Review. 58: 482-495.

Qian, Zhenchao, Daniel T. Lichter, and Leanna Mellott. 2005. “Out-of-Wedlock Childbearing, Marital Prospects, and Mate Selection.” Social Forces. 84: 473- 491.

Raley, R. Kelly and Jenifer Bratter. 2004. “Not Even if You Were the Last Person on Earth!” Journal of Family Issues. 25: 167-181.

Richards, Marcus, R. Hardy and M. Wadsworth. 1997. “The Effects of Divorce and Separation on Mental Health in a National UK Birth Cohort.” Psychological Medicine. 27: 1121-1128.

Richman, Naomi. 1977. “Behaviour Problems in Pre-School Children: Family and Social Factors.” British Journal of Psychiatry. 131: 523-527.

Rockwell, Richard C. 1976. “Historical Trends and Variations in Educational Homogamy.” Journal of Marriage & Family. 38: 83-95.

Rodgers, Bryan. 1994. “Pathways Between Parental Divorce and Adult Depresion.” Journal of Child Psychology and Psychiatry. 35: 1289-1308.

Rodgers, Bryan, Chris Power, and Steven Hope. 1997. “Parental Divorce and Adult Psychological Distress: Evidence from a National Birth Cohort: A Research Note.” Journal of Child Psychology and Psychiatry. 38: 867-872.

Rose, Elaine. 2004. “Education and Hypergamy in Marriage Markets.” CSDE Working Paper No. 04-03. Center for Studies in Demography and Ecology, University of Washington.

Rosenfeld, Michael J. 2005. “A Critique of Exchange Theory in Mate Selection.” American Journal of Sociology. 110: 1284-1325.

150

Sayer, Liana C., Suzanne M. Bianchi, and John P. Robinson. 2004. “Are Parents Investing Less in Children? Trends in Mothers’ and Fathers’ Time with Children.” American Journal of Sociology. 110: 1-43.

Schoen, Robert. 1992. “First Unions and the Stability of First Marriages.” Journal of Marriage and the Family. 54: 281-284.

Schwartz, Christine R. and Robert M. Mare. 2005. “Trends in Educational Assortative Marriage From 1940 to 2003.” Demography. 42: 621-646.

Schwebel, Andrew I., Mark A. Fine, and Maureena A. Renner. “A Study of Perceptions of the Stepparent Role.” Journal of Family Issues. 12: 43-57.

Selten, Peter. 1993. “The Religious Formation of Youth. Catholic Youth Movements in the Netherlands from 1900 to 1941.” Paedagogica Historica 29: 165-187.

Shafer, Kevin and Zhenchao Qian. Forthcoming. “Marriage Timing and Educational Assortative Mating.” Journal of Comparative Family Studies.

Shapiro, Adam. 1996. “Explaining Psychological Distress in a Sample of Remarried and Divorced Persons: The Influence of Economic Distress.” Journal of Family Issues. 17: 186-203.

Simon, Robin W. and Kristen Marcussen. 1999. “Marital Transitions, Marital Beliefs, and Mental Health.” Journal of Health and Social Behavior. 40: 111-125.

Smits, Jeroen, Wout Utless, and Jan Lammers. 1998. “Educational Homogamy in 65 Countries: An Explanation of Differences in Openness Using Country-Level Explanatory Variables.” American Sociological Review. 63: 264-285.

Smock, Pamela J. 1990. “Remarriage Patterns of Black and White Women: Reassessing the Role of Educational Attainment.” Demography. 27: 467-473.

______. 1994. “Gender and the Short-Run Economic Consequences of Marital Disruption.” Social Forces. 73: 243-262.

______. 2000. “Cohabitation in the United States: An Appraisal of Research Themes, Findings, and Implications.” Annual Review of Sociology. 26: 1-20.

Smock, Pamela J., Wendy D. Manning, and Sanjiv Gupta. 1999. “The Effect of Marriage and Divorce on Women’s Economic Well-Being.” American Sociological Review. 64: 794-812.

151

South, Scott J. 1991. “Sociodemographic Differentials in Mate Selection Preferences.” Journal of Marriage and the Family. 53: 928-940.

______. 1993. “Racial and Ethnic Differences in the Desire to Marry.” Journal of Marriage and the Family. 55: 357-370.

South, Scott J. and Kim M. Lloyd. "Spousal Alternatives and Marital Dissolution." American Sociological Review. 60: 21-35.

Stewart, Susan D. 1999. “Nonresidential Mothers’ and Fathers’ Social Contact with Children.” Journal of Marriage and the Family. 61: 894-907.

______. 2001. “The Effect of Stepchildren on Childbearing Intentions and Births.” Demography. 39: 181-197.

______. 2002. “Contemporary American Stepparenthood: Integrating Cohabiting and Nonresidential Stepparents.” Population Research and Policy Review.20: 345-364.

Stewart, Susan D., Wendy D. Manning, and Pamela J. Smock. 2003. “Union Formation Among Men in the U.S.: Does Having Prior Children Matter?” Journal of Marriage and Family. 65: 90-104.

Stokes, Joseph P. and Judith S. Peyton. 1986. “Attitudinal Differences Between Full- Time Homemakers and Women Who Work Outside the Home.” Sex Roles. 15: 299-310.

Sweeney, Megan M. 1997. “Remarriage of Women and Men After Divorce: The Role of Socioeconomic Prospects.” Journal of Family Issues. 18: 479-502.

______. 2002. “Two Decades of Family Change: The Shifting Economic Foundations of Marriage.” American Sociological Review.67: 132-147.

Sweeney, Megan M. and Maria Cancian. 2004. “The Changing Importance of White Women’s Economic Prospects for Assortative Mating.” Journal of Marriage and Family.66: 1015-1028.

Sweet, James A. and Larry L. Bumpass. 1987. American Families and Households. New York: Russell Sage Foundation.

Teachman, Jay D. 2002. “Childhood Living Arrangements and the Intergenerational Transimssion of Divorce.” Journal of Marriage and Family. 64: 717-729.

152

Teachman, Jay D. and Alex Heckert. 1985. “The Impact of Age and Children on Remarriage: Further Evidence.” Journal of Family Issues. 6: 185-203.

Teachman, Jay D., Kathleen Paasch, and Karen Carver. 1996. “Social Capital and Dropping Out of School Early.” Journal of Marriage and the Family. 58: 773- 783.

Teachman, Jay D., Lucky M. Tedrow, and Kyle D. Crowder. 2000. “The Changing Demography of America’s Families.” Journal of Marriage and the Family. 62: 1234-1246.

Thomson, Elizabeth, Sara S. McLanahan, and Roberta Braun Curtin. 1992. “Family Structure, Gender, and Parental Socialization.” Journal of Marriage and the Family. 54: 368-378.

Thornton, Arland. 1977. “Children and Marital Stability.” Journal of Marriage and the Family. 39: 531-540.

______. 1989. “Changing Attitudes Toward Family Issues in the United States.” Journal of Marriage and the Family. 51: 873-893.

Tynes, Sheryl R. 1990. “Educational Heterogamy and Marital Satisfaction Between Spouses.” Social Science Research. 19: 153-174

Umberson, Debra and Christine L. Williams. 1993. “Divorced Fathers: Parental Role Strain and Psychological Distress.” Journal of Family Issues. 14: 378-400.

United States Census Bureau. 2007. Statistical Abstract of the United States, 2006. Department of the Treasury, Washington DC.

Van Poppel, Frans and Inez Joung. 2001. “Long-Term Trends in Marital Status Mortality Differences in the Netherlands 1850-1970.” Journal of Biosocial Science. 33: 279-303.

Vera, Herman, Donna H. Berardo, and Felix M. Berardo. 1985. “Age Heterogamy in Marriage.” Journal of Marriage and Family. 47: 553-566.

Waite, Linda J. 1995. “Does Marriage Matter?” Demography. 32: 483-507.

Wallerstein, Judith S. and Sandra Blakeslee. 1989. Second Changes: Men, Women, and Children a Decade After Divorce. Boston: Houghton Mifflin.

Watkins, Susan Cotts. 1984. “Spinsters.” Journal of Family History. 9: 310-325.

153

Wetzel, James. 1995. “Labor Force, Unemployment, and Earnings.” Pp. 59-105 in State of the Union: America in the 1990s, vol. 1, Economic Trends, edited by Reynolds Farley. New York: Russell Sage Foundation.

White, Lynn and Stacy J. Rogers. 2000. “Economic Circumstances and Family Outcomes: A Review of the 1990s” Journal of Marriage and Family. 62: 1035- 1051.

Whyte, Martin King. 1990. Dating, Mating, and Marriage. Aldine de Gruyter: New York.

Wilson, Barbara Foley and Sally Cunningham Clarke. 1992. “Remarriages: A Demographic Profile.” Journal of Family Issues. 13: 123-141.

Wolfinger, Nicholas H. 1999. “Trends in the Intergenerational Transmission of Divorce.” Demography. 36: 415-420.

Wu, Zheng. 1994. “Remarriage in Canada: A Social Exchange Perspective.” Journal of Divorce and Remarriage. 21: 191-224.

Wu, Zheng and T.R. Balakrishnan. 1994. “Cohabitation After Marital Disruption in Canada.” Journal of Marriage and the Family. 56: 723-734.

Wu, Zheng and Chritoph M. Schimmele. 2005. “Repartnering After First Union Disruption.” Journal of Marriage and Family. 67: 27-36.

Xie, Yu, James M. Raymo, Kimberly Goyette, and Arland Thornton. 2003. “Economic Potential and Entry into Marriage and Cohabitation.” Demography. 40: 351-367.

Xu, Xiaohe, Clark D. Hudspeth, and John P. Bartkowski. 2006. “The Role of Cohabitation in Remarriage.” Journal of Marriage and Family. 68: 261-274.

Yeung, W. Jean, Greg J. Duncan, and Martha S. Hill. 2000. “Putting Fathers Back in the Picture: Parental Activities and Children’s Adult Outcomes.” Marriage and Family Review. 29: 97-113.

Zill, Nicholas, Donna Ruane Morrison, and Mary Jo Coiro. 1993. “Long-Term Effects of Parental Divorce on Parent-Child Relationships, Adjustment, and Achievement in Young Adulthood.” Journal of Family Psychology. 7: 91-103.

154

Appendix A: Supplemental Results

155 Men

Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. educated vs. educated vs. educated vs. husband more not remarried not remarried not remarried homogamy homogamy educated Time Duration since divorce 1.336 *** 1.406 ** 1.363 *** 1.052 1.019 0.969 Duration since divorce, squared 0.985 *** 0.977 ** 0.981 ** 0.991 0.996 1.004 Age, centered 0.901 *** 0.895 ** 0.904 *** 0.994 1.004 1.004

Socioeconomic status Log of income 1.038 ** 1.243 * 1.051 * 1.198 * 1.012 0.845 *

First marriage ties Co-residential child/children from divorce

156 1.052 0.927 0.861 0.881 0.819 0.929 (ref.: no co-residential child)

Currently enrolled in school 1.502 0.576 0.576 0.556 0.378 0.980

Race/ethnicity non-Hispanic Black 0.734 * 0.260 0.364 0.659 * 0.881 0.913 Hispanic 0.639 * 0.300 0.399 0.926 1.306 1.332

Table 10: Discrete-Time Multinomial Logistic Regression Results for Educational Assortative Mating, Full Results

Table 10 continued

Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. educated vs. educated vs. educated vs. husband more not remarried not remarried not remarried homogamy homogamy educated Socioeconomic background, age 14 Mother high school educated 0.732 * 0.357 0.982 1.109 1.515 1.100 Mother college educated 0.948 0.242 0.286 0.994 1.048 1.446

Family structure, age 14 Two parents present 1.077 0.524 0.5 0.793 0.736 0.605 Stepparent present 1.593 * 0.536 0.889 0.863 0.542 0.936 Other family structure 0.764 0.542 0.733 0.854 1.118 1.079

157 Religious background, age 14 Catholic 1.054 0.386 0.772 0.830 0.787 0.750 Conservative Protestant 1.049 0.313 0.859 1.056 1.006 1.120

Urban 0.941 0.247 0.513 0.907 0.964 1.105 South 1.146 0.312 0.65 0.978 0.853 0.863

First marriage characteristics Duration of first marriage 1.073 *** 0.047 0.621 1.078 * 1.005 1.054 First marriage educationally homogamous 1.307 * 0.300 0.491 0.554 ** 0.424 *** 0.466 * Cohabited prior to first marriage 0.798 0.405 0.489 0.758 1.038 1.166 N-individuals 1,179 N-person years 8,706 -2 Log likelihood 252.61 Psuedo R-squared 0.048

Table 10 continued

Women Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. educated vs. educated vs. educated vs. husband more not remarried not remarried not remarried homogamy homogamy educated Time Duration since divorce 1.367 *** 1.587 *** 1.423 ** 1.161 1.041 0.897 Duration since divorce, squared 0.985 *** 0.982 *** 0.976 *** 0.997 0.991 0.994 Age, centered 0.891 *** 0.825 *** 0.984 0.925 * 1.103 ** 1.192 ***

Socioeconomic status Full-time, year-round employed 0.894 0.863 0.896 0.965 1.003 1.039 (ref.: not full-time, year-round employed) 158 First marriage ties Co-residential child/children from divorce 0.852 0.916 0.678 * 1.076 0.797 0.741 (ref.: no co-residential child)

Currently enrolled in school 0.561 * 1.487 0.668 2.652 * 1.191 0.449

Race/ethnicity non-Hispanic Black 0.639 ** 0.487 * 0.217 *** 0.763 0.340 ** 0.445 * Hispanic 0.842 0.760 0.598 0.903 0.711 0.788

Table 10 continued

Wife more Husband more Wife more Husband more Wife more educated vs. Homogamy vs. educated vs. educated vs. educated vs. educated vs. husband more not remarried not remarried not remarried homogamy homogamy educated Socioeconomic background, age 14 Mother high school educated 0.903 1.094 1.285 1.212 1.423 1.174 Mother college educated 1.102 1.498 1.519 1.359 1.378 1.014

Family structure, age 14 Two parents present 1.240 1.281 1.284 1.033 1.035 1.002 Stepparent present 1.431 * 0.826 1.117 0.578 0.781 1.352 Other family structure 1.076 0.979 1.117 0.910 1.038 1.140

159 Religious background, age 14 Catholic 0.952 1.233 0.759 1.295 0.797 0.616 Conservative Protestant 1.066 0.800 0.934 0.751 0.877 1.168

Urban 0.948 0.726 0.786 0.766 0.829 1.082 South 1.268 1.064 1.568 * 0.839 1.237 1.474

First marriage characteristics Duration of first marriage 1.078 *** 1.104 * 0.977 1.024 0.906 ** 0.885 * First marriage educationally homogamous 1.070 0.580 ** 0.604 * 0.542 ** 0.565 * 1.042 Cohabited prior to first marriage 0.890 0.983 1.337 0.905 1.502 1.659 N-individuals 1,245 N-person years 10,438 -2 Log likelihood 338.35 Psuedo R-squared 0.058 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

Appendix B: Supplemental Results

160 Men Homogamy vs. Husband Wife older vs. not older vs. not not Husband older Wife older vs. Wife older vs. remarried remarried remarried vs. homogamy homogamy husband older Duration since divorce 1.305 *** 1.062 1.524 *** 0.814 * 1.168 1.435 *** Duration since divorce-squared 0.989 ** 0.989 ** 0.978 *** 0.999 0.988 * 0.989 * Age, centered 0.839 *** 1.024 0.881 *** 1.220 *** 1.050 0.860 ***

Current socioeconomic status Log of income, lagged 1.084 * 1.008 1.034 0.929 0.953 1.026

Long-term socioeconomic status (ref.: high school graduate) Less than high school 0.612 * 0.639 * 0.662 * 1.045 1.081 1.035 College degree or more 1.429 * 1.520 * 1.114 1.063 0.779 0.733

161 First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.897 1.353 * 1.049 1.507 * 1.169 0.775

Currently enrolled in school 1.680 0.540 1.154 0.321 * 0.687 2.137

Race/ethnicity non-Hispanic Black 0.779 0.576 ** 0.783 0.739 1.005 1.359 Hispanic 0.479 ** 1.287 0.707 2.690 ** 1.477 0.549

Table 11: Discrete-Time Multinomial Logistic Regression Results for Age Assortative Mating, Full Results—Main Effects

Model

Table 11 continued

Homogamy vs. Husband Wife older vs. not older vs. not not Husband older Wife older vs. Wife older vs. remarried remarried remarried vs. homogamy homogamy husband older Socioeconomic background, age 14 Mother high school educated 0.811 1.221 0.974 1.506 1.201 0.797 Mother college educated 0.973 1.043 1.043 1.073 1.073 1.000

Family structure, age 14 Two parents present 1.119 0.647 * 0.983 0.578 * 0.878 1.520 Stepparent present 1.477 1.045 1.188 0.707 0.804 1.137 Other family structure 0.641 0.482 * 1.195 0.752 1.865 2.478 *

Religious background, age 14

162 Catholic 1.339 0.801 0.922 0.598 * 0.688 1.151 Conservative Protestant 1.156 0.927 1.285 0.802 1.111 1.386

Urban 0.924 0.797 0.967 0.863 1.047 1.213 South 1.257 1.099 1.027 0.874 0.818 0.935

First marriage characteristics Duration of first marriage 1.089 ** 0.953 * 1.140 *** 0.876 *** 1.047 1.196 *** First marriage educationally homogamous 0.784 1.014 0.788 1.293 1.005 0.777 Cohabited prior to first marriage 0.864 0.710 0.782 0.821 0.905 1.102 N-individuals 1,179 N-person years 8,780 -2 Log likelihood 293.05 Psuedo R-squared 0.048

Table 11 continued

Women Homogamy vs. Heterogamy Wife older vs. not vs. not not Heterogamy vs. Wife older vs. Wife older vs. remarried remarried remarried Homogamy Homogamy husband older Time Duration since divorce 1.404 *** 1.174 ** 1.369 *** 0.836 0.975 1.166 * Duration since divorce-squared 0.983 *** 0.985 *** 0.984 *** 1.001 1.000 0.999 Age at divorce, centered 0.873 *** 1.037 * 0.922 ** 1.188 1.056 0.889 ***

Current socioeconomic status Full-time, year-round employed (ref.: not full-time, year-round employed) 0.920 1.085 0.687 * 1.179 0.747 0.634 *

Long-term socioeconomic status 163 (ref.: high school graduate) Less than high school 0.774 1.202 0.820 1.552 1.058 0.682 College degree or more 1.001 0.913 1.203 0.912 1.202 1.317

First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.798 0.967 1.003 1.211 1.257 1.038

Currently enrolled in school 1.679 0.555 1.211 0.330 0.721 2.182

Race/ethnicity non-Hispanic Black 0.763 0.582 ** 0.790 0.763 1.036 1.357 Hispanic 0.480 ** 1.292 0.708 2.691 ** 1.474 0.548 *

Table 11 continued

Homogamy vs. Heterogamy Wife older vs. not vs. not not Heterogamy vs. Wife older vs. Wife older vs. remarried remarried remarried Homogamy Homogamy husband older Socioeconomic background, age 14 Mother high school educated 0.802 1.218 0.968 1.519 1.208 0.795 Mother college educated 0.975 1.034 1.032 1.061 1.059 0.998

Family structure, age 14 Two parents present 1.112 0.647 * 0.983 0.582 * 0.884 1.519 Stepparent present 1.451 1.044 1.189 0.720 0.820 1.139 Other family structure 0.634 0.492 * 1.204 0.776 1.900 2.448 *

Religious background, age 14

164 Catholic 1.337 0.796 0.917 0.595 * 0.686 1.152 Conservative Protestant 1.166 0.926 1.292 0.794 1.108 1.396

Urban 0.927 0.801 0.975 0.865 1.052 1.216 South 1.236 1.090 1.012 0.882 0.818 0.928

First marriage characteristics Duration of first marriage 1.088 ** 0.954 * 1.140 *** 0.877 *** 1.048 1.195 *** First marriage educationally homogamous 0.786 1.006 0.784 1.280 0.998 0.779 Cohabited prior to first marriage 0.862 0.716 0.787 0.831 0.969 1.099 N-individuals 1,292 N-person years 10,498 -2 Log likelihood 338.35 Psuedo R-squared 0.050 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05

Appendix C: Supplemental Results

165 Homogamy vs. Husband older Wife older vs. Wife older not vs. not Mennot Husband older vs. remarried remarried remarried vs. Homogamy Homogamy Time Wife older vs. Duration since divorce 1.326 *** 1.058 1.546 *** 0.798 ** 1.165 1.461 *** husband older Duration since divorce-squared 0.989 ** 0.989 ** 0.977 *** 1.001 0.989 * 0.988 * Age, centered 0.841 *** 1.022 0.889 *** 1.215 *** 1.057 0.869 ***

Current socioeconomic status Log of income, lagged 1.086 * 1.007 1.036 0.927 0.953 1.028

Long-term socioeconomic status (ref.: high school graduate) Less than high school 0.703 0.632 * 0.632 * 0.899 0.899 0.999 College degree or more 1.308 1.443 * 1.197 1.103 0.915 0.829 166

Interactions Less than high school * age, centered 1.033 0.996 0.977 0.964 0.946 0.981 College degree or more * age, centered 0.901 ** 1.028 0.894 ** 1.141 ** 0.992 0.830 *

First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.907 1.345 * 1.069 1.482 * 1.177 0.795

Currently enrolled in school 1.644 0.547 1.138 0.333 * 0.693 2.083

Race/ethnicity non-Hispanic Black 0.765 0.585 ** 0.779 0.765 1.018 1.330 Table Hispanic 12: Discrete-Time Multinomial Logistic 0.465 Regression 1.302 Results 0.686for Age Assortative 2.802 ** Mating, 1.476 Full Results—Interactive 0.527 Model Table 12 continued

Table 12 continued

Homogamy vs. Husband older Wife older vs. Wife older not vs. not not Husband older vs. Wife older vs. remarried remarried remarried vs. Homogamy Homogamy husband older Socioeconomic background, age 14 Mother high school educated 0.795 1.238 0.962 1.556 1.209 0.777 Mother college educated 0.918 1.059 1.005 1.154 1.095 0.949 *

Family structure, age 14 Two parents present 1.088 0.649 * 0.950 0.597 * 0.873 1.462 Stepparent present 1.511 1.044 1.171 0.691 0.775 1.122 Other family structure 0.640 0.475 * 1.192 0.742 1.861 2.509 * 167 Religious background, age 14 Catholic 1.361 0.799 0.942 0.587 * 0.692 1.179 Conservative Protestant 1.142 0.917 1.262 0.803 1.105 1.376

Urban 0.905 0.805 0.941 0.890 1.040 1.169 South 1.247 1.108 1.022 0.888 0.819 0.922

First marriage characteristics Duration of first marriage 1.090 ** 0.953 ** 1.142 *** 0.874 *** 1.048 1.199 *** First marriage educationally homogamous 0.976 1.181 1.086 1.211 1.113 0.919 Cohabited prior to first marriage 0.926 0.710 0.834 0.766 0.900 1.175 N-individuals 1,179 N-person years 8,780 -2 Log likelihood 307.69 Psuedo R-squared 0.047

Table 12 continued

Women Homogamy vs. Husband older Wife older vs. Wife older not vs. not not Husband older vs. remarried remarried remarried vs. Homogamy Homogamy Wife older vs. Time husband older Duration since divorce 1.403 *** 1.176 ** 1.370 *** 0.838 * 0.976 1.164 * Duration since divorce-squared 0.983 *** 0.984 *** 0.984 *** 1.001 1.001 0.999 Age, centered 0.874 *** 1.046 ** 0.926 ** 1.197 *** 1.059 * 0.885 ***

Current socioeconomic status Full-time, year-round employed (ref.: not full-time, year-round employed) 0.919 1.079 0.684 * 1.175 0.744 0.633 *

Long-term socioeconomic status

168 (ref.: high school graduate) Less than high school 0.803 1.089 0.932 1.336 1.161 0.856 College degree or more 0.983 0.971 1.157 0.937 1.177 1.256

Interactions Less than high school * age, centered 1.005 0.971 1.025 0.967 1.020 1.056 College degree or more * age, centered 0.982 0.958 0.939 * 0.976 0.956 1.256

First marriage ties Co-residential child/children from divorce (ref.: no co-residential child) 0.803 0.974 1.041 1.213 1.297 1.069

Currently enrolled in school 0.721 0.891 0.482 * 1.236 0.668 0.540

Race/ethnicity non-Hispanic Black 0.768 0.331 ** 0.594 0.431 ** 0.773 1.791 * Hispanic 0.823 ** 0.507 1.349 0.616 * 1.638 2.660 ** Table 12 continued

Table 12 continued

Homogamy vs. Husband older Wife older vs. Wife older not vs. not not Husband older vs. Wife older vs. remarried remarried remarried vs. Homogamy Homogamy husband older Socioeconomic background, age 14 Mother high school educated 1.018 1.443 0.932 1.417 0.916 0.646 Mother college educated 1.162 1.530 1.275 1.316 1.097 0.833

Family structure, age 14 Two parents present 1.390 1.066 * 2.029 0.766 1.460 1.904 * Stepparent present 1.432 1.030 2.082 0.719 1.454 2.021 * Other family structure 1.228 1.204 * 0.988 0.980 0.804 0.821 169 Religious background, age 14 Catholic 1.316 0.705 0.897 0.536 * 0.681 1.272 Conservative Protestant 1.072 0.962 0.941 0.898 0.878 0.979

Urban 0.787 1.047 0.831 1.330 1.056 0.794 South 1.203 1.261 1.453 1.048 1.208 1.153

First marriage characteristics Duration of first marriage 1.063 ** 0.952 * 1.044 *** 0.895 *** 0.981 1.096 ** First marriage educationally homogamous 1.012 0.915 0.804 0.904 0.794 0.878 Cohabited prior to first marriage 1.015 0.667 0.765 0.657 0.754 1.147 N-individuals 1,292 N-person years 10,498 -2 Log likelihood 324.46 Psuedo R-squared 0.050 Source: National Longitudinal Survey of Youth, 1979 cohort ***p<.001; **p<.01; *p<.05