Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

6.1 Introduction

Following the chapter on relationship dissolution, this chapter examines repartnering over a 10-year period. Many people who have experienced relationship breakdowns go on to have new relationships. Repartnering can be defined as forming a new intimate relationship after the dissolution of a previous one, and can take a number of forms. This is because of the widespread changes in the types of relationships available to people in Western-industrialized countries like Australia. Until the 1970s, repartnering almost exclusively took the form of remarriage.

Non-marital cohabitation was evident at low levels in the 1971 (0.6 % of families) and 1976 censuses (2.2 % of families) (Saratankos 1983). Over the last 30 years this has increased considerably, with around 11 % of adults living in a cohabiting relationship in 2009–2010 (ABS 2012a). As discussed in Chap. 2, these relationships include cohabitation without marriage, cohabitation followed by marriage, and cohabitation after relationship dissolution. This increase in cohabitation, and its varying forms, is important to consider in this chapter on repartnering in Australia. To date, a substantial amount of research has considered remarriage, but little focuses on repartnerships in the context of the contemporary trends in relationship formation.

This chapter starts with a literature review and provides a theoretical lens that incorporates individual histories and social context for investigating repartnering. This is followed by a description of the data used to measure repartnering over a 10-year period. The analysis presented will be based on the retrospective and prospective longitudinal information available from the Household, Income and Labour Dynamics in Australia (HILDA) survey (2001–2010). These data provide an opportunity to incorporate past relationships, family and fertility histories with current socio-demographic characteristics into understanding patterns of repartnering. The main emphasis of the results of this chapter will be on ‘who, when and how’ people repartner.

6.2 Repartnering Pathways

In countries like Australia which have experienced a second demographic transition, repartnering takes the form of both remarriages and cohabiting partnerships, but also ‘living apart together’ (LAT) relationships (see Chapter 2). Lesthaeghe (1995), in his discussion of second demographic transition theory, argues that remarriage probabilities may decline, but that alternative repartnership forms such as cohabitation and LAT relationships will increase. It has also been noted that repartnering is becoming increasingly important because of relatively high divorce rates as well as increases in the percentage of cohabiting relationships that break up (de Vaus 2004).

Given the importance of both divorce, and cohabitation breakdown, there are now four discernible pathways of repartnering (Fig. 6.1). These can be illustrated as follows:

Fig. 6.1
figure 1

Repartnering pathways

Full size image

These pathways can be described as: (1) consecutive marriages; (2) cohabitation with a new partner following a marriage; (3) marriage to a new partner following cohabitation with another partner; and (4) consecutive cohabitations. Widowhood is another potential pathway into repartnering, and although occurring across all ages, is most likely to affect those aged 60 years or more. In that age group, the repartnering rate is 9.9 per 1,000 for men and 2.0 per 1,000 for women in 2011 (ABS 2012b).

National statistics provide information on remarriage rates and age at remarriage over time. While these statistics cannot provide information on patterns 2, 3, or 4, a substantial number of repartnerships fall into pattern 1. Overall, marriage is still more common than cohabitation, although a different pattern is evident for those who are repartnering: the results presented in this chapter show that people may be more likely to opt for pattern 2 (divorce followed by cohabitation) than pattern 1. Nevertheless, these official statistics provide details on remarriage trends that predominated in the past, even though post-marriage cohabitation is more common now.

Figure 6.2 shows the substantial decline in remarriage rates over the period 1976–2011. At both times, remarriage rates were much higher for women and men in the prime marriage age groups than for older age groups. In 1976, remarriage rates were higher for men than for women in all age groups except 20–24 years, which is due to women marrying earlier than men. In 2011 there is not such a dramatic difference between age groups, although there is a clear downward trend at the older ages.

Fig. 6.2
figure 2

Remarriage rates, men and women, 1976 and 2011 (ABS 3310.0 various years and online Census table builder 2011)

Full size image

This general flattening of remarriage rates over the period is also evident in the increase in the median age at remarriage. Figure 6.3 provides the median age for men and women from 1979 to 2011. This pattern also reflects the general increase in median age at marriage over the same period (see Chap. 3).

Fig. 6.3
figure 3

Median age at remarriage, men and women, 1979 to 2011, Australia (ABS 3310.0, various years)

Full size image

6.3 Factors Associated with Repartnering

Repartnering is an opportunity to embark on a new stage of life with a new partner, and as described, it can take the form of either remarriage or cohabitation. To date, there has been little written on who repartners, the timing of repartnering, and whether people get married or cohabit when repartnering in Australia.

Some notable exceptions include the study by Weston and Khoo (1993) who looked at both cohabitation and remarriage of divorced parents over a three year period, de Vaus (2004) who provided detailed statistics on remarriage for the divorced and widowed, and Skew et al. (2009) who compared repartnering in the UK and Australia. De Vaus’ report found very large differences in the likelihood of remarrying between people who were divorced and those who were widowed, with the divorced much more likely to remarry than those who were widowed. He further found that the gap between men and women in median age at remarriage was greater than that for first marriage, and further that men were more likely to remarry than women. Weston and Khoo, whose study was based on parents, also found that men were more likely to repartner, and repartner faster than women.

Despite the relative paucity of research on remarriage or repartnering in Australia, there is a vast amount written – particularly about remarriage – from North America and Europe. It is important to acknowledge that most of this research focuses on remarriage, even though in many countries the majority of second unions take the form of cohabitation (e.g. Wu and Schimmele 2005 for Canada). Most of the research on remarriage has focussed on understanding socio-demographic factors. However, person histories are also important, and I argue that previous relationships and childbearing histories matter in terms of understanding repartnering. Coleman and Ganong (1990) in their 1980s decade review found that few studies included prior relationship history, a sentiment echoed by Poortman (2007). Sweeney (2002, p. 411) expressed surprise at this omission in the literature given that much research on family transitions is based on a life-course approach, which typically takes into account past experiences. Cherlin provides one explanation. In his review of the first decade of the twenty-first century, he noted that family demographers ‘moved further away from the framework of a conventional, uniform family life cycle’ (2010, p. 403). He attributed this to the increasing divergence in family patterns, often attributable to disadvantage (education and income), as a reason for rethinking what affects family processes like family formation and dissolution.

In previous research, my co-authors and I argued that both individual life experiences and socio-demographic background influence repartnering by: (1) affecting a person’s own behaviour or attitude towards forming a new union, and (2) affecting their attractiveness as a potential partner to others (Skew et al. 2009). This is a similar framework to that used by Ivanova et al. (2013) who refer to need, attractiveness, and opportunity. These concepts will be used to provide context for the factors that have been found to be associated with repartnering (remarriage).

6.3.1 Age and Gender

Age and gender are inextricably related when it comes to patterns of repartnering. Age is often used in demographic research, not just as a measure of chronological age, but potentially also of generational differences in attitudes toward repartnering. It has been found that people who end their relationships at younger ages are more likely to remarry than people who end their relationships at older ages, and this holds for both men and women (de Vaus 2004). However, the pattern is much stronger for women.

Overall, women are less likely to repartner than men (Poortman 2007; Wu and Schimele 2005), but there is a much greater difference at later ages. One explanation is that men are more likely to be involved in the labour market at these ages than women, and work is a common way to meet people (de Graaf and Kalmjin 2003). It has also been argued that the marriage market plays a role here (Dean and Gurak 1978). As women tend to marry men who are a few years older than themselves, over time women’s pool of potential partners diminishes faster than men’s. This is also associated with greater longevity for women; hence women are more likely to experience widowhood than men.

6.3.2 Previous Unions

As described above, some past studies have lacked information on previous unions, such as duration of the previous relationship. However, there are studies that included this type of life course measure, including several from the latter years of the 20th century (Bumpass et al. 1990; Koo et al. 1984; Mott and Moore 1983). The main finding from these studies was that relationship duration had little effect on the likelihood of repartnering. More recent studies show that length of previous relationship is positively associated with repartnering (De Graaf and Kalmijn 2003; Poortman 2007; Wu and Schimmele 2005).

6.3.3 Children from Previous Unions

An important contribution to the literature on repartnering is a recent paper by Ivanova et al. (2013). As mentioned, this paper discusses why children might affect repartnering, and focuses specifically on the concepts of need, attractiveness, and opportunity. The authors argue that these concepts provide a theoretical position for explaining how children can affect repartnering. In sum, they find that childless men and women do not differ in the probability of repartnering, and that these findings hold across different institutional and cultural settings. Further, as children age, the chances of entering a new union increase.

This paper is useful for a number of reasons. Firstly, the effect of children on repartnering is a specific focus, although this is not necessarily unique to that paper. Other studies that specifically account for the role of children include: Bernhardt and Goldscheider (2002), Koo et al. (1984), Lampard and Peggs (1999), Stewart et al. (2003), Teachman and Heckert (1985).

Secondly, the paper does not only focus on the effect of children on women’s repartnering. A number of recent papers include the effect of children on men’s repartnering (Bernhardt and Goldscheider 2002; Goldscheider and Sassler 2006; Skew et al. 2009; Stewart et al. 2003). The results of past fertility seem to differ for men and women. Although the presence of children is consistently found to be associated with lowering repartnering rates for women, for men the effect is more mixed and not always significant (De Graaf and Kalmijn 2003).

Thirdly, Ivanova et al. (2013) provide a framework with three dimensions for considering the effect of children: needs, attractiveness, and opportunities. For example, mothers without a partner may have an economic need to repartner and may form new partnerships as a strategy to relieve pressure on their households (Duncan and Hoffman 1985; Smock 1990; Weston and Khoo 1993). However, most research has found that the presence of children from prior relationships has a negative effect on repartnering (Coleman et al. 2000). The chance of forming a new union decreases as the number of children increases.

Bumpass et al. (1990) argue that having children from a previous partnership may decrease one’s attractiveness as a partner due to its association with various costs, both direct financial costs and indirect costs associated with the complexities of step-families. This illustrates Ivanova et al.’s second concept: attractiveness. And lastly, opportunity: the presence of children may act as a barrier to repartnering by decreasing the chance for social interaction and the possibility of finding a new partner (de Graaf and Kalmijn 2003; Ermish et al. 1990; Wallerstein and Blakeslee 1989).

6.4 Data

This study uses Waves 1–10 of the HILDA survey, 2001–2010. Details on this household-based panel study are available in the Technical Appendix. In this chapter, the analysis is based on individual characteristics, and on the factors discussed previously. The research focuses on the respondent’s socio-demographic characteristics (various measures), their previous relationship type, whether they have children, and religiosity.

Table 6.1 provides details on these measures for the analytical sample, that is, respondents who were observed for more than one wave of the survey, and who were observed to separate (and stayed separated) from their partner. The sample selection is described in further detail below.

Table 6.1 Sample descriptives: Individuals included in event history models of repartnering
Full size table

The sample includes more women than men, a fairly evenly spread across the age ranges (although relatively fewer in the 50–59 year age group), fewer respondents who have completed university or secondary education than other levels of education, and more respondents who were married than previously cohabiting.

About 7 out of 10 respondents (67.9 %) did not repartner in the time they were observed. Respondents were much more likely to enter into a cohabiting repartnership than a marriage (27 % v 4 %). Of the 446 respondents who repartnered into cohabitation, one quarter were subsequently observed to have married their partners (N = 114).

6.4.1 Sample Selection

This analysis is based on a representative sample of Australians who have experienced relationship dissolution, and are observed over a period of up to 10 years to determine characteristics associated with repartnering. Hence, it is based on contemporary relationship (re)formation.

As the analysis follows people over time, respondents had to be observed for at least two waves between the periods of collection 2001–2010. There are a number of exclusions listed in Appendix 6.1. Respondents who experienced relationship dissolution were included in the sample up until they experienced a repartnership or were censored (that is, the final wave where they were observed but did not repartner). The final sample size is 1,643 respondents who were observed for a total of 6,506 person years. The average number of waves observed for was 3.17.

6.5 Method

I start by providing some indicators of the level of repartnerships by age in 2005, since this is the mid-point of the data collection period. The indicators provided are (1) percentage repartnered; (2) percentage married more than once; (3) percentage repartnered following death of partner and (4) percentage cohabited more than once.

The main analysis is based on descriptive survival analysis, followed by discrete-time event history analysis of those observed to have a relationship dissolution to determine the factors associated with repartnering.

Survival analysis is used to describe the timing to repartnering following relationship dissolution for both those previously married and those previously cohabiting. Overall levels of repartnering are provided, as well as disaggregating the new partnership by whether it is a cohabitation or marriage. This analysis is then compared for men and women. Given the striking findings of Ivanova et al. (2013) about the effect of children on the differential repartnering of men and women, time to repartnering is also analysed by whether respondents have children or not.

The analytical event history models are run separately for overall repartnering, repartnering through marriage, and repartnering through cohabitation. These models are run separately for men and women. There are some issues with sample size that affect the modelling of marriage for men, but results are available for overall repartnering and for cohabitation. Discrete-time event history models (or discrete-time hazard models) are appropriate when there are individuals who are ‘right-censored’, that is, respondents who have not experienced the event while under observation, but may still experience the event in the future. Standard statistical techniques such as logistic regression cannot handle censored data adequately, however discrete-time event history models simply apply standard logistic regression to person-period data, making the results quite easy to interpret (Singer and Willett 2003). Hence, the HILDA data have been set up as person-period data for this purpose.

Three models are run, separately by sex, which include duration of last relationship, type of relationship that ended, age (time varying), and highest education level (time varying). The first and third models include whether the respondent has a resident child aged <5, whether the respondent has a resident child aged 5–14, and whether the respondent has a resident child aged 15–24. These variables are not mutually exclusive – respondents can have a resident child in more than one of these age groups. Model 2 uses number of children ever born instead of resident children. Model 3 (the model which will be discussed) also includes how important religion is to the respondent.

6.6 Results

6.6.1 Indicators of Repartnering by Age and Sex

Table 6.2 provides indicators of the percentages of people experiencing various types of repartnering. Collecting relationship histories is a difficult task: there is often missing information on dates, or forgotten relationship information; recall error is a major issue (Gaskell et al. 2000; Hayford and Morgan 2008; Reimondos et al. 2011). In the case of HILDA, only a partial cohabitation history is collected (Reimondos et al. 2011). This means that there is a somewhat limited range of indicators available using retrospective relationship measures.

Table 6.2 Repartnering indicators by age and sex, 2005 (HILDA 2001–2005, retrospective information included)
Full size table

Distinct patterns are evident even with these limited indicators. Consider first the percentage that have ever experienced relationship dissolution, and who have been or are at risk of repartnering. The proportion is around 50 % for all age groups, but it is noticeable in the group aged 60+ that men are less likely to have experienced relationship dissolution than women. This is related to two demographic processes: (1) women tend to marry men a few years older than themselves; and (2) men have a shorter life expectancy. This means that women are much more likely to have experienced a dissolution than men in the older age groups.

Secondly, in the younger age groups there is little difference between men and women in the percentage who repartner. However, by age group 40–49 a greater percentage of men have repartnered than women, and this gap widens in the older age groups.

Perhaps surprisingly, there is little difference in the percentage of men and women who have married more than once, however there are some differences in the percentage that have cohabited more than once. In their 20s women are more likely than men to cohabit more than once, while men in their 40s and 50s are more likely than women to cohabit more than once.

Finally, looking at the percentage remarried after the death of the spouse, we again see the effects of the age difference in marriage and the longevity of men in these results. Women are affected earlier by spousal death than men; 22 % of those aged 50–59 have repartnered following a spouse dying. Too few men in this age group have experienced a spousal death to make an estimate of repartnering. In the 60+ age group, despite more women having experienced a spousal death, women are much less likely to repartner (10 %) than men who have experienced a spousal death (27 %).

6.6.2 Time to Repartnering

The survival graphs (Figs. 6.4, 6.5 and 6.6) provide information about the time to repartnering for people who have experienced relationship dissolution since 2000. Figure 6.4 shows that those people who experienced a marital dissolution had a longer survival time (that is, they were single for longer) than those who had a cohabitation breakdown. Of those who were previously cohabiting, about half had repartnered within 5 years of the previous relationship. For those who were previously married, almost half had repartnered 9 years after the previous relationship.

Fig. 6.4
figure 4

Survival analysis: time to repartnering (a) by previous relationship type; (b) by sex for previous marriage; (c) by sex for previous cohabitation

Full size image
Fig. 6.5
figure 5

Survival analysis: time to repartnering by previous relationship type

Full size image
Fig. 6.6
figure 6

Survival analysis: time to repartnering by sex and parenthood

Full size image

Looking at the repartnering times separately for men and women and by previous relationship type, it is evident that for those who were previously cohabiting there was no difference in the time to entering a subsequent relationship for men and women. However, among those who were repartnering after a marriage, men have a faster time to repartnering than women (about half of men had repartnered within 7 years, while only 40 % of women had repartnered 9 years post-marriage).

While the initial type of relationship is important for repartnering timing, more important is the type of relationship that people are entering. There is a much slower time to entering marriage than cohabitation, and this pattern holds whether people were leaving a marriage or cohabitation (Fig. 6.5). For those who were previously married, around 10 % entered a new marriage without cohabiting, while around 40 % entered a cohabitation not followed by marriage. This pattern was more extreme for those previously cohabiting: only a small percentage (less than 5 %) entered directly into a marriage, while over 60 % were observed to cohabit without marrying.

Although not shown here, these survival curves were compared by sex. There was no significant difference in the patterns observed for those whose previous relationship was cohabitation. However, for repartnering after marriage, men were more likely to either cohabit or remarry than women.

As expected, children played a significant role in the time to repartnering, which differed for men and women. Figure 6.6 shows the time to repartnering for men and women by whether they have a child or not. The solid lines show that for men and women with no children, there is no significant difference in the survival time to repartnering. However, for fathers and mothers there is a statistically significant difference. Fathers repartner faster than mothers, and are more likely to repartner over the period. Fifty percent of fathers have repartnered in around 7 years following dissolution.

The pattern for those with a resident children is more nuanced. Resident children matter in different ways for men and women (Fig. 6.7). This figure is based on respondents who are parents, but whose children may or may not be resident. It is interesting that having resident children matters for men; those who have a resident child aged 15 years or less have a slower progression to repartnering, and are less likely to repartner over the period than men who do not have a resident child. For women, the pattern is opposite, although not statistically significant. Of course, these results do not control for other factors, so this may be partly an age effect whereby women who do not have resident children are older than women with resident children.

Fig. 6.7
figure 7

Survival analysis: time to repartnering by sex and residence of children

Full size image

6.6.3 Event History

Model 3 (Table 6.3) provides the results regarding ‘who’ and ‘how’ people repartner, as it includes the characteristics associated with repartnering, as well as whether people repartner into a marriage or into a cohabitation. As mentioned previously, most respondents did not repartner in the time that they were observed, and of those that did repartner, most were likely to cohabit. However, there are different characteristics associated with who marries and who cohabits.

Table 6.3 Event history analysis of overall repartnering, marrying and cohabiting, logistic regression (HILDA 2001–2010)
Full size table

As evident in the survival analysis, women are less likely to repartner than men. They are much less likely to marry than men (odds = 0.30), but are not significantly less likely to cohabit.

Duration of previous relationship is associated with marriage but not cohabitation; those whose previous relationship was of 1–2 years being much more likely to marry than those with other relationship durations. The type of relationship that ended was also associated with marriage but not cohabitation, with those whose previous relationship was cohabitation having a significantly lower odds (0.17) of marrying than those who were previously married.

Age was negatively associated with both marrying and cohabiting. For marriage, those aged 50–59 (odds = 0.38) and 60+ (odds = 0.27) had a lower odds of marrying than those aged 30–39, while for those who moved into a cohabitation, those 40–49 (odds = 0.65), 50–59 (odds = 0.47), and 60+ (0.04) had a lower odds of cohabiting. There was no apparent effect of education.

Having resident children showed mixed results, and nothing consistent is evident. As the effect of having children was thought to be different for men and women, these models were run separately by sex (results in Table 6.4).Footnote 1 These results show that men who have resident children under age five are most likely to repartner compared to men with no resident children or older resident children. The effect was very large for cohabiting, with men who had a resident child less than five having an odds 13 times greater than men with no resident child. These results also showed that there was a lower propensity to repartner for women who had a resident child present. The odds are lower for both repartnering and cohabitation, with all the estimates for cohabitation being statistically significant. These results, for both men and women, are counter to the survival curves shown earlier. Disaggregating the results by more specific age of children, as well as controlling for other factors, shows the common finding that children have an effect on repartnering for women, but a more surprising result for men – that fathers with young children are more likely to repartner.

Table 6.4 Event history analysis of overall repartnering and cohabiting, by sex (HILDA 2001–2010)
Full size table

Patterns of repartnering by age also show some differences between men and women (Table 6.4). While from age 40 both men and women are less likely to repartner, the odds are substantially lower for women than for men in the 50–59 year age group.

Finally, religiosity was also associated with marriage, with those who stated that religion was ‘important’ having odds 3.17 times higher than those who stated that religion was ‘somewhat important’.

6.7 Conclusion

Repartnering in twenty-first century Australia differs for men and women, for different age groups, and by whether people have children or not. Other factors such as the duration of the previous relationship, and the type of relationship which ended, seem to have little effect on the likelihood of repartnering, at least when controlling for other factors. It is interesting that the descriptive survival curves showed that there are patterns depending on which type of relationship ended, but when controlling for other aspects of life, these patterns were not apparent.

In discussing previous research on repartnering, the concepts of ‘need, attractiveness and opportunity’ were raised (Ivanova et al. 2013; Skew et al. 2009). This is a useful frame in which to consider these results based on the first decade of the twenty-first century in Australia. Socio-economic factors such as poor living conditions were not considered here, but taking education as one measure of individual opportunity, there is no evidence that people with lower labour market attractiveness use repartnering as a way to boost economic position. As noted, earlier research found that this was a strategy that could be used to relieve pressure on female-headed households (Duncan and Hoffman 1985; Smock 1990; Weston and Khoo 1993).

Age may be a measure of attractiveness and opportunity, and its effect varies substantially by sex. Gendered notions of attractiveness tend to diminish for women at older ages, certainly more so than for men. Further, women are less likely to have opportunities for meeting partners due to a lower involvement in paid work, which is a prime place for meeting potential partners (de Graff and Kalmijn 2003).

The effect of having resident children is also highly gendered. Men who have young children were found to have a substantially higher probability of repartnering than other men. However, men are much less likely to have resident children of that age compared to women. As noted, women with resident children of any age are less likely to repartner than women without children. Further, men and women without children have a similar propensity to repartner. Children have a substantial effect on the different patterns of repartnering for men and women.

There are many other variables that are not accounted for in this analysis, mainly due to a lack of data. For example, in the USA, Sweeney (2002) found that those who initiated the relationship dissolution entered new relationships more quickly than non-initiators. This was a stronger result for older women, perhaps wanting to make sure of their future before leaving a relationship. This type of research is useful to help us better understand why and when people leave relationships and enter new ones.

An important finding from this research is that those who are repartnering are not necessarily remarrying. In the case of Australia, it is more likely that they will be cohabiting in their new relationship, whether their last relationship was a marriage or cohabitation. This is not unique to Australia, and is further evidence of the continuing change in modern family forms. Future research should consider cohabitation, as well as other relationship types such as those ‘living apart together’ as forms of repartnering and not focus solely on remarriage.