The Social Stratification of Partnership Trajectories to Childlessness in Germany
Emanuela Struffolino, WZB Berlin and Humboldt University of Berlin
Marcel Raab, Mannheim University
Using representative German panel data and sequence analysis, we identify five distinct patterns which we labeled according to the predominant partnership status: (1) married, (2) cohabiting, (3) living apart together, (4) previously partnered, (5) never partnered. Our study contributes to the previous literature by using rich retrospective accounts on partnership trajectories which allow to distinguish between unpartnered persons and persons in LAT relationships and which include information on life satisfaction, religiosity and self-reported fecundity. Furthermore, our analysis comprises women and men, whereas most research on childlessness is exclusively focusing on women. Finally, we are able to access context effects on childlessness by examining differences between the former socialist East Germany and the rather (gender) traditional West Germany. Preliminary results show that men are clearly overrepresented in the “never partnered” cluster while women have an increased probability of sorting into the marital pathway. The single clusters (4 & 5) are associated with a lower life satisfaction compared to the three remaining groups. Marriage among childless is much more common in the West and among low educated women. Interestingly more than 50% in this cluster report that they or their partner are infertile. This indicates that involuntary childlessness might be a concern among this group.
The social stratification of partnership trajectoriesto childlessness in Germany
In recent decades childlessness increased across manyEuropean countries. This spurred a lot of research on the determinants andconsequences of childlessness. Much of the previous research, however, isutilizing cross-sectional data or is only considering the childlessness ofwomen. Building on recent studies which addressed these shortcomings we userich retrospective and panel data to study partnership trajectories ofchildless men and women.
Next to socioeconomic characteristics partnershiptrajectories are among the most important determinants of childlessness. Yet,still little is known about the partnership biographies of childless persons.Our study addresses this research gap and adds to recent evidence from Finland providedin Jalovaara & Fasang (2017)in the following ways. First, we study childlessness in a different countrycontext, i.e. Germany. Because of the long-standing historical differences infertility behavior between East and West Germany that were aggravated by thepost-war separation we consider Germany to be an interesting study case foranalyzing the impact of normative and institutional differences on theassociation of partnership trajectories and childlessness. Second, in contrastto the paper on Finland, our study draws on survey data instead of registerdata. Although this reduces the statistical power of our analysis, survey dataalso bear notable advantages. For instance, register data are not suited foridentifying living apart together relationships and therefore tend to overstatethe extent of singleness among the childless. Moreover, we can considerindicators which cannot be accessed in official register data such as lifesatisfaction, frequency of church attendances, and self-reported infertility.
Our analysis draws on data from the Germany FamilyPanel and its East German supplement DemoDiff. We reconstruct the partnershiptrajectories from age 18-40 of respondents born between 1970-1973 and analyzethem using sequence, cluster, and multinomial regression analysis. This allowsus to identify different patterns of partnership trajectories and theirdeterminants. In addition, we present a brief descriptive comparison of thepartnership trajectories of childless persons and parents. Our analysis revealsfive distinct pathways to childlessness, which are presented in Figure 1. Inthis preliminary version of the analysis the clusters are simply namedaccording to their modal state, i.e. the most prevalent partnership status.
Table 1 provides descriptive information on thesequences and on variables that are considered determinants or consequences ofthe different partnership trajectories. It further comprises a comparison ofchildless and parents. This comparison shows that the trajectories of childlesspersons include slightly more sub-episodes because they experience more LATspells than their counterparts with children. Interestingly, however, bothgroups are rather similar in terms of the time they spend in coresidentialunions. This is also true for higher order cohabitation spells. Indeed, serialcohabitation (defined as more than one cohabitation spell) seems to be arather rare phenomenon experienced by less than 25% of the respondents. In linewith previous research, childless are less likely to be married, more oftenmale, better educated, and less religious.
Turning to the description of the heterogeneity in thepartnership trajectories our results indicate that roughly half of thechildless clusters 4 & 5 hardly have any partnership experience at age40 and spend more than 70% of the observation period without a partner. Acomparison of the two single groups illustrates a notable difference: whereas mostpersons in the never partnered group remain without any partnershipexperience up to age 30+ the members of the previously partnered cluster reportedat least one previous LAT relationship before they stayed single for anextended period. Interestingly, we witness a massive overrepresentation of men inthe first group while the gender ratio is more balanced in the second one. Inaddition, we find a cluster in which the respondents spent a considerableamount of time in a long-term LAT relationship without making the transition tocohabitation, let alone marriage. Due to data restrictions prohibiting adetection of LATs most previous research would have mislabeled these respondentsas unpartnered. Note, however, that this cluster is indeed similar to the twoprevious ones in terms of the below average life satisfaction reported by thisgroup. The two remaining clusters are characterized by long-term cohabitationand by marriage. Within the marriage cluster we observe an overrepresentationof women, low educated persons, and West Germans. Interestingly more than 50%in this cluster report that they or their partner are infertile. This and thetraditional partnership trajectory indicate that involuntary childlessnessmight be a concern among this group.
Further analysis will build on these stylized facts inmultivariate regression models. We will predict cluster membership with amultinomial regression analysis. Furthermore, we plan to predict outcomes (e.g.life satisfaction) based on cluster membership and to consider parallel processessuch as employment trajectories.
Figure 1 Sequence-index-plots
Presented in Session 1158: Fertility