"زناشویى بسته" در میان افراد طلاق گرفته و هرگز ازدواج نکرده در تاریخ تاهل در دهه های اخیر : شواهد از داده های حیاتی آمار

Abstract I investigate whether divorced and never married persons tend to marry within their own marital history group. This analysis is a step toward assessing any distinctions that may exist between the never married and the divorced, which informs the distinctions between first marriages and remarriages, across which inequality among coresident children has been observed. Using log-linear models applied to data of marriages formed in a year from the Vital Statistics Marriage Files, I find evidence of a tendency toward marital history homogamy beyond that accounted for by relative group size, education, and age. Specifically, the never married and the divorced are more likely to marry within their marital history group than to intermarry. Results also indicate that, although the tendency toward marital history homogamy unaccounted for by group size, age, and education persisted throughout the period 1970–1988 (i.e., the years for which educational data are available in the Vital Statistics data), it did diminish somewhat.

1. Introduction Sociological, economic, and demographic researchers have long sought to specify factors that contribute to inequalities in a population. In recent years, an increasing number of researchers have focused on understanding sources of inequality among children. While many have emphasized the role of parental socioeconomic status, others argue and find evidence that the type of parental marriage, such as first marriage and remarriage, contributes to inequality among children even when accounting for parental socioeconomic status (e.g., Hofferth and Anderson, 2003 and McLanahan and Sandefur, 1994). For example, McLanahan and Sandefur (1994) have shown that children who coreside with remarried parents are academically worse off than children who coreside with parents in their first marriage. More recent data indicate that children living with parents in “mixed” marriage (i.e., one coresident parent is in a first marriage and the other is in a second or higher order marriage) are situated between children living with parents in first marriage and children living with parents in a remarriage, wherein both spouses have been married before. To illustrate, Appendix A contains the mean academic-cognitive test scores (Woodcock-Johnson and digit span tests) of 6–12 year olds from the Panel Study of Income Dynamics, Child Development Supplement, 1997. The scores adjust for the child’s age, the child’s race, and the coresident father’s education; they indicate that, on average, children whose coresident parents are in first marriages achieve the highest scores, followed by those whose parents are in mixed marital history marriages (intermarriages), and then by those whose parents are in remarriages. Although marriage types—first marriage, marital history intermarriage, and remarriage—have been linked to inequality among coresident children, little is known about how these diverse marriage types that structure inequality are formed. One way to gain insight into the formation of these marriage types that structure inequality among children is to assess the processes that may contribute to the tendency toward marital history homogamy (as opposed to intermarriage) among the never married and the divorced. By marrying within their marital history groups, the never married and the divorced form first marriages and remarriages (i.e., the two marriage types across which inequality in child well-being is most notable) rather than mixed marriages. In this study, two hypotheses about the processes contributing to the tendency toward marital history homogamy are developed and tested. The compositional hypothesis is that little tendency toward marital history homogamy is present beyond that accounted for by the relative group size of the never married and the divorced, and the influences of spousal education and age. This hypothesis is developed from the perspective that marital history homogamy is primarily a by-product of these conditions rather than a direct result of substantive differences between the marital history groups. Specifically, some previous sociological studies suggest that any major observed tendency toward marital history homogamy may be an artifact of the larger group size (Blau et al., 1982 and Kalmijn, 1998) of the never married relative to that of the divorced. The relative group size argument is described further in the next section. Also, empirical evidence and previous studies suggest that a tendency toward marital history homogamy is a by-product of additional sociodemographic conditions that include the educational and age differences between the never married and the divorced and couples’ homogamous tendencies on these dimensions. For example, statistics from the 1% Integrated Public Use Microdata Series (IPUMS), 1970–1990, indicate that, on average, the divorced have less education and are older than the never married (Ono, forthcoming and Ruggles and Sobek, 1997). Previous studies also note preferences of couples to marry persons of like-education (Mare, 1991) and like-age (Atkinson and Glass, 1985 and Qian and Preston, 1993).1 The second hypothesis is that the processes that differentiate an average never married person from an average divorced person yield a tendency toward marital history homogamy unaccounted for by relative group size, and education and age of the spouses. For example, unlike the never married, the divorced have preferences, rights, and obligations (Masheter, 1997) to share resources with family members from the former marriage.2 In addition, in some cases, divorce may stigmatize a person as a mate (Gerstel, 1987), may be more likely to occur among persons with selected characteristics (Bumpass and Sweet, 1972), and may alter individuals’ personality, attitude, or behavior in specific ways (Wolfinger, 2003). A theory of assortative mating suggests that, due to these group distinctions, if the never married and the divorced were to intermarry, in many cases, persons in marriages between the never married and the divorced would experience: (a) inefficiencies in their marriage due to conflicts (Becker, 1981 and Posner, 1986); and (b) a slower development of social bond (e.g., DiMaggio and Mohr, 1985). This perspective is also discussed further in the next section. If results support the compositional hypothesis, this would imply that the two marital units across which notable inequalities in child well-being are observed (i.e., first marriages and remarriages) are products of relative group size, and differentials and homogamous tendencies on parental socioeconomic status and age. If results support the alternative hypothesis of substantive group differences, this would imply that the two marriage types are also in part the product of processes arising from spouses’ distinct marital histories. In this paper, I first apply log-linear models to the Vital Statistics Marriage File, 1970–1988, to create an index of the tendency toward marital history homogamy when removing the influences of relative group size, spousal educational levels, and wives’ age. I investigate the sensitivity of the results to: (a) husbands’ age relative to wives’ age; and (b) the inclusion/exclusion of spousal parent indicators (i.e., indicators of whether a spouse is a parent). The Vital Statistics data set cannot be used to conduct the test of sensitivity to the inclusion of parent indicators because it does not contain the parent status information. However, I assess how much of the estimated tendency toward marital history homogamy is attributable to spousal parent status composition and homogamy by analyzing another data set, the panel study of income dynamics (PSID). Second, if I find evidence of a tendency toward homogamy when accounting for group size, education, and age, I then examine whether it has declined between 1970 and 1988. The tendency may have diminished during this historical period, due to, for example, the possibility that the never married and the divorced are becoming more similar with increased cohabitation rates. Unfortunately, information on education of the spouses is not available in the Vital Statistics data after 1988. Hence, the last year in which the spouses’ educational measures are included in the model is 1988. However, I note patterns observed from the 1989–1995 data when not controlling for spousal education.3

5. Conclusion In this study, I find evidence of a tendency for the never married and the divorced to marry within their marital history group when removing the tendency attributable to relative group size, education and age. Specifically, between 1970 and 1988, never married and once divorced persons were more likely to be homogamous than to intermarry with respect to marital history. The results suggest that marital history homogamy is not simply an artifact of the relative group size of the never married and the divorced, or of differentials and homogamous tendencies on socioeconomic status and age; factors other than relative group size of the marital history groups, and spousal education and age contribute to the tendency toward marital history homogamy. As I argued earlier, these factors likely include resource sharing with family members from former marriages among the divorced, the stigma of divorce, selection into divorce, and the impact of divorce on personality, attitudes or behaviors; the differences between the never married and the divorced may reduce the productivity and slow the development of social bond (Becker, 1981 and DiMaggio and Mohr, 1985) between spouses in a marital history intermarriage, thereby encouraging homogamy on marital history. Additional analyses drawing on the PSID and patterns in the Vital Statistics data suggest that the conclusion is not altered by the influences of parenthood composition and homogamy. One implication of this study is that when the never married and the divorced marry within their marital history group and form first marriages and remarriages, two distinct environments are created in which coresident children are raised (e.g., Hofferth and Anderson, 2003 and McLanahan and Sandefur, 1994). An example of the distinction in the environment is that spouses in a remarriage tend to distribute some of their resources to non-resident family members from the former marriage in addition to the coresident family members. In contrast, spouses in a first marriage more exclusively distribute their resources to coresident family members. Therefore, homogamy on marital history produces inequality in the resources available to coresident family members, having a potentially notable impact on child well-being. Future research contrasting resource allocation patterns among couples who are married for the first time, couples in marital history intermarriage, and couples who are remarried (Seltzer, 1991) would yield further insights into the specific ways in which marriage types are linked to inequality among children. Another implication is that differences between the never married and the divorced unaccounted for by group size, education, and age may partly explain the higher rate of cohabitation among divorced persons than never married persons (e.g., Bachrach, 1987). If a couple consists of a divorced person and a never married person, they may be more likely than couples comprised of two never married persons to anticipate that a marital union, in which spouses pool resources and make joint decisions (Treas, 1993), involves too many stresses, institutional complexities, and ambiguities (Cherlin, 1978); as a result, the couple may opt for long-term cohabitation. Because the number of unmarried never married persons is larger than the number of unmarried divorced persons (Bachrach, 1987; Table 2), cohabitation as a result of pairing across marital history may raise the rate of cohabitation disproportionately among the divorced compared to the never married. Although data limitations do not allow investigation of the extent to which cross-marital history coupling accounts for the difference in cohabitation rates between the never married and the divorced, this issue may be investigated with other data sets. The second question asked is whether the tendency toward marital history homogamy beyond that explained by relative group size, education and age is declining over historical time. My results indeed provide some evidence of a general decline in the tendency between 1970 and 1988. This is consistent with the argument that, although marital history groups continue to be distinct, they are becoming less so with: (a) the growing similarity of the never married and the divorced in part due to the rising rate of cohabitation (Casper and Cohen, 2000); (b) the increased acceptance of divorce as part of everyday life (Thornton, 1985); and (c) the increased preference to match on spousal socioeconomic characteristics rather than on other qualities (Kalmijn, 1991 and Mare, 1991). The persistently present but declining differentiation between the never married and the divorced implies that the differentiation among the three marriage types–first marriage, intermarriage and remarriage—has also diminished. Because the marriage types would be distinguished, in part, by the quantity and quality of resources made available to the coresident family members, the diminishing distinctions among them would also imply that their impact on inequality in child well-being might be weakening. A decline in the influence of marriage type on inequality among children could potentially help offset the inequality inducing influence of the growing economic inequality among adults in the US (Acemoglu, 2003). In the age-specific analysis, I find an exception to the decline in the level of marital history homogamy among couples with younger wives (i.e., ages 23–28) between 1979 and 1988, for whom the relative odds of homogamy remained constant. This result implies that, among the young in those years, the distinction between the never married and the divorced may not have diminished. A possible explanation is that the previously increasing assimilation between the never married and the divorced slowed around that time period among the young. Consistent with this possibility, the level of cohabitation, a union type most commonly observed among the young (Thornton, 1988), was nearly constant in the early to mid 1980s, even though it increased in the 1970s (Casper and Cohen, 2000). Some researchers note that the increasing openness to liberal attitudes and behaviors that was observed among youths in the prior years stagnated in the 1980s in response to social issues notable in the 1970s (Corbett, 1988), which probably include divorce. Another explanation is that, with the delay in average age at marriage (Goldstein and Kenney, 2001), couples who persist to marry young, possibly including the religious (Lehrer, 2000), are increasingly composed of couples who have closed preferences on many dimensions, including marital history. Whether youths of the early 1980s and later contributed to the lack of a decline in the level of marital history homogamy in the 1990s—a possibility suggested by the analysis of the 1989–1995 Vital Statistics data that did not control for spousal education–may be investigated in future studies. Although the current analysis does not empirically separate the several theoretically proposed processes that contribute to the homogamous tendency on marital history, it provides evidence that such explication would be worth pursuing in future studies. Sufficient evidence is available over recent decades to indicate a tendency toward marital history homogamy beyond the influences of relative group size, education and age, and to suggest the need to examine specific processes supporting this homogamous tendency in future work. For example, what kinds of ties involving a former marriage or spouse contribute to the homogamous tendency? More analysis is needed in this direction to better understand how parental marriage type impacts inequality in child well-being.