Copyright Sociological Research Online, 2001


Clare Holdsworth and Jane Elliott (2001) 'The Timing of Family Formation in Britain and Spain'
Sociological Research Online, vol. 6, no. 2, <>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 2001/3/15      Accepted: 2001/8/15      Published: 2001/8/31


Both fertility and marriage behaviour have changed considerably throughout Western Europe. While fertility has declined across the continent there has been an increase in the age of marriage, accompanied by an increase in cohabitation rates and marriage dissolution. These cross- national trends have been suggested to be indicative of a second demographic transition. Both Beck and Giddens have attempted to locate these changes in family formation within a wider context of social change associated with late modernity. However, in this paper we argue that in trying to provide a universal theory to understand the second demographic transition it is important not to overlook important cross- national differences. Previous comparative research has established that in the South young people tend to marry later, but have children earlier in partnerships. While in the North, marriage and leaving home occur earlier but the link between marriage and childbirth is weaker. In this paper we use data from the British National Child Development Study and the Spanish 1991 Sociodemographic Survey in order to compare the processes of family formation in Britain and Spain. The emphasis is on understanding the sequencing and timing of a) leaving home, b) forming a partnership, and c) giving birth to a first child in relation to each other rather than as independent events.

Britain; Family Formation; Life Course Transitions; Spain.


This paper compares the 'early' processes of family formation - from leaving home to birth of first child - for a cohort of British and Spanish young people aged in their early thirties in 1991. We focus on the sequencing and timing of these events in each country, rather than just looking at the events in isolation. The analysis explores the extent to which these life course events are linked among young people in both countries. By focusing on the relationship between family formation events, we address the ways in which the processes of family formation are characterised by a breakdown in the traditional life course sequencing in Europe - and the extent to which we can define a 'traditional' life course trajectory for the early stages of family formation - leaving home, forming a partnership and having children.


Recent convergence in fertility trends and partnership formation throughout Western European is well documented (Boh 1989; Roussel 1992; Lesthaeghe 1995; Blossfeld 1995; Kuijsten 1996). Following the peak of the post-War baby boom in the mid-1960s fertility has declined throughout the continent. As can be seen in Figure 1a, whereas in the 1960s the total fertility ratios of countries such as the UK, Spain, Italy and Sweden were between 2 and 3, by the mid 1990s these total fertility ratios had declined to be in the range 1.18 (Spain and Italy) to 1.81 (Denmark). The timing of the start of this decline varies, with Northern European countries, especially in Scandinavia, experiencing a decline earlier than countries, such as Italy and Spain, in the South. In particular, it can be seen from Figure 1a that whereas Spanish fertility did not start to decline significantly until the mid- 1970s, British fertility had already started to decline a decade earlier. However, the fertility decline has been much steeper in the South and has bottomed out at a lower level, particularly in Italy and Spain which currently return an all-time Western European fertility low (in 1998 Total Fertility was estimated to be 1.20 in Italy and 1.15 in Spain: ISTAT 2000; INE 2000).

Figure 1a: Total Fertility Ratios in Selected Western European Countries: 

Marriage behaviour has also changed considerably throughout Western Europe, with an increase in age at first marriage since the 1970s, an overall decline in the marriage rate and a concomitant increase in cohabitation and marriage dissolution (Kiernan 1999). As can be seen from Figure 1b, there is convergence in the age of marriage in the UK, Spain, France and Italy, while the Scandinavian countries of Sweden and Denmark show a distinctively older age of first marriage, some two years older than other Western European countries. As a consequence of these changes in marital behaviour more children are born outside of marriage (24 per cent of births in 1996 throughout the European Union, Eurostat 1999), though this varies considerably between North and South - 54 per cent of births in Sweden compared to only 11 per cent in Spain. When we consider these changes collectively we observe a breakdown of established trajectories in Northern Europe, such as leaving home to get married, have a child etc. Young people in the North are increasingly experiencing events 'out-of order' and are less likely to follow traditional pathways; as signified by the increase in extra-marital births and prolonged transitions out of the parental home. In contrast their counterparts in the South appear to be following more traditional life course trajectories, though in a delayed fashion (Iacovou 1998; Kuijsten 1999).

Figure 1b: Mean Age at First 
Marriage in Selected Western European Countries: 1960-1997

These cross-national trends in demographic rates (i.e. fertility and nuptiality) may be taken as indicative of a second demographic transition (van de Kaa 1987; Lesthaeghe 1995). Trends in the process of family formation observed throughout Western Europe and North America are evidence of fundamental shifts in attitudes towards family formation, influenced by factors such as women's education and employment (Becker, 1981; Blossfeld, 1995); technological change (particularly with reference to contraception); and an 'ideational shift' associated with the rise of individualisation, secularisation, loosening of moral sanctions, and weakening of institutional constraints - such as marriage (Lesthaeghe 1995).

Recent developments in sociological theory, which focus on individual agency and the construction of the self, have also reinvigorated sociological interest in the family (Beck 1992; Giddens, 1992; Beck and Beck-Gernsheim 1995). Giddens' notions of pure relationships and confluent love are based on a negotiated framework within which individuals make decisions about how to live together. Unlike romantic love, which Giddens associates with the idealised notion of falling in love 'for-ever', the idea of confluent love does not presume that falling in love is sufficient alone, rather it incorporates an 'active, contingent love', without any expectation of permanency (Giddens 1992, p. 61). In Risk Society Beck focuses on the ways in which social actors are engaged in a reflexive negotiation of their own individual biographies (Beck 1992). The demands of the labour market pressurise individuals to plan their own biographies around their work careers (assuming that they have both the qualifications and mobility to achieve this). This internalisation of labour market pressures applies to both men and women. In trying to resolve these tensions, men and women are increasingly negotiating not only the timing and sequencing of family formation events but also whether they take place and how. Partners have to reconcile decisions about living together, whether to marry or cohabit, and when and if to have children, with their labour market position. Both Beck and Giddens argue that the traditional certainties of older generations - that most young people would get married, establish a family home, and raise children - no longer hold. Yet this very uncertainty also drives a continual desire for security and closeness in individuals' relationships with others (Beck and Beck-Gernsheim 1995). Both Beck's and Giddens' approaches locate changes that are happening to the family within a wider context of social change associated with late modernity: individualisation, globalisation, high unemployment etc, and as such reflect a general process of 'de- traditionalisation'. This does not imply a social order in which traditions no longer exist, but one where traditions, particularly those affecting gender and the family, are routinely subject to interrogation (Giddens 1994). While both Giddens and Beck are writing from a distinctive Northern European perspective, their treatment of relationships and the family as part of the agenda of late modernity, transcends national boundaries and provides a theoretical framework within which the convergence of family formation may be addressed.

However, in striving to arrive at a universal theory of the second demographic transition, it is all too easy to overlook important cross-national differences. Lesthaeghe (1998) has recently cautioned against trying to reach a universal theory and argues for a multi-causal approach. This is particularly salient when considering recent trends in family formation observed in Southern Europe (Iacovou 1998). Despite converging trends in European fertility and nuptiality rates, important differences in family systems remain, particularly regarding patterns of adult child/parent coresidence, cohabitation, and the proportion of children born outside of marriage (Jurado Guerrero and Naldini 1997; Iacovou 1998). These differences illustrate the difficulties in identifying a European family system and in generalising an explanation for the second demographic transition. They also pose some intriguing puzzles if we are to assume that theories developed in a Northern European or North American setting can be directly transferred to Southern Europe. In particular, while the South may be converging with the North, in terms of fertility and marriage rates, there is less evidence of non-traditional behaviour, particularly cohabitation. One approach to understanding these different behaviour patterns is to evoke a 'logic of modernisation' thesis that assumes that Southern European countries will 'catch up' with their Northern neighbours. Hence it is only a matter of time before young people in the South embrace new ways of family life, such as leaving home prior to marriage and cohabitation.

In Spain after the death of Franco in 1975 and the conversion to democracy, a general consensus emerged among social scientists that a new European, family would develop in a new European and democratic Spain. Predictions were made regarding increases in lone parenthood, divorce, and cohabitation (Alberdi 1982; Iglesias de Ussel and Flaquer, 1993). Feminists were also optimistic about improvements in women's social and economic positions. Attitudinal changes to women's roles associated with the conversion to democracy and the establishment of an influential Women's Unit (Instituto de la Mujer) by the Socialist PSOE Government in the 1980s, have played an instrumental role in facilitating new opportunities for women and changes in family formation behaviour (Alberdi, Escario and Hainovich 1984; Threlfall 1996).

Yet, in more recent years, this logic has been challenged, as Jurado Guerrero and Naldini (1997) observe, family formation in the South remains stubbornly different, the institutionalisation of marriage remains strong, and young people choose to delay leaving home (a 'traditional' choice) rather than embrace alternatives to marriage. Moreover, the 'logic of modernisation' thesis prioritises individualisation or modernisation as the driving force of changing patterns of family formation and fertility decline, and ignores the social and cultural context of these demographic trends. Irwin (2000) has recently argued that the dominance of individualisation theses only offers a partial understanding of recent demographic changes. An alternative perspective is to locate these changes within changing reproductive regimes and 'through a focus on shifting patterns of inter-dependence and obligation amongst different social groups' (Irwin, 2000; 6.2). In Southern Europe patterns of generational inter-dependence reflect the familisitic structure of Mediterranean societies which is distinctive from Northern Europe (Dalla Zuanna, 2001, Micheli. 2000). Familism is associated with strong kinship ties (Reher, 1998) and a society where 'throughout their life most people seek their own happiness and at the same time that of their nuclear family and - if possible - their relatives' (Dalla Zuanna, 2001, p.139).[1]

Familism is not only characteristic of family structure and kin relations but is also closely bound up with the structure of Southern European economy. The proliferation of small family firms and the importance of family ties in securing entry into the labour market are closely inter-woven with familistic social relations (Dalla Zuanna, 2001). Familism is also closely associated with prevailing gender relations and welfare regimes in the South which during the recent economic crisis have had a profound impact on young people (Esping Andersen, 1999). For example, the reluctance to leave home and associated delay in marriage may be interpreted as evidence of the precarious nature of the youth employment market and its impact on young people's financial ability to form new households - either traditional or 'new forms'. Not only is unemployment higher in countries such as Italy and Spain than in Northern Europe, but the unemployed tend to be concentrated in younger age groups (in 1991 half of all unemployed in Spain were aged less than 30), and many unemployed have never worked and do not qualify for welfare support (Bettio and Villa 1998). Southern European commentators have identified these economic hardships as the main reason for delays in family formation, from leaving home through getting married and having children (Garrido and Requena, 1996; Vergés Escuín 1997; Fernández Cordón 1997). The consequences of limited economic opportunities for young Spanish couples have also influenced recent policy debates on the future of the Spanish family. The most striking examples of this are recent local initiatives to provide financial incentives for young Spanish people to marry and buy a house in their local village or town in an attempt to reverse recent population decline (El Pais, 1997). These initiatives accede to the general wisdom that it is financial predicaments that are the main impediment to family formation, rather than a rejection of 'traditional' values. The South also remains distinctive with respect to women's employment. Despite the optimism of Spanish feminists in the 1980s, women in the South have yet to reach parity with their Northern neighbours regarding employment opportunities. Given the priority ascribed to changing gender roles in the modernisation thesis, the status of women may prove to be the crucial determining factor influencing family and social change in the South (McDonald, 2000). Employment and motherhood remain irreconcilable for many women in the South. This can be understood as the result of a combination of number of different factors. Firstly the structure of the labour market and in particular the lack of part time employment, secondly the prevailing family culture and thirdly the reluctance by the state to aid women in combining these two roles (Bettio and Villa 1998; Esping-Andersen 1999). As Ruivo et al (1998) have argued, greater support from social institutions is required if women are to successfully reconcile paid work and family life. In particular, provision for child care services, flexible working time arrangements and maternity leave need to be extended. It is these that are lacking in Southern European countries such as Spain and Portugal.

The comparison of Northern and Southern Europe suggests that it is not enough to focus on individual life course events, but that it is important to investigate the relationships between different events, which is the subject of this paper. There are two themes which are important here, first, not only are trends in fertility influenced by trends in partnership formation, but, second, we need to re-establish how fertility decisions are intricately linked to those concerning partnership formation and leaving home. This approach may appear somewhat anachronistic from a Northern European perspective, where the breakdown of 'traditional' trajectories means that we no longer restrict our analyses of fertility to partnered couples, nor assume that leaving home is synchronous with marriage. As the traditional certainties of leaving home to get married and have children diminish, researchers have sought alternative approaches to understanding these events; in particular we no longer regard marriage as entry to 'risk' of childbirth. Yet approaches to understand linkages have been reinvigorated by use of the life course framework (Giele and Elder 1998; Liefbroer 1999). The emergence of life course analysis of events since the 1960s has emphasised the importance of both the timing and sequencing of life course events, their relationship with other transitions (which include transitions in education, employment as well as family) and the inter-familial dynamics of these transitions. Life course transitions are contingent on individuals' location in time (i.e. period factors such as labour market conditions), place, and their own past experiences. The contextual setting for life course transitions is based on external conditions and individual experiences. Individuals will also experience life course transitions with reference to normative expectations of 'appropriate behaviour'. These expectations may refer to the age limits in which an event occurs (e.g. whether one is too young to leave home or too old to be living with parents ) and the sequencing of events (e.g. if a couple get married before they have established careers and economic independence).

The analysis presented here examines the relationship between timing of different life course events, starting with leaving home. Our interest in looking at these transitions is influenced by the varying assumptions of the relationship between different life course events incorporated into theories of family change and fertility decline. From a Northern European perspective the emphasis is now towards the break-up of traditional trajectories and identifying non-standard pathways (Leifbroer 1999). This appears somewhat at odds with the Southern emphasis on delayed leaving home and entry into marriage as important determinants of fertility decline (Cabré Pla 1993). We propose that both approaches are appropriate as they reflect very different patterns of family formation in Northern and Southern Europe.

The increasing availability of data allows for detailed comparative studies to extend our understanding of the complexities of family formation throughout Europe. Iacovou's (1998) recent analysis of the first wave of the European Community Household Panel is one such comparative study. In this she identifies distinctive Northern and Southern patterns of family formation and hints at the possibility of a third distinctive Scandinavian type. In the South young people marry later, but have children earlier in partnerships, while in the North, marriage and leaving home occur earlier, though they are less likely to be associated events and the link between marriage and childbirth is weakened. In this paper we provide a more detailed breakdown of these two distinctive models of family formation based on retrospective data for two countries: Britain and Spain. Though, according to Iacovou (1998), these two countries are representative of the Northern and Southern patterns, both countries have distinctive demographic characteristics, particularly Britain (high teenage fertility and lone motherhood: Coleman 1999), hence we should be cautious about inferring general Northern and Southern patterns from these two specific cases. Rather the life course trajectories which we can identify for each country are representative of a Northern and Southern model for Britain and Spain respectively, though they may be modified by certain distinctive national characteristics, such as high teenage fertility and early home leaving for education in Britain. The purpose of this paper is to use these two case studies to illustrate the extent to which family formation processes vary between these two specific contexts.


This analysis is based on comparison of a cohort, in each country, whose members were aged in their early thirties in 1991. The National Child Development Survey (NCDS) is the second of three British Cohort Studies (City University Social Statistics Research Unit). It has followed all those born in the first week of March 1958 so that data has been collected from the same individuals at intervals through childhood and into adult life. To date, five subsequent sweeps of the cohort have been carried out. In 1965 when the cohort members were aged 7, and then again in 1969 (age 11); 1974 (age 16); 1981 (age 23); and 1991 (age 33). The data, which form the basis of the analyses reported in this paper, are predominantly drawn from the third sweep (1974 interview), the fourth sweep (1981 interview), and the fifth sweep (1991 interview) of the study. In addition to the face-to-face interviews carried out in 1991, cohort members were asked to fill in a self-completion questionnaire entitled "Your life since 1974". This collects retrospective event histories of partnerships (marital and cohabiting); childbearing; paid work; periods not in a job; and housing. The NCDS therefore represents a valuable data source with very detailed life histories available for a relatively large sample of men and women aged 33. The analysis is restricted to those present in 1991, giving 10,478 cases.

The Spanish data are taken from the 1991 Sociodemographic Survey (SDS). This is a retrospective survey of 160,000 individuals aged 10 and over which was conducted in 1991 (INE 1993). The survey included questions on current household composition and employment, and retrospective questions on family members (parents and siblings), partnerships, children, education, employment, housing, and migration. A cohort born between 1956 and 1960 has been extracted to compare with the NCDS cohort, giving 14,582 cases in total. Both datasets provide retrospective information on life course transitions: leaving home, partnerships (up to four in each survey), and childbirths. For NCDS we also have the option of corroborating data from the later sweep though comparison with the earlier sweeps. The structure of the questions on childbirths and partnerships are quite similar in both surveys, though more information is collected in NCDS. For example, the minimum duration for a partnership is one month in NCDS compared to one year for cohabiting partnerships in SDS, and the status of partnerships (married or cohabiting) is fixed in SDS, while is allowed to vary in NCDS. The implications of these differences are discussed below. Information about leaving home is more complex. For the Spanish respondents age at leaving home is taken from a question on when respondents finished living with their parents. In NCDS a direct question on leaving home is only asked in the 1981 sweep. For respondents leaving home after 1981 we have therefore used housing histories to identify moves out of the parental home.


In the first stage of the analysis we compare the proportion of men and women in each country who have moved through the three stages of family formation: leaving home, partnership formation, and childbearing. We then use simple event history analysis techniques, based on Kaplan Meier estimates, to identify the main differences in family formation behaviour and, if possible, normative patterns for Britain and Spain. Finally these normative patterns are formalised for women using a discrete-time event history analysis model.

Descriptive Statistics

Leaving Home

Women from the cohort born in the late 1950s left home earlier than their male counterparts in both Great Britain and Spain (Table 1). The median ages at leaving home also underline the fact that, among this cohort, young people left home at younger ages in Great Britain than in Spain. Indeed, once we focus on the proportion of the cohort who had left home by their early thirties, it can be seen that whereas practically all women in Britain had left home, over 10 per cent of Spanish women had never left home. For British men only one in twenty had never left, compared to 17.5 per cent of Spanish men.

Table 1: Summary Statistics of Life Course Events

Life Course Event
Great Britain
% Cohort Ever Left Home
Mediana Age of Leaving Home26.023.721.720.0
% Cohort with at least one Partnership81.987.888.693.4
Median Age at First Partnership26.523.924.121.8
% Cohort with at least one child71.280.366.476.3
Median Age at Birth of First Child29.426.529.827.0
% Cohort with One Child Onlyb43.834.027.821.7
% Cohort in Childless Partnershipsc13.19.625.419.4

a Median age is the age at which half of the cohort have either left home, formed a partnership or had a child.
b Percentage expressed for those with children only.
c Percentage expressed for those in partnerships only.

Base: All cohort members living with parents at age 14. All analyses exclude British respondents whose first partnership was a cohabiting partnership of less than one year.

Source: Spain - 1991 Sociodemographic Survey ; Britain - National Child Development Survey.

Partnership Formation

In addition to leaving home at earlier ages, young people in Britain were also found to have formed partnerships at earlier ages than their Spanish counterparts. It can be seen from Table 1 that a larger proportion of British men and women had had at least one partner by 1991, and that the median age of first partnership was just over two years younger for men and women in Britain than in Spain. Moreover, a larger proportion of British men and women had formed cohabiting partnerships. However, it is difficult to compare these data directly due to differences in survey design. The SDS did not record change in partnership status, that is, while information is collected on cohabiting partnerships, partnerships are defined as either marital or cohabiting. We would expect a large number of cohabiting partners to marry (as is the case in Britain), hence it is likely that the number of cohabiting partnerships recorded by the SDS is underestimated and it is not really possible to compare partnership status in the two countries.

We also found a greater frequency of serial monogamy in Britain. Though each survey uses a different minimum period of co-residence for partnerships, one month in NCDS compared to one year for cohabiting partnerships in the SDS (there is no minimum period for marital partnerships in Spain), and this may affect the total number of partnerships. We controlled for this by selecting all cohabiting partnerships in Britain that lasted for longer than one year (cohabiting partnerships of less than one year duration account for five per cent of all partnerships in NCDS). Yet, we still found that 18.1 per cent of men and 20.6 per cent of women in the British cohort reported serial partnerships, compared to only 1.6 per cent and 1.7 per cent of Spanish men and women, respectively.

Birth of First Child

Despite the differences in median age at leaving home and first partnership between the two countries, median ages at the birth of first child are similar for both Spanish and British women (around 27 years) and men (around 29 years). It is also interesting to note that despite the smaller proportion of the Spanish cohort who had left home and formed a partnership, a slightly larger proportion had become parents by 1991 (Table 1). The corollary of this is that childless partnerships were more common among the British Cohort (Table 1).

Linking Life Course Events

A greater proportion of the cohort of Spanish men and women had not left home nor formed a partnership by their early thirties compared to their British counterparts, though in both countries this behaviour was more common for men than women. Of those young people who have left home - the modal behaviour in both countries is to form a partnership at the same time (the 'traditional' behaviour of leaving home to get married) though this pattern is more common in Spain. We find that in Spain 70.8 per cent of men and 75.1 per cent of women, who left home, formed a partnership at the same time, compared to 44.4 per cent and 52.4 per cent of men and women in Britain.

One 'traditional' Spanish pathway is to get married before leaving home, particularly among women, and we find 9.5 per cent of the Spanish female cohort chose this route and 7.5 per cent of men. However, this pattern also occurred among British young people and was at a similar magnitude - 7.7 per cent of both men and women. The main difference between the two countries was that by the time of the survey about half of the Spanish group remained living with a partner in the parental home (and may never leave) while the majority of British respondents had left. At the other extreme, one in five of the British male cohort had left home five years before forming a partnership, compared to only seven per cent of Spanish men.

Turning to the timing of partnership formation and birth of first child, of those who had become parents the modal behaviour, in both countries, is to have a child one to two years after forming a partnership. We find that 54.9 per cent and 56.3 per cent of Spanish fathers and mothers and 29.8 per cent and 30.1 per cent of British fathers and mothers had their first child one to two years after forming a partnership.

When we put all three life course events together we find that in both countries the modal behaviour is leaving home to form a partnership in the same year, followed by having a child one to two years later. This trajectory is more common in Spain, accounting for around one-third of the Spanish cohort. Whereas 29.5 per cent and 34.5 per cent of the Spanish cohort left home to get married and had their first child one to two years later, this compares to 12.0 per cent and 14.9 per cent of the British cohort (men and women respectively).

Kaplan Meier Analysis

One of the problems of using data such as that provided by NCDS and the SDS is that most respondents included in the analysis have not completed their family formation and their biographies are censored in 1991 - i.e. respondents who had not left home, formed a partnership or had a first child by 1991, but may go on to do so at a later date. As the process of family formation starts at older ages in Spain (beginning with leaving home) it is problematic to compare the two countries directly, without controlling for the number of censored cases. A straightforward way to do this is to use Kaplan Meier survival estimates (see Tuma and Huinink 1990 for application of the technique to birth intervals). These give estimates of mean survival times and first quartile, median and third quartile (Table 2) - i.e, the time interval by which 25 per cent, 50 per cent and 75 per cent of subjects had left home, censored cases contribute to the number of observed person years. An additional advantage of using Kaplan Meier estimate, over the simple percentages reported in the previous section, is that they allow us to quantify the variability in behaviour in the two countries. A wider inter- quartile range can be interpreted as greater heterogeneity in the timing of family formation transitions for a particular cohort.[2]

Table 2: Kaplan-Meier Estimates.
Survival Time from Age 14 to Leaving Home
Base: All Respondents at Home at Age 14
1st Quartile
3rd Quartile
N of Cases
N of cases censored
Survival Time from Leaving Home to First Partnership
Base: All Respondents who have Left Home
1st Quartile
3rd Quartile
N of Casesa
N of cases censored
Survival Time from First Partnership to First Child
Base: All Respondents who Form a Partnership
1st Quartile
3rd Quartile
N of Casesb
N of cases censoredc

a Excludes those who form a partnership prior to leaving home.
b Includes those who form a partnership but do not leave home.
c Includes those whose first partnerships end before birth of first child.

All analyses exclude British respondents whose first partnership was a cohabiting partnership of less than one year.

Source: Spain -1991 Sociodemographic Survey ; Britain - National Child Development Survey.

The Kaplan Meier estimates corroborate the evidence presented above, and illustrate distinctive British and Spanish patterns. First, leaving home occurs at older ages for the Spanish cohort, and is also distributed over a wider age group - as indicated by the first and third quartiles. In contrast the younger median age at leaving home in Britain is associated with greater homogeneity in the timing of leaving home. Though it should be noted that this homogeneity in timing is associated with greater heterogeneity in terms of reasons for leaving compared to Spain (see Holdsworth 2000).

Second, when we turn to look at the transition from leaving home to partnership we find greater homogeneity in timing among the Spanish cohort. Spanish young people form a first partnership more quickly after leaving home than their British counterparts, with the majority leaving home to form a partnership (hence the zero estimates, as the two events occur in the same year). In Britain we find a longer mean survival time, and in particular a later third quartile. This difference is due to the two distinctive patterns of leaving home in Britain. The majority of young people in this cohort leave home to form a partnership, though a substantial minority leaves home for other reasons, such as to go to university, work elsewhere or live independently, and subsequently form partnerships a few years after leaving home.

For the third transition - that is from partnership to first child - we find shorter median survival times for Spanish young people, and the smaller inter- quartile range indicates greater homogeneity in this transition. Conversely in Britain, the median survival time to first child is much longer (approximately twice as long as for Spanish young people) and the wider inter-quartile range suggests greater heterogeneity in the timing of starting a family.

In summary, therefore, the data suggest that the normative Spanish pattern is to delay leaving home (and there is a relatively heterogeneous age of leaving home), but once young people have left they form a partnership and start a family relatively quickly. In contrast to this, British men and women tend to leave home at earlier ages but then exhibit greater heterogeneity in the timing of forming partnerships and having a first child.

Modelling first births

In the final stage of the analysis we focus in more detail on the relationship between the timing of first partnership and first birth. The Kaplan Meier estimates suggest that these two events are more closely linked in Spain than in Britain. However, we might expect this relationship to differ by age at first partnership. For example is the time difference between first partnership and first child shorter for those who either marry or form a cohabiting partnership at very young ages (maybe because of an unplanned pregnancy) or at older ages? To investigate this relationship we have modelled the probability of having a first child for women from each country for different partnership cohorts. We fit a discrete-time event history model (Allison 1984). To generate the dataset for the model, we select all women who have started a partnership. For simplicity we have selected women in first partnerships only. Each woman contributes a number of years - starting with the year she enters her first partnership and ending either at the birth of her first child, the end of the first partnership or by being censored in 1991 (i.e. she has not had a child by the time of the survey in 1991, but was still in her first partnership), whichever event occurs earliest. In total 30 per cent of the British cohort were censored (13 per cent by their being childless at the time of the survey and 17 per cent by partnership dissolution) and 10 per cent of the Spanish cohort (9 per cent by being childless and 1 per cent by partnership dissolution). The dependent variable is whether the respondent has a first birth or not. This is coded one in the year that a woman's first child is born, and zero in all other years. In the first model we include three variables:
  1. Time in Partnership - this variable gives the survival time for the model. As the probability of having a child does not linearly increase with time in partnership, both time and time squared are fitted in the final model.
  2. Age at Start of Partnership - a categorical variable divided into the following age groups:16-18, 19-21, 22-24, 25-27, 28-30.
  3. Dummy variable to capture year entered partnership. This variable is included in addition to time in partnership, as year started (year zero of the partnership) does not necessarily equate to a full year's duration.
It should be noted that as events are recorded in years we cannot distinguish premarital conceptions in our analysis.

We test to see whether the hazard is proportional by fitting interaction terms between age at start of partnership and time in partnership.[3] These interaction terms proved to be significant, demonstrating that the probability of having a first birth by time in partnership varies for different partnership cohorts.

The parameter estimates for the models are given in Table 3, though the easiest way to examine the results is by plotting the estimated probabilities (Figures 2a and 2b). A clear pattern emerges for the relationship between age at partnership and time of partnership. In Spain for the youngest three partnership cohorts the probability of having a child peaks in the first full year of partnership and then declines steeply. For the older partnership cohorts there is a more even distribution of first births during the first three years, although there is a steep decline after this. The probability of having a child in the first year does, though, decline with older ages at first partnership. These probabilities for Spanish women may be compared with similar analysis undertaken by Castro Martín (1992) using an older dataset. This analysis illustrated that the relationship between first partnership and having a first birth has weakened for women marrying in the 1980s compared to women marrying in the 1960s and 1970s.

Figure 2a: Estimated probability 
of having a first birth by time in first partnership and age at start of first 
partnership: Spanish women

Figure 2b: Estimated 
probability of having a first birth by time in first partnership and age at 
start of first partnership: British women

Table 3: Parameter Estimates for Model of First Births within First Partnerships.
All women in first partnerships; Numbers refer to parameter estimates, standard errors in parentheses.
VariableSpain ßBritain ß

Age Start Partnership: Base category 19- 21

16-180.43 (0.10)0.87 (0.12)
22-24-0.46 (0.17)-0.44 (0.12)
25-27-0.72 (0.10)-0.49 (0.19)
28-30-0.85 (0.17)-0.56 (0.33)
31-330.31 (0.24)-0.14 (0.52)
34-36-0.01 (1.06)NA
Time in partnership (years)-0.13 (0.04)0.28 (0.04)
Time in partnership squared-0.01 (0.00)-0.03 (0.00)

First year of partnership

-1.76 (0.07)-0.51 (0.09)

Mother's age at first birth (time varying)

0.00 (0.00)-0.01 (0.01)

Higher Qualification: Base no high qual

University-level qualification

-0.43 (0.05)-0.57 (0.07)

Interaction Terms:
Age start partnership* time in partnership

16-18*time-0.18 (0.07)-0.42 (0.06)
22-24*time0.21 (0.05)0.28 (0.06)
25-27*time0.43 (0.07)0.49 (0.13)
28-30*time0.57 (0.16)1.29 (0.33)
31-33*time-1.06 (0.39)0.95 (1.11)
34-36*time1.17 (1.63)NA

Age start partnership* time in partnership sqda

16-18*time20.02 (0.00)0.02 (0.01)
22-24*time20.02 (0.01)-0.03 (0.01)
25-27*time2-0.05 (0.01)-0.07 (0.02)
28-30*time2-0.10 (0.03)-0.33 (0.08)
31-33*time20.22 (0.12)-0.57 (0.54)


-0.21 (0.07)-2.00 (0.08)
a The final interaction of age group 34-36*time2 for Spain is aliased

In Britain we see more variation by age at first partnership. The youngest cohort appears to follow the same pattern as their Spanish counterparts, with a peaked probability distribution - though the probability of having a first child is much lower than is observed in Spain. For the middle cohorts (19-21, 22-24 and 25-27 years at first partnership) the distribution is more dispersed; women in Britain are more likely to have a child in the fourth or fifth years than are women in Spain, and are less likely to have a child in their first year. The estimated probability of having a first birth for women forming a partnership at ages 28 to 30, peaks in the third year and is similar to that observed for the equivalent Spanish cohort. This suggests that British women who form a partnership at older ages accelerate the timing of first births in comparison with younger partnerships cohorts, while older Spanish women delay their first birth slightly. Overall, however, the estimated probabilities illustrate greater heterogeneity in the timing of first births within partnerships for British women, compared to Spanish women.

In the final stage of the model we examined other influences on the probability of having a first birth within first partnership. We identified a number of factors of interest: employment history, education and age of respondent's mother on her first birth. Unfortunately, due to limitations with the Spanish dataset, it was only possible to fit a dummy for possession of higher qualification (degree or equivalent) and a time- varying variable for the difference between respondents' age and mothers' age at first birth.[4] This variable captures the potential influence of a mother's age at starting a family on the timing of her daughter's childbearing. It is coded zero when the respondent is the same age as her mother at her first birth, one when a year older or younger, etc. Of the two additional variables fitted, only possession of a higher qualification proved to be significant (Table 3). We then extended the model by fitting interaction terms between higher qualification, age at first partnership, and time in partnership. These interaction terms were not significant in the British model, but revealed some interesting differences for the Spanish cohort. We plot the estimated probabilities of having a first birth by level of qualification for the modal partnership cohort (22-24) using the interaction terms for the Spanish cohort (Figure 2c). The results of this model suggest that the impact of having a higher qualification in Britain is to lower the probability of having a first birth regardless of time in partnership. However for the Spanish women, those with higher qualifications appear to be delaying having a first birth within partnerships.

Figure 2c: Estimated probability 
of having a first birth by time in partnership and level of qualification: women 
in Great Britain and Spain, aged 22-24 at  first partnership

Life Course Trajectories

The analysis presented confirms previous research on the distinctive nature of family formation in Southern Europe (Jurado Guerrero and Naldini 1997;Iacovou 1998). The findings echo Kuijsten's (1999, p. 99) description of young people either choosing 'new options' of family formation or embracing the opportunity of an '"old" script, but perhaps with deliberate postponement'.

However, while there is evidence of more British young people embracing alternative family trajectories (leaving home prior to family formation, higher prevalence of cohabitation, serial monogamy) if we focus our attention on those young people who reject traditional patterns we risk ignoring significant variation in behaviour among those whom, at first glance, are following traditional trajectories. To illustrate the variety of trajectories, both 'traditional' and 'non-traditional' among the British and Spanish cohorts we have selected those that account for at least five per cent (rounded-up) of all respondents, treating men and women separately (Figures 3a to 3b). These trajectories are based on the relationships between leaving home, partnership formation, and birth of first child, identifying those where the events occur in the same year or in a time interval of within one to two, three to four, or greater than five years (to simplify these trajectories we do not differentiate by age at leaving home, though we indicate those trajectories associated with early leaving). The modal trajectory for all four groups is to leave home and form a partnership at the same time and have a first child one to two years later, though this accounts for a far greater proportion of Spanish men and women. Besides the option of remaining in the family home, all the trajectories highlighted for the Spanish cohort involve leaving home and forming a partnership in the same year.

The main differentiating factor is the timing of having a child. The six pathways highlighted in Figures 3a and 3b account for 70 per cent of Spanish men and women and this high coverage is associated with a high proportion of young people following the modal trajectory. The highlighted trajectories for the British cohort account for fewer respondents than observed in Spain, less than 60 per cent of men and women, despite the fact that for British men eight trajectories are identified with a frequency at or above 5 per cent. By contrast with Spain there are trajectories where leaving home and partnership are distinct events, particularly for British men. Moreover we find a more even distribution of respondents classified by the trajectories which involve leaving home for partnership and having a child either one to two years, three to four years, or five years later. This finding was confirmed for British women from the model of timing of first births within partnerships. The model also demonstrated that we would expect women who have a child within one year of forming a partnership to have entered partnerships at younger ages than women who delay first births. The trajectories highlighted in Figures 3c to 3d confirm the greater heterogeneity in the timing of events for British young people following 'normative' routes (i.e. leaving home to get married followed by a child), but moreover, that there is greater diversity in the different types of trajectories observed, as these principal pathways account for just over half of the British cohort. Some of these alternative trajectories which might account for the remaining respondents are illustrated in Figure 4, for example young people in Britain may leave home early, but return at a later date, sometimes of as a result of partnership dissolution. We might also expect a greater frequency of serial monogamy. Finally as this cohort is observed at age 33, many respondents may have a child at much older ages after a long period of childless partnership(s).


While young people in both countries continue to follow traditional life course trajectories, the diversity of life course transitions observed for the British cohort, particularly regarding the timing of first births, coincides with Beck's (1992) description of the breakdown of traditional certainties and the emergence of individualised biographies. Conversely, young people in Spain are postponing leaving the parental home to form an independent household but this in turn compresses the time between living together as a couple and starting to have children. Explanations for recent delays in fertility may therefore be linked with familistic characteristics of Southern European societies which are associated with delays throughout youth to adult transitions, starting with leaving home (Dalla Zuanna, 2001). The postponement of leaving home, getting married and starting a family among both men and women in the South is associated with strong inter-generational ties, a high level of consumption within the family home and a general reluctance on the part of both parents and adult children for children to leave home (the concept of 'chicos cómodos' literally 'comfortable boys' who remain in the parental home until their early thirties). This is reinforced by prevailing opportunity structures, including the importance of both men's and women's economic stability (Bettio and Villa 1998), residual welfare state (Esping-Andersen 1999), and an inaccessible housing market (Miret Gamundi, 1997).

We may speculate about the future of family formation in Spain and Southern Europe in general. Are the Southern states in a period of flux and, given time, will the convergence of European labour markets and, more contentiously, welfare states, bring about the social changes identified in Northern Europe and the eventual rise of individualisation and breakdown of traditional life course trajectories? Predictions of convergence to a Northern European model of family formation are not new in Spain, though many of the predications made in the early 1980s have yet to be realised. There is though important evidence of changing behaviour in the South. For example Castro Martín (1992) identifies a shift towards delayed first births for younger generations, though, as demonstrated in this analysis, when compared with other countries modern Spaniards appear very 'traditional' in their behaviour. Moreover, in the analysis presented here we find evidence that more highly qualified women are delaying first births.

It would be naive to assume that family practices and mores have remained unchanged in Spain throughout the dramatic social changes that have occurred since 1975. Modern Spaniards report increasingly liberal attitudes towards divorce, cohabitation, and lone parenthood (Alberdi 1999). Moreover, current difficulties in linking the demands of family and employment for men and especially women that are identified in Northern Europe are just as relevant in the South. Differences in the timing of first birth for more highly qualified Spanish women in particular suggest support for the modernity thesis, which may have a trickle down effect. We might therefore expect that, as the analysis presented here refers to transitions made during the 1980s, these transitions would have been modified during the 1990s. The aggregate data available suggest this not to be the case, with the underlying trend towards even later ages of leaving home, delayed marriage, and low fertility. (Fernández Cordón 1997). As this trend continues we need a greater understanding of young people's motivation to delay family formation on the one hand, and to reject alternative trajectories on the other. Do the opportunity structures that restrict leaving home to get married and have a family, such as high youth unemployment and an inaccessible housing market, also impede alternative living arrangements? Or is both the delay in leaving home to get married and rejection of alternative forms (e.g. leaving home to live alone, cohabitation) a choice influenced by prevailing familistic norms? The question of changing and diverse family practices can be addressed with new family and household datasets available for Europe, such as the ECHP (Iacovou 1998) and the Family and Fertility Surveys (Corijn and Klijzing, forthcoming). Yet we also need greater understanding of how young people are making these choices in the light of structural opportunities and cultural norms.

The irrepressible momentum of modernity and its consequences for family formation have tended to dominate studies of family formation in recent years, with the emphasis on convergence and universal theories of social change (Roussel 1992). This approach has now been questioned as the greater availability of comparative data illustrates the diversity of family forms throughout Europe (Boh 1989; Reher 1998). Social scientists are faced with an intriguing challenge; that is to reconcile this diversity with theories of modernity, such as those developed by Beck and Giddens, in which the apparent break from traditional norms and expectations of family life are treated as central to the process of social change itself. The situation in Southern Europe is particularly intriguing and one that is extremely difficult to predict. The timing and intensity of family formation in the South will change in the future, though it is not possible to predict an inevitable cultural transition of Southern family formation patterns to a Northern European model, hence the direction of this change remains unknown.


1 There are important regional variations in family structure throughout Spain. In particular the North is more strongly associated with stem- family formation in contrast to nuclear family formation in the South (Holdsworth, 1998). As Micheli (2000) argues it is difficult to generalise a uniform Southern European family culture. In this paper we focus on features of Spanish families which are consistent throughout all regions: older ages of leaving home, low incidence of cohabitation and divorce and low fertility, though the latter is more characteristic of Northern Spain.

22 One drawback of using these datasets for this technique is that the Spanish data are only available in completed years, rather than completed months, and the time periods for some of the transitions considered are, on average, less than one year hence we have a few zero estimates.

3The interaction terms are fitted to investigate whether the effect of time in partnership is constant for all partnership cohorts. Significant interaction terms indicate that the impact of time in partnership on the odds of having a first birth varies for different partnership cohorts.

4Data collected on employment history in the SDS were not detailed enough to provide any insight on the relationship between maternity and employment.


The research for this paper was supported by the Economic and Social Research Council Grants R00023658 and R000236549 whose support is gratefully acknowledged. We would also like to thank Peter Sheppard from the Centre of Longitudinal Studies, Institute of Education for help and advice in analysis of the NCDS; the ESRC Data Archive for providing the NCDS data; and Instituto Nacional to Estadística for permission to analyse the Sociodemographic Survey. We would also like to thank Professor R.I. Woods for comments on the original draft of this paper.


ALBERDI, Inés (1982) "Un nuevo modelo de familia" Papers: Revista de Sociología, Vol 18, pp. 87-112.

ALBERDI, Inés. (1999) La Nueva Familia Española. Madrid: Taurus.

ALBERDI, Inés, ESCARIO, Pilar and HAINOVICH, Perla (1984) "Actitudes de las mujeres hacia el cambio familiar" Revista Española de Investigaciones Sociológicas. Vol 27, pp. 41-59.

ALLISON, Paul D. (1984) Event History Analysis: Regression for Longitudinal Event Data Beverly Hills: Sage.

BECK, Ulrich. (1992) Risk Society: Towards a New Modernity. London: Sage.

BECK, Ulrich. and BECK-GERNSHEIM, Elisabeth (1995) The Normal Chaos of Love. London: Polity Press.

BECKER, Gary S. (1981) A Treatise on the Family. Cambridge MA: Harvard Univeristy Press.

BETTIO, Francesca and VILLA, Paola (1998) "A Mediterranean perspective on the breakdown of the relationship between participation and fertility" Cambridge Journal of Economics, Vol 22, pp. 137-172.

BLOSSFELD, Hans-Peter. (1995) "Changes in the process of family formation", in Hans-Peter Blossfeld (Editor) The New Role of Women: Family Formation in Modern Societies. Oxford: Westview Press, pp. 3-32.

BOH, Katja (1989) "European family life patterns - a reappraisal" in Katya Boh et al (Editors) Changing Patterns of European Family Life: A Comparative Analysis of 14 European countries. London: Routledge, pp. 265-298.

CABRÉ PLA, Anna. (1993) "Volverán tórtolos y cigüeñas" in Luis Garrido Medina and Enrique Gil Calvo (Editors) Estrategias Familiares. Madrid: Alianza Editorial pp. 113-131.

CASTRO MARTÍN, Teresa. (1992) "Delayed childbearing in contemporary Spain: trends and differentials" European Journal of Population, Vol 8, pp. 217- 246.

CITY UNIVERSITY SOCIAL STATISTICS RESEARCH UNIT, National Child Development Study Composite File including selected perinatal data and sweeps one to five, 1958 - 1991 [computer file]. National Birthday Trust Fund, National Children's Bureau, City University Social Statistics research Unit [original data producers]. Colchester, Essex: The Data Archive [distributor], 21 June 1994. SN: 3148.

COLEMAN, David (1999) "Britain's place in Europe's population" in Susan McRae (ed) Changing Britain: Families and Households in the 1990s. Oxford: OUP pp. 34-67.

CORIJN, M. and KLIJZING, E. (eds) (forthcoming) Transitions to adulthood in Europe. Dordrecht: Kluwer Academic Publishers.

COUNCIL OF EUROPE (1998) Recent demographic developments in Europe. Council of Europe Publishing: Strasbourg.

DALLA ZUANNA, Gianpiero (2001) 'The Banquet of Aeolus: A familistic interpretation of Italy's lowest low fertility' Demographic Research, 4, no 5. <http://www.demographic->

EL PAIS (1997) "Un pueblo vicaíno brindó 100,000 pesetas por tener niños y espera que nazcan siete", 2nd February 1997.

ESPING-ANDERSEN, Gøsta (1999) Social Foundations of Postindustrial Economies. Oxford: Oxford University Press.

EUROSTAT 1999, "European lifestyles as the new millennium dawns" Press Release 105/99, Luxembourg: Eurostat Press Office. < product/EN?catalogue=Eurostat&product=1-14101999-EN-BP&type=pdf>

FERNÁNDEZ CORDON, Juan Antonio (1997) "Youth residential independence and autonomy: a comparative study" Journal of Family Issues, Vol 6, pp. 572- 5.

GARRIDO, Luis and REQUENA, Miguel (1996) La Emancipación de los Jóvenes en España. Madrid: Instituto de la Juventud.

GIDDENS, Anthony (1992) The Transformation of Intimacy. Cambridge: Polity Press.

GIDDENS, Anthony (1994) "Living in a post-traditional society" in Ulrich Beck, Anthony Giddens and Scott Lash Reflexive Modernisation. Cambridge: Polity Press, pp. 56-109.

GIELE, Janet Z. and ELDER jnr, Glen H. (1998) (Editors) Methods of Life Course Research: Qualitative and Quantitative Approaches. London: Sage.

JURADO GUERRERO, Teresa and NALDINI, Manuela (1997) "Is the South so different? Italian and Spanish families in comparative perspective" South European Society and Politics, Vol 1, pp. 42-66.

HOLDSWORTH, C. (1998) 'Leaving home in Spain: a Regional Analysis' International Journal of Population Studies, 4, 341-360.

HOLDSWORTH, Clare (2000) "Leaving home in Britain and Spain" European Sociological Review, Vol 16, pp. 201-222.

IACOVOU, Maria (1998) "Young people in Europe: Two models of household formation" Institute for Social and Economic Research Working Paper 98-13. Colchester: University of Essex. <>

IGLESIAIS DE USSEL, Julio and FLAQUER, Luís (1993) "Familia y analisis sociologico: el caso de España" Revista Española de Investigaciones Sociológicas, Vol 61, pp. 57-75.

INE (1993) Encuesta Sociodemográfica 1991. Madrid: Instituto Nacional de Estadística.

INE (2000) España en cifras 1999. Madrid: Instituto Nacional de Estadística. < /espcifes/pobl00.pdf >

IRWIN, Sarah (2000) 'Reproductive Regimes: Changing Relations of Inter-dependence and Fertility Change' Sociological Research Online, 5, no 1. <http://www.socresonline>

ISTAT (2000) Italy in Figures. Rome: Istituto Nazionale de Statistica. <>

KIERNAN, Kathleen (1999) "Child bearing outside marriage" Population Trends, Winter 98, pp. 11-20.

KUIJSTEN, Anton C. (1996) "Changing family patterns in Europe: A case study of divergence?" European Journal of Population, Vol 12, pp. 115-143.

KUIJSTEN, Anton C. (1999) "Households, families and kin networks" in Leo J.G. van Wissen and Pearl A. Dykstra. (Editors) Population Issues: An Interdisciplinary Focus. New York: Kluwer Academic/Plenum Publishers.

LEIFBROER, Aart (1999) "From youth to adulthood: understanding changing patterns of family formation from a life course perspective" in in Leo J.G. van Wissen and Pearl A. Dykstra. (Editors) Population Issues: An Interdisciplinary Focus. New York: Kluwer Academic/Plenum Publishers.

LESTHAEGHE, Ron (1995) "The second demographic transition in western countries: an interpretation" in Karen Oppenheim Mason and An-Magritt Jensen (Editors) Gender and Family Change in Industrialised Countries. Oxford: Clarendon Press, pp. 17- 62.

LESTHAEGHE, Ron (1998) "On theory development: applications to the study of family formation" Population and Development Review, Vol 24, pp. 1-14.

MCDONALD, P. (2000) 'Gender Equity in theories of fertility transition' Population Development Review, Vol 26, pp. 427-439.

MICHELI, Giuseppe A. (2000) 'Kinship, Family and Social Network: The anthropological embedment of fertility change in Southern Europe', Demographic Research, 3, no 13.< http://www.demographic- 13.pdf>

MIRET GAMUNDI, Pau (1997) "Nuptiality patterns in Spain in the eighties" Genus, Vol 53, pp. 183-198.

REHER, David Sven (1998) "Families ties in Western Europe: Persistent contrasts" Population Development Review, Vol 24, pp. 203-234.

ROUSSEL, Louis (1992) "La famille en Europe occidentale: divergences et convergences" Population, Vol 47, pp. 133-152.

RUIVO, M., Do Pilar Gonzalez, M., and Varejao, J.M. (1998) 'Why is part-time work so low in Portugal and Spain?' in O'Reilly, J. and Fagan, C. (Editors). Part-Time Prospects: An International Comparison of Part-time Work in Europe, North America and the Pacific Rim. London: Routledge, pp199-213.

THRELFALL, Monica (1996) "Feminist politics and social change in Spain" in Monica Threlfall (Editor) Mapping the Women's Movement. London: Verso, pp. 115-151.

TUMA, Nancy Brandon and HUININK, Johannes (1990) "Post- war fertility patterns in the Federal Republic of Germany", in Karl Ulrich Mayer and Nancy Brandon Tuma. (Editors) Event History Analysis in Life Course Research. Madison: The University of Wisconsin Press, pp. 146-169.

VAN DE KAA, Dirk (1987) "Europe's second demographic transition" Population Bulletin Volumn 42, pp. 1-57; Washington D.C., Population Reference Bureau.

VERGÉS ESCUIN, R. (ed) (1997) La edad de emancipacion de los jovenes. Barcelona: Centre de Cultura Contemporània.

Copyright Sociological Research Online, 2001