Keywords

3.1 Introduction

Since the 1960s, Western societies are marked by a postponement of marriage and childbearing, along with an increase in unmarried cohabitation and divorce. These developments resulted in a disconnection between marriage and parenthood (Billari and Liefbroer 2010) and a substantial increase in the number of births that happen outside of marriage (also referred to as nonmarital births or nonmarital fertility). In the 1980s and 1990s, Van de Kaa and Lesthaeghe developed the Second Demographic Transition theory (SDT) in order to explain the observed changes in family-related demographic behavior (Lesthaeghe and Van de Kaa 1986). According to the SDT, the increase in nonmarital fertility is the result of the enhanced freedom of choice in organizing one’s individual life and in following non-traditional family formation patterns (Lesthaeghe 2010; Lesthaeghe and Van de Kaa 1986; Van de Kaa 2001). Non-traditional family patterns are expected to spread from a small group of people with more progressive value orientations – such as the younger cohorts and the higher educated – to the whole society (Van de Kaa 2001).

Today, the SDT is a widely used framework to study changes in family formation and childbearing. Nevertheless, the theory has also been criticized, and American scholars have done so in particular with regard to nonmarital fertility. In the USA, women with fewer opportunities and resources are more likely to follow the trajectory of nonmarital births (McLanahan 2004; Musick 2002). The fact that nonmarital fertility is more common among the lower than among the higher educated contrasts with the assumption of the SDT that new demographic behavior tends to be more common among the higher educated during the early stages of the SDT, after which it spreads across different educational groups during later stages. Apart from one’s own socio-economic status (SES), parental SES plays a pivotal role in explaining nonmarital fertility in the USA. Research has shown that those growing up with lower-SES parents are more likely to have a nonmarital birth when reaching adulthood than those growing up with higher-SES parents (Aassve 2003; Amato et al. 2008; Högnäs and Carlson 2012; McLanahan 2004; Musick 2002; Wu 1996). Because of the role that the socio-economic background of the childhood family plays in one’s adult family-related experiences, scholars have concluded that in the USA nonmarital fertility is part of a “pattern of disadvantage” which plays an important role in the reproduction of inequality from one generation to the next (McLanahan 2009). In the remainder of this chapter I will refer to this view as the Pattern of Disadvantage perspective.

A few single-country studies have examined the influence of socio-economic status of the childhood family on nonmarital fertility outside of the USA. These studies, conducted in the UK (Berrington 2001; Ermisch 2001; Ermisch and Francesconi 2000; Rowlingson and McKay 2005) and Sweden (Bernhardt and Hoem 1985), have come to the same conclusion as the American scholars, namely that those growing up with parents with a lower socio-economic status are more likely to have a nonmarital birth when reaching adulthood. A recent multi-country study examined the influence of parental SES on partnership status at first birth in 16 European and North-American countries (Koops et al. 2017). Results of this study showed that substantial country differences exist in the effect of parental SES on nonmarital fertility. In North-American and East-European countries men and women growing up with lower-SES parents have a higher risk of experiencing a birth in cohabitation as compared to those with higher-SES parents. In West-European countries, this effect was only found in Norway, but not in France, Belgium and Austria. In the Netherlands a positive effect of parental SES on the risk of having a birth in cohabitation was found. Specifically, Dutch individuals growing up with higher-SES parents have a higher risk of experiencing a birth in cohabitation as compared to those with lower-SES parents. The study furthermore shows that in North-American countries and in half of West- and East-European countries, those growing up with lower-SES parents also have a higher risk of experiencing a birth while being single as compared to those with higher-SES parents.

Parental SES is expected to have a direct effect on family-formation behavior through socialization and the availability of resources. In addition, a more indirect pathway is assumed through the intergenerational transmission of a person’s own socio-economic status. Those growing up with higher-SES parents may have a higher SES themselves and this may influence their preferences and availability of resources, and, therefore, their family formation behavior. Research in the USA suggests that the effect of parental SES on partnership context at birth is only partially mediated through a person’s own SES (Aassve 2003; Amato et al. 2008). A multiple-country study examined the influence of own as well as parental SES on the likelihood of having a first birth in cohabitation or in marriage in 16 European countries (Lappegård et al. 2018). After controlling for own SES, there was no significant effect of father’s education on women’s likelihood of having their first birth in cohabitation or in marriage. But although this study included multiple countries, analyses were conducted on the pooled data only. This may have masked the fact that in some countries the effect of socio-economic status of the childhood family is fully mediated through own SES, while in other countries a direct effect of parental SES is found in addition. This is supported by the study of Koops et al. (2017), which showed that in some countries the effect of parental SES on partnership status disappeared after own education was included in the model, while in other countries this was not the case.

To examine mediation through own SES, Koops et al. (2017) compared the effect of parental SES on partnership status at birth between a model excluding and including own SES. This method suffices to answer the question if there is a direct effect of parental education on partnership status once mediation through own SES has been taken into account. However, it does not allow to examine in detail to what extent the effect is mediated through own SES, because of the non-linear character of the models (Karlson et al. 2012). In the current chapter, I aim to make additional steps in unravelling the mechanism behind inequalities found in family formation behavior, by providing a more detailed picture of whether and to which extent the effect of parental education is mediated through own education. I will do so by using the Karlson-Holm-Breen (KHB) method which allows for an unbiased decomposition of the percentage reduction in the effect of parental SES caused by the mediation through own SES for non-linear models (Karlson et al. 2012). Coefficients obtained from (multinomial) logistic regression models, such as odds ratios and relative risk ratios, do not immediately allow for a meaningful interpretation of the reduction in effect sizes. Therefore, in addition to reporting the effect of parental SES on the outcome variable as relative risk ratios, effect sizes will be reported as differences in the predicted probabilities of having a birth in marriage, cohabitation or while being single. This will allow for easy interpretation of the effect size of parental SES before and after mediation through own SES has been taken into account.

3.2 Background

Family sociologists have identified several mechanisms through which parents can influence their children’s family formation. These mechanisms can roughly be separated into those related to socialization, the availability of resources, and the intergenerational transmission of SES. These mechanisms will be further explained below.

3.2.1 Influence of Parental SES on Nonmarital Fertility Through Socialization

Socialization can be described as “the process by which children acquire the social, emotional, and cognitive skills needed to function in the social community” (Grusec and Davidov 2010, p. 691). Much of the socialization process occurs at a young age in the context of one’s family of origin. Parents and caregivers are therefore among the most important social agents through which children acquire their social, emotional, and cognitive skills (Grusec and Davidov 2010). Through socialization, children can internalize the values and attitudes of their parents, which in turn influence children’s family-related preferences and behaviors, not only during childhood, but also during their adult life (Axinn et al. 1994; Barber 2000; Barber and Axinn 1998; Liefbroer and Elzinga 2012; Starrels and Holm 2000).

According to the SDT, the increase in cohabitation at the expense of marriage reflects individuals’ need for autonomy and self-actualization in the context of romantic relationships. This view is supported by qualitative studies which have shown that people tend to associate cohabitation with personal freedom, financial independence, and distancing oneself from social norms advocating marriage (Perelli-Harris et al. 2014). The SDT does not specifically address how SES may influence the likelihood of becoming a cohabiting parent. However, research examining differences in parenting styles have shown that generally, higher-SES parents emphasize autonomy more when raising their children, while lower-SES parents tend to underscore conformity (Gauthier 2015; Kohn 1969; Park and Lau 2016; Weininger and Lareau 2009). One could therefore argue that via the socialization mechanism, children of higher-SES parents are more likely to cohabit when starting a family than children growing up with lower-SES parents.

With regard to single parenthood, it has been argued that parents’ SES influences their preferences on the importance of parenthood in relation to other life goals such as educational and occupational attainment. Through socialization, parents may transmit these beliefs to their children. Children growing up with higher-SES parents may therefore have higher educational and occupational aspirations, while those growing up with lower-SES parents may value parenthood more (Friedman et al. 1994; Musick et al. 2009). Due to the difficulty for single parents to combine family life with other socio-economic goals (Härkönen 2016; Lichter et al. 2014), those with higher-SES parents are assumed to be more motivated to avoid unplanned pregnancies and single parenthood than those with lower-SES parents (Brown and Booth 1996; Frisco 2005). One could therefore argue that via the socialization mechanism, children of lower-SES parents may be more likely to be single when starting a family than children of higher-SES parents.

3.2.2 Influence of Parental SES on Nonmarital Fertility Through the Availability of Resources

When it comes to the influence of parental SES on becoming a cohabiting parent, American literature has sought an explanation in the different meanings that are attached to cohabitation and marriage. It is argued that in the USA, couples hold on to certain goals and objectives which they feel they have to achieve before they are ready to get married (Edin and Kefalas 2005; Gibson-Davis 2011; Gibson-Davis et al. 2005). Several authors (Cherlin 2004; Smock et al. 2005) claimed that the list of prerequisites for marriage includes aspects related to material resources (purchase of a house), financial resources (adequate savings for a ‘proper’ wedding), and financial stability (stable income). Research has shown that parents tend to transfer money or real estate around the period when children leave the parental home, which suggests that parental assets play a role in children’s family transitions (Albertini et al. 2007; Albertini and Kohli 2013; Kohli 1999; Ploeg et al. 2004). Through the transfer of these resources, parents may affect their children’s likelihood of having a marital birth, for example by influencing their ability to afford a wedding (Albertini and Kohli 2013; Ploeg et al. 2004). People with lower-SES parents may have more difficulty to meet the requirements for marriage than those with higher-SES parents due to a lower likelihood of receiving direct financial transfers from their parents (Albertini and Kohli 2013; Fingerman et al. 2015). Those with lower-SES parents may therefore be more likely to decide to postpone the wedding until later in their relationship and perhaps after becoming parents, while people with higher-SES parents do not have to postpone the wedding and are therefore more likely to marry before starting a family (McLanahan and Percheski 2008; Oppenheimer 2003). Thus, one could argue that via the resources mechanism, children of lower-SES parents are expected to be more likely to cohabit when starting a family than children of higher-SES parents.

Note that the expectations derived from the socialization mechanism and the resources mechanism clearly differ. Based on the SDT and the socialization mechanism, one would expect that those with higher-SES parents are more likely to experience a birth in cohabitation, while based on the Pattern of Disadvantage perspective and the resources mechanism, one would expect that those with lower-SES parents are more likely to have a birth in cohabitation.

Regarding single parenthood, it has been argued that parental SES may influence children’s ability to successfully prevent single parenthood. American studies have shown that children growing up with lower-SES parents have reduced access to medical care, abortions, as well as certain type of modern contraceptive methods (Boussen 2012; Musick et al. 2009; Silverman et al. 1987). Reduced access to family planning methods may in turn have a negative effect on contraceptive use which can increase the chance of an unplanned pregnancy. Compared to those with higher-SES parents, those growing up with lower-SES parents may also be less likely to move in with a partner before the birth of the child when faced with an unplanned pregnancy because they receive less material support to establish their own household (Albertini and Kohli 2013; Fingerman et al. 2015). Based on these arguments, one would expect that via the resources mechanism, those with lower-SES parents are more likely to have a birth while being single than those with higher-SES parents.

3.2.3 Influence of Parental SES on Nonmarital Fertility Through the Transmission of SES

The intergenerational transmission of SES refers to the consistently found correlation between the SES of parents and their children (Breen and Jonsson 2005; Ganzeboom et al. 1991; Kurz and Müller 1987). Education plays an important role within this mechanism (Breen and Jonsson 2005; Ganzeboom et al. 1991; Jerrim and Macmillan 2015). Via socialization, parents are expected to shape their children’s educational aspirations and school performance (Erikson and Jonsson 1996). Parental material and financial resources may also influence the quality of the school attended by children, as well as their participation in extra-curricular activities and courses (Garner and Raudenbush 1991; Lehmann 2012; Mayer 2002). In addition, genetics play a role in status transmission of parents to children, by affecting for example cognitive ability and personality (Nielsen 2006).

SES of the childhood family may influence the likelihood of having a birth in cohabitation or while being single indirectly through the intergenerational transmission of SES. For example, those growing up with higher-SES parents may have higher educational and occupational aspirations which can increase their motivation to avoid single motherhood. This link may become even stronger due to the actual higher educational and occupational performance of children of higher-SES parents.

3.3 Data and Method

This chapter builds on the research of Koops et al. (2017). For more information on sample selection and analytical approach, I refer the reader to this article.

3.3.1 Data

This chapter uses data from the Generations and Gender Survey Wave 1 (Generations and Gender Programme 2019; Vikat et al. 2007). In addition, it used data of the Canadian General Social Survey (Béchard and Marchand 2008), and the Dutch Survey on Family Formation (CBS 2012). Post-harmonized data of the American National Survey on Family Growth and the British Household Panel Survey were taken from the Harmonized Histories dataset (Perelli-Harris et al. 2010).

3.3.2 Variables

Parental SES can be captured with various indicators. The most commonly used are parental education, occupational status, and income (Breen and Jonsson 2005). All three indicators have a clear link with economic capital and therefore with parental resources. However, especially education is commonly mentioned as a proxy for cultural capital which may capture parental preferences better (Bourdieu 1986; Kraaykamp and van Eijck 2010). In addition, parental education is rather stable over time, and may therefore give a better reflection of the general situation during childhood. Therefore, this chapter follows the strategy of Koops et al. (2017) by capturing parental SES with parental education.

Father’s and mother’s education are recoded from the categorical International Standard Classification of Education (ISCED) into a continuous variable (ranging from 0 to 100) based on the International Standard Level of Education (ISLED) coding system, a method developed by Schröder and Ganzeboom (2014). A coding scheme was available for all countries, except for Georgia, the USA, and Canada (Brons and Mooyaart 2018). For these three countries, the general conversion scheme – based on the correspondence between ISCED and ISLED in all available countries – was used instead (Schröder and Ganzeboom 2014). Parental education was constructed to represent the mean value of father and mother’s ISLED, or the educational level of only one of the parents if the information was not available for both.

Own education was also converted from ISCED to ISLED. The variable is added as a time-varying covariate in the model by using information on the highest educational level attained by the respondent in combination with information on the year and month in which this highest level was reached. First it was established in which year individuals finished primary school, and which ISLED level was reached at this point. From secondary education onwards, individuals are assumed to be enrolled in school continuously, until they reached their highest attained educational level. After this point in time, education remains at a constant level. Missing information on year and month of reaching the highest educational level was imputed by taking the median age of reaching that level in that country.

Cohort is a continuous variable reflecting respondent’s birth year. The variables age, age2, and age3 are time-varying and refer to the respondent’s age in months. They are included in the models to correct for the nonlinear effect of age on the risk of having a first birth.

3.3.3 Analytical Approach

Discrete-time competing risk models were used to estimate the influence of parental education on the monthly risk of having a first birth in marriage, in cohabitation, or while being single. This model examines the monthly risk of experiencing these events, starting at the age of 15. Those who did not have a biological child at the time of the interview or before the age of 45, were right-censored. Having a birth in marriage is taken as the reference category. The models test if an increase in parental education alters the risk of becoming a cohabiting or a single parent as compared to a married parent. Two models were estimated for each country, one excluding own education and one including own education. The results of the multinomial logistic regressions reported as relative risk ratios are presented in Tables 3.1 and 3.3 of the Appendix. Note that these results are the same as reported in Tables 3 and 5 of Koops et al. (2017).

The multinomial logistic regressions answer the question if parental education influences the risk of having a birth in marriage, cohabitation, or while being single, before and after including own education. However, it is not immediately clear from the coefficients how relevant the effect sizes of parental education are. In order to make the effect sizes easier to interpret, the margins of the dependent variable were estimated for the lowest and for the highest reported level of parental education in a country. The other indicators were set at their average. With these specifications, the margins estimate the predicted probability for a certain outcome given the specified level of parental education. Thereafter, the predicted probability of having a birth in cohabitation (or while being single) was divided by the predicted probability of having a birth in marriage, cohabitation or while being single. With the KHB-method, the percentage reduction in the effect of parental education caused by the mediation through own education was obtained (Karlson et al. 2012). The KHB-method provides unbiased decompositions for non-linear models. Together, this information provides a detailed picture of the extent to which the effect of parental education is partially or fully mediated through own education and how large the effect of parental education on partnership status at first birth is, once own education is accounted for. The results are summarized in Figs. 3.2, 3.3, 3.4, 3.5 and 3.6.

3.4 Results

Figure 3.1 gives an overview of the distribution of first births across partnership contexts. In all countries, most births happen in the context of marriage. At the same time, the actual percentage of births in marriage differs substantially between countries. In some Eastern European countries such as Hungary and Romania, close to 90% of first births happen in the context of marriage, while in some Western European countries such as Austria and Norway, about 60% of first births happen in marriage. First births to single people are the least common, about 5–10% of births happen in this partnership context. Compared to the other countries, the USA is an outlier when it comes to births to singles. In the USA more than 20% of births happen to men and women who are living single. In Eastern Europe and North-America the percentage of first births that happen in unmarried cohabitation tends to be low and is comparable with the percentage of first births that happen to those who are single. However, in Western Europe and also in some Eastern European countries, such as Georgia and Estonia, the percentage of births to cohabiting couples clearly exceeds the percentage of births to singles.

Fig. 3.1
The bar graph illustrates that the percentage of the respondent having their first child after marriage is higher than that of cohabiting, and after that, single.

Weighted percentage of respondents having a first child in marriage, cohabitation, and while being single

3.4.1 First Births in Marriage

Figure 3.2 provides a graphical representation of the influence of parental education on the probability of having a birth in marriage, when own education is not taken into account. The results are based on multinomial logistic regressions, see Tables 3.1 (for women) and 3.2 (for men) of the Appendix. The dots reflect the predicted probability of having a birth in marriage among those with the lowest level of parental education in that country (black dot) and those with the highest level of parental education in that country (grey dot). The length of the line between the dots reflects the maximum difference in predicted probability of having a birth in marriage based on parental education. A line attached to the right side of the black dot indicates a positive effect of parental education on the probability of having a birth in marriage, while a line attached to the left side of the black dot indicates a negative effect.

Fig. 3.2
The graphical representation of the probability of the influence of parental education on men and women having children through marriage.

Differences in the predicted probability of having a first birth in marriage between those with the lowest level of parental education (black dot) and those with the highest level of parental education (grey dot)

Note: Results are based on the multinomial logistic regressions excluding own education, presented in Tables 3.1 and 3.2 of the Appendix

The results show that women with higher educated parents have a higher predicted probability of having a birth in marriage as compared to those with lower educated parents. Differences in the probability of having a birth in marriage are largest in Norway, the USA, and Estonia. In these countries, the predicted probability of having a birth in marriage is about 25 percentage points higher for those with the highest level of parental education as compared to those with the lowest level of parental education. In Bulgaria, the Czech Republic and France the difference is close to 20 percentage points. The difference between educational levels is around 12 percentage points in Canada, Russia, Georgia, Romania, Lithuania, and Hungary. In Austria, Belgium, Poland, and the Netherlands the difference is smaller than 10 percentage points.

For men, the difference in the probability of having a birth in marriage between parental educational levels is largest in Bulgaria, the USA, and Norway, with distinctions of approximately 25 percentage points. In Romania, Georgia, and Hungary the difference is close to 15 percentage points, and in the Czech Republic, Estonia, and Canada close to 12 percentage points. In Poland, Lithuania, Belgium, France, Russia, Austria, and the Netherlands, differences are smaller than 10 percentage points. For women in France and Russia, substantial differences were found in the predicted probability of having a birth in marriage by parental education, but this is not found for men. Apart from these two countries, patterns for men and women are relatively similar.

3.4.2 First Births in Cohabitation

Figure 3.3 displays the influence of parental education on the probability of having a birth in cohabitation, when own education is not taken into account. In the majority of countries, women with lower educated parents have a higher probability of having a birth in cohabitation. The difference between educational levels is largest in Bulgaria, Georgia, and Estonia. In these countries the probability of having a first birth in cohabitation is about 15 percentage points higher for those with the lowest level of parental education as compared to those with the highest level of parental education. In the USA, Norway, and Romania the differences are around 12 percentage points. In all other countries, the differences are smaller than 10 percentage points.

Fig. 3.3
The graphical representation of the probability of the influence of parental education on men and women having children in cohabitation when education is not taken into account.

Differences in the predicted probability of having a first birth in cohabitation between those with the lowest level of parental education (black dot) and those with the highest level of parental education (grey dot)

Note: The asterisks indicate if parental education significantly influences the risk of having a birth in cohabitation or in marriage (* p < 0.05; ** p < 0.01; *** p < 0.001). Results are based on the multinomial logistic regressions excluding own education presented in Tables 3.1 and 3.2 of the Appendix

For women it is found that in Norway, North American and East European countries (except Poland) higher parental education is related to an increased risk of having a birth in cohabitation. In the Netherlands and Austria, women with lower educated parents have a lower predicted probability of having a birth in cohabitation compared to women with higher educated parents. However, the differences by educational level are small, and it is only statistically significant for the Netherlands. In France, Belgium, and Poland, parental education does not significantly influence the risk of having a birth in cohabitation or marriage.

For Bulgarian men, the difference in predicted probability of having a first birth in cohabitation is 28 percentage points higher for those with the lowest level of parental education as compared to those with the highest level of parental education. The difference in Estonia is 20 percentage points, and in the USA, Romania, Georgia, and Norway close to 15 percentage points. In all other countries the difference between educational levels is smaller than 10 percentage points. Generally, the influence of parental education on the risk of having a birth in cohabitation or marriage is similar for men and women. Only in some Eastern European countries the conclusions for men and women are different. In the Czech Republic, Lithuania, and Russia, the negative effect of parental SES on the risk of having a birth in cohabitation is small for men and not statistically significant.

Figure 3.4 shows the influence of parental education on the probability of having a birth in cohabitation once own education is included in the model. In Russia, Lithuania, the Czech Republic, and the Netherlands the effect of parental education on the risk of having a birth in cohabitation or marriage is fully mediated through own education. In Norway, Georgia, Bulgaria, Estonia, the USA, Romania and Canada, the effect of parental education is partially mediated through own education. The partial reduction in the effect of parental education that can be attributed to the mediation through own education ranges from 18% in Norway to 51% in Romania. When taking mediation through own education into account, the difference between parental educational levels in the probability of having a birth in cohabitation is largest in Norway, Georgia, and Bulgaria, with close to 10 percentage points. In all other countries, the difference is smaller.

Fig. 3.4
The graphical representation of the likelihood that parental education will have an impact on men and women who give birth in cohabitation when education is taken into account.

Differences in the predicted probability of having a first birth in cohabitation between those with the lowest level of parental education (black dot) and those with the highest level of parental education (grey dot)

Note: The asterisks indicate if parental education significantly influences the risk of having a birth in cohabitation or in marriage (* p < 0.05; ** p < 0.01; *** p < 0.001). Results are based on the multinomial logistic regressions including own education presented in Tables 3.3 and 3.4 of the Appendix. The graphs also report on the percentage reduction in the effect of parental education that can be attributed to the mediation effect through own education. This percentage is only reported for those countries for which a significant effect of parental education was found in the models excluding own education

For the model including a respondent’s own education, different results for men and women are found in three countries. For men in the Netherlands, the positive effect of parental education on the risk of having a birth in cohabitation is only partially mediated through own education (while it is fully mediated for women). In Estonia, the negative effect of parental education on the risk of having a birth in cohabitation is fully mediated through own education (while it is partially mediated for women). Notably, in France, only after controlling for own education, a positive and significant effect of parental education on the risk of having a birth in cohabitation is found for men (while a negative – non-significant - effect is found for women). The largest difference of parental educational levels on the probability of having a birth in cohabitation is found in Bulgaria (20 percentage points), and Estonia and Romania (around 12 percentage points). In all other countries the difference is smaller than 10 percentage points.

3.4.3 First Births While Being Single

In the case of births to singles, one has to be wary of data quality issues for men. Previous research has shown that men tend to underreport the number of nonmarital children, especially when they do not share a household with these children (Joyner et al. 2012). It is therefore possible that men who had their first birth while being single are less accurate in providing birth dates or might not report these births. Research has shown that underreporting can affect the influence of background variables on fertility outcomes, by attenuating the coefficients (Joyner et al. 2012). Figures 3.5 and 3.6 show the results of the influence of parental education on the probability of having a birth while being single, for both men and women. However, the results for women will be considered more reliable and will therefore be the focus of the next discussion.

Fig. 3.5
The graphical representation of the likelihood of parental education has an impact on men and women giving birth while unmarried when education is not taken into account.

Differences in the predicted probability of having a first birth while being single between those with the lowest level of parental education (black dot) and those with the highest level of parental education (grey dot)

Note: The asterisks indicate if parental education significantly influences the risk of having a birth while being single or in marriage (* p < 0.05; ** p < 0.01; *** p < 0.001). Results are based on the multinomial logistic regressions excluding own education presented in Tables 3.1 and 3.2 of the Appendix

Fig. 3.6
The graphical representation of the probability of parental education has an impact on men and women giving birth when unmarried when education is taken into account.

Differences in the predicted probability of having a first birth while being single between those with the lowest level of parental education (black dot) and those with the highest level of parental education (grey dot)

Note: The asterisks indicate if parental education significantly influences the risk of having a birth while being single or in marriage (* p < 0.05; ** p < 0.01; *** p < 0.001). Results are based on the multinomial logistic regressions including own education presented in Tables 3.3 and 3.4 of the Appendix. The graphs also report on the percentage reduction in the effect of parental education that can be attributed to the mediation effect through own education. This percentage is only reported for those countries for which a significant effect of parental education was found in the models excluding own education

In the majority of countries, women growing up with lower educated parents have a higher probability of having a birth while being single (see Fig. 3.5). The difference between educational levels is largest in France, Norway and Austria, where the probability for a birth while being single is about 15 percentage point higher for those with the lowest level of parental education compared to those with the highest level of parental education. In the USA, the Czech Republic and Estonia, the difference is close to 11 percentage points. In all other countries the difference is less than 10 percentage points.

In North-America, women with lower educated parents have a higher risk of having a birth while being single than in marriage, as compared to those with higher educated parents. In Western Europe, the negative effect of parental education on the risk of having a birth while being single is statistically significant in France, Norway, and Austria but not in Belgium and the Netherlands. In Central and Eastern Europe, a negative effect of parental education is found for women in the Czech Republic, Estonia, Hungary, and Russia. Parental education does not significantly alter the risk of having a birth while being single or in marriage in Lithuania, Bulgaria, Poland, Romania, and Georgia.

Figure 3.6 shows the effect of parental education on the probability of having a birth while being single when own education has been taken into account. For women in France, Estonia, the Czech Republic, Russia, Canada, and Hungary, the effect of parental education on the risk of having a birth in cohabitation or in marriage is fully mediated through own education. In Austria, the USA, and Norway, the effect of parental education is partially mediated through own education. The reduction in the effect that can be attributed to this mediation ranges from 19% in Austria to 40% in the USA. When taking mediation through own education into account, the difference between parental educational levels on the probability of having a birth in cohabitation is close to 9 percentage points in Austria, the USA, and Norway.

3.5 Discussion

Previous research has shown that parental SES is an important factor in explaining partnership status at first birth. Research suggests that parents influence their children’s family formation through the mechanisms of socialization, availability of resources, and the intergenerational transmission of SES. In this chapter, the methodological approach of Koops et al. (2017) is followed by examining the effect of parental SES on partnership status at first birth in 16 countries. Particular focus has been placed on whether the influence of parental SES on the dependent variable is mediated through own SES.

In most countries investigated (11 out of 16), women with lower-SES parents are more likely to have a birth in cohabitation than women with higher-SES parents. Differences in the predicted probability of having a birth in cohabitation between women with the lowest level of parental SES and women with the highest level of parental SES can be up to 15 percentage points. These results align with the Pattern of Disadvantage perspective which views cohabitation as a ‘poor-man’s marriage’ (Edin and Kefalas 2005; Gibson-Davis et al. 2005). Those with lower parental SES may be less likely to meet the requirements for marriage, and are therefore more likely to have a first birth in cohabitation.

In 4 out of 11 countries (Russia, Lithuania, the Czech Republic, and Hungary), the negative effect of parental SES on the likelihood of having a birth in cohabitation or marriage is fully mediated through own SES. In the remaining 7 countries (Canada, USA, Norway, Bulgaria, Estonia, Georgia, and Romania), the negative effect of parental SES on the likelihood of having a birth in cohabitation is partially mediated through own SES. The partial reduction in the effect of parental SES that can be attributed to the mediation through own SES ranges in these countries between 18–51%. This is a first indication that the influence of parental SES on the likelihood of having a first birth in cohabitation and marriage, is at least partly explained by the intergenerational transmission of SES. At the same time, regardless of the attenuation in the effects, a direct effect of parental SES is also found. A possible explanation is that in these countries, women with lower-SES parents receive fewer financial transfers from their parents (resources mechanism), which hamper them from being married when having their first child.

Only in one country (the Netherlands), a positive effect is found of parental SES on the likelihood of having a first birth in cohabitation. However, the differences in the predicted probability of having a birth in cohabitation between women with the lowest level of parental SES and women with the highest level of parental SES is only 3 percentage points. The results indicate that after including own SES in the model, the positive effect of parental SES becomes smaller and is no longer statistically significant. This indicates that for women in the Netherlands, the effect of parental SES may, for a large part, be explained through the intergenerational transmission of SES.

Existing fertility-related studies tend to focus on women. One of the reasons for not including men in fertility research is their tendency to underreport nonmarital children or to misreport birth dates. Under- and misreporting are especially prevalent when men have not shared a household with their children (Joyner et al. 2012). It is therefore not problematic to include men in research focusing on births that happen within cohabitation or marriage. The results in this chapter show that the influence of parental SES on the likelihood of having a birth in cohabitation or marriage is comparable between men and women. However, in the Netherlands and France some interesting gender differences appear once the models include own SES. As mentioned in the previous paragraph, for women in the Netherlands the effect of parental SES on the likelihood of having a birth in cohabitation is nonsignificant after including own SES in the model. In contrast, for men the positive effect of parental SES becomes stronger after own SES is included in the model and remains statistically significant. A similar pattern is found for men in France. However, in this case the effect of parental SES on the likelihood of having a birth in cohabitation becomes positive and is statistically significant only after own SES is included in the model. These results may suggest that in the Netherlands and in France, men with higher-SES parents are more likely to have a birth in cohabitation, because these parents are more positive about cohabitation than lower-SES parents, and they transmit these preferences to their sons. It is remarkable that these effects are only found for men, and not for women. Some research on the intergenerational transmission of living arrangements has found that the socialization mechanism is stronger for daughters than for sons (Axinn and Thornton 1993). However, the findings in this chapter are more in line with research showing that parents transmit more traditional family norms to their daughters and more liberal norms to their sons (Barber 2000). The effect sizes are not large. The probability of having a first birth in cohabitation is 7–8 percentage points higher for men with the highest level of parental SES compared to men with the lowest level of parental SES.

This research also contrasted first births occurring while being single and in marriage. The results presented in this chapter show that in 9 out of 16 countries, women with lower-SES parents are more likely to have a birth while being single. In 6 countries (France, Estonia, the Czech Republic, Russia, Canada, and Hungary), the negative effect of parental SES on the likelihood of having a birth while being single or in marriage is fully mediated through own SES. In the USA, Austria, and Norway, the effect of parental SES on the likelihood of having a birth while being single is only partially mediated through own SES. The reduction in the effect of parental SES that can be attributed to this mediation is between 19–40%. These findings may indicate that those growing up with higher-SES parents are more motivated to avoid single parenthood because they may attribute more importance to other life goals than motherhood (socialization mechanism and intergenerational transmission of SES). In addition, they may be better able to prevent unplanned pregnancies and they may be more likely to move in with a partner when confronted with a pregnancy (resources mechanism and intergenerational transmission of SES).

To sum up, the research in this chapter provides consistent evidence that parental SES influences partnership status at first birth in western societies. Overall, lower parental SES is related to a higher likelihood of having a first birth in cohabitation and while being single. It is possible that unequal occurrence of births in cohabitation and while being single aggravate inequalities between groups. Previous research shows that single motherhood has negative socio-economic consequences for mothers and their children (Kollmeyer 2013). Moreover, cohabiting parents are often treated differently than married parents. For example, cohabiters may be entitled to a lesser extent to tax and social security benefits. Also, after separation or death of a partner, the division of property and entitlement to inheritance may depend on the partnership status (Perelli-Harris and Gassen 2012). It is therefore important to understand if inequalities that exist at the start of parenthood could be prohibited. What may help in this regard is understanding why inequalities that are found in some western societies are not found in others. The research in this chapter suggests that the influence of parental SES on the likelihood of having a birth outside of marriage can be partly explained through the intergenerational transmission of SES. In addition, a direct – mostly negative – effect of parental SES is found. Possibly, differences in parental support in the transition to adulthood can explain these differences. In the case of single motherhood, it is also possible that parental SES influences the ability to prevent pregnancies.