View, Methods, Results, and Discussion Sections

A Cohort Comparison of the Transition to Adulthood in the United States

DISSERTATION

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

Bohyun Jang, M.S.

Graduate Program in Human Development and Family Science

The Ohio State University

2014

Dissertation Committee:

Dr. Anastasia Snyder, Advisor Dr. John Casterline Dr. Claire Kamp Dush Dr. Zhenchao Qian

Copyrighted by

Bohyun Jang

2014

Abstract

The transition to adulthood is a critical period of human development when young adults gain a sense of independence and accumulate human capital through various life experiences. Demographically, this is also an important period when many major life transitions occur. With economic globalization and rapid technological changes, young adulthood has extended and become diverse in recent years. Although studies on the transition to adulthood are abundant, some dynamics of the life course during this period are not completely understood. Studies on this topic are often limited because most only consider one or two domains of the life course, even though complexity among various life events has increased during young adulthood. The current study, therefore, investigates changes in the transition to adulthood by comparing two cohorts in the U.S.

Various life course events, including family formation, education, employment, and home-leaving, are taken into consideration to account for complexity of life course events. Particularly, this study includes geographic mobility which has been understudied but has significant implications for obtaining independence and future outcomes.

I use data from the public and geocode files from both the National Longitudinal

Survey of Youth 1979 (NLSY79) and the National Longitudinal Survey of Youth 1997

Cohort (NLSY97) to examine the transition to adulthood and compare it across cohorts.

These data are well suited for this dissertation because both contain a wide range of life

ii course information, and respondents from each dataset have undergone the same developmental stages at different historical times (i.e. their 20s during 1980s and in the

2000s for the NLSY79 and 97 respectively).

This dissertation is separated into three independent studies; first, in chapter 2, I use Latent Class Analysis to investigate distinct patterns in the transition to adulthood for men and women. Results show that young adults in the NLSY97 are more disproportionately distributed to different classes, which indicates their diverse paths to adulthood compared to those of the NLSY79. In the following chapter, I examine the complexity of life course transitions by focusing on mobility and union formation.

Findings reveal that life course events are closely related to each other but the relationship differs by cohort, pointing to contextual influence on young adults’ life courses. As a decision on the life course is likely made in concert with other life events, chapter 4 examines endogeneity between life course transitions. I find that unobserved characteristics affect the estimation of life course events in both cohorts, and therefore ignoring the factors could misrepresent the actual relationship between life events. From these findings, I address implications of theory, methodology, and social policy for those in the transition to adulthood in chapter 5.

iii

Acknowledgements

I would like to extend my sincere appreciation to my advisor, Dr. Tasha Snyder, for her continued mentorship, patience, and support throughout my graduate career. From the very beginning of my graduate program at OSU, she has provided me with great opportunities for research and career. Her guidance and encouragement has helped me develop my professional expertise and navigate research paths. I particularly thank her for believing in my ability and giving such endless support and encouragement. It was at times hard for me to deal with challenges but she was always accountable. She offered me intelligent advice and opportunities to facilitate external resources to pass through the challenges. Further, she helped me adapt to a new environment in the United States which is different than my home, South Korea. It is not sufficient to express my gratitude to her in the space allotted here. I am sincerely grateful to you, Tasha!

I also wish to thank my dissertation committee members, Drs. John Casterline,

Claire Kamp Dush, and Zhenchao Qian, for their continuous support and guidance. John has given me such great advice to write down my idea in a ‘do’ file and sharpen my analytic skills. He also provides insightful advice for my short- and long- term careers, which I deeply appreciate. Claire has inspired me to develop professional expertise by offering great amount of feedback and practical tips. I have learned tremendously from working with her, both her expertise and passion. Further, I am grateful to Zhenchao who

iv facilitated my understanding of advanced statistical methods. I took his method course in my first year at OSU. I remember him taking me a tour to a researcher lab and making sure that I do not have any problem to do a research project. Until now, he has given me continuous support. I truly appreciate all of their advice and support.

I am grateful to my friends from HDFS, IPR, and church and academic colleagues who supported and encouraged me over the years. They are good friends, counselors, and consultants. I also want to thank my friends in Korea who have been there for me whenever I need them. Although we have a time difference of 14 hours, they have shown great amount of support for me. Further, I thank the Institute for Population Research and

Department of Human Sciences at the Ohio State University for giving me various research opportunities and a space to grow as a scholar.

I express heartfelt gratitude to my family - mom, dad, sister, brother, and husband-to-be, Hong-Min. Without their love, prayer, and support, I would have not made it. They made me laugh and smile. Even though I do not see you as often as I used to, I still love you and miss you so much every day. Thank God for giving me this great opportunity to do my Ph.D. at OSU and having all these wonderful people in my life!

Vita

February 2000 ...... Hye-Sung High school, Seoul, S. Korea 2004...... B.S. Human Development and Family Science and Sociology, Kyung-Hee University, Seoul, S. Korea 2006...... M.S. Family Studies, Kyung-Hee University, Seoul, S. Korea 2009 to present ...... Graduate Research and Teaching Associate, Graduate Fellow, Department of Human Sciences, The Ohio State University

Publications

Jang, B., Casterline, J., & Snyder, A. (2014). Migration and marriage in the United States: Modeling the Joint Process, Demographic Research, 30(47), 1339-1366.

Fields of Study

Major Field: Human Development and Family Science

Table of Contents

Abstract ...... ii

Acknowledgements ...... iv

Vita ...... vi

Table of Contents ...... vii

List of Tables ...... viii

List of Figures ...... x

Chapter 1: Introduction ...... 1

Chapter 2: Differences in the transition to adulthood by gender and cohort ...... 14

Chapter 3: Moving and union formation: a cohort comparison of the association between

life course events ...... 45

Chapter 4: Modeling the joint process between life course events ...... 76

Chapter 5: Conclusions ...... 120

Appendix A: Measurement of cohabitation in the NLSY79 ...... 128

References ...... 129

vii

List of Tables

Table 2.1 Model Fit Information used in selecting the LCA model ...... 40

Table 2.2 Item-response probabilities for men (a) and women (b) ...... 41

Table 2.3 Multinomial logistic regressions for predicting latent class membership ...... 43

Table 3.1 Description of sample by first union formation ...... 71

Table 3.2 Description of mobility by first union formation...... 72

Table 3.3 Comparison of mobility in the NLSY97 by different measurement units ...... 73

Table 3.4 Cox proportional hazard models predicting first union ...... 74

Table 3.5 Cox competing risks models predicting first union type ...... 75

Table 4.1 Description of migration and marriage histories ...... 109

Table 4.2 Description of covariates included in the model ...... 110

Table 4.3 Estimated Random-effects ...... 111

Table 4.4 Estimates from models for migration (NLSY79) ...... 112

Table 4.5 Estimates from models for migration (NLSY97) ...... 113

Table 4.6 Estimates from models for marriage (NLSY79) ...... 114

Table 4.7 Estimates from models for union formation (NLSY97) ...... 115

Table 4.8 Robustness of the multi-process model estimation with different identification

variables for migration (NLSY79) ...... 116

Table 4.9 Robustness of the multi-process model estimation with different identification

variables for marriage (NLSY79) ...... 117 viii

Table 4.10 Robustness of the multi-process model estimation with different identification

variables for migration (NLSY97) ...... 118

Table 4.11 Robustness of the multi-process model estimation with different identification

variables for union formation (NLSY97) ...... 119

List of Figures

Figure 1.1 Contextual changes during the transition to adulthood in the NLSY79 and 97

...... 13

Figure 2.1 Completing life course transitions, by gender and cohort ...... 39

Figure 2.2 Class membership probabilities in each latent class for men(a) and women(b)

...... 42

CHAPTER 1: INTRODUCTION

The transition to adulthood1 is recognized as a critical period of life in which young people experience numerous life changes over a relatively short time (Arnett,

2004; Furstenberg, Rumbaut, & Settersten, 2005; Rindfuss, 1991). Furthermore, life course decisions during this period have subsequent influences on later life outcomes such as poverty, earnings, and general well-being (Dahl, 2010; Mouw, 2005). Several demographic markers i.e., leaving the parental home, completing schools, entering the labor force, and establishing one’s own family have delineated the entry adulthood, which used to be brief and predictable for most young adults (Furstenberg, Rumbaut, &

Settersten 2005). In most industrialized societies, however, the phase to become an adult, which contemporary young adults encounter, is no longer transitory (Arnett 2000;

Furstenberg, Rumbaut, & Settersten 2005). In the United States, for example, the average age at first marriage increased to the late 20s by 2011 compared to the early 20s in the

1970s (US Census Bureau 2011). Moreover, a growing number of young adults attend college after high school (Bound, Lovenheim, & Turner 2009), which in turn postpones an entry into the labor force.

1 Although young adults who delay the entry into adulthood are recently referred to as “emerging adults” in literature, I use the term “transition to adulthood” because 1) emerging adulthood only embraces young adults in the 2000s and 2) major features of emerging adulthood (e.g. exploration of life possibility) have been limited to a selective group of young adults who are white, from middle-class family, and college students but has not applied to general population (Cӧte & Bynner 2008). 1

The prolonged transition to adulthood has drawn research attention to the significance of this period as a human development process (Arnett 2000; Tanner &

Arnett 2011). From a developmental perspective, the transition to adulthood is a time when individuals begin a new independent stage of life with an identity that began to be formed during adolescence (Erikson 1980). Young adults construct life trajectories in education and occupation and establish intimate relationships with significant others during the period. As the transition to adulthood has extended, however, developmental tasks from adolescence often spill over into this stage and young adults continue to explore their identity and life possibilities (Arnett 2000; Arnett & Tanner 2006). The changing features of the transition to adulthood coincide with recent changes regarding individual expression and the weakening of social norms confining individual’s values and behaviors (Beck & Beck-Gernsheim 2002; Shanahan 2000). Self- achievement and autonomy have been emphasized, and thus the life course of contemporary young adults has become ill-timed, less normative, and disordered (Furstenberg 2005; Macmillan

2007). As a result, variability increases across cohorts and even within the same cohort, which make existing institutional support and policies ill-suited for recent young adults

(Settersten 2005). Furthermore, these may contribute to a smooth or uneasy transition to adulthood (Cook & Furstenberg 2002; Settersten 2005).

Although studies on the transition to adulthood are abundant, the dynamic process of the life course during this period is not yet completely understood. Studies on this topic are limited because most only consider one or two domains of the life course, even though complexity among multiple life events has increased in the transition to adulthood 2 in most societies (Billari & Liefbroer 2010; Furstenberg, Rumbaut, & Settersten 2005).

Given the enormous future significance of this period, additional research needs to investigate the complexity and changes in the transition to adulthood, across multiple domains.

The life course perspective

Human development has been understood as a complex interaction between personal agency and economic and social conditions (Bronfenbrenner 1977; Elder 1986;

1998; Shanahan 2000). According to Bronfenbrenner (1977)’s ecological perspective, immediate and broad systems surrounding an individual are critical to developmental trajectories. That is, place, time, physical features, and particular roles in a certain environment interact with individuals and create developmental pathways of the individuals (Bronfenbrenner 1989; Bronfenbrenner & Evans 2000). In line with this ecological perspective, Elder (1998) suggested four theoretical principles that are pivotal to human development: historical time and place, linked life events, developmental stage at which historical event occurs, and human agency. Historical time and place emphasizes the shared historical events, opportunities and constraints that individuals experience in their society at a given time (Alwin & McCammon 2003). For example, during the post-war baby boom and bust era in the United States (1945-1980), the U.S. population revealed unique patterns of family formation (Easterlin 1961; Pampel &

Peters 1995). As the baby boom cohort aged, late marriage, high divorce rates, and lower birthrates characterized the period as a consequence of growing cohort size and reduced

3 economic opportunities (Pampel & Peters 1995). Moreover, the changes in family formation affect living arrangement of the baby boomers in later life such as an increase in living alone for the elder (Macunovich, Easterlin, Schaeffer, & Crimmins 1995).

Another example of the importance of historical time and place is the impact that the

Great Depression crisis had on the life course (Elder 1998; Shanahan, Elder, and Miech

1997). Young men who were about to enter the labor market during the economic chaos of the 1930s delayed this transition, and many thus entered postsecondary education

(Shanahan, Elder, and Miech 1997). Recently, this ‘warehousing hypothesis’ appeared among young adults during the Great Recession in the late 2000s; more young adults live with parents and are more likely to return to the parental home (Billari & Liefbroer 2007;

Goldscheider & Goldscheider 1999; Qian 2012).

Second, the concept linked lives refers to the interdependency of individual’s life course events. Studies have shown that developmental change, which extends through the life span, is distinguishable from the former and following phases (Erikson 1980). Yet, life experiences in a given developmental stage are linked to the previous and later outcomes. For example, educational and occupational trajectories for immigrants in early adulthood are related to their experiences during adolescence such as incarceration and teen-birth (Rumbaut 2005). Moreover, Walesmann, Geronimus, and Gee (2008) found that educational disadvantages during adolescence are significantly relevant to health disparities in middle age, pointing to cumulative effects across the life course. When it comes to the outcomes of linked life course, the time when a life event occurs is critical as each developmental stage has a unique task to accomplish (Erikson 1968; Elder 1998). 4

During the Great Depression, men in different developmental stages revealed distinct educational and occupational trajectories (Shanahan, Elder, and Miech 1997). Those in the younger cohort (born in 1911-1917) were graduating from high school at the beginning of the economic crisis, and the economy started to recover when they were in college. This contextual shift enabled the young men to choose either to continue or drop out of college. In other words, the younger cohort men had options to take different paths in life. On the other hand, those from the older cohort (born in 1904-1910) encountered the economic collapse when they were finishing college. As a result of the different developmental stages during the same historical time, the older cohort of men achieved higher educational and first-time occupational premium, while those from the younger cohort experienced lower premium for first-time occupations but greater life-long promotion (Elder et al. 2003).

While social and economic structures are significant factors shaping the life course, human agency exerts internal potentials to modify the given context (Clausen

1991; Elder 1998; Settersten & Gannon 2005). Human agency has been described in the literature as self-efficacy, self-confidence, and planful competence that plays a critical role in carrying out plans and achieving short and long-term goals in the life course

(Clausen 1991; Crockett & Beal 2012). For example, Mortimer, Staff, and Lee (2005) found that work behavior during high school is related to previous life goals and values, and further predict future trajectories of work and education. In addition, competent individuals appear to set realistic goals for life and prepare to accomplish them (Clausen

1991). In turn, these internal characteristics are associated with personality stability, marital stability, and later life satisfaction (Clausen 1991).

A large body of research has built upon the life course perspective and demonstrates that both structural context and individual capability determine variability in the life course among recent cohorts of young adults (Cόte & Bynner 2008; Mayer

2009; Mortimer, Staff, & Lee 2005; Shanahan 2000). In particular, research on the transition to adulthood has paid attention to the role of human agency and contextual influences in educational attainment and family formation behaviors (Mortimer et al.

2005; Rumberger 2010; Settersten & Gannon 2005). Despite these efforts, few studies have focused on examining multiple transition events together and how the transition to adulthood has changed across cohorts (Furstenberg, Rumbaut, & Settersten, 2005). Most studies have paid attention to each transition separately, ignoring the possibility of dependency (Furstenberg, Rumbaut, & Settersten, 2005). Moreover, demonstrating the effects of historical time and social context, most studies have employed cross-sectional data (Fussell & Furstenberg, 2005; Smith, 2005) and thus our knowledge about changes in the transition to adulthood has been limited. Given the growing complexity of transition events and the potential implications of change and continuity in the transition to adulthood, further research using longitudinal information on multiple life transitions will enrich our understanding of the changes in the transition to adulthood.

Second Demographic Transition

Changes in the transition to adulthood coincide with the second demographic shift in most developed countries (Billari 2004). Whereas the first demographic transition was driven by the decline in mortality, the second transition is driven by below replacement fertility levels (van de Kaa 1987; 2002). Moreover, the second demographic transition embraces socioeconomic, cultural, and technological changes, which have transformed various life domains (van de Kaa 2002). First, the decline in fertility rates coincides with a prevalence of contraceptive use and increasing female labor force participation since the mid 20th century. With the introduction of more effective contraception, individuals were able to control the timing of childbirth and even choose to forego children. This contraceptive revolution further separated sex and childbearing from marriage which resulted in an increase in nonmarital births and diverse living arrangements that emphasize intimacy in romantic relationships (Lesthaeghe 2010).

The decline in fertility during the second demographic transition reflects fundamental changes in the family that signals a weakening of the institution of marriage

(Lesthaeghe 1995; 2010; van de Kaa 2002). For example, in societies that undergo the second demographic change, it is found that divorce rates increase and cohabitation is prevalent as a precursor or an alternative to marriage (Lesthaeghe 2010). Economic security, emotional support, and childbearing are no longer considered essential components of marriage while mutual love, satisfaction, and self-fulfillment are considered essential (van de Kaa 2002). Moreover, in this highly individualized society,

7 people are freed from traditional norms and constraints and have more autonomy to manage their life course (Shanahan 2000). On the one hand, the weakened social norms boost uncertainty (Beck 1992; Giddens 1991) which in turn makes youth and young adults delay significant life events (Billari & Liefbroer 2010).

Studies have found that the life course of young adults during the second demographic transition can be characterized by a mixture of de-standardization and standardization and increasing individualization (Billari & Liefbroer 2010; Brückner &

Mayer 2005; Liefbroer & de Jong Gierveld 1995; Shanahan 2000). First, with regard to de-standardization, the association between life course events has become fuzzier although convergence has also been found as the majority of young adults experience delayed, late, and complex life course trajectories. For example, Billari and Liefbroer

(2010) find that family formation, in particular marriage and entry into parenthood, has been postponed in 25 European countries although the onset of delay varies by countries.

Moreover, among young women in a recent birth cohort in the United States, four distinct family formation patterns are detected: remaining single, early marriage, single parents, and cohabitors (Schoen, Landale, & Daniels 2007). Shanahan (2000) also points to the compactness of the life course among young adults, i.e. a relatively short time to complete a range of life course transitions. Despite the converging patterns, the growing individualism and the weakening of an institution contribute to variability and divergence during young adulthood (Shanahan 2000; Setterstern & Ray 2010). In the past the transition to adulthood had been delineated by several key life course events, but in recent years individual’s perspective of being independent has become important to 8 reaching an adult status (Furstenberg et al. 2005; Kins & Beyers 2010). Imposing more autonomy on individuals, the transition to adulthood has become less predictable and more diverse in terms of the timing and sequences of life events.

In the United States, massive changes have occurred during the past few decades with regard to family values and economy (Thornton & Young-DeMarco 2001). The approval of oral contraceptives (the Pill) and legalization of abortion in 1960 and in 1973

(Guldi 2008), a shift from fault to no fault divorce in the 1970s (Nakonezny, Shull, &

Rodgers 1995; Glenn 1997), and marriage promotion under the Personal Responsibility and Work Opportunity Act (PRWORA) in the 1990s (Lichter, Graefe, & Brown 2003) may have disproportionately affected individuals depending on their life stages when these social events occurred (Elder 1998). Moreover, the United States had experienced economic turmoil over the last few decades (e.g. “double-dip” recession during 1980-

1982, “jobless recovery” recession in 2001, and “Great Recession” during 2007-2009,

Hout, Levanon, & Cumberworth 2011). These social and economic changes may have influenced individual developmental pathways both directly and indirectly (Elder,

Johnson, & Crosnoe 2003; Alwin & McCammon 2003; Shanahan 2000).

The United States, however, reveals an exceptional demographic pattern - maintaining relatively stable birth rates around replacement level and a great emphasis on marriage (Cherlin 2010; Lesthaeghe & Neidert 2006). Arguably, the exception is derived from the higher degree of religious identification and behaviors (Carlson 2005) and high fertility of immigrants (Lesthaeghe & Neidert 2006). Moreover, substantial variability in

9 demographic patterns by regions is found within the U.S., which distinguishes the U.S. from other developed countries (Lesthaeghe & Neidert 2006). Despite the uniqueness of the US demographic shift, little is known about how the transition to adulthood is contingent on the recent demographic changes in the U.S. Although studies have shown that young adults in the U.S. reveal late and complex trajectories in the transition to adulthood (Furstenberg, Rumbaut, Settersten 2005), it is not clear to what extent the transition to adulthood has changed over time during the substantial social and economic shifts.

The current study, therefore, aims to first investigate changes in the transition to adulthood by comparing two birth cohorts in the U.S. Second, the complexity of life course events will be examined by including multiple events together and expanding the first life course transition to sequential transitions in young adults’ lives. Third, I include geographic mobility among young adults which is understudied but has significant implications for current and future outcomes. By examining these, I expect 1) to ratify the main principles of the life course perspective – historical time and place, linked lives, importance of developmental stages, and agency, 2) to contribute empirical evidence from young adults in the U.S. to the debate on changes in the transition to adulthood which raises concerns about extended youth and failure to launching as an independent person, and 3) to expand the methodological capacity studying the transition to adulthood. Overall, the findings will contribute to a better understanding of the transition to adulthood in the U.S. and help young adults successfully launch their independent adult lives. 10

Structure of dissertation

The current dissertation project is separated into three independent studies, with each containing its literature review, methods, results, and discussion sections. Chapter 2

(study1) uses Latent Class Analysis to investigate whether distinct patterns in the transition to adulthood can be identified across cohorts. The traditional markers of entry into adulthood include leaving the parental home, completing school, entering the labor force, getting married and having children; these are compared across cohorts. In chapter

3 (study2), the complexity between life course events is examined, particularly focusing on the relationship between mobility and union formation. The timing and type of union formation (i.e., cohabitation, marriage, no union) are investigated in relation to moving experiences using Cox proportional hazard models. In chapter 4 (study3), the interrelationship between mobility and union formation is accounted for by allowing for unobserved heterogeneity, which captures preference or selection biases affecting both mobility and union formation processes. A multi-process model (Lillard & Waite 1993) is used and the findings are compared by cohort. Finally, chapter 5 ends with my concluding thoughts and reflections on the results from each study. I address implications of theory, methodology, and social policy for those in the transition to adulthood.

Data

The public and geocode files from both the National Longitudinal Survey of

Youth 1979 (NLSY79) and the National Longitudinal Survey of Youth 1997 Cohort

(NLSY97) are used in this dissertation. The NLSY79 has interviewed 12,686 individuals

11 who were born in 1957 to 1964 since 1979, when the respondents were ages 14 to 21, through 2010, when they reached their late 40s and early 50s. Among those having longitudinal information, I include life experiences of 9,763 respondents, omitting a military sample of 1,280 individuals having unusual moving patterns and a subsample of economically disadvantaged non-black/non-Hispanic men and women of 1,643 respondents who have not been interviewed since 1990 (Center for Human Resources

2013). The NLSY97 includes panel data from 8,984 individuals who were born between

1980 and 1984 and have been interviewed annually since 1997, when the respondents were ages 12 to 18. In the latest wave of 2011, they were age 25 to 31. The NLSYs contain detailed information on various life course events such as family formation, education and employment in public data sets, and the geocode files particularly provide detailed information about the residence of respondents every survey year. These data are well suited to cohort comparison as both contain a wide range of life course information, and respondents from each dataset have undergone the same developmental stages at different historical times (see Figure 1.1). Respondents from the NLSY97 who encountered young adulthood in the 2000s are probably exposed to more diverse life choices since social institutions that define normative life course events have weakened in recent years. For example, cohabitation presents as a means of marriage compatibility for recent young adults while it is considered an alternative to marriage for older people in the U.S. (Sassler 2010). Comparing the life course between the NLSY79 and 97, therefore, will help us understand changes in young adult’s lives and draw policy implications for those in the transition to adulthood.

Chapter 1 figures

Figure 1.1 Contextual changes during the transition to adulthood in the NLSY79 and 97

CHAPTER 2: DIFFERENCES IN THE TRANSITON TO ADULTHOOD BY GENDER AND COHORT

The life course of young adults has changed over time. It has extended, become de-institutionalized and de-standardized while diverse and individualized patterns have emerged (Brückner & Mayer 2005; Furstenberg, Rumbaut, and Settersten 2005; Osgood et al. 2005). Although the changing features of the transition to adulthood have received much research attention, empirical evidence has been anecdotal and illustrative (Brückner

& Mayer 2005). Moreover, only one or two domains of life courses are explored at a time such as family formation and employment trajectories (Brückner & Mayer 2005; Elder,

Johnson, & Crosnoe 2003), and data have been largely drawn from European countries such as Switzerland (Widmer & Ritschard 2009), West Germany (Brückner & Mayer

2005), and Italy (Billari 2001). As a result, previous literature provides a limited understanding of dynamics in the life course of young adults in the U.S., and resulted in mixed findings about changes in the transition to adulthood (Billari & Liefbroer 2010;

Brückner & Mayer 2005).

Elder (1998) highlighted how the historical time and place in which an individual is embedded can impact their life course outcomes. For example, historical evidence suggests that women adjust the timing of births according to economic conditions (Mare

1995; Rindfuss et al. 1996) and people tend to protract education during a period of 14 socioeconomic crisis (Elder, Johnson, & Cronsoe 2003). In addition, life course experiences of a certain cohort are distinguishable from those of the proceeding and succeeding cohorts (Alwin & Mccammon 2003) as individuals in the same birth cohort share historical events and challenges (Shanahan 2000). As an example, Sandefur,

Eggerling-Boeck, and Park (2005) found that those born in 1974 were less likely to marry but more likely to obtain postsecondary education and live independently compared to those born in 1964.

Though extant research has illuminated some patterns of the transition to adulthood and compared these patterns across cohorts, additional research is needed for several reasons. First, changes in family formation behaviors have been profound during the past few decades (Cherlin 2010; Schoen et al. 2007). Research on cohort comparison, however, does not often take cohabitation into account in the analysis and instead focuses on employment, education, marriage, childbearing, and living independently (Sandefur et al. 2005). This study extends the existing literature by including cohabitation. Second, while the de-standardization of the transition to adulthood has been long debated, few studies have used data from the recent birth cohorts. By comparing those who were born in 1954-1957 to those born in 1980-1984, this study adds empirical evidence of contemporary young adults. Finally, as various life course events are interrelated (Elder

1998), it is beneficial to inform the underlying association between the life courses and identify patterns of life choices for individuals. Using Latent Class Analysis (LCA), this study attempts to discover the dynamics between life course events during the transition to adulthood. LCA has been used in studies on substance use, prevention and treatment 15

(Cleveland, Collins, Lanza, Greenberg, & Feinberg 2010; Coffman, Patrick, Palen,

Rhoades, & Ventura 2007; Lanza & Rhoades 2013) whose goals are to detect multidimensional patterns. Recently, the method has been considered in research on life course transitions of young adults (Amato et al. 2008; Lui, Chung, Wallace, &

Aneshensel 2013; Macmillan & Copher 2005; Osgood et al. 2005; Sandefur, Eggerling-

Boeck, & Park 2005).

The current study focuses on young adults aged 18 to 25, a time during which most transition to adulthood events are completed. Life events include leaving the parental home, completing school (more than college), entering the full-time workforce, and family formation (marriage, cohabitation, ad childbirth). Given the growing diversity of life course experiences by gender (Widmer & Ritschard 2009), separate models for males and females are examined.

Changes in the transition to adulthood

The transition to adulthood has become diverse and de-standardized across many life domains. In particular, significant changes have occurred with regard to family formation among contemporary young adults (Brückner & Mayer 2005; Schoen, Landale,

& Daniels 2007). The rates of nonmarital childbearing and cohabitation have climbed since the 1980s, which raises concerns about a retreat from marriage (Goldstein &

Kenney 2001). For example, in the United States, the median age at first marriage increased from 22 and 21 in 1970 to 28 and 27 in 2009 for men and women, respectively

(Elliott & Simmons 2011), and marriage rates gradually decrease over the past several

16 decades (Copen, Daniels, Vespa, & Mosher 2012). Cohabitation, on the other hand, has become prevalent and is substituted for marriage, resulting in the growth of non-marital births (Chandra, Martinez, & Mosher 2005; Copen et al. 2012; Martin et al. 2007).

Indeed, the meaning of cohabitation as a precursor or an alternative to marriage and the implications of cohabitation has changed among contemporary young adults (Sandberg-

Thoma & Kamp Dush 2014; Schoen et al. 2007). Cohabitation now seems to reflect an alternative to singlehood as few young adults have children while cohabiting and then transition into marriage (Rindfuss & VandenHeuvel 1990; Schoen et al. 2007). These prior studies demonstrate how complex family formation events have become.

Nevertheless, few studies have taken a holistic view of changing family trajectories. One exception is when Amato et al. (2008) used latent class analysis and found seven distinct latent classes that describe the transition to adulthood: college-no family formation, high school-no family formation, cohabiting without children, married mothers, single mothers, cohabiting mothers, and inactive, among young women in the U.S. The study, however, only takes into account the family formation dynamics of females, and thus limits the full understanding of dynamics between various life course transitions including employment and education.

The transition into the labor market is considered necessary to achieve financial and residential independence for young adults. As the economy is subject to globalization and post-industrialization, higher education, training, and advanced skills are required for decent earnings and adequate positions in recent years. Accordingly, employment trajectories of young adults have changed over time. For example, during the Great 17

Recession of 2007 to 2009, young men’s employment rates were considerably lower than those in the 1970s and even in early 2000 (Sum, Khatiwada, McLaughlin, & Palma

2011). Their earnings also plummeted during the adverse economy. Although this degeneration was prevalent in all educational groups, those with less education fared the worst (Sum et al. 2011). Wages for those with a high school degree have stagnated since the 1970s in part due to intense labor market competition which has reduced job security.

As a result, it has become harder for young adults to maintain a solid standard of living with a high school degree, a problem even experienced by some with a college degree in the 2000s (Settersten & Ray 2010). As a result of these changes in labor market conditions, young adults have invest more in their education, and adjusted other life decisions (Settersten & Ray 2010).

Regarding education, the proportion of 25 to 29 year olds who completed a bachelor degree or higher has risen to almost one third of the entire population during the past few decades (Ryan and Siebens 2012). A college education is thought to provide young adults with a range of knowledge, skills, and social networks which are positively related to career prospects (Brock 2010). Moreover, the college years provide an opportunity to explore various life experiences (Arnett 2004; Brock 2010) and are considered by some to be the most important factor differentiating the transition to adulthood (Fussell, Gauthier, and Evans 2007). College students tend to delay family formation until completing school (Sandefur, Eggerling-Boeck, & Park 2005) but non college bound young people take a different route of entry into adulthood, notably marrying and having children earlier (Sandefur, Eggerling-Boeck, & Park 2005). In 18 addition, those with a bachelor degree or higher earn nearly twice as much as those with only a high school diploma over the lifetime (Day & Newburger 2002; Hout 2012).

Finally, leaving the parental home is an important indicator of independence, especially for young adults, because living independently helps them learn to manage financial and emotional independence (Aquilino 1997; Garasky, Haurin, & Haurin 2001;

Goldscheider & Goldscheider 1999). Increasing time spent in school and the delay in the labor force entrance substantially affect the age at leaving home over time (Corcoran &

Matsudaira 2005). In the late 2000s, more than 40% of men and slightly less than 40% of women aged 20 to 24 were found to live with their parents (Settersten & Ray 2010).

Moreover, a greater number of young adults return to the parental home at some point after leaving the home (Stone, Berrington, & Falkingham 2014). As the launch to independent living is relevant to other life course transitions such as marriage and employment, home-leaving needs to be considered, especially in studies of differences in the transition to adulthood by cohort.

Approach of the present study

An individual’s life course consists of multiple discrete transitions in various domains. The transitions are interdependent of each other, which increases the complexity of the life course (Zollinger & Elder 1998) and calls for a holistic and continuous view to study the multiple life course trajectories (Aisenbrey & Fasang 2010;

Elder 1998). Moreover, the growing variability in the life course has correspondingly raised questions about how to examine the time and sequence of a single event

(Aisenbrey & Fasang 2010). For the analysis of the life course, event history analysis had been disproportionately used to understand relationships between various life events

(Schoen, Landale, & Daniels 2007; Schoen, Landale, Daniels, & Cheng 2009). Yet, event history analysis focuses on the timing of events (Cleves, Gutierrez, Gould, & Marchenko

2010), and thus studies using this method have produced partial knowledge, that is - the timing of the life course and movement from one state to another over time (e.g. singlehood to married state) (Schoen, Landale, & Daniels 2007). Despite its contributions, the event history analysis is lacking a holistic view on the life course

(Aassave, Billari, & Piccarretta 2007) and rarely takes into account multidimensional associations at a time (Aisenbrey & Fasang 2010).

Latent class analysis has been introduced to the life course literature to account for possible dependency between events (Clogg and Goodman 1984; Lanza, Collins,

Lemmon, & Schafer 2007; Collins & Lanza 2010). It is useful to assess the underlying association between the variables of interest and identify a set of mutually exclusive latent classes (Lanza et al. 2007). Traditionally latent class analysis has been used in research on drug abuse, HIV, and other risk outcomes (Cleveland et al. 2010; Coffman et al. 2007; Lanza & Rhoades 2013), but has recently expanded to life course transitions of young adults for the following reasons. First, the method is useful for discovering patterns and illustrating data (Collins & Lanza 2010; Lanza, Collins, Lemmon, & Schafer

2007). Osgood and colleagues (2005) discovered six paths through the transition to adulthood among young adults in Detroit suburb areas (i.e. fast starter, parents without careers, educated partners, educated singles, working singles, and slow starters). Each 20 group was distinguished by their experiences of family formation, employment, education, and residence histories during ages 18 to 24. For example, compared to other pathways, “educated singles”, the most common pathway among the young adults (39% of the entire sample), were those who do not live with romantic partners, are more likely to live with parents, spend time gaining education, and are on a upward trajectory because of their high educational level. Considering all the key life course transitions, young adults in that group fit into the emerging adults who had a slower entry into adulthood due to continuing exploration, delaying commitment to adult roles, and gathering personal capital (Arnett, 2000; Osgood et al. 2005).

Second, latent class analysis is well suited to reducing the number of possible pathways that young people experience during the transition to adulthood (Amato et al.

2008). Growing diversity in the transition to adulthood may create numerous trajectories.

Transforming them into meaningful units of analysis may offer insight to understand the dynamics between life course events and find possible factors placing individuals in specific latent classes (Lanza, Tan, & Bray 2013). As patterns describing individual’s pathways are identified, latent class analysis is well suited to ask how life course transitions are influenced by social context and historical time, which is the main point of the cohort comparison. Sandefur, Eggerling-Boeck, and Park (2005) used data on young adults from two cohorts in the United States and compared their life course trajectories using latent class analysis. They found that those from the younger cohort (who were born in 1974) were better off with regard to educational attainment and women’s labor force participation although fewer have engaged in union formation, compared to those 21 who were born in 1964. Afterward, the study obtained gender-distinct patterns in the transition to adulthood by separating the latent class model by gender. This allowed for a deeper look at how the patterns differ by groups. This study, however, does not incorporate cohabitation in the analysis and thus likely misrepresent the family formation trajectories of recent young adults.

Building upon the previous literature, the current study examines the following research questions using latent class analysis (Lanza et al. 2007):

1. Are there differences in the completion of life course transitions by gender and cohort?

2. Can a model of the typical transition to adulthood be identified by gender? What are the types of latent classes for males and females?

3. Are there cohort differences in the measurement of latent classes for the transition to adulthood? How does the probability distributed to each latent class differ by cohort?

4. How do race/ethnicity and household characteristics predict the probability of

being in a certain latent class in each cohort?

Data and Methods

Data from the National Longitudinal Survey of Youth 1979 and 1997 are used for the current study to examine life course transitions i.e., leaving the parental home, completing college, entering the full-time labor force, getting married, cohabiting, and childbirth. The life course experiences are limited to age 18 to 25 in both cohorts in this study for the following reasons. First, most young adults expect to experience at least one

22 major transition event before age 25 (Osgood et al. 2005). The majority of young adults graduate from high school before age 20 (Fussell & Furstenberg 2005), and then either continue their education to college or enter the labor force. The mean age at first marriage is about 29 and 27 for men and women in 2011, yet a considerable share of marriages still occurs before age 25 and cohabitation compensates for the rising age at first marriage

(Copen et al. 2012). Second, given the relatively younger ages of the NLSY97 respondents, data is not available for many of them after age252. The age restriction may leave out possible significant events after age 25 but it appears that most respondents from the NLSY79 and 97 have experienced the six critical life transitions before age 25

(see figure 2.1).

Measures

Family formation history To construct union formation transitions (i.e. marriage and cohabitation), the partnership history files are drawn from both cohorts. The files include a sequential identification number of spouse/partner, and a code reflecting whether the person listed is a spouse or a partner (Center for Human Resources 2013). Using the information, the age at first marriage and cohabitation3 is created and included in the analysis. For childbearing history, the NLSYs provide refined data on the fertility history every survey year. For the analysis, respondents’ age at first birth is created.

Full-time employment Weekly employment status is used to create the employment history in both cohorts. Those who work on average more than 39 hours per week in a

2 Only 25.4% of the NLSY97 sample were age30 or more in last interview of 2011. 3 Regarding the cohabitation history in the NLSY79, refer to Appendix A. 23 month for more than or equivalent to 50 weeks are coded as being employed full-time, those who work less than 39 hours per week in a month for less than 50 weeks are defined as being employed part-time and others are considered being unemployed.

Completing school Education is measured by whether respondents complete a postsecondary degree each year. Both NLSY79 and 97 provide the exact month and year of school completion. Those of higher than Associate degree (i.e. AA, BA or higher) are included in the model.

Leaving the parental home The first exit from the parental home is an important life transition which is relevant to becoming independent. Both data sets do not directly ask respondents about their home leaving and returning events. To create this history, I rely on the household roster information from each cohort which contains details of residents aged 14 or older in the same household with the respondent. Home-leaving is measured by living without parents - respondents no longer report their parents as a household member. Biological, step, and other parent figures are taken into consideration the definition of parents.

Other covariates Race/ethnicity is categorized into three groups: African-American,

Hispanics, and non-Black non-Hispanic whites and others. Dummy variables for African-

American and Hispanics are included in the model, treating whites and others as a reference category. Household characteristics include 1) whether the respondent had lived with their biological parents during childhood (1=yes, 0=no) and a continuous variable of 2) mother’s educational attainment.

Analytic strategy

The current study uses latent class analysis (LCA) to identify latent class membership and item-response probabilities by gender (Collins & Lanza 2010; Lanza,

Collins, Lemmon, & Schafer 2007; Lanza & Rhoades 2013). After identifying the latent classes for men and women, the multiple groups LCA is used to explore group differences in class membership probabilities and item-response probabilities by cohort.

The LCA assumes that each identified latent class follows local independence which guarantees independence of items within a class (Lanza et al. 2007). Then, the multiple groups LCA estimates all parameters conditional on group membership. To test the measurement invariance across cohorts meaning that the latent classes are adequately applied to each cohort, I compare G2 difference to a chi-square distribution. For data analysis, the LCA Stata Plugin4 is used in Stata 13 (Lanza, Dziak, Huang, Wagner, &

Collins, 2013). The full information is included in the estimation as missing variables are assumed to be missing at random (MAR). Relative model fits are compared by G2, AIC and BIC. After obtaining the latent classes for young males and females, a multinomial logistic regression model is used to investigate how individual and household characteristics predict the latent classes. It is not recommended to assign individuals to a certain latent class because the identified classes are based on a probability of belonging to each subgroup instead of the true subgroup membership (Lanza and Rhoades 2011).

To account for the probability distributed to a latent class, the multinomial logit model is

4 http://methodology.psu.edu/downloads/lcastata 25 estimated with a weight using a posterior probability for each individual (Vermunt &

Magidson 2004).

Results

Descriptive Findings

Figure 2.1 depicts the proportion of men and women who completed the six life course transitions at each age. For male’s life course transitions, differences are found in family formation by cohort. In both cohorts, the proportion of being in a union, either cohabitation or marriage, and the proportion of childbirth increase with age. By age 25, about 40% of those from the NLSY79 are married while the similar portion of NLSY97 males is in a cohabiting relationship. Cohabitation is not prevalent for NLSY79 respondents; only 15% of the NLSY79 males have experienced cohabitation by age 25.

On the other hand, the retreat from marriage appears as distinct among NLSY97 men as only 20% have married by age 25. While more NLSY97 males report childbearing before age 20, by age25 those from the NLSY79 are more likely to experience childbirth (35.8% and 31.3% of males from the NLSY79 and NLSY97, respectively).

In the bottom panel of the Figure 2.1, the completed life course transitions for females are described by age. Similar to men’s family formation, females from the

NLSY97 are more likely to be in a cohabiting relationship than marriage while the older cohort women are more involved in a marital relationship. More than a half and about

45% of the NLSY79 and NLSY97 females, respectively, gave a birth before age 25.

Regarding educational attainment, about 38% and 22% of females from the older and

26 recent cohorts finish schooling by age 25. Moreover, more females from the NLSY97 participate in full-time work compared to their counterpart from the earlier cohort.

Almost 90% of the NLSY79 females leave the parental home before age 25 as do about

77% of the NLSY97 females. The descriptive findings provide distinctive differences across cohort and gender. However, these findings only account for group averages at each age, and thus group dynamics and the actual pathways of any particular person are not revealed.

Model Selection

Table 2.1 presents the summary of model fit for identifying an optimal baseline model. The results reveal that five latent classes for males and six classes for females would be the most parsimonious since they have lower values of AIC and BIC. For males

(see table 2.2, panel (a)), class 1 can be labeled as “traditional transition, without higher education” because about 80% of those in class1 married and gave a birth, and about

90% and 97% were employed full-time and left the parental home by age 25. But less than 1% of those in class 1 finished any postsecondary education. Class 2 is characterized as “procrastinated transition” since very few individuals (less than 2%) in this class were involved in either marital or cohabiting relationships. And only 14% gave a birth by age

25. A more striking feature of this class is that more than 70% had not left their parental home until age 25 while around 90% of individuals in the other classes made that transition. Moreover, a relatively lower proportion of men in class 2 had been employed full-time by age 25. In class 3, a large share cohabited, slightly more than a quarter

27 married, and more than half had a child by age 25. Most entered the full-time labor force and left the parental home, but very few had finished college. Class 3 is, therefore, named as “contemporary family formation pathway, without higher education.” Those in class 4 reveal very unique patterns of family formation, similar to males in class 2. Less than

10% either cohabited or married while more than 90% had full-time work and left the parental home. Less than 1% of males in class 4 had a child. This group can be referred to as “delayed family formation, self-focused pathway.” Finally, class 5 is comparable to the class1 in that male respondents in both classes reveal similar family formation and home- leaving trajectories. The only difference is in the educational pathway. Most males in class 5 completed at least a college degree while less than 1% of those in class 1 did. The class 5 is, therefore, labeled as “traditional transition, with higher education.”

For females (see table 2.2, panel (b)), six latent classes appear best to describe the transition to adulthood for both cohorts. Women in class 1 are characterized by their high proportions of cohabitation, childbirth, full-time employment, and leaving the parental home. It can be labeled as “contemporary family formation pathways, without higher education” and is comparable to the same class for males (class 3 for males). Similar to males in this class, very few females completed a postsecondary education by age 25 in class 1. In class 2, females are least likely to be in a union, and involved in any adult role acquisition compared to those in other classes. They represent the “procrastinated transition” class. As with the male procrastinated transition class, only 20% of women in this class left home. Women classified in class 3 and 4 are comparable with each other regarding their rare childbirth, employment, education, and home-leaving trajectories. 28

However, those in class 3 seem reluctant to take a traditional family formation path by age 25. Instead, about 34% in this class cohabited. On the other hand, about 34% of those in class 4 married while very few cohabited. These groups are, therefore, defined as

“independent cohabitors and independent married, respectively.” Most females in class 5 experienced cohabitation, were employed full-time, and left their parental home by age

25. They also married and gave a birth, and therefore can be named as “contemporary family formation pathways, with higher education,” which is comparable to those in class

1 who show similar life course paths without the education. Finally, females in class 6 married and gave a birth before age 25, but a small share cohabited. Most of them left their parental home and a comparable proportion were employed. They can be referred to as “traditional pathway.”

Multiple Groups LCA

As different life trajectories are expected across cohorts, the following analyses use the multiple-groups LCA, in which all parameters are conditional on group membership (Lanza et al. 2007). In other words, class membership probabilities are allowed to differ by cohort in the LCA model. As a result of the measurement invariance test, the grouping does not significantly influence the meaning of the latent classes (item- response probability), which thus indicates that the baseline models have equivalent meanings for both NLSY79 and 97.

Figure 2.2 shows that men (panel a) and women (panel b) from the NLSY79 and

NLSY97 are disproportionately distributed to each class. For males, about 31% and 38%

29 of the NLSY79 are classified into ‘class 1: Traditional transition, without higher education (31.4%)’ and ‘class 4: Delayed family formation, self-focused pathway

(37.8%)’ while about 30% and 40% of the NLSY97 males are distributed to ‘class 2:

Procrastinated transition (29.8%)’ and ‘class 3: Contemporary family formation pathway, without higher education (39.9%).’ Few males from the recent cohort are included in ‘class 5: Traditional transition, with higher education (1.1%)’ and few from the earlier cohort are in ‘class 3: Contemporary family formation pathway, without higher education (5.3%).’

For females, the results also reveal very distinct patterns across cohorts. For those from the NLSY79, more than a half are distributed in ‘class 6: Traditional pathway

(53.2%)’ and about 30% in ‘class 4: Independent married (29.0%).’ These two classes are only prominent for the NLSY79 females, and less than 10% belong to other classes.

For the NLSY97 females, on the other hand, the most prevalent class appears as ‘class 1:

Contemporary family formation pathways, without higher education.’ About 50%

(48.8%) of the females from the recent cohort are included in that class, and 15.9% and

17.3% are in ‘class 2: Procrastinated transition’ and ‘class 3: Independent cohabitors’ respectively.

Multinomial Logistic Regressions

The following analysis involves a multinomial logistic regression in which the identified classes for men and women are predicted by race/ethnicity and household characteristics. Table 2.3 reports the coefficients and the relative risk ratios (RRR) from

30 the multinomial models. First, for males (panel (a) in table 2.3), in both cohorts African

Americans are more likely than non-black non-Hispanic whites to be included in Class 2

(procrastinate transition) compared to class 1 (traditional transition without higher education). Moreover, in the NLSY79 only, being African Americans increases the odds of belonging to class 3 (contemporary family formation pathway without higher education) and class 5 (traditional transition with higher education). Hispanics from the

NLSY79 are more likely to be included in classes 3, 4 and 5 with extremely high odds of belonging to class 3 (contemporary family formation pathway without higher education).

Yet, the NLSY97 Hispanics are less likely to be distributed to class 4 (delayed family formation, self-focused pathway). Having lived in an intact family increases the odds of being in class 5 (traditional transition with higher education) in both the NLSY79 and 97.

Moreover, in the NLSY79 the intact family structure increases the odds of belonging to class 2 (procrastinated transition) and class 4 (delayed family formation, self-focused pathway) but the odds being in class 3 (contemporary family formation pathway without higher education) decrease for the NLSY97 males. For both cohorts, increasing mother’s educational attainment increases the odds of being in class 4 (delayed family formation, self-focused) and class 5 (traditional transition with higher education). Only for the

NLSY97 men, the positive relationship persists for the class 2 (procrastinated transition).

Second, the panel (b) in Table 2.3 shows results from the multinomial model of females from both cohorts. In the NLSY79, African Americans are more likely to be distributed to class 2 (procrastinated transition) compared to class 6 (traditional pathway), but negative relationships appear with class 4 (independent married) and class 5 31

(contemporary family formation pathways with higher education). Female Hispanics in the NLSY79 are more likely to be in class 4 (independent married) versus class 6

(traditional pathway). Having lived in an intact family decreases the odds of belonging to class 1 (contemporary family formation pathways without higher education) in the

NLSY97 but increases the odds in class 2 (procrastinated transition) and class 4

(independent married) for the NLSY79 women. Finally, higher mother’s educational attainment increases the odds of being classified in class 4 (independent married) for both cohorts. For only the NLSY97 females, the positive relationship remains in class 3

(independent cohabitors) and class 5 (contemporary family formation pathways with higher education).

Discussion

Using comparable longitudinal data from the NLSY79 and 97, the current study identified distinct patterns in the transition to adulthood by gender and compared them across cohorts. By encompassing six major life course transitions and using the recent cohort young adults, this study adds empirical evidence to the existing literature. Since the current study had four research goals, each point will be discussed in turn. First, to illustrate life experiences during the transition to adulthood, age-specific life course transitions are compared by cohort. Several distinct patterns are found in family formation and home leaving. Consistent with previous literature (Schoen, Landale, &

Daniels 2007), cohabitation is prevalent among young adults in the NLSY97 (both men and women) while few are involved in a marital relationship by age 25, compared to that

32 of the NLSY79. Despite the changes, few changes are found in childbirth for both men and women, which drives the increasing births out of wedlock and within a cohabiting relationship (Chandra et al. 2005; Copen et al. 2012). In addition, changes in home- leaving patterns are pronounced as both men and women in the recent cohort are less likely than their earlier cohort counterparts to leave home, and if exiting home, they do so at much slower rates. In contrast, completing schooling and full-time employment do not differ by cohort as much as the union formation and home-leaving patterns. Family formation often requires a certain level of commitment to others (spouse or partner, and child) and individuals are expected to have responsibilities upon leaving home. On the other hand, employment and education are considered a process of acquiring human capital (Sum et al. 2011) which is an important factor that is necessary to achieve autonomy in life. Therefore, these descriptive findings suggest that compared to an earlier cohort, contemporary young adults spend more time in self-exploration.

Second, using latent class analysis, this study aimed to detect general patterns of the transition to adulthood for men and women. Five latent classes and six latent classes appeared optimal for men and women, respectively, and each latent class reveals unique trajectories of individuals. There were comparable classes corresponding to both males and females: contemporary family formation pathway without higher education and procrastinated transition. Individuals in the class of contemporary family formation without higher education are distinguished by their high rates of completing family formation (particularly cohabitation), engaging in full-time work, and exiting the parental home. However, this group reveals very low rates of completing postsecondary 33 education. For both men and women, the procrastinated transition class is identified, which mostly occurs in the NLSY97 cohort and is therefore evidence of the delayed transition to adulthood. Other than those two similar classes, the LCA findings show how the transition to adulthood differs by gender. For males, a self-focused group is found for those who delay family formation while focusing on themselves through education, employment and independent living. The comparable self-focused class for females is independent cohabitors and independent marrieds. The females in these two classes hold similar trajectories in childbearing, education, employment, and residence status but are involved in either cohabiting or marital relationships. In other words, females tend to be in a union during young adulthood. Previous studies show that females who are in a union often sacrifice their career prospects for a spouse or partner’s life course pathways

(e.g. family migration Boyle, Feng, & Gayle 2009; Quinn & Rubb 2011). Because the transition to adulthood is an important developmental stage with regard to augmenting human capital and forming a union, these two patterns might limit female’s future life trajectories. Future research is needed to examine how the life trajectories in career and union formation differ by gender after the early adulthood.

Third, the current study compares the transition to adulthood between the

NLSY79 and NLSY97 to comprehend the dynamics of young adulthood across cohorts.

The findings suggest that individuals from the NLSY79 and 97 are disproportionately distributed to each class of the baseline model (five and six classes for men and women).

For males, a large share of those from the NLSY79 classified into delayed family formation (lower probabilities of family formation with high rates of full-time 34 employment and home leaving) or traditional transition without higher education

(married, had a child, and high rates of full-time employment and home leaving). On the other hand, men from the NLSY97 mostly belong to classes of contemporary family formation paths without higher education, procrastinated transition, or delayed family formation. These indicate that young men in the earlier cohort are divided by either taking traditional pathways or delaying family formation and focusing on accumulating human capital. Yet, very few (less than 8%) from the recent cohort reveal the traditional pathway of family formation, employment, and home leaving. Young men in the recent cohort are mostly placed in either contemporary family formation paths which involves cohabitation or delaying family formation. With regard to female’s transition to adulthood, more than half of the earlier cohort females follow a traditional pathway and about 30% are in the independent married class. In other words, females in the earlier cohort either take the traditional transition to adulthood or gain independence by employment, education, and residence while being married. On the other hand, the recent cohort young women reveal a variety of paths although most are placed in contemporary family formation pathways without higher education (about 50%). Considerable proportions of females are distributed in the procrastinated transition or independent cohabitors classes. The findings suggest a de-standardization of the transition to adulthood in recent years as young men and women follow less traditional but more complicated pathways compared to those from the former cohort.

Finally, since changes in the transition to adulthood have become diverse by social groups (Widmer & Rischard 2009), this study tests how the distribution is relevant 35 to race/ethnicity and household characteristics using a multinomial logistic regression model. The findings suggest that male African Americans and Hispanics from the

NLSY79 are more likely to follow the contemporary family formation trajectory without higher education compared to the traditional pathway. In addition, these two race/ethnic groups in the NLSY79, compared to whites, are more likely to be in the traditional transition with higher education when compared to the traditional transition without higher education. The significant relationships, however, disappear among men from the

NLSY97 except that blacks are more likely to be in the procrastinated class and

Hispanics are less likely to be in the delayed family formation self-focused class. For females, African Americans in the NLSY79 are more likely to be classified in the procrastinated transition class but less likely to be in the independent married or contemporary family formation with higher education classes when compared to traditional pathways. Female Hispanics from the earlier cohort are more likely to be in the independent married class, but the significance disappears for those from the recent cohort. Taken together, these findings reveal that race/ethnicity effects on differentiating latent classes have been weakened across cohorts. In other words, the changes in the transition to adulthood in the NLSY97 cohort occur to most young adults, regardless of race/ethnicity.

The effects of household characteristics have remained significant across cohorts in different ways. For example, males from an intact family are more likely to be in the procrastinated transition, the delayed family formation self-focused pathway, or traditional pathways with higher education classes in the NLSY79. Yet, their counterparts 36 in the NLSY97 are significantly less likely to be in contemporary family formation without higher education but more likely to be in the traditional family formation with higher education class. Similar changes are found among females: those from an intact family are less likely to be in the contemporary family formation without higher education class in the recent cohort. Regarding mother’s education, increasing mother’s educational attainment helps young adults in both earlier and recent cohorts have higher education and self-exploration (i.e. delayed family formation self-focused pathway and traditional transition with higher education for males, and independent cohabitors or married, and contemporary family formation with higher education for females).

Although the household indicators do not directly measure family resources, the findings suggest that young adults, regardless of their birth cohort or gender, benefit from household resources during the transition to adulthood. Those with greater maternal educational attainment and from an intact family are more likely to belong to either traditional pathways with higher education or self-focused classes.

Despite the significant implications, the current study has several limitations.

First, six life course events included in this study may have presented incompatible meanings for young adults in the NLSY79 and 97. For example, the premium of AA or

BA degrees in the labor force has changed because college attendance in the 1980s was not as prevalent as that in the 2000s; college enrollment rates of high school graduates have increased since 1980s (Bound, Lovenheim, & Turner 2009). Moreover, it is possible that a larger proportion of childbirth among NLSY97 respondents are out of wedlock compared to that of the NLSY79. Future research that can capture the underlying and 37 changing meaning would contribute to better understanding of change of the transition to adulthood over time. Second, though latent class analysis is well suited to detect underlying relationships between various life course transitions, the findings do not take into account sequences of the transitions. Individuals distributed to the same class, however, may have experienced diverse patterns with regard to the order and sequences of family formation, education, employment, and home-leaving. Additional research on the sequences is needed to advance our knowledge about the variability and de- standardization of the transition to adulthood. Third, the current study does not reveal any particular relationship between the six life course transitions. Life course events during the transition to adulthood are, however, interrelated to each other; disregarding the complex associations may leave behind an important piece of the changing transition to adulthood. The next study, therefore, examines the association between life course events; particular research attention is given to geographic mobility that has been relatively understudied.

Chapter 2 tables and figures

NLSY79 NLSY97

(a) Male completing transition, by age

NLSY79 NLSY97

(b) Female completing transition, by age Figure 2.1 Completing life course transitions, by gender and cohort

Table 2.1 Model Fit Information used in selecting the LCA model

Classes G2 df AIC BIC CAIC Entropy Male 1 4481.9 57 4493.9 4536.8 4542.8 1.00 2 1741.4 50 1767.4 1860.3 1873.3 .58 3 729.4 43 769.4 912.5 932.5 .77 4 122.6 36 176.6 369.7 396.7 .61 5 88.1 29 156.1 399.3 433.3 .65 6 74.4 22 156.4 449.7 490.7 .62 Female 1 4663.3 57 4675.3 4718.1 4724.1 1.00 2 2099.5 50 2125.5 2218.3 2231.3 .66 3 830.4 43 870.4 1013.2 1033.2 .74 4 213.2 36 267.2 460.0 487.0 .68 5 142.9 29 210.9 453.6 487.6 .63 6 78.0 22 160.0 452.7 493.7 .65 7 36.4 15 132.4 475.1 523.1 .71

Table 2.2 Item-response probabilities for men (a) and women (b) (a) 5 Classes Model for Males

Latent Class Class1 Class2 Class3 Class4 Class5 Traditional Procrastinated Contemporary Delayed Traditional transition, transition family family transition, without higher formation formation, with higher education pathway, self-focused education without higher pathway education NLSY79 31.40% 13.10% 5.30% 37.80% 12.40% NLSY97 6.40% 29.80% 39.90% 22.70% 1.10% Cohabitation 0.17 0.00 0.90 0.10 0.19 Marriage 0.78 0.02 0.28 0.10 0.77 Birth 0.76 0.14 0.56 0.00 0.53 Fulltime emp. 0.89 0.74 0.95 0.93 0.95 Finishing school 0.00 0.07 0.04 0.49 0.95 Home leaving 0.97 0.28 0.86 0.90 0.98

(b) 6 Classes Model for Females

Latent Class Class1 Class2 Class3 Class4 Class5 Class6 Contemporar Procrastinated Independen Independen Contemporar Tradition y family transition t cohabitors t married y family al formation formation pathways pathways, pathways, without with higher higher education education NLSY79 1.7% 7.7% 0.1% 29.0% 8.4% 53.2% NLSY97 48.8% 15.9% 17.3% 6.1% 7.9% 3.9% Cohabitation 0.76 0.03 0.34 0.00 0.97 0.16 Marriage 0.41 0.02 0.04 0.34 0.42 0.79 Birth 0.77 0.22 0.00 0.05 0.21 0.85 Fulltime 0.92 0.58 0.95 0.96 0.98 0.73 emp. Finishing 0.01 0.11 0.56 0.73 0.54 0.21 school Home- 0.90 0.19 0.74 0.87 1.00 0.98 leaving

(a) Class membership probabilities for males

(b) Class membership probabilities for females

Figure 2.2 Class membership probabilities in each latent class for men(a) and women(b) 42

Table 2.3 Multinomial logistic regressions for predicting latent class membership

(a) Males Comparison Class: Traditional transition without higher education (Class1) Class2 Class3 Class4 Class5 [procrastinated transition] [contemporary family formation [delayed family formation, self- [traditional transition with without higher education] focused pathway] higher education] B (SE) RRR B (SE) RRR B (SE) RRR B (SE) RRR NLSY79 Black .80*** (.14) 2.23 1.93*** (.28) 6.86 .08 (.09) 1.08 .36** (.13) 1.44 Hispanics -.23 (.21) .79 1.37*** (.35) 3.92 .27* (.12) 1.31 .56*** (.16) 1.75 Intact family .72*** (.13) 2.06 -.25 (.20) .78 .30*** (.08) 1.35 .41*** (.11) 1.50 Momedu -.02 (.02) .98 .01 (.03) 1.01 .17*** (.01) 1.18 .20*** (.02) 1.23 NLSY97 Black .70** (.24) 2.01 .27 (.23) 1.31 -.10 (.24) .90 -.05 (.78) .95 Hispanics -.00 (.22) 1.00 -.33 (.22) .72 -.52* (.22) .60 .02 (.67) 1.02 Intact family .32 (.17) 1.38 -.47** (.17) .63 .18 (.17) 1.20 1.79* (.81) 6.01 43 Momedu .11*** (.03) 1.12 .05 (.03) 1.05 .26*** (.03) 1.29 .27*** (.07) 1.30 *p<.05, ** p<.01, *** p<.001 (continued)

Table 2.3 (continued)

(b) Females Comparison Class: Traditional pathway (Class6) Class1 Class2 Class3 Class4 Class5 [contemporary family [procrastinated transition] [independent cohabitors] [independent married] [contemporary family formation without higher formation with higher education] education] B (SE) RRR B (SE) RRR B (SE) RRR B (SE) RRR B (SE) RRR NLSY79 Black - - - 1.02*** (.15) 2.76 - - - -.20* (.09) .82 -.81*** (.15) .44 Hispanics - - - .40 (.22) 1.49 - - - .22* (.11) 1.24 -.33 (.19) .72 Intact family - - - .90*** (.14) 2.45 - - - .75*** (.08) 2.11 -.16 (.12) .85 Momedu - - - -.02 (.02) .98 - - - .22*** (.02) 1.25 .21 (.02) 1.23 NLSY97 Black .49 (.38) 1.63 .90* (.38) 2.46 .26 (.38) 1.30 -.90 (.53) .41 -.18 (.46) .84 Hispanics -.12 (.30) .88 .02 (.32) 1.02 -.40 (.31) .67 -.49 (.45) .61 -.32 (.40) .73 Intact family -1.26*** (.29) .28 -.41 (.30) .67 -.30 (.30) .74 .70 (.42) 2.01 -.55 (.35) .60 44 Momedu -.10 (.06) .90 -.04 (.06) .96 .15* (.06) 1.16 .19** (.07) 1.21 .25*** (.07) 1.28 *p<.05, ** p<.01, *** p<.001

CHAPTER 3: MOVING AND UNION FORMATION: A COHORT COMPARISON OF THE ASSOCIATION BETWEEN LIFE COURSE EVENTS

Distinct patterns appear in the transition to adulthood across cohorts in Chapter 2.

For example, delays in the timing of life course events and the prevalence of cohabitation as a first family formation type are a pattern observed among the recent cohort of young adults (Manning, Brown, & Payne 2013; Schoen, Landale, & Daniels 2007). These changes have contributed to the de-standardization of the life course that is more complicated in recent young adult lives. According to the life course perspective, various life transitions are linked to each other such that decisions about one life event have implications for other events (Elder 1998). In addition, prior studies find that interrelated life courses have subsequent influences on later life trajectories, e.g. effect of early marriage and school dropout on later life poverty status (Dahl 2010; Mouw 2005). Few studies, however, have considered geographic mobility and its implications for other life transitions of young adults. Moving is a common life event that individuals experience multiple times across the life course. Moreover, it is a significant life course event that often alters one’s socioeconomic context and can provide new opportunities or presents new constraints (Clark 2013; Elder, King, & Conger 1996; Sharkey 2012). Particularly for young adults, residential change is a learning process that promotes independence and autonomy (Garasky, Haurin, & Haurin 2001; Goldscheider & Goldscheider 1999; Mulder

& Clark, 2002). After leaving the parental home, young adults begin to manage budgets and life decisions independently. Although young adults may receive a fair amount of support from family and later return home, moving expedites the entry into adulthood by providing at least a semi-independent period. It is therefore important to understand moving trajectories of young adults over the life course. Prior studies on moving as a life course event, however, have not fully provided the information because they have incorporated static measures of mobility such as ever moved (Jampaklay 2006) or were restricted to only married or cohabiting couples (Boyle, Kulu, Cooke, Gayle, & Mulder

2008; Clark & Withers 2007; Jacobsen & Levin 1997). Furthermore, no study has exclusively focused on young adults who undergo many life course transitions in a short time.

This chapter, therefore, examines moving experiences in the transition to adulthood and their association with other life course transitions. Among various life transitions that young adults encounter, particular attention is given to union formation in this study since tremendous changes have occurred in the timing and type of family formation in recent years (Schoen, Landale, & Daniels 2007). Findings from this study will add to existing knowledge on the importance of geographic mobility in the life course as follows. First, using longitudinal information from nationally representative samples, the current study analyzes the association between moving and union formation in relation to various life course transitions such as educational and occupational changes.

Studies on migration often use cross-sectional data which does not account for how diverse factors interplay with the association between union formation and mobility 46

(Geist & McManus 2008). Second, this study examines young adulthood from two comparable cohorts which contain contemporary young adults in the United States (i.e. those born in 1980-84). Despite growing variability in the young adult life course recently, no study has compared the implications of moving for other life transitions by cohort. Finally, with regard to residential moves, prior studies find that different types of moves have distinct motivations and implications for life course transitions (Schachter

2001). Thus, this study disaggregates moving events in great detail by distance moved and economic conditions in origin and destination places.

Migration as a Life Course Event

The life course perspective emphasizes the timing and sequence of multiple interrelated events (Elder 1998). Including migration in the life course paradigm has been useful since residential moves are linked to other important life changes (Clark & Withers

2007; Clark 2013). For example, changes in family size and marital status often trigger residential moves (Kulu & Milewski 2007; Michielin & Mulder 2008), and individuals consider various aspects of residence and how residence impacts quality of life such as the quality of neighborhoods and school districts for families with young children or climate for retirees (Clark & Huang 2003). Moreover, a residential move motivates other life changes (Geist & McManus 2008). Movers are more likely than stayers to experience changes in marital status and household composition in the year following a move, even after controlling for age effects (Geist & McManus 2008). In addition, mobility influences

47 income and poverty status, which either improves or jeopardizes movers’ economic circumstances (Geist & McManus 2008).

With regard to mobility, the distance moved has implications for the outcome of a move (Boyle et al. 2008; Cadwallader, 1992; Clark 2013; Clark & Withers 2007; Geist &

McManus 2008). Although moving trajectories are much more complicated with higher order moves (Clark & Withers 2007), long-distance moves in general involve greater mental and physical stress for movers (Cadwallader 1992; Magdol 2002). And the mental and physical stress of moving could discourage other life course transitions such as courting or dating partners. Boyle et al. (2008) found that two or more moves increase the likelihood of union dissolution for Austrian couples, especially if they are long-distance moves. It is, however, also possible that a long-distance move is a positive experience if it provides better economic opportunities or just places movers in a new environment. For example, moving outside the home village is positively related to first marriage in

Thailand as it expands mover’s marriage markets (Jampaklay 2006). Moreover, a first long-distance move appears to bring excitement for movers and enhance a couple’s relationship (Boyle et al. 2008). The distance of a move is further linked to life course trajectories since long distance moves may alter the socioeconomic context in which a mover is embedded, and thus is more likely to impact individual’s life course (Elder 1998;

Sharkey 2012). However, a short-distance move hardly alters mover’s surroundings with regard to socioeconomic and family context, and is thus less likely to influence other life course transitions.

According to the U.S. Census Bureau, about 12 to 18% of the U.S. population has changed residence every year during the last three decades (U.S. Census Bureau, 2012).

While individuals move for various reasons and the patterns vary by developmental stages, young people aged 15 to 29 are the most frequent movers (Clark 2013; McAuley & Nutty

1982). Most moves of young adults coincided with marriage in the past (Long 1973), whereas for contemporary young adults moving may be related to other life choices exploring possibilities in partnership, work, and education. In other words, as the entry into adulthood has delayed, the close association between moving and family formation may change.

Union Formation in the transition to adulthood

The transition to adulthood has extended and now includes more time in school, delays in family formation and entry into the labor force (Furstenberg, Rumbaut, &

Settersten 2005; Rindfuss 1991). The most remarkable change in young adulthood is increasing variation in the timing of family formation and composition of the families formed (Schoen, Landale, & Daniels 2007). The number of children born to unmarried parents has increased from about 30% in 1980 to 48% in 2010 (U.S. National Center for

Health Statistics 2011) and marriage has been delayed until the late 20s in recent years

(ages 28.9 and 26.9 for men and women in 2011, US Census Bureau 2011). These demographic shifts parallel growing family diversity and the prevalence of nonmarital cohabitation, particularly among young adults. Moreover, the changes reflect a transformation of young adulthood into an intensive self-focused period as part of identity

49 exploration (Arnett 2004; Arnett et al. 2011). Cohabitation represents an exploration of intimate relationships for young adults because cohabitation in general requires a lower level of commitment and responsibility than marriage (Cherlin 2010; Smock 2000). For example, Schoen, Landale and Daniels (2007) find that fewer young cohabiting couples have children together or transition to marriage, which implies that cohabitation is more of an alternative to singlehood rather than a substitute for or antecedent to marriage for this group.

The association between moving and union formation can be addressed by several theoretical approaches. Migration theories suggest that moving is an investment for those who are rational and have an ability to calculate the costs and benefits of mobility

(DaVanzo 1983; Massey et al. 1993; Tienda & Wilson 1992). Therefore, moving likely improves one’s human capital (Massey et al. 1993), which could influence the timing and type of union formation; for example, individual’s economic status promotes marriage among singles and cohabiting couples (Carlson, McLanahan, & England 2004;

Oppenheimer 1988, 2003; Smock & Manning 1997). Cohabitation is considered a transitory state for those with unstable economic positions (Oppenheimer 2003) while marriage is desired by most cohabitors once they establish economic stability (Cherlin

2004; Oppenheimer 1988). Therefore, positive gains from moving (e.g. higher earnings) can increase the likelihood of union formation, especially marriage. It is also possible that mobility alters socioeconomic conditions in which an individual resides, and thus reflects variability of potential mates in a local marriage market (Lewis & Oppenheimer 2000;

Lichter, McLaughlin, Kephart, & Landry 1992; Lichter et al. 1995). Mate seekers search 50 for partners within a local marriage market, considering those who fall in the scope of acceptability as a potential partner (Lichter et al. 1995). The quality of available partners in a local area is thus important to the mate searching process (Lichter et al. 1992). However, if the area fails to provide a sufficient pool, people expand their markets by adjusting the criteria for potential mates (Lichter et al. 1995; Qian & Lichter 2011). In such a situation, people are assumed to basically ‘cast a wide net’ while staying in the same marriage market. Yet, due to constrained market conditions people may instead move to another market, similar to job seekers who move from low to high wage places in order to maximize their income (Massey et al. 1993). If it is the case, moving to a better marriage market (e.g. economic contexts) should increase the likelihood of union formation.

Despite the deliberate process of mate selection, it is plausible that young adults are not necessary rational, which would be consistent with their developmental stage and emphasis on exploration (Arnett 2004; Arnett et al. 2011). For young adults, particularly those in recent years, identity formation during adolescence has extended into the 20s

(Arnett 2004). Moreover, the primary involvement of young adults begins to expand to peer, romantic partners, and work colleagues (Tanner 2006). Expanding social relationships can reshape worldviews and life goals based on new resources and opportunities (Freund & Baltes 2002). In line with these processes, moving may be one of the life events that young adults explore as part of identity formation or opportunity expansion. It may then become less important whether a move improves young adults’ marriage markets or socioeconomic prospects; instead, moving itself functions as a means to explore other life courses. 51

Research Questions

This study aims to answer the following research questions:

1) What are the patterns of union formation and mobility among young adults in the U.S.? 2) How are moving events related to first union formation? 3) How do different types of mobility predict the timing and type of first union formation? 4) How does the association between moving and union formation differ by cohort?

Data and Measures

Panel data of young adults from the NLSY79 and 97 are used in the current study.

Both data sets limit the period of observation from ages 16 to 30. Data are transformed into person-year files for the NLSY79, and person-month files for the NLSY97 to estimate an event history model. Since the unit of measurement differs between the

NLSY79 and 97, separate models for each cohort are operated while including the same set of covariates.

The outcome variable in this study is union formation. Union formation is measured by either first cohabitation or marriage without prior cohabitation. Although individual’s dating status is important to the decision of first union formation, it is not possible to trace the previous relationship status in the NLSY. For example, the NLSY97 asked all respondents whether they have ever been on a date or social outing, how often the respondents have dated and the number of different people the respondent has gone out with in the past year (wave 1) or since the last interview (all subsequent waves)

(Center for Human Resources 2013). These questions, however, do not correspond to any 52 particular person and thus make it impossible to match the information to the first cohabiting or marital partner. Moreover, the dating histories are yearly measures which eliminate the possibility of multiple partners within a year. Thus, among those with no prior union experience, dating history before the date of first cohabitation or marriage is not available.

Respondent’s moving experiences are created using both the public and geocode files. The exact dates of migration are not available in the NLSY79 prior to 2000, and thus the county and state FIPS codes (Federal Information Processing Standard) and the actual distance of a move are drawn from the geocode file. A long-distance move is measured by changes in the county of residence between the annual survey interviews, and is referred to as a migration event. A short-distance move, on the other hand, is equivalent to a within-county move whose FIPS codes are not changed but has some distance moved. This type of move is called residential mobility. The mobility measures are inferred indirectly and available only yearly in the NLSY79 and thus, the chronology of moves and union formation that occur in the same year is uncertain. To be certain of the relationship between mobility and union formation, each event is lagged one year with respect to another. To account for contextual changes after moving, migration events are further categorized by economic conditions in origin and destination places. Using the

Local Area Unemployment Statistics (LAUS) of the Current Population Survey (CPS)

(Bureau of Labor Statistics 2013), the FIPS codes of the origin and new places are matched to county unemployment rates. From this procedure, migration events are sorted into three categories: moving to another county with lower unemployment rates, moving 53 to another county with higher unemployment rates, and moving to another county with the same employment rates (see Table 3.2). International migration, moving from or to a foreign country, is excluded from the analysis because it is not possible to assign employment conditions to the moves.

The NLSY97, on the other hand, provides detailed information on respondent’s migration history, classifying moving into 4 categories in the public file; move within the same county, move within the same state but different county, move between states, and move to or from a foreign country (Center for Human Resources 2013). A within-county move is considered a short distance move called residential mobility. A long distance move is defined as a between-county move including those who change their residence to different county either within the same state or in different states, and is considered a migration event. Using the geocode file, I match the origin and destination of each move to the FIPS codes. Then, as the NLSY79, migration experiences are elaborated by economic conditions of origin and destination places in the NLSY97. The Local Area

Unemployment Statistics (LAUS) of the CPS is again used (Bureau of Labor Statistics

2013) (see Table 3.2). Again, international migration is dropped from the analysis. In the

NLSY97, both moving and union formation events are measured monthly, and it is not possible to determine which occurs first in a given month. To be certain about the timing and to examine the possibility of a time gap between the two events, all migration and residential mobility variables are lagged 6 months in the analyses. It is possible that the discrepancy in measurement units in NLSY79 and 97 influences the estimation of relationships between moving and union formation. Table 3.3 shows that yearly variables 54 underestimate the moving incidents in the NLSY97 when it compares to those with monthly measurements. I will return to this point in the discussion.

The association between union formation and mobility of young adults from the

NLSY79 and 97 is predicted with the same set of individual and household characteristics. Individual characteristics include respondent’s gender, race/ethnicity, educational attainment, employment status, and childbearing experiences. Race/ethnicity has two categories; Black and Hispanic, with non-Black non-Hispanic Whites and others being a reference category. Education is measured by several time-varying covariates; being enrolled in school, earning a high school diploma (or GED), and a postsecondary degree each year for the NLSY79 or each month for the NLSY97. The postsecondary education is distinguished by the institutional type (i.e., two-year colleges, an Associate

Degree; four-year colleges, a bachelor degree) because different life trajectories are expected by the school type (Bozick & DeLuca 2005). Employment status is measured using data from the employment status history files in both datasets. For the NLSY79, those who work on average more than 39 hours per week for more than or equivalent to

50 weeks are coded as being employed full-time, those who work less than 39 hours per week for less than 50 weeks are defined as being employed part-time and others are considered being unemployed. For the NLSY97, individuals working more than 39 hours per week in a month are defined as full-time employees while those who work less than

39 hours per week in a month are part-time employees. All others are considered unemployed. In both cohorts, a time-varying childbearing variable is included (i.e. the year and month of childbirth). Furthermore, household characteristics are included in the 55 model as control variables. These include mother’s educational attainment which is a continuous variable representing the highest grade completed and a dummy variable for whether the respondent lived with both biological parents during childhood. All time- varying variables are lagged one year and one month for the NLSY79 and 97, respectively, to ensure that the explanatory events occur prior to union formation.

With regard to contextual influences, the analyses include county unemployment rates in the residence at the first interview and a time-varying measure of residence of nonmetropolitan and metropolitan areas. Due to changes in the standard measuring metropolitan and nonmetropolitan statistical areas in the NLSY5, 2003 Urban Influence

Codes from USDA ERS (the United States Department of Agriculture Economic

Research Service) were merged with both NLSY data. According to the US Census

Bureau, about 16.6% of the U.S. population lives in nonmetropolitan areas (USDA

2007); in my sample, about 20.8% and 23.6% of the NLSY79 and NLSY97, respectively, lived in nonmetropolitan areas in the first survey year.

5For variables of classification in nonmetropolitan and metropolitan areas, the NLSY79 use the 1973 City Reference File (CRF) in 1979 through 1982, the 1982 CRF for 1983 variables, the 1983 CRF for 1984 through 1987, the 1987 CRF for 1988 through 1992, the 1992 CRF for 1993 through 1998, and a slight different calculation process from 2000 to 2006 (Center for Human Resources, NLSY79 Codebook Supplement, Appendix 6 2013). In the NLSY97, the MSA code scheme from the 1994 County and City Data Book is used in rounds 1-7, Core-Based Statistical Areas (CBSA) statistical geographic entities are used from round 8 (Center for Human Resources, NLSY97 Geocode Codebook Supplement Attachment 101), A Due to the change, some respondents can appear to move from metropolitan to nonmetropolitan areas though they have not changed their residence. 56

Analysis

The relationship between mobility and the outcome, first union formation, is examined using a Cox proportional hazards model. Separate analyses are performed by cohort and each cohort involves 2 models; first, the first union regardless of its type is predicated in relation to all control variables and detailed moving events. Following the analysis, a Cox proportional hazards competing risks model is used to present first union formation type (i.e., marriage or cohabitation) as competing risks (Box-Steffensmeier &

Jones 2004). In this study, the onset of risk of union formation is set to age16. About

2.8% (272 individuals) and 2.0% (181 respondents) of the NLSY79 and 97 are married or cohabited before age16 and are excluded from the analyses. The final sample includes

9,491 and 8,803 individuals who contribute 60,045 person years and 799,377 person months to analyses of the NLSY79 and NLSY97, respectively. The Breslow method of the Cox proportional hazards model is used to take into account the tied cases in the dataset (Box-Steffensmeier & Jones 2004). The method assumes that the tied cases occur sequentially, and is therefore adequate for a model with few tied cases. All analyses are conducted using the survey setting command in Stata to account for the complicated sampling strategy of the NLSY (Cleves et al. 2010; Center for Human Resources 2013).

Results

In Tables 3-1 and 3-2, descriptive statistics are reported by cohort and first union formation type i.e. no union, marriage, and cohabitation. Results from the Cox models of the timing of first union formation are shown in Tables 3-3 and 3-4. The findings reveal

57 that for young adults in the recent cohort, union formation has been delayed and cohabitation is substituted for marriage as a common first union. Table 3.1 shows that approximately 28% of the NLSY97 sample remains single while only 8% do so in the

NLSY79. For the NLSY79 cohort, 62% married and 30% cohabited by age 30, whereas in the NLSY97 about 57% experienced cohabitation as a first union type and only 14% married without premarital cohabitation. Young adults in the NLSY97 on average cohabit earlier than those from the NLSY79 cohort, which results in later entry into union formation for the NLSY79 cohort. The mean age at the first union is 23.2 and 22.1 for those from the NLSY79 and 97, respectively.

In Table 3.1, demographic and household characteristics appear to differ by first union formation type. A higher percent of females are in a union in both cohorts. More than half of those who marry are women in both the NLSY79 and 97, and 47% and 52% of the NLSY79 and 97 cohabitors are females. A larger share of African Americans remains single while non-Hispanic non-Black whites and others tend to be in a union in both cohorts. During young adulthood, a large share of those who marry or cohabit in both cohorts are enrolled in any type of school compared to their counterparts who remain single. Moreover, about 60% of cohabitors from the NLSY79 complete high school while more individuals with a degree marry in the NLSY97. A larger proportion of singles in both NLSY79 and 97 have completed college compared to cohabitors or married. With regard to employment status, about half of cohabitors from the NLSY79 are employed part-time, as are about 40% of those who remain single and marry. For the

NLSY97, a larger share of those who marry are employed part-time compared to singles 58 and cohabitors. In both cohorts, a large share of married and singles are employed full- time. In addition, no difference is found in childbearing experiences by union type in the

NLSY97 but singles in the NLSY79 are less likely to have a child. Household characteristics also appear different by first union type. For the NLSY79 and 97, about

64% and 53% had lived with both biological parents during childhood. Yet, less than

60% and 50% of cohabitors from each cohort are from an intact family, which suggests that more individuals who cohabit are from other family structures. On average, mother’s educational attainment is higher for those from the NLSY97 than those from the

NLSY79. Maternal educational attainment of cohabitors is higher than that of singles or married in the NLSY79 but in the NLSY97, it appears higher for singles than that of married or cohabitors. Significant differences are also found in residence characteristics by cohort. About 80% of respondents from both cohorts lived in metropolitan areas in the first interview year. A larger share of the NLSY79 singles and cohabitors lives in the metro areas while more singles do in the NLSY97. Economic conditions in county of residence differ by first union type in the NLSY79 while no difference is found in the

NLSY97. For those from the earlier cohort, unemployment rates of county of residence for singles are lower than those of their counterparts who marry or cohabit.

Table 3.2 reports individual’s mobility patterns aggregated by first union formation type. About 23% and 27% of the NLSY79 and 97, respectively, have changed their residence at least once, and young adults in the earlier cohort are more likely to move within the same county while those in the recent cohort are more likely to move to another county. About 12% and 7% of the NLSY79 and 97 samples change their 59 residence within the same county and approximately 11% and 20% move to another county. The first move occurs on average at ages 22.5 and 20.7 for the NLSY79 and 97, respectively. For young adults from the NLSY79, migration events occur later than residential mobility while their counterparts from the NLSY97 migrate to another county earlier than residential mobility. Comparing the moving experiences by union status, distinct patterns appear across cohorts. For the NLSY79 young adults, cohabitors are significantly more likely to move either within the same county or across counties. On the other hand, singles from the NLSY97 are significantly more likely to change their residence. In the NLSY97, about 41% of singles have ever moved compared to approximately 20% of those who marry or cohabit. Singles from both cohorts, however, move significantly later than those who marry and cohabit. When considering the socioeconomic conditions of moving experiences, I find that more young adults from both cohorts migrate to better economic circumstances. About 55% and 60% of migration in the NLSY79 and 97 (i.e. 6% and 12% of the entire samples) represent county changes from high to low unemployment rates. For the NLSY79 respondents, a larger share of cohabitors move to new counties with higher unemployment rates although no significant difference is found between migration to counties with lower unemployment rates and the same level of unemployment. For NLSY97 movers, however, a large share of singles changes their residence to a new county regardless of its unemployment rates.

Table 3.3 presents the coefficients from the Cox models predicting the timing of first union formation. For the NLSY79, migration to another county with the same unemployment rates decreases the likelihood of union formation by more than 60%. For 60 the NLSY97 young adults, on the other hand, residential mobility (i.e. short distance move within the same county) and migration to counties with higher unemployment rates increases the hazard of union formation. Other factors show that in both cohorts, females are more likely to be involved in union formation while African Americans are less likely to establish any romantic relationship. Hispanics in the NLSY79 also reveal a lower hazard of union formation. The NLSY79 respondents who are enrolled in school are less likely to be involved in union formation while those who completed a bachelor degree in the NLSY97 are more likely to be in any union. Moreover, in both cohorts, full-time employment significantly increases the likelihood of union formation (hazard ratios: 1.59 and 1.33 for the NLSY79 and 97). However, only in the NLSY79 does part-time employment increase the hazard of union formation. In both cohorts, childbearing significantly increases the hazard of union formation. Similar patterns appear across cohorts in household characteristics. Living in an intact family decreases the likelihood of union formation by about 20% and 40% in the NLSY79 and 97, respectively, and increasing mother’s educational attainment decreases the hazard of union formation in both cohorts. Finally, living in metropolitan areas decreases the likelihood of union formation only for the NLSY97 young adults.

Findings from the Table 3.4 reveal that the associations between mobility and first union formation significantly differ by first union type. First, in the NLSY79, migration to another county with the same unemployment rates decreases the likelihood of marriage versus remaining single while moving to new counties with lower unemployment rates increases the relative risks of cohabitation versus no union. However, if marriage is 61 compared to cohabitation, residential mobility and migration to new places with better or worse economic conditions decreases the likelihood of marriage. On the other hand, in the NLSY97, residential mobility and moving to another county with higher unemployment rates increase the relative risks of cohabitation compared to remaining single. These results suggest that residence changes deter permanent union formation for the NLSY79 young adults while encouraging the NLSY97 individuals to form a less permanent union (cohabitation).

Second, most other covariates produce similar results shown in Table 3.3. In general, females are more likely than men to be involved in union formation in both cohorts, and if marriage is compared to cohabitation, females are more likely to marry.

Similarly, African Americans are less likely than their non-black non-Hispanic whites and others to either marry or cohabit. In the NLSY79, Hispanics are less likely to cohabit than remain single. In the NLSY97, however, Hispanics are significantly less likely to cohabit compared to no union but more likely to marry regardless of the competing risks.

Educational attainment and employment status appear critical for young people’s union formation. Those who are enrolled in school are less likely to either marry or cohabit in the NLSY79 although school enrollment is not related to union formation in the

NLSY97. In the NLSY97, those who have completed high school are more likely to marry regardless of the competing risks. Completing a bachelor degree increases the likelihood of marriage and cohabitation for the NLSY97 young adults compared to remaining single. In the NLSY79, finishing the degree increases the relative risk of marriage compared to cohabitation. Full-time employment increases the relative risks of 62 marriage and cohabitation compared to remaining single in both cohorts. For the

NLSY79 only, part-time employment increases the relative risks of marriage and cohabitation compared to remaining single, and if the marriage is compared to cohabitation, full-time employment increases the likelihood of marriage. Childbearing increases the likelihood of marriage and cohabitation versus no union in the NLSY79. In the NLSY97, it only increases the hazard of cohabitation compared to no union.

Regarding household characteristics, those who are from an intact family in the NLSY79 are less likely to be involved in union formation. In the NLSY97, on the other hand, living in an intact family decreases the hazard of cohabitation regardless of competing risks. Higher maternal educational attainment significantly decreases the hazard of marriage regardless of the competing risks in both cohorts. In the NLSY97 only, increasing maternal educational attainment also decreases the hazard of cohabitation.

Finally, residence characteristics have relatively small impacts on the relative risks of marriage and cohabitation. For the NLSY79, living in metro areas decreases the relative risk of marriage compared to cohabitation while those who lived in metropolitan areas in the NLSY97 are less likely to cohabit versus remaining single.

Discussion

Using two comparable samples of different cohorts, this study examines moving experiences and their relationship with first union formation among young adults in the

United States. Existing research has given much attention to the union formation of young adults, yet less is known about its relationship with geographic mobility. Taking

63 advantage of the NLSY data sets that contain longitudinal information on various life course transitions, the association between union formation and moving is examined in conjunction with other life course transition in great detail. In what follows, I summarize findings from this study and discuss each point.

First, it is found that more young adults in the NLSY79 are involved in any union whereas about a quarter of those from the NLSY97 have not been in a union up until age

30. In addition, cohabitation has become prevalent as a first union type for young adults in the NLSY97 while few marry without prior cohabitation. Although the NLSY79 young adults marry earlier than their counterparts from the NLSY97 on average, the mean age at first union is higher for those from the earlier cohort. This is because of the earlier entry into cohabiting relationships for young adults in the NLSY97. These findings reflect the recent union formation changes such as the delay of marriage and the increase of cohabitation (Schoen et al. 2007). Moreover, the results may indicate that young adults in the recent cohort take more time for self-exploration rather than committing to others, given that cohabitation is considered a temporary and less stable relationship choice in terms of its duration and commitment level (Cherlin 2010).

Second, descriptive statistics of mobility reveal that a larger share of young adults from the NLSY97 change their residence and start to move earlier than those from the

NLSY79 cohort. Assuming mobility is a deliberate behavior optimizing educational and employment opportunities, the results suggest that young adults in the recent cohort are more likely to explore various opportunities during young adulthood. Moreover,

64 differences are found in moving type between cohorts; young adults in the NLSY79 are more likely to move within the same county, while those from the NLSY97 cohort tend to migrate to another county. Migration events, long-distance between county moves, are often considered an investment behavior related to employment and careers because moving across a certain jurisdictional boundary often changes one’s labor markets

(Cadwallader 1992; Clark 2013). The findings, therefore, reveal that young adults from the recent cohort are more likely to be involved in human capital exploration. It is also found that moving experiences differ by first union type in both cohorts. For young adults in the NLSY79, cohabitors are more likely than married or singles to move. The earlier cohort cohabitors in the current study predominantly held a high school degree rather than a college degree and are more likely to be employed part-time. Moreover, fewer were in an intact family during childhood although maternal educational attainment is significantly higher for the cohabitors in the earlier cohort. Their large proportion of moves may thus indicate a search for stable residence that is a sign of their unstable socioeconomic status. On the other hand, for those from the NLSY97, singles tend to change their residence the most. Similar to cohabitors in the NLSY79 cohort, singles are probably less likely to settle down because of their unstable positions. This finding may also reveal a tendency toward life exploration of young adults by moving to new places in the recent cohort.

Third, findings from Cox models suggest that moving significantly influences the likelihood of first union formation in both cohorts although the direction differs by cohort. In the NLSY79, moving is hardly associated with union formation except that 65 migration to another county with the same unemployment level decreases the hazard of first union. During the 1980s when the earlier cohort individuals entered young adulthood, the economy had recently recovered from the second worst recession since the

Great Depression (Hout, Levanon, & Comberworth 2011) and female and college graduates’ economic standings have improved (Jacobsen & Levin 1997). At that time, the outcomes of a move varied from potentially great economic gains to much riskier investment than staying (Jacobsen & Levin 1997). The non-significant coefficients in the earlier cohort therefore likely reflect the extent of variation in moving and its socioeconomic outcomes. Boyle et al. (2008) found that the first long-distance migration event does not affect women’s separation or dissolution because couples often view that move as an exciting moment to intensify their relationship. For single young adults, long- distance moves could be a similar experience, but may not necessarily coincide with an expectation of new union formation. It is also possible that those who migrate to similar economic conditions are a small proportion of the earlier cohort; they are probably a selective group of young adults who are reluctant to form a union. For young adults in the

NLSY97, on the other hand, residential mobility and migration to new counties with higher unemployment rates are positively related to first union formation. Studies have shown that the first residential move is a positive experience that helps to promote independence for couples (Boyle et al. 2008). For contemporary single young adults who tend to take more self-focused time, moving may play the same role as for couples practicing physical and psychological independence. Regarding the positive hazard of migration to new places with worse economic circumstances on union formation, it may

66 indicate a strategy of young adults to reduce the impact of economic recession. Although young adults from the NLSY79 cohort had experienced an economic collapse during young adulthood, that of the recent cohort appeared much worse with regard to its recovery pace and growing inequality (Grusky, Western, & Wimer 2011). As young adults from the recent cohort experienced more severe economic recession, to cope with the hardship in a new place with the worse economic conditions they married or cohabited with partners to share living expenses.

Fourth, the association between union formation and mobility has become more complicated by the type of first union in both cohorts. The negative effect of migration to new counties with similar unemployment rates on union formation is likely because moving events significantly decrease the relative risk of marriage compared to no union in the NLSY79 cohort. When marriage is compared to cohabitation, mobility decreases the relative risk of marriage, except that migration to a place with similar economic conditions does not differentiate the risk of marriage from cohabitation. Moreover, moving to another county with better economic conditions increases the likelihood of cohabitation compared to remaining single among young adults from the NLSY79. These results reveal that for those in the earlier cohort moving represents an experience that discourages marriage. Cohabitation was not as prevalent for young adults from the

NLSY79 compared those in the NLSY97. Moreover, cohabitation used to serve as either a precursor or alternative to marriage for the earlier cohort of young adults. In both cases, the positive relationship between cohabitation and moving to better economic conditions suggests that singles cast a wider net for union formation by moving to better marriage 67 markets. Since marriage is largely preceded by cohabitation, singles first cohabit in the new place. Marriage, however, is not affected by moving to better places for singles in the NLSY79 since marriage requires greater commitment and firm connections with each other. The meaning of cohabitation, however, has changed over time and many young adults are now more likely to be involved in cohabitation than marriage. As a result, no significant association is found between mobility and marriage among those from the

NLSY97 cohort. Cohabitation, on the other hand, appears significantly related to residential mobility and moving to worse economic counties. As described, mobility likely serves as a learning process to become independent because moving within the same county increases the likelihood of cohabitation compared to remaining single. The economic reason (i.e. saving living expenses) of the positive association between union formation and moving to another county with a worse economic situation is again supported; moving increases the likelihood of cohabitation but not marriage among those from the recent cohort. I also admit that differences in the association between moving and union formation by cohort may be due to a discrepancy in measurement units. As table 3.3 revealed, mobility with yearly variables underestimates monthly moving experiences which may disregard significant moving events related to union formation in the NLSY79. Future research with more refined event history data will contribute to more accurate portrait of association between moving and union formation.

In sum, findings from this study reveal that the life course transitions are closely interrelated with each other in both cohorts and the complicated association differs by changing socioeconomic context. Previous research found that the economic recession in 68 the late 2000s hardly affected marriage and fertility in the U.S. despite some anecdotal concerns about deleterious effects on individual behavior (Morgan, Cumberworth, &

Wimer 2011). However, this study finds that the socioeconomic changes may have influenced an underlying association between life course transitions in significant ways.

Although it is not clear from this study what mechanisms contribute to the close associations between moving and union formation and differences by cohort, this study provides empirical evidence of complicated life course transitions embedded in social changes. Future research may be enriched by incorporating diverse life course transitions and their deliberate decision making process. Moreover, it is possible that other dynamics such as housing markets in residence and distance from the parental home affect the association between moving and union formation for young adults (Clark 2013;

Michielin, Mulder, & Zorlu 2008). For example, young adults who are in an early career stage and have few resources may prefer living close to parents until or even after union formation. I also acknowledge that those who have a greater likelihood of marriage and cohabitation could be a selective group of people who have already been in a dating relationship before moving. In addition, some people tend to move and marry earlier than others. Despite the significant coefficients in the multivariate models, therefore, endogeneity may have existed between mobility and union formation. Life course events such as employment, moving, and union formation often occur simultaneously (Guzzo

2006); a decision to move may be jointly determined with a decision to form a union

(Schachter 2001). A multi-process model, which jointly estimates the processes of both moves and union formation, should be used to obtain refined estimates by allowing for

69 unobserved factors that influence both transitions (Lillard & Waite 1993; Steele, Kallis,

Goldsten, & Joshi 2005; Steele, Joshi, Kallis, & Goldstein 2006). The next chapter uses this approach to address possible endogeneity between migration and union formation.

Chapter 3 tables

Table 3.1 Description of sample by first union formation

No union Married Cohabited Total 79 97 79 97 79 97 79 97 Proportion .08 .28 .62 .14 .30 .57 1.00 1.00 Mean age at first union - - 22.77 23.79 23.98 21.98 23.15 22.09 (years) (.05) (.09) (.07) (.07) (.04) *** (.08)*** Women .41 .39 .51 .52 .47 .52 .49*** .48*** Race Black .32 .21 .11 .10 .16 .14 .14*** .16*** Hispanic .07 .12 .06 .16 .06 .12 .06 .13*** Non-Black and Non- .61 .67 .83 .73 .78 .74 .80*** .72*** Hispanic whites and others Education Enrolled in school .10 .23 .18 .51 .19 .45 .17*** .41*** Less than high school .14 .53 .27 .49 .22 .73 .24*** .66*** High school completion .54 .37 .53 .44 .60 .24 .56*** .29*** AA degree completion .08 .02 .04 .02 .04 .01 .04** .01*** Bachelor degree completion .24 .08 .16 .05 .14 .02 .16*** .04*** Employment Not employed .15 .64 .18 .47 .09 .64 .15*** .62*** Employed part-time .37 .14 .40 .28 .49 .19 .43*** .19** Employed full-time .48 .22 .42 .25 .42 .17 .42* .19*** Childbearing .02 .05 .03 .04 .04 .04 .03** .04 Living in intact family .64 .59 .68 .66 .57 .47 .64*** .53*** Maternal education (years) a 11.51 13.10 11.58 12.83 11.79 12.76 11.64 12.86 *** (.17) (.10) (.09) (.14) (.10) (.08) (.08) (.08) Living in metro .81 .86 .77 .80 .84 .76 .79*** .79*** Unemployment rate in county 3.05 6.81 3.51 6.83 3.51 6.82 3.47*** 6.82 of residence (%) a (.09) (.18) (.09) (.22) (.09) (.18) (.08) (.18) [NLSY79] n=9,491 (60,045 person years), Population size= 32,041,472 [NLSY97] n=8,803 (799,377 person months), Population size= 17,666,347

Note: Numbers are proportions except those indicated. Numbers in parentheses are standard errors. All variables are adjusted under survey setting. The Chi-Square test is used to test the sample distribution except the variables with a which use one-way ANOVA to test mean differences between groups. * p < .05, ** p < .01, *** p < .001

Table 3.2 Description of mobility by first union formation

No union Married Cohabited Total 79 97 79 97 79 97 79 97 Ever experienced mobility Any moves .22 .41 .21 .23 .28 .21 .23*** .27*** Residential mobility .13 .12 .10 .04 .14 .05 .12*** .07*** Migration .09 .28 .11 .19 .13 .15 .11** .20*** to high unemployment rates .04 .08 .04 .06 .07 .05 .05*** .06*** to low unemployment rates .05 .19 .06 .11 .06 .09 .06 .12*** to same unemployment rates .00 .02 .00 .01 .00 .01 .00 .01*** Mean age at first mobility (yrs) Any moves 29.8 21.4 22.1 20.5 22.1 20.1 22.5 *** 20.7 *** (.09) (.11) (.16) (.16) (.27) (.08) (.19) (.07) Residential mobility 29.9 22.6 21.7 21.2 22.0 20.5 22.1 *** 21.6 *** (.07) (.22) (.21) (.31) (.33) (.19) (.21) (.16) Migration 29.8 21.4 22.6 20.6 22.4 20.2 23.0 *** 20.8 *** (.16) (.11) (.31) (.18) (.45) (.09) (.30) (.07) Note: Numbers in parentheses are standard errors. All variables are adjusted under survey setting. The Chi-Square test is used to test the sample distribution. ** p < .01, *** p < .001

Table 3.3 Comparison of mobility in the NLSY97 by different measurement units

No union Married Cohabited Total month year month year month year month year Ever experienced mobility Any moves .41 .08 .23 .14 .21 .16 .27*** .14*** Residential mobility .12 .02 .04 .03 .05 .05 .07*** .04*** Migration .28 .07 .19 .11 .15 .12 .20*** .11*** to high unemployment rates .08 .04 .06 .07 .05 .08 .06*** .07*** to low unemployment rates .19 .03 .11 .05 .09 .06 .12*** .05*** to same unemployment rates .02 .003 .01 .004 .01 .004 .01*** .004** Mean age at first mobility (yrs) Any moves 21.4 22.2 20.5 22.2 20.1 21.7 20.7 *** 21.8 *** (.11) (.11) (.16) (.13) (.08) (.08) (.07) (.07) Residential mobility 22.6 23.5 21.2 23.6 20.5 22.9 21.6 *** 23.1 *** (.22) (.19) (.31) (.17) (.19) (.09) (.16) (.09) Migration 21.4 22.1 20.6 22.3 20.2 21.9 20.8 *** 22.0 *** (.11) (.11) (.18) (.14) (.09) (.08) (.07) (.07) Note: Numbers in parentheses are standard errors. All variables are adjusted under survey setting. The Chi-Square test is used to test the sample distribution. ** p < .01, *** p < .001

Table 3.4 Cox proportional hazard models predicting first union

NLSY79 NLSY97 Coef. H. R. Coef. H. R. Moving factors tv residential mobility -.01 (.05) 1.00 .45 (.13)** 1.57 migration from low to high unemployment rates -.03 (.08) .97 .30 (.09)** 1.34 from high to low unemployment rates .07 (.06) 1.08 .10 (.11) 1.11 to same unemployment rates -1.06 (.42) * .35 .13 (.37) 1.13 Women .35 (.03) *** 1.42 .48 (.04)*** 1.62 Black -.50 (.04) *** .61 -.43 (.05)*** .65 Hispanic -.10 (.05) * .90 -.10 (.05) .91 Education tv enrolled in school -.22 (.04) *** .80 -.03 (.04) .97 completed high school -.04 (.05) .96 .02 (.04) 1.02 completed AA degree .00 (.09) 1.00 .09 (.10) 1.10 completed Bachelor degree .08 (.07) 1.09 .28 (.06)*** 1.33 Employment tv employed part-time .20 (.06) *** 1.22 .01 (.05) 1.01 employed full-time .46 (.06) *** 1.59 .28 (.04)*** 1.33 Childbearing .48 (.08) *** 1.62 .14 (.07) * 1.15 Household characteristics living in an intact family -.20 (.03) *** .82 -.44 (.04)*** .65 maternal education -.03 (.01) *** .97 -.07 (.01)*** .93 Residence characteristics living in metro area tv -.08 (.04) .93 -.21 (.05)*** .81 county unemployment rates .01 (.01) 1.01 -.01 (.01) . 99 [NLSY79] Strata = 2, PSUs = 195, Design df = 193 Number of person years= 60,045; Population size= 20,158,267 [NLSY97] Strata = 1, PSUs = 199, Design df = 198 Number of person months= 692,075; Population size= 15,669,446 Note: All variables are adjusted under survey setting. tv indicates time-varying variables whose values vary by time. * p < .05, ** p < .01, *** p < .001

Table 3.5 Cox competing risks models predicting first union type

Marriage vs No Union Cohabitation vs No Union Marriage vs cohabitation 79 97 79 97 79 97 b eb B eb b eb b eb b eb b eb Moving factors tv residential mobility -.04 .96 -.25 .76 .10 1.11 .59*** 1.80 -.24** .79 -.27 .76 migration from low to high unemploy rates -.12 .89 .00 1.00 .16 1.17 .38*** 1.46 -.38** .68 -.05 .95 from high to low unemploy rates .01 1.01 .37 1.45 .21* 1.23 .02 1.02 -.20* .82 .34 1.41 to same unemploy rates -1.06* .35 -.67 .51 -.89 .41 .29 1.33 -.94 .39 -.66 .52 Women .36*** 1.44 .57*** 1.77 .35*** 1.43 .46*** 1.59 .27*** 1.31 .48*** 1.61 Black -.62*** .54 -.70*** .50 -.54*** .59 -.38*** .68 -.33*** .72 -.42*** .66 Hispanic -.12 .89 .25* 1.28 -.18* .84 -.17** .84 -.01 .99 .29* 1.34 Education tv enrolled in school -.22*** .80 .09 1.09 -.21** .81 -.06 .94 -.13** .88 .08 1.08 75 completed high school -.05 .95 .36*** 1.43 -.02 .98 -.07 .93 .06 1.06 .25** 1.29

completed AA degree -.05 .95 .33* 1.39 .08 1.08 .00 1.00 .16 1.17 .12 1.13 completed Bachelor degree .15 1.16 .46*** 1.59 .01 1.01 .20** 1.23 .55*** 1.73 .09 1.09 Employment tv employed part-time .20** 1.22 -.07 .93 .34** 1.41 .02 1.03 .13 1.14 -.12 .88 employed full-time .54*** 1.72 .20* 1.23 .48*** 1.61 .30*** 1.35 .53*** 1.70 .07 1.07 Childbearing .54 *** 1.71 .17 1.19 .56 *** 1.76 .14* 1.16 -.19 .83 .07 1.07 Household characteristics living in an intact family -.17*** .85 .16 1.17 -.33*** .72 -.55*** .58 .04 1.04 .23* 1.26 maternal education -.04*** .96 -.08*** .92 -.02 .98 -.07*** .93 -.05*** .95 -.08*** .92 Residence characteristics living in metro area tv -.10 .90 -.02 .99 .07 1.08 -.24*** .78 -.20** .82 .03 1.03 county unemployment rates .02 1.02 .00 1.00 -.00 1.00 -.01 .99 .01 1.01 .00 1.00 Note: All variables are adjusted under survey setting. tv indicates time-varying variables whose values vary by time. * p < .05, ** p < .01, *** p < .001

CHAPTER 4: MODELING THE JOINT PROCESS BETWEEN LIFE COURSE EVENTS

Migration decisions are made in concert with decisions about other life course events (Groot et al. 2011; Kulu & Milewski 2007; Schachter 2001). Young adults leave their parental home and settle in new places in order to pursue educational and occupational goals (Garasky, Haurin, & Haurin 2001), where some will find a partner and form families (Kulu 2008). Changes in union status (formation or dissolution) and in childbearing can also motivate changes in residence, in part due to space requirements

(Flowerdew & Al-Hamad 2004; Kulu & Milewski 2007). These and other plausible life- course scenarios suggest an interplay between migration and other life-course transitions that is hardly unidirectional (Kulu & Milewski 2007). Nevertheless, most empirical research to date has examined unidirectional effects only. Certainly this has been the case with research on migration and union formation – most of this research has simplified the role of migration with migration serving as an explanatory variable that either promotes or delays union formation (Feijten & Mulder 2002; Guzzo 2006). However, the inter- relations between migration and marriage may be more complicated. It is plausible that migration affects union formation, and also that marriage or cohabitation affects migration; ignoring this more complicated inter-dependency may yield biased estimates

76 of the causal effects of interest, including the effects of other explanatory variables (e.g. schooling history) (Mulder & Wagner 1993; Kulu & Milewski 2007; Steele et al. 2005).

In addition, selectivity could be an issue, e.g. migrants could be selective of persons who are prone to marry or cohabit (or vice versa) (Kulu & Milewski 2007; Mulder & Wagner

1993).

Despite these concerns, the comprehensive set of plausible inter-relations between migration and union formation has rarely been examined in empirical research to date.

And the few studies that have considered this range of inter-relations between migration and union formation have relied on data from Europe (Michielin & Mulder 2008). While it may be reasonable to draw implications for the inter-dependency in the United States from findings in European countries, it is the case that migration and union formation in the U.S. are distinct. For example, rates of internal mobility in the U.S. have generally been higher than in European countries (Molloy, Smith, & Wozniak 2011), and marriage retains an importance in the U.S. (Cherlin 2004) which is not observed in some European societies (especially northern Europe). Therefore one must question if the nature and magnitude of effects evident in Europe (e.g. Michielin & Mulder 2008 for the

Netherlands and Mulder & Wagner 1993 for West Germany) characterize the U.S.

Most young adults experience first union formation and residential change in the transition to adulthood, and decisions about these transitions often depend on each other.

As union formation and moving trajectories become diverse, the inter-relations between moving and union formation may appear more complicated in recent years. Moreover,

77 within-group differences in recent young adults’ life course grow (e.g. Silva 2013) and thus conventional demographic and household characteristics may not fully account for the dynamics of the transition to adulthood. No study, however, investigates how endogenous features of life course transitions have changed across cohorts in the U.S. To fill the gap in the literature, the current study compares the inter-dependency between moving and union formation by cohort in the U.S. using panel data from the National

Longitudinal Survey of Youth 1979 and 1997. Since individuals are expected to experience these two events several times over the life course, the series of events are taken into consideration by employing a multi-level model which accounts for the individual propensity to move or to be in a union (Steele et al. 2006). Moreover, to allow for an association between the two life course trajectories that are not accounted for by the measured explanatory variables, a multi-process model is used in which equations for both types of events are estimated simultaneously (Lillard & Waite 1993).

This chapter extends the previous literature in several important respects. First, a better fix on the associations between mobility and union formation contributes to a better understanding of interrelationships between family transitions and residential change over the life course. Although a growing body of research has investigated the association between various life course transitions, when it comes to migration most existing research has been limited to the effects of migration on reproductive transitions

(Clark & Withers 2009; Kulu & Steele 2013; Michielin & Mulder 2008). Consequently, it is not clear from past research how union formation transitions affect changes in residence, and in turn how mobility influences union formation decisions. In the existing 78 literature, moves have been understood as an adjustment to and a trigger of family events, especially childbirth (e.g., subsequent moves in response to family growth) (Kulu &

Steele 2013). Parallel to this literature, I examine inter-relations between migration and union formation; it cannot be assumed that these will take the same form as migration- fertility inter-relations. Second, analyzing histories with multiple occurrences of migration and union formation strengthens estimates of the effects of each on the other.

Both transitions – geographic moves, entrance to a union – are affected by unmeasured individual propensities which possibly are associated with measured explanatory variables of interest. For example, some people prefer to live in specific areas (Benson &

O’Reilly 2009) while some are attracted by other environments. Moreover, research on marital decisions acknowledges the likely impact of variables which are often unmeasured, such as physical appearance or personality (Lichter, Anderson, & Hayward

1995). Estimations relying on histories with multiple events of both types provide some leverage for controlling for unmeasured propensities to transition (move or form a union).

There is the further threat to estimation related to decisions to marry or cohabit and to move, which are made jointly; not allowing for this association can also result in biased estimates. This research addresses both of these threats to recovering valid estimates of the causal effects of interest. A final contribution of this research is the empirical territory which it encompasses. I compare longitudinal information on nationally representative samples of individuals in the United States from two cohorts (i.e. those who were born in

1957 - 1964 and born in 1980-1984). These data cover each individual’s lifetime from late teens to early 50s during the historical period 1979-2010 for the NLSY79 and from

79 teens to late 20s during 1997-2011 for the NLSY97. That is, this study examines migration-union formation inter-relations over several decades of the life course and about three decades of historical experiences.

Moving and union formation

Residential changes are closely related to family events in part because changes in family size require residential adjustment (Kulu & Mileski 2007; McAuley & Nutty

1982; Michielin & Mulder 2008). More fundamentally, family-building events are well- recognized as a major cause of migration, and in particular migration is closely related to marriage (Guzzo, 2006; Speare & Goldscheider 1987). For example, Michielin and

Mulder (2008) found that the hazard of moving in the Netherlands increases within 6 months of becoming married. Although the study only considered first marriage, the results confirm a close relationship between migration and marriage. With regard to the association between the two life course transitions in the U.S., marriage increases the likelihood of migration as newlyweds settle in new places or at least one partner moves in with the other (Speare & Goldscheider 1987). This positive relationship is observed in both first-marriage and remarriages, and it persists for several years (Speare &

Goldscheider 1987). This particular study used a geographically restricted U.S. sample

(Rhode Island) and findings should be confirmed using a national sample from more recent years. On the other hand, family accumulation (partner, children) can dampen the risk of migration, due to the increased financial costs of moving as well as the costs of breaking ties (Long 1973; Michielin & Mulder 2008).

Prior studies find that migration influences union formation in two contradictory ways (Guzzo 2006; Jampaklay 2006). First, migration can encourage union formation to the extent that the move improves an individual’s socioeconomic status (Cadwallader

1992; Massey et al. 1993). Empirical research confirms that migrants from nonmetropolitan counties obtain higher educational attainment and earnings after migrating to a metropolitan county (Mills & Hazarika 2001), thereby improving their marriage market prospects. More directly, migration can improve marriage market prospects if individuals move to a location where there is a larger supply of marriageable mates (Lichter et al. 1992; Lichter, Anderson, & Hayward 1995; Oppenheimer 1988;

South & Lloyd 1992). If a marriage market fails to provide favorable conditions matched to a mate seeker’s criteria, one strategy is to compensate by adjusting the criteria and in effect casting a wider net (Lichter et al. 1992; Qian, Lichter, & Mellott 2005). Moving to a new place may be an alternative strategy to casting a wider net in one’s current marriage market, just as job seekers move from low to high wage places to improve income (Massey et al. 1993). An empirical finding of increased likelihood of marrying after migration would support the proposition that migration is, among other things, a strategy for expanding marital opportunities.

Even so, the short-term impact of migration on union formation may be negative, because time is required to adjust to new environments, including becoming familiar with opportunities to meet potential partners. Unless the move occurs with a partner, meeting and courting potential partners in new places may be difficult at first. In studies on relationships between migration and fertility, for instance, the fertility levels of migrants 81 are low in the immediate post-migration period but then eventually catch up to non- migrants later (Goldstein 1973; Kulu 2005). Similar to fertility, marital behaviors may be disrupted by the migration process temporarily. Thus, it is reasonable to expect that the probability of union formation decreases within the short period after a migration event, later rebounding as the length of residence increases, and possibly exceeding the probability of marrying in the place of origin.

The joint process and cohort difference

Research on migration has disproportionately focused on the first move although most people migrate several times throughout the life course (DaVanzo 1983; Molloy et al. 2011). The initial move is, however, less likely to satisfy movers’ needs because they may have had little information about new environments and be disappointed by the discrepancy between their expected and actual gains (McHugh, Hogan, & Happel 1995).

As a result, some of the initial movers seek additional destinations while others return to their original places (McHugh et al. 1995; Wilson et al. 2009). The sequential migration patterns vary by location-specific capital and length residing in the place (DaVanzo &

Morrison 1981) and also by personal traits which are not captured with explanatory variables typically available in major data-sets (DaVanzo 1983; Gabriel & Schmitz

1995). Similarly, some individuals experience multiple union transitions (formation and dissolution) over their lifetime (Cherlin 2010), and the likelihood of multiple transitions is itself associated with socioeconomic status and race/ethnicity, as well as personal values and personality factors (Smith 2005; Thornton & Young-DeMarco 2001). Such

82 personal traits, unmeasured in most data that offer good measurement of migration and union formation transitions, affect marital choices and differentiate those who are involved in a succession of marriages or cohabitations from those who are not. In regression analysis with appropriate data, these unmeasured personal traits can be represented by individual-specific random effects (for the propensity to move, the propensity to marry or cohabit, and the correlation between the two).

In addition, unobserved traits probably play a significant role in other life course transitions (Kulu & Milewski 2007; Mulder & Wagner 1993). For example, using data from West Germany, Mulder and Wagner (1993) investigate the inter-relation between migration and marriage employing a model that allows for an interaction between the two events (largely ignored in the prior literature). They find that estimates from a model which allows for this interaction differ in important respects from estimates that do not allow for it. More specifically, if the model does not allow for an interaction, it appears that women younger than age 25 are more likely than young males to move a short distance. However, once the interaction is allowed, sex differences in short-distance moves disappear and age-specific moving patterns become less pronounced (Mulder &

Wagner 1993). In other words, the sex differences in mobility are largely attributed to its inter-relation with marriage. Despite this demonstration two decades ago of the importance of allowing for residual co-occurrence of the two events, few empirical studies since then have done so.

As life course decisions are simultaneously made, failure to incorporate unobserved characteristics which affect both processes has produced upward biases of the actual relationship (Kulu & Steele 2013). Studies have shown that unmeasured features such as a desire for a large family (Kulu & Steele 2013) or prior dating history which are rarely measured by conventional surveys have individuals prone to move or enter in a union. Moreover, life events which occur repeatedly over the life course need to be taken into consideration, in particular during young adulthood. The transition to adulthood has become subjective and based more on personal characteristics than normative expectations in delayed life transitions during young adulthood (Settersten & Ray 2010;

Shanahan, Porfeli, Mortier, & Erickson 2005). Studies have found a large variability in how and when young adults feel entirely like an adult with regard to life transitions

(Shanahan et al. 2005), although the traditional transition markers still remain salient delineating the entry into adulthood (Rindfuss 1991; Arnett 2004; Furstenberg, Rumbaut,

& Settesten 2005). These indicate that various life transitions become more endogenous and unobserved factors intervene the estimation of one life transition with others. In this study, the simultaneous relationships between life course transitions is estimated using a multi-process model (Lillard & Waite 1993; Steele et al. 2005). A model with separate equations for moving and union formation are estimated with each equation having a random effect for unmeasured person-specific propensities. The model also allows for correlation between these two random effects, i.e. association between propensities to move and to form a union. Consistent estimation of the two random effects and their correlation is achieved via joint estimation of the two equations. The multiple episodes

84 for each person of risk of union transition and risk of migration transition is the basis for identification of the model (Rabe-Hesketh & Skrondal 2012; Steele et al. 2006). To display how the inter-dependency between moving and union formation has changed over time, two separate estimation procedures are included: a multi-process model for the

NLSY79 and 97.

Informed by this previous literature, this study addresses the following research questions:

1) Does union formation affect the hazard of mobility?

2) Does mobility affect the hazard of union formation?

3) Do unmeasured person-specific traits affect the co-occurrence of moving and union formation? How does the relationship differ by cohort?

Data and Methods

This study uses public and geocode data from the National Longitudinal Survey of Youth 1979 and 1997 (NLSY79 and NLSY97), which contain nationally representative samples of individuals in the United States. Union formation is defined as marriage in the NLSY79, and both marriage and cohabitation are included in the analysis of NLSY97 because cohabitation has become prevalent in the recent cohort (see Table

3.1). For marriage in the NLSY79, information on relationship status (i.e., spouse, partner, or single) is drawn from the partner-specific characteristics files every survey year (Center for Human Resources 2013). To create a complete marriage history, I also use the actual starting and ending dates of marriages as provided in the public files.

About 79% of the respondents experience first marriage and about 25% of them marry again (20% of the entire sample) in the NLSY79 (see Table 4.1). Regarding union formation in the NLSY97, the month that either cohabitation or marriage begins is utilized to measure the timing of union formation. During the survey period of 1997 to

2011, about a quarter of the NLSY97 have not been involved in any union while about

75% have experienced either marriage or cohabitation (see Table 4.1).

In the NLSY79, the exact dates of migration are not available prior to 2000.

Therefore, the county and state FIPS codes (Federal Information Processing Standard) are used, and a migration event is defined as changes in county of residence between the annual survey interviews. Migration is therefore inferred indirectly in the NLSY79, and some short-term moves are missed, namely those that are followed by another move before the next interview (respondent is correctly classified as a migrant, but one destination is missed) and short-term moves that are followed by return to residence at previous interview prior to the next interview (respondent is incorrectly classified as non- migrant). On the other hand, the NLSY97 asks respondents about their detailed migration history every survey year. Using this information, moving is defined as changes in county of residence as NLSY79. Since international migration has different implications and impacts on life courses than internal migration (Molloy, Smith, & Wozniak 2011), it is excluded from both datasets. As the year and month of every move are utilized to create an event history file of mobility for analysis, the chronology of moves and union formation that occur in the same year or month is uncertain. This study assumes that responses of marriage to migration occur with some lag, and vice versa, and therefore lag 86 each event one year and six months with respect to the other in the NLSY79 and 97, respectively. Moreover, it is possible that premarital cohabitation (a common precursor to marriage) distorts the relationship between migration and marriage in the NLSY79. In the data, about 7% of moves follow premarital cohabitation in the NLSY79; that is, migration intervenes between cohabitation and marriage, and should be regarded as occurring after the partnership has started (and therefore it would be inappropriate to attribute a causal effect of the moves on marriage). To avoid bias due to this phenomenon, I drop from the analysis moves which follow immediately after premarital cohabitation in the NLSY79. In the resulting samples, about 50% of the sample has moved at least once from their county of residence in 1979 (see Table 4.1). For the

NLSY97, about 36% have moved to different county (i.e., migration) in the NLSY97 (see

Table 4.1).

Since the analyses take an event history approach, data are transformed into person-year and person-month files for the NLSY79 and NLSY97, respectively, which generate 134,204 person-years and 1,185,485 person-months. As the unit of analysis differs between the NLSY79 and NLSY97, separate models are estimated by cohort. Two equations need to be operated simultaneously in each cohort, and thus I restrict the sample to those who were age 16 at the first interview for the NLSY79, which is the onset of risk for both first migration and first marriage. The final dataset includes 7,827 individuals who provide 87,931 person years for analysis. As the NLSY79, the NLSY97 sample is restricted to those who were age 16 or younger at the first interview in 1997.

The final data includes 8,154 individual who contribute 1,045,943 person months to the analysis.

The same sets of individual, household, and county characteristics are included in the model of each cohort. As distinct developmental tasks are required at each stage of life course (McAuley & Nutty 1982; 1985), age effects are controlled for in the models with squared and cubic terms. Previous empirical research on life course events shows mixed findings about the nature of gender differences. Females marry earlier than males in virtually all societies and sub-groups (Kreider & Ellis 2011), but gender differences in migration vary by study. Males are in general more likely than females to migrate, probably because historically men invest more in human capital (Quinn & Rubb 2011).

On the other hand, young female adults leave their parental home earlier than their male counterparts (Buck & Scott 1993; Garasky 2002) and unmarried females move significantly more than their male counterparts, especially in the case of long distance moves (Mulder & Wagner 1993). This is partly because of early family formation of females (Long 1973). To sort this out, the effects of gender on migration and marriage are examined in the current study.

Race and ethnicity also influence the timing of marriage and migration (Glick et al. 2006). Whites tend to delay marriage until they have achieved a first set of career goals but the lifetime risk of marriage is higher for Whites than Blacks and Hispanics

(Goldstein & Kenney 2001). Studies on racial differences in moving patterns also suggest that mobility rates vary by the characteristics of places such as racial distribution

(Crowder 2000) and racial differences in wage rates (Wolaver & White 2006). In addition, individual socioeconomic status represents the capability to manage life course transitions (Oppenheimer 1988). For example, those who have completed their schooling and are employed full-time are more likely to marry but less likely to move, whereas those who are in school and unemployed tend to show unstable life course transitions including frequent moves as part of job searches (DaVanzo 1983). To the extent that economic characteristics affect life course decisions, I also include local economic conditions in the current county of residence as control variables, assuming that individuals in places with worse economic conditions are more likely to move (Massey et al. 1993). The indicators of local economic conditions are unemployment rates, poverty rates, and the share of the population having a college or higher degree. Migration is also expected to be influenced by comfort or concern with local public safety (Cadwallader

1992; McAuley & Nutty 1982), and therefore the county-level crime rate is included in the equations. Finally, the local demographic structure may influence either migration or marriage, the latter by affecting the availability of potential mates of the opposite sex

(South & Lloyd 1992); to account for this, the sex ratio (percentage of opposite sex population in the county) is a further explanatory variable.

Household characteristics can also affect life course decisions. Greater household socioeconomic resources promote independence and autonomy for household members, resulting in earlier migration and independent living (Avery et al. 1992). Unstable family structure encourages individuals to move frequently and form their own family earlier, although the opposite has also received empirical support such that economic and 89 emotional support from a stable family encourages their adult children to be independent

(Avery et al. 1992). These hypothesized relationships are tested by including explanatory variables of maternal educational attainment and whether respondents had lived during childhood. In addition, I include an indicator of type of place of residence, i.e. nonmetropolitan or metropolitan areas, because previous research reveals significant variation in life course experiences by type of place (Cromartie 1993; Snyder, Brown, &

Condo 2004). A yearly time-varying measure of nonmetropolitan and metropolitan residence is merged from the 2003 Urban Influence Codes from USDA ERS (United

States Department of Agriculture Economic Research Service) due to changes in standards measuring the metropolitan and nonmetropolitan statistical areas in the NLSY

(see footnote 3).

Analytical Strategies

Equations for the discrete-time hazards of migration and union formation are estimated as follows (Allison 1984). The fact that many NLSY respondents experience more than one union formation and migration event provides a basis for estimation of a person-specific random effect. The equations for each process can be specified as:

푀퐼퐺 푀퐼퐺 푀퐼퐺 푀퐼퐺 (1) log[ℎ (푡)] = 훼0 퐷(푡) + 훼1 퐹(푡) + 훼2푋 + 훼3푈푛𝑖표푛(푡−1 표푟 푡−6) +

푢푀퐼퐺

푈푛𝑖표푛 푈푛𝑖표푛 푈푛𝑖표푛 푈푛𝑖표푛 (2) log[ℎ (푡)] = 훽0 퐷(푡) + 훽1 퐹(푡) + 훽2푋 +

푈푛𝑖표푛 훽3푀𝑖𝑔푟푎푡𝑖표푛(푡−1 표푟 푡−6) + 푢

Equation (1) is for the hazard of migration at time t (log [hMIG (t)]); D (t) represents the duration-pattern of migration following the onset of the risk. Once individuals move, they are at risk of a next move. F(t) denotes a time-varying covariate whose values change over time: educational attainment, employment status, living in metro areas, and county characteristics in this study. X denotes time-constant variables such as demographic and household factors at the first interview. The equation also includes selected facets of the union formation history, namely total number of unions and whether a union formed in year t-1 or month t-6; both are time-varying. As described, a person-specific residual, uMIG, is included to represent the person-specific propensity to move that is not captured by measured explanatory variables (Allison 1984;

Rabe-Hesketh & Skrondal 2012).

Similar to the migration equation (1), equation (2) is for the hazard of union formation and consists of a set of terms D capturing the duration-pattern and a large set of time-varying (F(t)) and time-constant (X) covariates. And, analogous to the migration equation, equation (2) contains selected facets of the migration history, namely total number of moves and whether a move occurred in year t-1 or month t-6 for the NLSY79 and 97, respectively. In the analysis of union formation, the onset of risk for first union is defined as age 16. Whereas after each move an individual is at risk of another move, in the case of union formation the subsequent episode is the risk of union dissolution. Only after union dissolution are individuals at risk of transitioning into another union. A person-specific random effect, uUNION, controls for any unobserved heterogeneity for the same individual, affecting union formation and being constant across subsequent unions. 91

Turning to the random effects, these are assumed to follow the normal distribution, with a variance specific to each effect to be estimated from the data:

uevent ~ N(0, σ2)

It is first assumed that the two person-specific propensities (to move and to form a union) are independent of each other. This specification is correct if the facets of the two histories included as explanatory variables fully capture the effects of each history on the other and if there are no unobserved explanatory variables that affect both histories.

These are very strong assumptions, and it seems unlikely they are satisfied; therefore it is safer to posit a correlation between these two random effects, i.e. an association between the propensity to move and to form a union that is not captured by the measured right- hand-side variables. This form of endogeneity can be accounted for by estimating the two equations simultaneously and allowing the two disturbances to be correlated (Steele et al.

2005; Upchurch et al. 2002).

2 mig marr 휎푚푖푔 u = (u , u ) ~ N (0, 0) ( 2 ) 휎푚푖푔,푚푎푟푟 휎푚푎푟푟

Identification

Identification of a two-equation system with random effects and correlated random effects typically requires covariates that are included in one equation but excluded from the other, i.e. instrumental variables (Brien, Lillard & Waite 1999; Lillard

& Waite 1993; Steele et al. 2005; 2006). But if respondents experience repeated events of

92 each type (i.e. multiple marriages and multiple moves), this provides an alternative to covariate exclusion as a means of model identification (Lillard, Brien & Waite 1995;

Steele et al. 2005; Upchurch, Lillard, & Panis 2002). All sources of correlation between migration and union formation are accounted for by person-specific random effects that are constant across replications for the same individual. Then, the remaining variation after accounting for the correlation across two processes represents the effects of previous moving (or marriage/cohabitation) on the current episode, which is exogenous from the other process (Upchurch, Lillard, & Panis 2002). This means of model identification has been utilized in previous research on interdependent life course histories because it is often difficult to locate variables that satisfy the strict exclusion requirements (Steele et al. 2005).

Results

Table 4.1 describes respondents’ migration and union formation experiences over survey years. For the NLSY79, about half of the respondents never move from their county of residence in 1979, while the other half migrate at least once. On the other hand, about 64% of young adults from the NLSY97 have not changed their county of residence in 1997 over the survey years. In both cohorts, very few individuals are involved in higher order migration (i.e. 5 or more). With regard to union formation, about 79% of the respondents marry once during the survey period and about 20% experience a second marriage in the NLSY79. In the NLSY97, about 75% have formed a union during the survey years and less than 10% of them are involved in third or higher-order union. The

93 specifications of the two-equation model presented above (equations (1) and (2)) are considered in this study. In the first, the migration and union formation equations are estimated independently, i.e. as a pair of independent processes (although the explanatory variables for both processes contain aspects of the parallel history). The second specification is a multi-process model that includes correlation between the person- specific random effects. If this correlation emerges as significant, the conclusion is that unobserved propensities to move and to form a union are associated, and the coefficients from the single-process model are biased (Steele et al. 2005).

Correlation between Random Effects

Table 4.3 presents the random effects estimated from the single-process and multi-process models for each cohort. In both cohorts, significant correlations between moving and union formation are found in the multi-process model (σmig *union = .24, .30 for the NLSY79 and 97). This indicates that some components not included in the models of migration and union formation make people more likely to both move and form a union.

Moreover, the coefficient on the union formation variables in the migration equations and that on the migration variables in the union formation equations are different in the multi- process model for both the NLSY79 and 97 (see top rows in Tables 4-4, 4-5, 4-6, and 4-

76). This suggests upward bias in the coefficients in the single process model, possibly because of shared unobserved determinants. In other words, the estimated effects of union formation on migration (and vice versa) in the single process specification in part

6 The differences in coefficients between the single-process and multi-process models are tested using the Hausman Test. Both differences are statistically significant. 94 reflect selection of individuals with a tendency towards migration and union formation, rather than causal impact of one history on the other. The discrepancy between single- process and multi-process models in Tables 4-4 and 4-6 for the NLSY79 and Tables 4-5 and 4-7 for the NLSY97 indicates that not allowing for the correlation of disturbances results in a distorted impression of the causal relationships between moving and union formation.

Modeling the Hazard of Migration

Tables 4-4 and 4-5 present coefficients for effects on the hazard of migration in the NLSY79 and 97, respectively. In each table, the left-hand column presents the single- process estimates and the right-hand column presents the multi-process estimates. The random effects of migration in the single-process models (σmig = .13 and .31 for the

NLSY79 and 97, respectively, in Table 4.3) indicates that individuals who changed their residence in the past are significantly more likely to move again due to unmeasured person-specific characteristics in both cohorts.

Controlling for the unmeasured personal propensity to migrate in the single- process model, marriage increases the hazard of migration by about 82% in the NLSY79.

Even in the multi-process model of the NLSY79, marriage is a significant determinant of migration although the size of coefficient decreases by more than one half (b= .60 and b=

.28 in single-process and multi-process models, respectively). The positive relationship between migration and marriage is consistent with the well-known process of newlyweds establishing their household in a new place. In contrast, the total number of marriages is

95 negatively related to the hazard of migration in the single-process model (b= -.08, p=.073) in the NLSY79. But this variable becomes much smaller and loses statistical significance once the correlation between the propensity to move and to marry are accounted for in the multi-process model (b= -.01, p=.564), which indicates that unmeasured factors contribute to the significant relationship between the total number of marriages and migration. In contrast to the NLSY79, Table 4.5 shows that union formation in the last 6 months significantly decreases the likelihood of migration for the

NLSY97 respondents (b= -.09, p≤.01) although the significance disappears once unmeasured correlation is accounted for in the multi-process model (b= -.02, p=.147).

The total number of unions formed is positively related to the hazard of migration in the

NLSY97 in both the single-process and multi-process models (b=.13, p≤.001 and b=.03, p≤.001 for the single and multi-process models). This indicates that in the NLSY97, the negative effect of union formation on migration is derived from unmeasured features rather than a genuine relationship while the total number of unions increases the likelihood of migration.

The coefficients for the other explanatory variables in the migration equation change once unobserved heterogeneity is accounted for in the multi-process model in both cohorts, although the signs of most coefficients remain the same. Females in the

NLSY79 are less likely than males to move and the statistical significance becomes stronger in the multi-process model (from b= -.06, p=.141 to b= -.05, p≤.01). On the other hand, no significant gender effect appears in the NLSY97. African Americans and

Hispanics are less likely than their non-Black non-Hispanic white counterparts to move, 96 and the racial differences persist after accounting for the correlation between the migration and marriage random effects in both cohorts. To the extent that migration is an investment for purposes of socioeconomic gain (Cadwallader 1992), these results suggest that women, Black and Hispanics are less likely to make human capital investments that require moves (Quinn & Rubb 2011; Shauman & Noonan 2007).

Greater individual and household socioeconomic resources increase the hazard of migration in both the single-process and the multi-process models in the NLSY79 (Table

4.4). Those who have completed high school or college and who are employed part-time are more likely to move compared to those with less than a high school diploma and those not employed, respectively. Full-time employment, however, decreases the hazard of migration (as compared to being unemployed). Taken together, these results are consistent with a change in residence as a job search strategy. Those who have acquired sufficient human capital (education, employment experience) tend to settle down rather than move around (Schachter 2001). In the NLSY97, however, those who have completed high school are less likely to move even after controlling for person-specific characteristics (Table 4.5). Similar to the NLSY79, those having a college degree are more likely to move in the NLSY97. Moreover, those who are employed either part-time or full-time are less likely than non-employees to move in the NLSY97. If moving is assumed as a job search or improving career prospects (Cadwallader 1992; Schachter

2001), individuals with a high school degree in the NLSY97 have fewer opportunities.

Otherwise, it is also possible that they settle in one place similar to those employed either part-time or full-time. Although those living in metropolitan areas are less likely to move 97 in the NLSY79, their counterparts in the NLSY97 are more likely to do. In both cohorts, childbearing is not linked to the likelihood of mobility but the total number of children is negatively related to moving in both the single-process and multi-process models.

The effects of household characteristics become weaker once the simultaneous relationship between migration and union formation is accounted for in both cohorts. For the NLSY79 (see Table 4.4), those from an intact family are less likely to move (b= -.08, p≤.001 in the multi-process model) although an increase in maternal educational attainment is positively related to the hazard of migration (b=.02, p≤.001 in the multi- process model). Similarly, in the NLSY97 (Table 4.5), living in an intact family decreases the hazard of moving and an increase in maternal educational attainment is positively related to residential changes. However, the significant effect of living in an intact family disappears in the multi-process model in the NLSY97. Finally, residence characteristics are significantly related to the hazard of migration of individuals in both cohorts. An increase in the share of population with a college degree or higher increases the likelihood of migration by 1% (b=.01, p≤.001 in the multi-process model) in the

NLSY79 although it is negatively related to moving in the NLSY97 (b= -.004, p≤.001 in the multi-process model). The hazard of moving is also higher where county-level poverty rates are higher in both cohorts (b=.004, p≤.001 and b=.002, p≤.001 in the multi- process model of the NLSY79 and 97, respectively). However, sex ratio, unemployment and crime rates in the county of residence do not affect the hazard of migration in either the single-process or multi-process specification in the NLSY79. In the NLSY97, the hazard of moving increases with an increase in opposite sex ratios in the county (b=.01, 98 p≤.05 in the multi-process model) although it decreases with higher unemployment rates and crime rates in the county (b= -.013, p≤.001 and b= -.012, p≤.001 in the multi-process model, respectively).

Modeling the Hazard of Union Formation

The results from both the single-process and multi-process models for union formation are shown in Tables 4-6 and 4-7 for the NLSY79 and 97, respectively. Again, the person-specific random effects from the equations are presented in Table 4.3. The random effect of union formation (σunion =.10, .13 for the NLSY79 and 97) indicates that unmeasured factors make some individuals more prone to marry and form a union in both cohorts (i.e. those with a higher hazard of union formation in the past have a higher hazard in the present episode of union formation).

It is found that migration is positively associated with the hazard of marriage in the NLSY79 (b=.11, p≤.05); those who moved 1 year ago are approximately 12% more likely to marry (Table 4.6). However, once correlation between the random effects for migration and marriage is accounted for in the multi-process model, the significant effect of migration on marriage disappears (b=.04, p=.584). This suggests that the positive effect of past migration on the hazard of marrying, according to the single process model, actually reflects association between the two propensities due to other unmeasured factors. The total number of moves in the NLSY79 does not appear significant in both the single and multi-process models. Frequent movers may face various new environments to which they must adapt and this process may hinder other major decisions such as

99 marriage, which in turn shows no association between marriage and migration in our model. On the other hand, migration and the total number of moves are positively related to the hazard of union formation in the NLSY97 (b=.05, p≤.05 and b=.05, p≤.001 in the multi-process model) (Table 4.7). Once controlling for the unobserved factors that affect both moving and union formation in the multi-process model, the negative effect of moving on union formation in the NLSY97 becomes positive and statistically significant.

Although the unmeasured residuals can be various facets of individual specific characteristics, the results show that moving itself serves as a positive life event for those in the NLSY97 with regard to union formation. The positive impact also appears in that those who have multiple migration experiences are significantly more likely to form a union in the NLSY97. These indicate that mobility improves movers’ marriage markets

(Jampaklay 2006) or helps the young become independent (Garasky, Haurin, & Haurin

2001).

Accounting for personal propensity also reduces the estimated effects of most explanatory variables in the union formation equation. The directions of effects are stable in the NLSY79 after controlling for the person-specific heterogeneity although distinct changes are found in the NLSY97. In both cohorts, females have a higher hazard of union formation (i.e. earlier marriage or cohabitation) than males, and African-Americans and

Hispanics have a lower hazard of marrying than non-Black non-Hispanic whites.

Individual socioeconomic characteristics are also significantly related to the hazard of union formation. Hispanics in the NLSY79 are less likely to marry in both single-process and multi-process models but their counterparts in the NLSY97 are significantly more 100 likely to form a union (b=.05, p≤.01 in the multi-process model). In the NLSY97, unobserved factors exist and influence Hispanic’s hazard of union formation in other direction (b= -.08 in the singles-process model to b=.05 in the multi-process model). In the NLSY79, those having a high school diploma or a college and higher degree have a higher hazard of marrying (i.e. earlier marriage) than those with less than a high school diploma. However, their counterparts in the NLSY97 are less likely to form a union although the significant effect of having a college degree on union formation disappears in the multi-process model. Moreover, employees (either part-time or full-time) are more likely than those not employed to marry in the NLSY79. Indeed, employment status in the preceding year is the most potent factor affecting the hazard of marriage in the

NLSY79; the odds ratio increases by 23% and 49% for part-time and full-time employees

(in the multi-process model), respectively. In the NLSY97, however, part-time employees are less likely to be involved in a union but full-time employees are more likely to form a union, compared to those who are not employed. Childbirth increases the hazard of marriage (b= .06, p=.051 in the multi-process model) and total number of children is positively related to the hazard of marriage (b= .01, p≤ .05 in the multi-process model) in the NLSY79. Although the total number of children is also positively related to the hazard of union formation, childbearing decreases the likelihood of union formation in the NLSY97 (b=-.11, p≤.01 in the multi-process model). Couples may have already been together before the baby is born (e.g. after pregnancy).

Household and residence characteristics are also significantly related to the hazard of union formation in both cohorts although the effects generally are smaller in 101 magnitude than the effects of individual characteristics. Maternal education and having lived in an intact family are negatively associated with the hazard of union formation in both cohorts. After accounting for the unobserved characteristics, however, the significant effect of family structure disappears in the NLSY97. In the NLSY79, the hazard of marriage is negatively related to the share of the population with a college degree or higher and to the local crime rate (b= -.01, p≤.01 and b= -.005, p≤.05 in the multi-process model, respectively). On the other hand, the higher poverty rates in the county of residence, the higher the hazard of marrying in the NLSY79 (b= .002, p≤.05 in the multi-process model). Sex ratio and unemployment rates in the county are not significantly related to the hazard of marriage in the multi-process model in the earlier cohort. Yet, in the NLSY97, increases in the opposite sex ratio, proportion of population with a college degree or more, and unemployment rates in the county of residence are related to lower odds of union formation. Moreover, it is found that the higher county crime rates, the higher the hazard of union formation in the NLSY97 (b=.01, p≤.01 in the multi-process model).

Robustness of the Results

Although identification without exclusion restrictions fares well, to ensure the robustness of the results I estimated the multi-process model with selected explanatory variables excluded (and therefore serving to identify the model). First, it is assumed that the sex ratio in the place of residence is directly related to individual marital decisions

(South & Lloyd 1992) but not to migration, and therefore can serve to identify the

102 marriage equation (see model 1 in Tables 4.8 to 4.11). Regarding migration, people take into consideration the local environment such as safety of the residence or milder weather when considering migration (Chen & Rosenthal 2008; Whisler et al. 2008). Therefore, county crime rates are served as an identifying variable (see model 2 in Tables 4.8 to

4.11). As it turned out, the regression results are remarkably robust to these alternative specifications; none of them results in markedly different coefficient values (see Tables

4.8 to 4.11). On the basis of this set of tests, I conclude that the results in this study are robust to alternative model identification strategies.

Moreover, the effect on estimates of the treatment of cohabitation-related moves is examined for the NLSY79. In the estimates presented above, I excluded moves occurring one year after the onset of cohabitation (i.e. 4,709 person-years of 66 individuals are dropped from 92,640 person-years of 7,893 individuals). Moves occurring in the same year as the onset of cohabitation may also intervene between cohabitation and marriage, and therefore should be excluded (i.e. 7,077 person-years of 127 respondents).

Excluding these additional moves has minimal impact on the estimates (coefficients differ only slightly). In short, the results are robust to the choice of criteria for identifying cohabitation-related moves.

Discussion

The current study investigates the relationships between migration and union formation in the United States allowing for complex inter-dependency between the two life course histories. A few studies using European samples have allowed for complexity

103 of the form incorporated in the present study, but no such study has been conducted using

U.S. data and cohort comparison. Thus, not only do the findings from this study improve our understanding of the association between migration and union formation, they also add a country case study that, when compared to results in other settings, will begin to illuminate cross-country variability in the nature and strength of the inter-dependencies between these two processes. Moreover, this study compares the inter-dependency between major life course transitions by cohort, and thus rectifies endogeneity over time and contributes to better understanding of the dynamic of the life course.

Findings from this study reveal that the multi-process model which accounts for unobserved propensity to move and to form a union yields different estimates than a single process model in both cohorts. This suggests that significant unmeasured correlation between migration and union formation exists, reflected in a tendency for individuals to migrate and form a union in a short time or even simultaneously. This is consistent with findings from a previous study using a West German sample (Mulder &

Wagner 1993). Accounting for endogeneity between life course transitions yields less biased estimates, whereas ignoring this form of inter-dependency leads to a mistaken impression of the direct effects of certain features of each history on the other. Future research on the life course will benefit from more complete accounting for interrelationships between life events in the model.

Second, after accounting for person-specific characteristics (e.g. propensity to move or form a union), it is found that marriage is positively related to the risk of

104 migration in the short-term, but over the longer-term the total number of marriages does not significantly affect the hazard of migration in the NLSY79. On the other hand, union formation is not related to the likelihood of moving in the NLSY97 although the total number of unions increases the hazard of moving. These findings suggest that migration occurs in anticipation of family events but over the long haul increased social ties via family events can dampen the risk of moving in the earlier cohort. However, the sample of persons experiencing multiple marriages is small in the NLSY79 so this result should be viewed with caution. In the NLSY97 where I include both cohabitation and marriage with 6 months lag, union formation does not trigger migration. Since the union formation is included with 6 months lag, this result may indicate that couples do not wait for 6 months to move in together. Nevertheless, the total number of unions increases the hazard of moving in the NLSY97. As many young adults from the NLSY97 cohabit rather than marry, the higher order unions likely are cohabitation. If it is the case, this result shows that cohabitation as an exploration of a union helps those in the NLSY97 become independent to move to a new place.

Although cohabitation is not as common in the NLSY79 as the NLSY97, some of marriages began with cohabitation in the earlier cohort, especially remarriages (Smock

2000; Cherlin 2010). There has been evidence of a growing tendency in the U.S. to remain in a cohabiting status rather than transition to marriage if the partnership is second or higher order and especially if one or the other partner is previously married (Smock

2000; Cherlin 2010). Hence incorporating the cohabitation history in research on migration-marriage inter-relations would be desirable. However, this study does not 105 include cohabitation in the NLSY79 for the following reasons. The cohabitation history was constructed through retrospective inquiry in 1990, which raises concerns about its completeness and accuracy. There is non-comparability introduced by changes in the definition of ‘partner’ across survey years (Center for Human Resources 2013).

Moreover, only since 2002 have respondents been asked detailed questions about brief cohabitation episodes (i.e. three months) (Center for Human Resources 2013). Although data analyses in the chapter 3 are hardly affected by the cohabitation measures (because it restricts to ages 16 to 30 when they are likely on the same measurement scheme), the current study, which includes important life course transitions over the three decades, may be more susceptible to the changes in measurement. Therefore, cohabitation histories are not incorporated in this research, except for removing cohabitation-related migration7. Nevertheless, for comparison of the NLSY79 and 97, I include marriage and cohabitation in the NLSY97 into the same category as only few individuals are involved in a marital relationship. Certainly these different definitions of union formation may have affected the estimation of the single-process and multi-process model in this study.

It would be desirable to have future research which investigates to what extent the relationship between migration and union transitions is affected by marriage and cohabitation experiences, thereby drawing a more complete portrait of inter-dependencies between life course transitions.

Regarding the effects of migration on union formation, I find no significant effect once unobserved heterogeneity has been controlled via the multi-process model with

7 For more information about cohabitation variables, refer to Appendix A. 106 random effects in the NLSY79. My expectation that migration is a strategy to rectify failed marriage markets is not supported in the earlier cohort. In the previous literature, effects of moving on union formation has been understood as an anticipation or plan for family changes (Feijten & Mulder 2002). In this study, the possible anticipating or planning effect is partially ruled out when I drop moves which are preceded by cohabitation. Furthermore, an inclusion of unobserved personal components in the model controls for possible endogeneity of the planned behaviors. The absence of significant effects of migration on marriage in the NLSY79, therefore, suggests that migration is not motivated much by pursuit of marriage opportunities, rather it is a life course transition driven by other life decisions in the earlier cohort. On the other hand, migration significantly increases the likelihood of union formation in the NLSY97 when unobserved characteristics are controlled for. It indicates that young adults in the recent cohort move to new places before forming a union. Otherwise, it is possible that moving improves a mover’s marriage market in the recent cohort, which is consistent with the findings from chapter 3. These conclusions must be treated warily, however: the time- metric is year and six months, and a clear portrayal of migration-marriage inter-relations may require more fine-grained measurement (months or even weeks). For example, using

Finish data, Kulu and Steele (2013) found that couples are more likely to get pregnant during the first months after a move but the risk of pregnancy decreases and becomes stable afterwards (i.e., one year after the move). In addition, the relationships between migration and marriage may vary by types of mobility e.g. within-county versus between- county moves. The two different types of mobility have had different purposes in the

107

U.S.: between-county moves are predominantly motivated by employment considerations while within-county moves often occur as a response to changes in the family configuration such as the birth of a child (Schachter 2001). More precise measurement of migration and union histories (e.g. clearer distinction among different forms of migration and types of partnership) will provide a stronger foundation for a valid assessment of the association between these two life course transitions.

Finally, findings from this study show that unobserved features likely have larger impacts on the estimation of the life course in the NLSY97 than that of the NLSY79.

After controlling for person-specific characteristics which are not included in the model, more coefficients are notably changed and even their direction is converted into the other in the recent cohort. This indicates that major demographic and household factors are not sufficient to elaborate the transition to adulthood in the recent cohort. In other words, life transitions among recent young adults have become more contingent on subjectivity. If so, failure to account for the unmeasured heterogeneity could draw misrepresented findings and distort a genuine relationship between life course transitions, particularly for contemporary young adulthood.

108

Chapter 4 tables

Table 4.1 Description of migration and marriage histories

Migration History Marital History NLSY79 NLSY97 NLSY79 NLSY97 Never moved .50 .64 Marr Union Marr Cohabit 1 .21 .13 0 .21 .25 .86 .39 2 .13 .11 1 .61 .29 .13 .39 3 .07 .07 2 .16 .27 .01 .15 4 .04 .05 3 .02 .11 .00 .05 5 .02 .002 more than 3 .01 .08 .00 .02 More than 5 .03 .000 Notes: 87,931 person-years for 7,827 respondents 1,185,485 person-months for 8,853 individuals for the NLSY79 and NLSY97, respectively, are included. Union duration of those who are still married until last interview was calculated using the year of last interview. All statistics were adjusted under survey setting in Stata. Numbers in parentheses are standard errors.

109

Table 4.2 Description of covariates included in the model

NLSY79 NLSY97 Female .47 .48 Race White .72 .69 Black .21 .17 Hispanic .07 .14 Education Less than high school .22 .11 High school or equivalent .60 .54 College or more .18 .35 Employment Employed part-time .44 .25 Employed full-time .44 .42 Birth of a child .30 .23 Total number of children (numbers) 2.24 (.03) 0.75 (.03) Household Characteristics Had lived in an intact family until age 18 .64 .53 Maternal educational attainment (years) 11.62 (.09) 12.80 (.09) Living in metro areas .84 .84 County characteristics Female population (%) 51.25 (.11) 51.21 (.10) Male population (%) 48.65 (.10) 48.82 (.11) Population with a college degree or higher (%) 10.96 (.21) 19.78 (.66) Unemployment rates (%) 3.87 (.08) 6.84 (.18) Poverty rates (%) 1.62 (.04) 13.25 (.49) Crime rate (%) 5.86 (.14) 5.47 (.20) Notes: 87,931 person-years for 7,827 respondents 1,185,485 person-months for 8,853 individuals for the NLSY79 and NLSY97, respectively, are included. All statistics were adjusted under survey setting in Stata. Numbers in parentheses are standard errors.

110

Table 4.3 Estimated Random-effects

NLSY79 NLSY97 Migration Union Formation Migration Union Formation Migration .13 .31 (.11, .15) (.29, .32) Union Formation .24 .10 .30 .13 (.22, .26) (.07, 13) (.27, .32) (.11, .14) Notes: Numbers in parentheses are 95% CI.

111

Table 4.4 Estimates from models for migration (NLSY79)

Variable Single process Multi process Coeff. O.R. Coeff. O.R. Marriage Married 1 year before migration .60 (.16) *** 1.82 .28 (.08) ** 1.33 Total number of marriages -.08 (.05) † .92 -.01 (.02) ns .99 Duration Duration to migration -.11 (.02) *** .90 -.09 (.01) *** .91 Squared duration to migration .01 (.00) ** 1.01 .01 (.00) *** 1.01 Cubical duration to migration -.00 (.00) *** 1.00 -.00 (.00) *** 1.00 Individual characteristics Female -.06 (.04) ns .94 -.05 (.02) ** .95 Black -.49 (.04) *** .61 -.21 (.02) *** .81 Hispanic -.31 (.05) *** .73 -.14 (.02) *** .87 High school or equivalent .52 (.04) *** 1.69 .26 (.02) *** 1.29 College or more .27 (.04) *** 1.31 .18 (.02) *** 1.20 Employed part-time .21 (.04) ** 1.24 .11 (.02) *** 1.12 Employed full-time -.35 (.05) *** .70 -.15 (.02) *** .86 Living in metro areas -.20 (.04) *** .82 -.09 (.02) *** .92 Birth of a child -.04 (.08) ns .96 -.04 (.04) ns .97 Total number of children -.13 (.02) *** .88 -.04 (.01) *** .97 Household characteristics R lived in an intact family until age 18 -.17 (.03) *** .84 -.08 (.01) *** .92 Mom’s education .05 (.01) *** 1.05 .02 (.00) *** 1.02 County characteristics Sex ratio -.00 (.01) ns 1.00 -.00 (.00) ns 1.00 Population with a college degree/higher .02 (.00) *** 1.02 .01 (.00) *** 1.01 Unemployment rates .01 (.01) ns 1.01 .00 (.00) ns 1.00 Poverty rates .01 (.00) *** 1.01 .00 (.00) *** 1.00 Crime rates -.01 (.00) ns .99 -.00 (.00) ns 1.00 Intercept -2.90 (.49) *** .06 -1.31 (.19) *** .27 Log likelihood -22900.9 -46092.4 Wald Chi2 (20) 1709.5 4303.4 Person-years 87,931 Number of observations 7,827 Notes: Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

112

Table 4.5 Estimates from models for migration (NLSY97)

Variable Single process Multi process Coeff. O.R. Coeff. O.R. Union formation Union formed 6 months before migration -.09 (.03) ** .91 -.02 (.01) ns .98 Total number of unions .13 (.01) *** 1.14 .03 (.00) *** 1.03 Duration Duration to migration -.00 (.00) ns 1.00 -.01 (.00) *** .99 Squared duration to migration -.00 (.00) ns 1.00 .00 (.00) *** 1.00 Cubical duration to migration .00 (.00) ns 1.00 -.00 (.00) *** 1.00 Individual characteristics Female -.02 (.05) ns .98 -.00 (.01) ns 1.00 Black -.25 (.05) *** .78 -.05 (.01) *** .95 Hispanic -.36 (.05) *** .70 -.09 (.01) *** .92 High school or equivalent -1.27 (.03) *** .28 -.41 (.01) *** .67 College or more .46 (.04) *** 1.58 .10 (.01) *** 1.11 Employed part-time -.29 (.03) *** .75 -.12 (.01) *** .89 Employed full-time -.21 (.02) *** .81 -.11 (.01) *** .86 Living in metro areas .36 (.04) *** 1.44 .14 (.01) *** 1.15 Birth of a child .03 (.05) ns 1.03 -.00 (.02) ns 1.00 Total number of children -.25 (.02) *** .78 -.07 (.01) *** .93 Household characteristics R lived in an intact family until age 18 -.08 (.04) * .93 -.01 (.01) ns .99 Mom’s education .07 (.01) *** 1.08 .02 (.00) *** 1.02 County characteristics Sex ratio .02 (.01) ns 1.02 .01 (.00) * 1.01 Population with a college degree/higher -.02 (.00) *** .99 -.00 (.00) *** 1.00 Unemployment rates -.04 (.01) *** .96 -.01 (.00) *** .99 Poverty rates .01 (.00) ** 1.01 .00 (.00) *** 1.00 Crime rates -.05 (.01) *** .96 -.01 (.00) *** .99 Intercept -5.91 (.63) *** .00 -2.21 (.12) *** .11 Log likelihood -72001.9 -89964.1 Wald Chi2 (22) 2777.6 7445.6 Person-months 1,045,943 Number of observations 8,154 Notes: Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

113

Table 4.6 Estimates from models for marriage (NLSY79)

Variable Single process Multi process Coeff. O.R. Coeff. O.R. Migration Moved 1 year before marriage .11 (.05) * 1.12 .04 (.03) ns 1.05 Total number of moves .02 (.02) ns 1.02 .00 (.01) ns 1.00 Duration Duration to marriage .18 (.02) *** 1.20 .07 (.01) *** 1.07 Squared duration to marriage -.01 (.00) *** .99 -.01 (.00) *** .99 Cubical duration to marriage .00 (.00) *** 1.00 .00 (.00) *** 1.00 Individual characteristics Female .29 (.04) *** 1.34 .12 (.02) *** 1.12 Black -.83 (.04) *** .43 -.36 (.02) *** .70 Hispanic -.16 (.05) ** .85 -.08 (.02) *** .93 High school or equivalent .34 (.04) *** 1.40 .16 (.02) *** 1.17 College or more .26 (.04) *** 1.30 .12 (.02) *** 1.13 Employed part-time .44 (.05) *** 1.55 .21 (.02) *** 1.23 Employed full-time .79 (.05) *** 2.21 .40 (.02) *** 1.49 Living in metro areas -.11 (.04) * .90 -.05 (.02) ** .95 Birth of a child .16 (.07) * 1.17 .06 (.03) † 1.06 Total number of children -.01 (.02) ns .99 .01 (.01) * 1.01 Household characteristics R lived in an intact family until age 18 -.06 (.03) † .94 -.03 (.01) * .97 Mom’s education -.02 (.01) *** .98 -.01 (.00) *** .99 County characteristics Sex ratio .02 (.01) † 1.02 .01 (.00) ns 1.01 Population with a college degree/higher -.01 (.00) ** .99 -.01 (.00) ** .99 Unemployment rates -.00 (.01) ns 1.00 .00 (.00) ns 1.00 Poverty rates .01 (.00) * 1.01 .00 (.00) * 1.00 Crime rates -.01 (.00) * .99 -.00 (.00) * 1.00 Intercept -4.05 (.49) *** .02 -1.99 (.20) *** .14 Log likelihood -23270.2 -46092.4 Wald Chi2 (20) 1396.4 4303.4 Person-years 87,931 Number of observations 7,827 Notes: Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

114

Table 4.7 Estimates from models for union formation (NLSY97)

Variable Single process Multi process Coeff. O.R. Coeff. O.R. Migration Moved 6 months before union formation -.12 (.03) ** .89 .05 (.02) * 1.06 Total number of moves .18 (.01) *** 1.20 .05 (.01) *** 1.05 Duration Duration to union -.03 (.00) *** .97 .00 (.00) ** 1.00 Squared duration to union .00 (.00) *** 1.00 -.00 (.00) *** 1.00 Cubical duration to union -.00 (.00) *** 1.00 .00 (.00) *** 1.00 Individual characteristics Female .42 (.03) *** 1.52 .13 (.02) *** 1.14 Black -.34 (.03) *** .71 -.19 (.02) *** .83 Hispanic -.08 (.04) * .92 .05 (.02) ** 1.05 High school or equivalent -.71 (.03) *** .49 -.34 (.03) *** .71 College or more -.53 (.03) *** .59 -.02 (.02) ns .98 Employed part-time -.03 (.03) ns .97 -.11 (.02) *** .90 Employed full-time .29 (.02) *** 1.34 .04 (.02) * 1.04 Living in metro areas .06 (.03) * 1.07 -.12 (.02) *** .89 Birth of a child .17 (.04) *** 1.18 -.11 (.03) ** .90 Total number of children -.10 (.01) *** .90 .09 (.01) *** 1.10 Household characteristics R lived in an intact family until age 18 -.30 (.03) *** .74 -.00 (.01) ns 1.00 Mom’s education -.02 (.00) *** .98 -.01 (.00) * .99 County characteristics Sex ratio -.02 (.01) *** .98 -.01 (.00) * .99 Population with a college degree/higher -.01 (.00) *** .99 -.01 (.00) *** .99 Unemployment rates -.03 (.01) *** .97 -.01 (.00) *** .99 Poverty rates .00 (.00) ns 1.00 .00 (.00) ns 1.00 Crime rates .00 (.00) ns 1.00 .01 (.00) ** 1.01 Intercept -2.63 (.46) *** .07 -2.02 (.25) *** .13 Log likelihood -69731.8 -89964.1 Wald Chi2 (22) 2978.5 7445.6 Person-months 1,045,943 Number of observations 8,154 Notes: Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

115

Table 4.8 Robustness of the multi-process model estimation with different identification variables for migration (NLSY79)

Variable Model 1 Model 2 Model 3 Marriage Married 1 year before migration .282 (.08) ** .282 (.08) ** .312 (.08) *** Total number of marriages -.011 (.02) ns -.012 (.02) ns -.036 (.02) † Duration Duration to migration -.093 (.01) *** -.093 (.01) *** -.080 (.01) *** Squared duration to migration .007 (.00) *** .007 (.00) *** .005 (.00) *** Cubical duration to migration -.000 (.00) *** -.000 (.00) *** -.000 (.00) *** Individual characteristics Female -.045 (.01) ** -.051 (.02) ** -.055 (.02) ** Black -.213 (.02) *** -.213 (.02) *** -.205 (.02) *** Hispanic -.142 (.02) *** -.142 (.02) *** -.141 (.02) *** High school or equivalent .257 (.02) *** .257 (.02) *** .253 (.02) *** College or more .181 (.02) *** .181 (.02) *** .182 (.02) *** Employed part-time .112 (.02) *** .112 (.02) *** .116 (.02) *** Employed full-time -.155 (.02) *** -.155 (.02) *** -.162 (.02) *** Living in metro areas -.088 (.02) *** -.087 (.02) *** -.091 (.02) *** Birth of a child -.035 (.04) ns -.035 (.04) ns -.050 (.04) ns Total number of children -.036 (.01) *** -.035 (.01) *** -.040 (.01) *** Household characteristics R lived in an intact family -.080 (.01) *** -.080 (.01) *** -.069 (.01) *** Mom’s education .019 (.00) *** .019 (.00) *** .021 (.00) *** County characteristics Sex ratio - -.003 (.00) ns -.003 (.00) ns Population with college degree or higher .007 (.00) *** .007 (.00) *** .007 (.00) *** Unemployment rates .004 (.00) ns .003 (.00) ns .001 (.00) ns Poverty rates .004 (.00) *** .004 (.00) *** .005 (.00) *** Crime rates -.003 (.00) ns -.002 (.00) ns -.002 (.00) ns Intercept -1.44 (.05) *** -1.31 (.19) *** -1.32 (.20) *** Log likelihood -46092.7 -46095.5 -44326.0 Wald Chi2 4302.7 4298.2 4207.1 Person-years 87,931 87,931 85,563 Number of Observations 7,827 7,827 7,766 Notes: Model1 (without sex-ratio in the migration model), Model2 (without crime rates in the marriage model), Model3 (drop migration events related to cohabitation in the same year or 1 year before the migration). Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

116

Table 4.9 Robustness of the multi-process model estimation with different identification variables for marriage (NLSY79)

Variable Model 1 Model 2 Model 3 Migration Moved 1 year before marriage .044 (.03) ns .044 (.03) ns .045 (.03) ns Total number of moves .004 (.01) ns .004 (.01) ns .005 (.01) ns Duration Duration to marriage .072 (.01) *** .072 (.01) *** .074 (.01) *** Squared duration to marriage -.006 (.00) *** -.006 (.00) *** -.006 (.00) *** Cubical duration to marriage .000 (.00) *** .000 (.00) *** .000 (.00) *** Individual characteristics Female .116 (.02) *** .116 (.02) *** .118 (.02) *** Black -.356 (.02) *** -.361 (.02) *** -.357 (.02) *** Hispanic -.077 (.02) *** -.083 (.02) *** -.080 (.02) *** High school or equivalent .159 (.02) *** .160 (.02) *** .160 (.02) *** College or more .118 (.02) *** .119 (.02) *** .116 (.02) *** Employed part-time .210 (.02) *** .212 (.02) *** .215 (.02) *** Employed full-time .396 (.02) *** .398 (.02) *** .402 (.02) *** Living in metro areas -.050 (.02) ** -.060 (.02) ** -.057 (.02) ** Birth of a child .062 (.03) † .063 (.03) † .062 (.03) † Total number of children .015 (.01) * .014 (.01) * .015 (.01) * Household characteristics R lived in an intact family -.027 (.01) * -.025 (.01) + -.024 (.01) + Mom’s education -.009 (.00) *** -.009 (.00) *** -.010 (.00) *** County characteristics Sex ratio .006 (.00) ns .006 (.00) ns .006 (.00) ns Population with college degree or higher -.005 (.00) ** -.006 (.00) *** -.005 (.00) * Unemployment rates .000 (.00) ns -.001 (.00) ns -.000 (.00) ns Poverty rates .002 (.00) * .002 (.00) * .002 (.00) * Crime rates -.005 (.00) * - -.005 (.00) * Intercept -2.00 (.20) *** -1.99 (.20) *** -2.00 (.20) *** Log likelihood -46092.7 -46095.5 -44326.0 Wald Chi2 4302.7 4298.2 4207.1 Person-years 87,931 87,931 85,563 Number of Observations 7,827 7,827 7,766 Notes: Model1 (without sex-ratio in the migration model), Model2 (without crime rates in the marriage model), Model3 (drop migration events related to cohabitation in the same year or 1 year before the migration). Numbers in parentheses are standard errors. ns: not significant, †p≤.10, * p≤.05, ** p≤.01, *** p≤.001

117

Table 4.10 Robustness of the multi-process model estimation with different identification variables for migration (NLSY97)

Variable Model 1 Model 2 Union formation Union formation 6 months before migration -.019 (.01) ns -.019 (.01) ns Total number of unions .025 (.00) *** .025 (.00) *** Duration Duration to migration -.009 (.00) *** -.009 (.00) *** Squared duration to migration .000 (.00) *** .000 (.00) *** Cubical duration to migration -.000 (.00) *** -.000 (.00) *** Individual characteristics Female .010 (.01) ns -.003 (.01) ns Black -.055 (.01) *** -.055 (.01) *** Hispanic -.086 (.01) *** -.085 (.01) *** High school or equivalent -.407 (.01) *** -.407 (.01) *** College or more .102 (.01) *** .102 (.01) *** Employed part-time -.118 (.01) *** -.118 (.01) *** Employed full-time -.114 (.01) *** -.115 (.01) *** Living in metro areas .143 (.01) *** .143 (.01) *** Birth of a child -.004 (.02) ns -.004 (.02) ns Total number of children -.073 (.01) *** -.073 (.01) *** Household characteristics R lived in an intact family -.010 (.01) ns -.010 (.01) ns Mom’s education .017 (.00) *** .017 (.00) *** County characteristics Sex ratio - .005 (.00) * Population with college degree or higher -.004 (.00) *** -.004 (.00) *** Unemployment rates -.013 (.00) *** -.013 (.00) *** Poverty rates .002 (.00) *** .002 (.00) *** Crime rates -.012 (.00) *** -.012 (.00) *** Intercept -1.94 (.03) *** -2.20 (.12) *** Log likelihood -89966.7 -89968.8 Wald Chi2 7441.7 7439.5 Person-months 1,045,943 1,045,943 Number of Observations 8,154 8,154 Notes: Model1 (without sex-ratio in the migration model), Model2 (without crime rates in the union model). Numbers in parentheses are standard errors. ns: not significant, * p≤.05, ** p≤.01, *** p≤.001

118

Table 4.11 Robustness of the multi-process model estimation with different identification variables for union formation (NLSY97)

Variable Model 1 Model 2 Migration Moved 6 months before union formation .055 (.02) * .054 (.02) * Total number of moves .045 (.01) *** .044 (.01) *** Duration Duration to union .003 (.00) ** .003 (.00) ** Squared duration to union -.000 (.00) *** -.000 (.00) *** Cubical duration to union .000 (.00) *** .000 (.00) *** Individual characteristics Female .133 (.02) *** .130 (.02) *** Black -.192 (.02) *** -.182 (.02) *** Hispanic .051 (.02) ** .061 (.02) ** High school or equivalent -.337 (.03) *** -.337 (.03) *** College or more -.019 (.02) ns -.020 (.02) ns Employed part-time -.106 (.02) *** -.106 (.02) *** Employed full-time .037 (.02) * .037 (.02) * Living in metro areas -.120 (.02) *** -.110 (.02) *** Birth of a child -.108 (.03) ** -.110 (.03) ** Total number of children .095 (.01) *** .096 (.01) *** Household characteristics R lived in an intact family -.002 (.01) ns -.003 (.01) ns Mom’s education -.006 (.00) * -.006 (.00) * County characteristics Sex ratio -.012 (.00) * -.011 (.00) * Population with college degree or higher -.008 (.00) *** -.007 (.00) *** Unemployment rates -.014 (.00) *** -.014 (.00) *** Poverty rates .001 (.00) ns .002 (.00) ns Crime rates .008 (.00) ** - - Intercept -2.01 (.25) *** -2.02 (.25) *** Log likelihood -89966.7 -89968.8 Wald Chi2 7441.7 7439.5 Person-months 1,045,943 1,045,943 Number of Observations 8,154 8,154 Notes: Model1 (without sex-ratio in the migration model), Model2 (without crime rates in the union model). Numbers in parentheses are standard errors. ns: not significant, * p≤.05, ** p≤.01, *** p≤.001

119

CHAPTER 5: CONCLUSIONS

The transition to adulthood is a critical time for young adults when major life changes occur intensively. For recent cohorts, rapid technological development and economic globalization require young adults to have advanced skills and knowledge to compete well in the labor market, which results in demographic shifts such as a delay of family formation (Beck & Beck-Gernsheim 2002; Settersten 2005). Moreover, the transition to adulthood has become extended and individualized in recent years.

Nevertheless, little empirical evidence has existed and thus underlying dynamics of the transition to adulthood have been less clear. To clearly comprehend how the transition to adulthood has changed over time, I use longitudinal information of young adults from two cohorts and compare their life course transitions in the current study.

In chapter 2, distinct patterns of the transition to adulthood are identified by gender and cohort. Five classes for males (i.e. traditional transition without higher education, procrastinated transition, contemporary family formation without higher education, delayed family formation self-focused pathway, and traditional transition with higher education) and six classes for females (i.e. contemporary family formation without higher education, procrastinated transition, independent cohabitors, independent married, contemporary family formation with higher education, and traditional pathways) are

120 appropriate to describe the key life course events during young adulthood. When I compare the identified classes by cohort, the NLSY79 males are more likely to belong to either the traditional transition without higher education or delayed family formation self- focused classes while their counterparts in the NLSY97 are more likely to be in procrastinated transition, contemporary family formation without higher education, and delayed family formation self-focused classes. On the other hand, more than half of females from the NLSY79 are involved in traditional pathways and those from the

NLSY97 belong to contemporary family formation without higher education and independent cohabitors. These results suggest that despite the general tendency toward a delayed transition to adulthood in the NLSY97, recent young adults show more complicated life course pathways. In addition, I find that a large proportion of young adults in the NLSY79, both men and women, take traditional pathways where six life course events (i.e. family formation, education, employment, and home-leaving) are completed by age 25. On the other hand, the most prevalent class for both men and women in the NLSY97 is the contemporary family formation without higher education.

This result is consistent with previous studies demonstrating that increasing cohabitation is the most important change in young adults’ life in recent years (Schoen et al. 2007).

The second largest class of both gender in the NLSY97 is the procrastinated group of people who hardly make any transition by age 25, and therefore represents an extended period of exploration for independence. From this chapter, it is evident that young adults in the recent cohort reveal more complex pathways in part due to the increasing prevalence of cohabitation and increasing self-focused time. Although this chapter 2

121 provides a holistic view on the life course events during the transition to adulthood, it does not necessarily identify specific changes in the relationship between life courses which probably hold greater variability (Brückner & Mayer 2005).

In chapter 3, I therefore examine young adults’ moving experiences and their relationship with first union formation and compare them by cohort. The findings reveal that the life course transitions (i.e. moving and union formation) are closely related with each other in both cohorts and the complicated association differs between the NLSY79 and 97. In the NLSY79, moving is hardly associated with union formation except that migration to another county with the same unemployment level decreases the hazard of marriage versus no union. When marriage is compared to cohabitation, mobility decreases the relative risk of marriage, although migration to a place with similar economic conditions does not differentiate the risk of marriage from cohabitation in the

NLSY79. Moreover, moving to another county with better economic conditions increases the likelihood of cohabitation compared to no union for the NLSY79 respondents. These results reveal that for those in the earlier cohort moving represents a negative experience for singles that discourages marriage. On the other hand, no significant association is found between mobility and marriage among those from the NLSY97. Cohabitation, on the other hand, appears significantly related to residential mobility and moving to the worse economic counties. Mobility likely serves as a learning process to become independent because moving within the same county increases the likelihood of cohabitation compared to no union in this cohort. As young adults in the NLSY97 go through the worst economic recession since the Great Depression, they may live with 122 partners to cope with economic hardships. This chapter therefore reveals that the relationship between moving and union formation varies by socioeconomic context that an individual is embedded. Despite the significant findings from this chapter, endogeneity may have existed between life courses and influenced the estimation. This possible endogeneity between migration and union formation is addressed and tested in the Chapter 4.

Chapter 4 investigates the interrelationship between migration and union formation allowing for complex inter-dependency between the two life course histories in a multi-process model. Moreover, the inter-dependency between the life course transitions is compared by cohort. Findings from this chapter reveal that the multi- process model with inter-dependency yields different estimates than a single process model without the inter-dependency in both cohorts. This suggests that significant unmeasured correlations between migration and union formation exist, reflecting a tendency for individuals to migrate and form a union in a short time or even simultaneously. I also find that unobserved features have larger impacts on the estimation of the life course in the NLSY97 compared to the NLSY79. After controlling for person- specific characteristics which are not included in the model, more coefficients are notably changed and even their direction is converted into the other in the recent cohort. This indicates that major demographic and household factors are not sufficient to elaborate the transition to adulthood in the recent cohort. In other words, life transitions among recent young adults have become more contingent on subjectivity. Therefore, failure to account for the unmeasured heterogeneity could draw misrepresented findings and distort a 123 genuine relationship between life course transitions, particularly for contemporary young adults.

The findings in the current study are in line with the recent theoretical approach to the transition to adulthood which is reflecting the Second Demographic Transition.

Studies have shown that young adulthood is de-standardized, de-institutionalized, and diverse despite some commonality by institutional and cultural determinants (Billari &

Liefbroer 2010; Brückner & Mayer 2005). The current study finds that the United States, where unique demographic shifts appear, also reveals de-standardization in young adulthood as both men and women in the recent cohort show a more varied distribution of the pathway to adulthood. Nevertheless, it is also found that the transition to adulthood of contemporary young adults is converged on some degree of standardization which is delayed and self-focused, particularly with respect to family formation. This is consistent with the changes occurring to young adulthood in European countries; postponement for entry into marriage and increasing prevalence of cohabitation characterize the transition to adulthood in 25 European countries over four decades (Billari & Liefbroer 2010).

Findings from this study also show that the nexus between life course events during the transition to adulthood has changed over time. The relationship between moving and union formation differs between cohorts, and this again supports the de- standardization theorem. In the de-standardized life course, universal or normative expectations about the timing and connection of the life courses have weakened, instead the life course experiences are varied by individuals (Brückner & Mayer 2005). In

124 addition, the dynamics observed in the relationship between moving and union formation reflect growing variability in young adult lives. I find in chapter 4 that the estimation of life course transitions is susceptible to the variability that is not accounted for by measured characteristics and it is more apparent in the NLSY97 than in the NLSY79. It indicates that variability in the young adult life course has increased across cohorts and we are not capturing this variability in our existing panel studies. When it comes to the de-institutionalization, I find that education and family seem to be the institutions differentiating the transition to adulthood for both men and women. For males, more than

40% of the NLSY79 follow a traditional pathway while they are differentiated by the experience of postsecondary education. Moreover, another 40% of the earlier cohort men are classified into the delayed family formation class. However, in the NLSY97, male’s education and family formation are not critical in identifying their transition to adulthood as those of the NLSY79. For females, on the other hand, family formation is the most important institution identifying differences in young adulthood. The earlier cohort women are largely categorized in traditional pathways while their counterpart in the recent cohort are more in the contemporary family formation pathways. This indicates that marriage does not necessarily characterize the transition to adulthood for young females in recent years. Taken together, findings in the current study support the three core theorem in the transition to adulthood – de-standardization, de-institutionalization, and diversity.

Methodologically, I incorporate multi-group latent class analysis and event history analysis (a Cox hazard model and a multi-process model) to comprehend 125 dynamics in the transition to adulthood. Each has pros and cons and has contributed to a large body of literature to date. For example, LCA does not illuminate an association between particular life course events which may restrict our understanding of complexity between them. Although event history analysis compensates for the limitation, this method does not take a holistic view on various life course transitions together. It would be ideal that the transition to adulthood is illuminated by various methods that can include multiple facets. Moreover, future research needs to be aware of advantages or drawbacks of each approach.

Finally, the transition to adulthood is likely a null period with regard to social policies and institutional support (Settersten 2005). Moreover, with growing emphasis on autonomy and individualism, no consensus has been even made on the need of social support for young adults (Beck 1992). However, as my dissertation shows, the entry into adulthood has become prolonged, and young people in recent years take more complicated pathways to adulthood. This change may significantly contribute to the early adult development of those who have enough resources (e.g. of their own or of parents and family, Sandefur, Eggerling-Boeck, & Park 2005). On the other hand, it may aggravate young adults’ future prospects, especially those from vulnerable population

(Osgood, Foster, & Courtney 2010; Silva 2013). For example, young adults who drop out of college often experience unstable employment with low wages and a growing college debt (Silva 2013). It is therefore important to establish safety nets for young people during the transition to adulthood. This does not necessarily mean that the young need social policies as dependents, but it may be more efficient to help them to become 126 independent such as assistance for school to work transition or strengthening remedial programs. For example, completing at least a high school and some college education can shift the trajectory of the early adult life that has lifelong implications. Reducing the burden of college debt can also help with young and middle adulthood by reducing economic stress and possibly promoting family stability in the process.

127

APPENDIX A: Measurement of Cohabitation in the NLSY79

Cohabitation information is available from 1990, 1992 and subsequent surveys in the NLSY79. It includes the month and year of cohabitation (i.e. respondent and the current opposite-sex partner or spouse began living together) and whether the respondent and spouse lived with together before marriage. Besides the information, the cohabitation history can be retrieved from the Household Record every survey year. Respondents listed their partner as a household member if they live together at the time of interview.

The household record, however, does not provide any retrospective information of cohabitation before the first interview, which may misrepresent early union formation of the NLSY79 respondents who were late teens and early twenties in the first interview.

Moreover, these two sources of information do not include cohabiting experiences with anyone other than the current partner or spouse, which underestimates the actual cohabitation experiences. Beginning in 2002, the NLSY79 gathers a detailed cohabitation experience, i.e. cohabitation for any period of three months or more. The problem with this method is that a large share of the NLSY79 sample has not been interviewed since

2002 (e.g. in 2002, only 61.3% of the sample was interviewed.). I created a longitudinal cohabitation history by utilizing all of the above cohabitation information, but it is still possible that the actual cohabitation experiences are underestimated in this dissertation.

128

REFERENCES

Aassave, A., Billari, R.C., & Piccarretta, R. (2007). Strings of Adulthood: A sequence of young British women’s work-family trajectories. European Journal of Population, 23, 369-388. Aisenbrey, S., & Fasang, A. E. (2010). New life for old ideas: The “second wave” of sequence analysis bringing the “course” back into the life course. Sociological Methods & Research, 38(3), 420-462. Allison, P. D. (1984). Event history Analysis: Regression for longitudinal event data. Sage University Paper. Alwin, D. F., & McCammon, R. J. (2003). Generations, cohorts, and social change. In J. T. Mortimer & M. J. Shanahan (Ed.), Handbook of the life course (pp.23-49). Kluwer Academic/ Plenum Publishers, New York. Amato, P. R., Landale, L.S., Havasevich-Brooks, T.C., Booth, A., Eggbeen, D.J., Shoen, R., & McHale, S.M. (2008). Precursors of young women’s family formation pathways. Journal of Marriage and Family, 70, 1271-1286. Aquilino, W.S. (1997). From adolescent to young adult: A prospective study of parent- child relations during the transition to adulthood. Journal of Marriage and Family, 59, 670-686. Arnett, J. J. (2000). Emerging Adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55(5), 469-480. Arnett, J. J. (2004). Emerging adulthood: The winding road from the late teens through the twenties. Oxford University Press. Arnett, J.J. & Tanner, J.L (2006). Emerging adults in America: Coming of age in the 21st century. American Psychological Association. Arnett, J. J., Hendry, L. B., Kloep, M., & Tanner, J. L. (2011). The curtain rises: a brief overview of the book. In Arnett, J. J., Kloep, M., Hendry, L. B., & Tanner, J. L. (Eds.), Debating emerging adulthood: Stage or process? (pp. 3-9). Oxford.

129

Avery, R., Goldschneider, F., & Speare, Jr., A. (1992). Feathered nest/gilded cage: parental income and leaving home in the transition to adulthood. Demography, 29 (3), 375-388. Beck, U. (1992). Risk Society towards a new Modernity. Sage Publications. Beck, U. & Beck-Gernsheim, E. (2002). Losing the traditional: Individualization and ‘precarious freedoms’. In Beck, U. & Beck-Gernsheim, E. Individualization: Institutionalized individualism and its social and political consequences (pp.1- 21). Sage Publications. Benson, M., & O’Reilly, K. (2009). Migration and the search for a better way of life: a critical exploration of lifestyle migration. The Sociological Review, 57(4), 608- 625. Billari, F.C. (2001). The analysis of early life courses: Complex descriptions of the transition to adulthood. Journal of Population Research, 18(2), 119-142. Billari, F.C. (2004). Becoming an Adult in Europe: A Macro(/Micro)-Demographic Perspective. Demographic Research, Special 3, 15-44. Billari, F.C., & Liefbroer, A.C. (2010). Towards a new pattern of transition to adulthood? Advances in Life Course Research, 15, 59-75. Billari, F.C. & Liefbroer, A.C. (2007). Should I stay or should I go? The impact of age norms on leaving home. Demography, 44(1), 181-198. Bound, J., Lovenheim, M. & Turner, S. (2009). Why have college completion rates declined? An analysis of changing student preparation and collegiate resources NBER Working Paper No. 15566, December 2009, JEL No. I2,I23. Box-Steffensmeier, J. M., & Jones, B. S. (2004). Event History Modeling. Cambridge. Boyle, P.J., Kulu, H., Cooke, T., Galye, V., & Mulder, C.H. (2008). Moving and union dissolution. Demography, 45(1), 209-222. Boyle, P.J., Feng, Z., & Gayle, V. (2009). A new look at family migration and women’s employment status. Journal of Marriage and Family, 71(2), 417-431. Bozick, R., & DeLuca, S. (2005). Better late than never? Delayed enrollment in the high school to college transition. Social Forces, 84 (1), 531-554. Brien, M. J., Lillard, L. A., & Waite, L. J. (1999). Interrelated family-building behaviors: cohabitation, marriage, and nonmarital conception. Demography, 36 (4), 535-551. Brock, T. (2010). Young adults and higher education: Barriers and breakthroughs to success. The Future of Children, 20(1), 109-132. 130

Bronfenbrenner, U. (1977). Toward an experimental Ecology of human development. American Psychologist, 32(7), 513-531. Bronfenbrenner, U. (1989). Ecological systems theory. Annals of Child Development, 6, 187-249. Bronfenbrenner, U. & Evans, G.W. (2000). Developmental science in the 21st century: Emerging theoretical models, research designs, and empirical findings. Social Development, 9, 115-125. Brückner, H., & Mayer, K.U. (2005). De-standardization of the life course: what it might mean? And if it means anything, whether it actually took place? Advances in Life Course Research, 9, 27-53. Buck, N., & Scott, J. (1993). She’s leaving home: but why? An analysis of young people leaving the parental home. Journal of Marriage and Family, 55 (4), 863-874. Bureau of Labor Statistics. (2013). Retrieved from http://www.bls.gov/lau/lauov.htm Cadwallader, M. (1992). Migration and residential mobility: Macro and Micro approaches. University of Wisconsin Press, Madison, WI. Carlson, A.C. (2005). The fertility gap: Recrafting American population, Family Policy Lectures, Family Research Council, 14 December:1-14. Carlson, M., McLanahan, S. S., & England, P. (2004). Union formation in fragile families, Demography, 41(2), 237-261. Center for Human Resources. (2013). NLSY Handbook. The Ohio State University, Columbus, OH. Chandra A., Martinez G.M., Mosher, W.D., et al. (2005). Fertility, family planning, and reproductive health of U.S. women: Data from the 2002 National Survey of Family Growth. National Center for Health Statistics. Vital Health Stat, 23(25). Chen, Y., & Rosenthal, S. S. (2008). Local amenities and life-cycle migration: Do people move for jobs or fun? Journal of Urban Economics, 64 (3), 519-537. Cherlin, A. J. (2004). The deinstitutionalization of American marriage. Journal of Marriage and Family, 66 (4), 848-861. Cherlin, A. J. (2010). Demographic Trends in the United States: A review of research in the 2000s. Journal of Marriage and Family, 72 (3), 403-419. Clark, W. A. V., & Withers, S. D. (2007). Family migration and mobility sequences in the United States: Spatial mobility in the context of the life course. Demographic Research, 17 (20), 591-622. 131

Clark, W. A. V., & Withers, S. D. (2009). Fertility, mobility and labor-force participation: A study of synchronicity. Population, Space and Place, 15, 305- 321. Clark, W. A. V., & Huang, Y. (2003). The life course and residential mobility in British housing markets. Environment and Planning A, 35, 323-339. Clark, W. A. V. (2013). Life course events and residential change: unpacking age effects on the proabiliy of moving. Journal of Population Research, 30, 319-334. Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American Journal of Sociology, 96(4), 805-842. Cleveland, M. J., Collins, L. M., Lanza, S. T., Greenberg, M. T., & Feinberg, M.E. (2010). Does individual risk moderate the effect of contextual-level protective factors? A latent class analysis of substance use. Journal of Prevention and Intervention in the Community, 38(3), 213-228. Cleves, M., Gutierrez, R.G., Gould, W., & Marchenko, Y.V. (2010). An Introduction to Survival Analysis using Stata, 3rd edition. Stata Press. Clogg, C.C., & Goodman, L.A. (1984). Latent structure analysis of a set of multidimensional contingency tables. Journal of American Statistical Association, 79, 762-771. Coffman, D.L., Patrick, M.E., Palen, L.A., Rhoades, B.L., Ventura, A.K. (2007). Why do high school seniors drink? Implications for a targeted approach to intervention. Prevention Science, 8, 241-248. Collins, L.M., & Lanza, S.T. (2010). Latent class and latent transition analysis: with application in the social, behavioral, and health sciences. New York: Wiley; 2010. Cook, T.D., & Furstenberg, F.F. (2002). Explaining aspects of the transition to adulthood in Italy, Sweden, Germany, and the United States: A cross-disciplinary, case synthesis approach. pp. 257-287 in Annals of the American Academy of Political and Social Science: Early Adulthood in cross-national perspective, edited by Furstenberg, F. F. London: Sage Publications. Copen, C.E., Daniels, K., Vespa, J., & Mosher, W.D. (2012). First marriages in the United States: Data from the 2006–2010 National Survey of Family Growth. National health statistics reports; no 49. Hyattsville, MD: National Center for Health Statistics.

132

Corcoran, M. & Matsudaira, J. (2005). “Is it getting harder to get ahead? Economic attainment in early adulthood for two cohorts.” In R.A. Settersten, F. F. Furstenberg, & R. G. Rumbaut (Ed.), On the frontier of adulthood: Theory, Research, and Public Policy (pp. 356-395). The University of Chicago Press. Chicago& London. Côté, J., & Bynner, J. M. (2008). Changes in the transition to adulthood in the UK and Canada: the role of structure and agency in emerging adulthood. Journal of Youth Studies, 11(3), 251-268. Crockett, L. J., & Beal, S. J. (2012). The life course in the making: gender and the development of adolescents' expected timing of adult role transitions. Developmental Psychology, 48(6), 1727-1738. Cromartie, J. (1993). Leaving the countryside: young adults follow complex migration patterns. Rural Development Perspective, 8 (2), 22-27. Crowder, K. D. (2000). The racial context of white mobility: an individual-level assessment of the White flight hypothesis. Social Science Research, 29, 223-257. Dahl, G. B. (2010). Early teen marriage and future poverty. Demography, 47(3), 689-718 DaVanzo, J. (1983). Repeat migration in the United States: who moves back and who moves on? The Review of Economics and Statistics, 65 (4), 552-559. DaVanzo, J., & Morrison, P. A. (1981). Return and other sequences of migration in the United States. Demography, 18 (1), 85-101. Day, J.C., & Newburger, E.C. (2002). The big payoff: educational attainment and synthetic estimates of work-life earnings. Current Population Reports, P23-210, US Census Bureau. Easterlin, R.A. (1961). The American baby boom in historical perspective. The American Economic Review, 51(5), 869-911. Elder, Jr. G. H. (1986). Military times and turning points in men’s lives. Developmental Psychology, 22(2), 233-245. Elder, Jr. G. H. (1998). The Life Course as Developmental Theory. Child Development, 69(1), 1-12. Elder, Jr. G. H., King V., & Conger, R. D. (1996). Attachment to place and migration prospects: A developmental perspective. Journal of Research on Adolescence, 6(4), 397-425.

133

Elder, Jr. G. H., Johnson, M. K., & Crosnoe, R. (2003). The Emergence and Development of Life Course Theory. In Mortimer, J. T., & Shanahan, M. J. (Ed). Handbook of the Life Course. Kluwer Academic/Plenum Publishers. Elliott, D.B., & Simmons, T. (2011). Marital Events of Americans: 2009, American Community Survey Reports, ACS-13. U.S. Census Bureau, Washington, DC. Erikson, E. H. (1968). Identity: Youth and Crisis. W.W. Norton & Company, Inc. Erikson, E. H. (1980). Identity and the life cycle. W.W. Norton & Company, Inc. Feijten, P., & Mulder, C. H. (2002). The timing of household events and housing events in the Netherlands: A longitudinal perspective. Housing Studies, 17 (5), 773-792. Flowerdew, R., & Al-Hamad, A. (2004). The relationship between marriage, divorce and migration in a British data set. Journal of Ethnic and Migration Studies, 30 (2), 339-351. Freund,A.M. & Baltes, P.B. (2002). Life management strategies of selection, optimization, and compensation: Measurement by self-report and construct validity. Journal of Personality and Social Psychology, 82, 642–662. Furstenberg, F.F., Rumbaut, R. G., & Settersten, R. A. (2005). On the frontier of adulthood: Emerging themes and new directions. In R.A. Settersten, F. F. Furstenberg, & R. G. Rumbaut (Ed.), On the frontier of adulthood: Theory, Research, and Public Policy (pp.3-25). The University of Chicago Press. Chicago & London. Furstenberg, F. F. (2005). Non-normative life course transitions: reflections on the significance of demographic events on lives. Advances in Life Course Research, 10, 155-172. Fussell, E., & Furstenberg, F. F. (2005). The transition to adulthood during the twentieth century: Race, nativity, and gender. In R.A. Settersten, F. F. Furstenberg, & R. G. Rumbaut (Ed.), On the frontier of adulthood: Theory, Research, and Public Policy (pp.356-395). The University of Chicago Press. Chicago& London. Fussell, E., Gauthier, A.H., & Evans, A. (2007). Heterogeneity in the transition to adulthood: the cases of Australia, Canada, and the United States. European Journal of Population, 34, 389-414. Gabriel, P. E., & Schmitz, S. (1995). Favorable self-selection and the internal migration of young white males in the United States. The Journal of Human Resources, 30(3), 460-471.

134

Garasky, S. (2002). Where are they going? A comparison of urban and rural youths’ locational choices after leaving the parental home. Social Science Research, 31, 409-431. Garasky, S., Haurin, J. R., & Haurin, D. R. (2001). Group living decisions as youths transition to adulthood. Journal of Population Economics, 14, 329-349. Geist, C., & McManus, P. A. (2008). Geographic mobility over the life course: motivations and implications. Population, Space, and Place, 14, 283-303. Giddens, A. (1991). Modernity and Self-identity: Self and society in the late modern age. Stanford University Press. Glenn, N. D. (1997). Feedback: A reconsideration of the effect of No-Fault Divorce on divorce rates. Journal of Marriage and the Family, 59, 1023-1025. Glick, J. E., Ruf, S. D., White, M. J., & Goldscheider, F. (2006). Educational engagement and early family formation: differences by ethnicity and generation. Social Forces, 84 (3), 1391-1415. Goldscheider, F. & Goldscheider, C. (1999). The changing transition to adulthood. Leaving and returning home. Thousand Oaks, CA: SAGE. Goldstein, J. R., & Kenney, C. T. (2001). Marriage delayed or marriage forgone? New cohort forecasts of first marriage for U.S. women. American Sociological Review, 66, 506-519. Goldstein, S. (1973). Interrelations between migration and fertility in Thailand. Demography, 10 (2), 225-241. Groot, de C., Mulder, C. H., Das, M., & Manting, D. (2011). Life events and the gap between intention to move and actual mobility. Environment and Planning A, 43 (1), 48-66. Grusky, D.B., Western, B., & Wimer, C. (2011). The consequences of the Great Recession. In Grusky, D.B., Western, B., & Wimer, C. (Eds.), The Great Recession, (pp.3-20). Russell Sage Foundation. New York. Guldi, M. (2008). Fertility effects of abortion and birth control pill for minors. Demography, 45(4), 817-827. Guzzo, K. B. (2006). The relationship between life course events and union formation. Social Science Research, 35, 384-408. Hout, M. (2012). Social and economic returns to college education in the United States. Annual Review of Sociology, 38, 379-400.

135

Hout, M., Levanon, A., & Comberworth, E. (2011). Job loss and unemployment. In Grusky, D.B., Western, B., & Wimer, C. (Eds.), The Great Recession, (pp.59-81). Russell Sage Foundation. New York. Jacobsen, J.P., & Levin, L.M. (1997). Marriage and migration: comparing gains and losses from migration for couples and singles. Social Science Quarterly, 78 (3), 688-709. Jampaklay, A. (2006). How does leaving home affect marital timing? An event-history analysis of migration and marriage in Nang Rong, Thailand. Demography, 43(4), 711-725. Kins, E. & Beyers, W. (2010). Failure to launch, failure to achieve criteria for adulthood? Journal of Adolescent Research, 25(5), 743-777. Kreider, R. M., & Ellis, R. (2011). Number, timing, and duration of marriages and divorces: 2009. Current Population Reports, P70-125. U.S. Census Bureau, Washington, DC, 2011. Kulu, H. (2005). Migration and fertility: Competing hypotheses re-examined. European Journal of Population, 21, 51-87. Kulu, H. (2008). Fertility and spatial mobility in the life course: evidence from Austria. Environment and Planning A, 40 (3), 632-652. Kulu, H., & Milewski, N. (2007). Family change and migration in the life course: An introduction. Demographic Research, 17, 567-590. Kulu, H., & Steele, F. (2013). Interrelationships between childbearing and housing transition in the family life course. Demography, 50, 1687-1714. Lanza, S.T., Collins, L.M., Lemmon, D.R., & Schafer, J.L. (2007). PROC LCA: A SAS procedure for Latent Class Analysis. Structural Equation Modeling: A multidisciplinary Journal, 14(4), 671-694. Lanza, S.T., & Rhoades, B.L. (2013). Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prevention Science, 14(2), 157- 168. Lanza, S.T., Tan, X., & Bray, B.C. (2013). Latent class analysis with distal outcomes: A flexible model-based approach. Structural Equation Modeling: A multidisciplinary Journal, 20(1), 1-26. Lanza, S.T., Dziak, J.J. Huang, L., Wagner, A.T., & Collins, L.M. (2013). LCA Stata plugin users’ guide (Version 1.0). University Park: The Methodology Center, Penn State. Available from methodology.psu.edu. 136

Lesthaeghe, R. (1995). The second demographic transition in Western countries: An Interpretation. In K.Oppenheim Mason A.M. Jensen (Eds.). Gender and family changes in industrialized countries (pp.17-62). Oxford: Clarendon. Lesthaeghe, R. (2010). The unfolding story of the Second Demographic Transition. Population and Development Review, 36(2), 211-251. Lesthaeght, R. & Neidert., L. (2006). The Second Demographic Transition in the United States: Exception or textbook example? Population and Development Review, 32(4), 669-698. Lewis, S. K., & Oppenheimer, V. K. (2000). Educational assortative mating across marriage markets: Non-Hispanic Whites in the United States. Demography, 37(1), 29-40. Lichter, D. T., McLaughlin, D. K., Kephart, G., & Landry, D. J. (1992). Race and the retreat from marriage: A shortage of marriageable men? American Sociological Review, 57, 781-799. Lichter, D. T., Anderson, R. N., & Hayward, M. D. (1995). Marriage markets and marital choice. Journal of Family Issues, 16(4), 412-431. Lichter, D. T., Graefe, D. R., & Brown, J. B. (2003). Is marriage a panacea? Union formation among economically disadvantaged unwed mothers. Social Problems, 50(1), 60-86. Liefbroer, A.C., & de Jong Gierveld, J. (1995). Standardization and individualization: The transition from youth to adulthood among cohorts born between 1903 and 1965. In H.van den Brekel & F. Deven (Eds.) Population and family in the low countries (pp.57-80). Dordrecht: Kluwer. Lillard, L. A., & Waite, L. J. (1993). A joint model of marital childbearing and marital disruption. Demography, 30(4), 653-681. Lillard, L. A., Brien, M. J., & Waite, L. J. (1995). Premarital cohabitation and subsequent marital dissolution: A matter of self-selection? Demography, 32 (3), 437-457. Long, L. H. (1973). Migration differentials by education and occupation: trends and variations. Demography, 10 (2), 243-258. Lui, C. K., Chung, P. J., Wallace, S. P., & Aneshensel, C. S. (2013). Social status attainment during the transition to adulthood. Journal of youth and adolescence, 1-17. Macmillan, R. (2007). ‘Constructing adulthood’: Agency and subjectivity in the transition to adulthood. Advances in Life Course Research, 11, 3-29. 137

Macmillan, R., & Copher, R. (2005). Families in the life course: interdependency of roles, role configurations, and pathways. Journal of Marriage and Family, 67, 858-879. Macunovich, D.J., Easterlin, R.A., Schaeffer, C.M. & Crimmins, E.M. (1995). Echoes of the baby boom and bust: Recent and prospective changes in living alone among elderly widows in the United States. Demography, 32(1), 17-28. Magdol, L. (2002). Is moving gendered? The effects of residential mobility on the psychological well-being of men and women. Sex roles, 47, 553-560. Manning, W. D., Brown, S. L., & Payne, K. K. (2013). Two decades of stability and changes in age at first union formation. Paper to be presented at the annual meeting of the Population Association of America, 2013. Retrieved from http://paa2013.princeton.edu/papers/132048. Mare, R.D. (1995). Changes in educational attainment and school enrollment. In State of the Union: America in the 1990s, Vol.1: Economic Trends, ed. R. Farley, pp. 155- 213. New York: Russell Sage Found. Martin, J.A., Hamilton, B.E., Sutton, P.D., Ventura, S.J., Menacker, F., Kirmeyer, S., & Munson, M.L. (2007). Births: Final data for 2005. National vital statistics reports; Vol 56 no 6. Hyattsville, MD: National Center for Health Statistics. Massey, D. S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., & Taylor, J. E. (1993). Theories of international migration: A review and appraisal. Population and Development Review, 19(3), 431-466. Mayer, K. U. (2009). New directions in life course research. Annual Review of Sociology, 35(1), 413-433. McAuley, W. J., & Nutty, C. L. (1982). Residential preferences and moving behavior: A family life-cycle analysis. Journal of Marriage and Family, 44 (2), 301-309. McAuley, W. J., & Nutty, C. L. (1985). Residential satisfaction, community integration, and risk across the family life cycle. Journal of Marriage and the Family, 47 (1), 125-130. McHugh, K. E., Hogan, T. D., & Happel, S. K. (1995). Multiple residence and cyclical migration: A life course perspective. Professional Geographer, 47, 251-267. Michielin, F., & Mulder, C. H. (2008). Family events and the residential mobility of couples. Environment and Planning A, 40, 2770-2790. Michielin, F., Mulder, C., & Zorlu, A. (2008). Distance to parents and geographic mobility. Population, Space, and Place, 14, 327-345. 138

Mills, B., & Hazarika, G. (2001). The migration of young adults from non-metropolitan counties. Journal of Agricultural Economics, 83 (2), 329-340. Molloy, R., Smith, C. L., & Wozniak, A. (2011). Internal migration in the United States. Journal of Economic Perspectives, 25 (3), 173-196. Morgan, S. P., Cumberworth, E., & Wimer, C. (2011). The Great Recession’s influence on fertility, marriage, divorce, and cohabitation. In Grusky, D.B., Western, B., & Wimer, C. (Eds.), The Great Recession, (pp.220-245). Russell Sage Foundation. New York. Mortimer, J. T., Staff, J., & Lee, J. C. (2005). Agency and structure in educational attainment and the transition to adulthood. Advances in Life Course Research, 10, 131-153. Mouw, T. (2005). Sequences of early adult transitions: A look at variability and consequences. In Settersten, Jr. R. A., Furstenberg, F. F., & Rumbaut, R. G. (Eds.), On the Frontier of Adulthood: Theory, Research, and Public Policy (pp. 256-291). The University of Chicago Press, Chicago, IL. Mulder, C. H., & Clark, W. A. V. (2002). Leaving home for college and gaining independence. Environment and Planning A, 34, 981-999. Mulder, C. H., & Wagner, M. (1993). Migration and marriage in the life course: a method for studying synchronized events. European Journal of Population, 9, 55-76. Nakonezny, P. A., Shull, R. D., & Rodgers, J. L. (1995). The effect of no-fault divorce law on the divorce rate across the 50 states and its relation to income, education, and religiosity. Journal of Marriage and the Family, 57, 477-488. Oppenheimer, V. K. (1988). A theory of marriage timing. American Journal of Sociology, 94, 563-591. Oppenheimer, V. K. (2003). Cohabiting and marriage during young men’s career- development process. Demography, 40(1), 127-149. Osgood, W.D., Ruth, G., Eccles, J.S., Jacobs, J.E., Barber, B.L. (2005). Six paths to adulthood: Fast starters, parents without careers, educated partners, educated singles, working singles, and slow starters. In R.A. Settersten, F. F. Furstenberg, & R. G. Rumbaut (Ed.), On the frontier of adulthood: Theory, Research, and Public Policy (pp.320-355). The University of Chicago Press. Chicago & London. Osgood, W.D., Foster, M.E., & Courtney, M.E. (2010). Vulnerable populations and the transition to adulthood. The Future of Children, 20(1), 209-229.

139

Pampel, F.C. & Peters, E.H. (1995). The Easterlin effect. Annual Review of Sociology, 21, 163-194. Qian, Z. (2012). During the Great Recession, more young adults lived with parents. Census Brief prepared for Project US2010. Qian, Z. & Lichter, D. T. (2011). Changing patterns of interracial marriage in a multiracial society. Journal of Marriage and Family, 73, 1065-1084. Qian, Z., Lichter, D. T., & Mellott, L. M. (2005). Out-of-Wedlock childbearing, marital prospects and mate selection. Social Forces, 84 (1), 473-491. Quinn, M. A., & Rubb, S. (2011). Spouse overeducation and family migration: Evidence from the U. S. Journal of Family and Economic Issues, 32(1), 36-45. Rabe-Hesketh, S., & Skrondal, A. (2012). Multilevel and longitudinal modeling using Stata. Volume II: Categorical responses, counts, and survival, Third Edition. Stata Press. Rindfuss, R.R. (1991). The young adult years: Diversity, structural change and fertility. Demography, 28(4), 493-512. Rindfuss, R.R., Morgan, P., & Offutt, K. (1996). Education and the changing age pattern of American fertility: 1963-1989. Demography, 33(3), 277-290. Rindfuss, R.R., & VandenHeuvel, A. (1990). Cohabitation: precursor to marriage or alternative to being single? Population and Development Review, 16, 703-726. Rumbaut, R.G. (2005). Turning points in the transition to adulthood: Determinants of educational attainment, incarceration, and early childbearing among children of immigrants. Ethnic and Racial Studies, 28(6), 1041-1086. Rumberger, R.W. (2010). Education and the reproduction of economic inequality in the United States: An empirical investigation. Economics of Education, 29, 246-254. Ryan, C. L., & Siebens, J. (2012). Educational attainment in the United States: 2009. Current Population Report, P20-566. U.S. Census Bureau, Washington, DC. Sandberg-Thoma, S.E., & Kamp Dush, C.M. (2014). Indicators of adolescent depression and relationship progression in emerging adulthood. Journal of Marriage and Family, 76(1), 191-206. Sandefur, G.D., Eggerling-Boeck, J. & Park, H. (2005). Off to a good start? Postsecondary education and early adult life. In Settersten, Jr. R. A., Furstenberg, F. F., & Rumbaut, R. G. (Eds.), On the Frontier of Adulthood: Theory, Research, and Public Policy (pp. 292-319). The University of Chicago Press, Chicago, IL.

140

Sassler, S. (2010). Partnering across the life course: Sex, relationships, and mate selection. Journal of Marriage and Family, 72, 557-575. Schachter, J. (2001). Why People Move: Exploring the March 2000 Current Population Survey. Current Population Reports, P23-204.U.S.Census Bureau, Washington, DC. Schoen, R., Landale, N.S., & Daniels, K. (2007). Family transitions in young adulthood. Demography, 44(4), 807-820. Schoen, R., Landale, N.S., Daniels, K., & Cheng, Y.A. (2009). Social background differences in early family behavior. Journal of Marriage and Family, 71(2), 384- 395. Settersten, R.A. (2005). Social policy and the transition to adulthood: Toward stronger institutions and individual capacities. In Settersten, Jr. R. A., Furstenberg, F. F., & Rumbaut, R. G. (Eds.), On the Frontier of Adulthood: Theory, Research, and Public Policy (pp. 534-560). The University of Chicago Press, Chicago, IL. Settersten, R.A. & Gannon, L. (2005). Structure, agency, and the space between: on the challenges and contradictions of a blended view of the life course. Advances in Life Course Research, 10, 35-55. Settersten, R. A. & Ray, B. (2010). What’s going on with young people today? The long and twisting path to adulthood. The Future of Children, 20(1), 19-41. Shanahan, M. J. (2000). Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective. Annual Review of Sociology, 26, 667-692. Shanahan, M. J., Elder, Jr. G. H., & Miech, R. A. (1997). History and agency in men’s lives: pathways to achievement in cohort perspective. Sociology of Education, 70, 54-67. Shanahan, M.J., Porfeli, E.J., Mortimer, J.T., & Erickson, L.D. (2005). Subjective age identity and the transition to adulthood. On the frontier of adulthood: Theory, Research, and Public Policy (pp.225-255). The University of Chicago Press. Chicago & London. Sharkey, P. (2012). Temporary integration, resilient inequality: Race and neighborhood change in the transition to adulthood. Demography, 49, 889-912. Shauman, K. A., & Noonan, M. C. (2007). Family migration and labor force outcomes: sex differences in occupational context. Social Forces, 85 (4), 1735-1764. Silva, J.M. (2013). Coming up short: Working-class adulthood in an age of uncertainty. Oxford University Press. 141

Smith, T. W. (2005). Generation gaps in attitudes and values from the 1970s to the 1990s. In R.A. Settersten, F. F. Furstenberg, & R. G. Rumbaut (Ed.), On the frontier of adulthood: Theory, Research, and Public Policy (pp.177-221). The University of Chicago Press. Chicago & London. Smock, P. J. (2000). Cohabitation in the United States: An appraisal of research themes, findings, and implications. Annual Review of Sociology, 26, 1-20. Smock, P. J., & Manning, W. D. (1997). Cohabiting partners’ economic circumstances and marriage. Demography, 34(3), 331-341. Snyder, A. R., Brown, S. L., & Condo, E. P. (2004). Residential differences in family formation: The significance of cohabitation. Rural Sociology, 69 (2), 235-260. South, S. J., & Lloyd, K. M. (1992). Marriage opportunities and family formation: Further implications of imbalanced sex ratios. Journal of Marriage and the Family, 54, 440-451. Speare, Jr., A., & Goldscheider, F.K. (1987). Effects of marital status change on residential mobility. Journal of Marriage and the Family, 49 (2), 455-464. Steele, F., Kallis, C., Goldstein, H, & Joshi, H. (2005). The relationship between childbearing and transitions from marriage and cohabitation in Britain. Demography, 42 (4), 647-673. Steele, F., Joshi, H., Kallis, C., & Goldstein, H. (2006). Changing compatibility of cohabitation and childbearing between young British women born in 1958 and 1970. Population Studies, 60 (2), 137-152. Stone, J., Berrington, A., & Falkingham, J. (2014). Gender, turning points, and boomerangs: Returning home in young adulthood in Great Britain, Demography, 51, 257-276. Sum, A., Khatiwada, I., McLaughlin, J., & Palma, S. (2011). No country for young men: deteriorating labor market prospects for low-skilled men in the United States. The Annals of the American Academy of Political and Social Science, 635(1), 24-55. Tanner, J. L. (2006). Recentering during emerging adulthood: A critical turning point in life span human development. In Arnett, J. J., & Tanner, J. L. (Ed) Emerging adults in America: coming of age in the 21st century (pp. 193-217) Washington, DC: American Psychological Association. Tanner, J. L. & Arnett J.J. (2011). Presenting “Emerging Adulthood”: what makes it developmentally distinctive? In J.J. Arnett, M.K.L.B. Hendry, & J.L.Tanner (Ed.),

142

Debating Emerging Adulthood: Stage or process? (pp.13-30). Oxford University Press. Thornton, A., & Young-DeMarco, L. (2001). Four decades of trends in attitudes toward family issues in the United States: the 1960s through the 1990s. Journal of Marriage and Family, 63 (4), 1009-1037. Tienda, M. & Wilson, F. D. (1992). Migration and the earnings of Hispanic men. American Sociological Review, 57(5), 661-678. Thornton, A., & Young-DeMarco L. (2001). Four decades of trends in attitudes toward family issues in the United States: The 1960s through the 1990s. Journal of Marriage and Family, 63, 1009-1037. Upchurch, D. M., Lillard, L. A., & Panis, C.W.A. (2002). Nonmarital Childbearing: influence of education, marriage, and fertility. Demography, 39 (2), 311-329. U.S. Census Bureau. (2012, March 14). Current Population Survey Data on Geographical Mobility/Migration. Retrieved from http://www.census.gov/hhes/migration/data/cps.html US. Census Bureau (2011). Retrieved from http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src =bkmk. USDA (2007). Rural population and migration: Trend 2_nonmetro population growth slows. Retrieved from http://www.ers.usda.gov/Briefing/Population U.S. National Center for Health Statistics. 2011. Retrieved from http://www.cdc.gov/nchs/data/hus/2011/007.pdf. van de Kaa, D. J. (1987). Europe’s second demographic transition. Population Bulletin, 42(1). van de Kaa, D. J. (2002). The idea of a second demographic transition in industrialized countries. Paper 6th Welfare Policy Seminar, National Institute for Population and Social Security, Tokyo, Japan, 29, January, 2002. Vermunt, J.K., & Magidso, J. (2004). Latent class analysis. In: Lewis-Beck, M., Bryman, A., Liao, T.F. (Eds.), The Sage Encyclopedia of Social Sciences Research Methods. Sage Publications, Thousand Oakes, CA. Walsemann, K.M., Geronimus, A.T., & Gee, G.C. (2008). Accumulating disadvantage over the life course. Research on Aging 30(2), 169-199.

143

Whisler, R. L., Waldorf, B. S., Mulligan, G. F., & Plane, D. A. (2008). Quality of life and the migration of the college-educated: A life course approach. Growth & Change, 39 (1), 58-94. Widmer, R.D., & Ritschard, G. (2009). The de-standardization of the life course: Are men and women equal? Advances in Life Course Research, 14, 28-39. Wilson, B. A., Berry, H. E., Toney, M. B., Kim, Y., Cromartie, J. B. (2009). A panel based analysis of the effects of race/ethnicity and other individual level characteristics at leaving on returning. Population Research and Policy Review, 28, 405-428. Wolaver, A. M., & White, N. E. (2006). Racial wage differences among young male job changers: The relative contribution of migration, occupational change, site characteristics, and human capital. Growth and Change, 37 (1), 34-59. Zollinger, G.J., & Elder, G.H., Jr. (1998). Methods of life course research: Qualitative and quantitative approaches. Thousand Oaks, CA:Sage.

144