SLIDE 1
Being born to a single mother in France: trajectories of father’s involvement over the first year
- f life
Introduction The de-standardization of family life across Western - - PDF document
Introduction The de-standardization of family life across Western - - PDF document
Being born to a single mother in France: trajectories of fathers involvement over the first year of life Marieke Heers (University of Lausanne), Ariane Pailh (Ined), Lidia Panico (Ined) Draft prepared for the IUSSP 2017 conference, do not
SLIDE 2
SLIDE 3
children's lives are important in this respect (Marsiglio et al., 2000; Aldous & Mulligan; 2002), but less explored. Across two primary dimensions of direct father involvement (frequency of positive engagement activities and aspects of parenting quality) and five dimensions of children’s early learning (representing social and cognitive domains), findings revealed small but significant associations (REF?). Understanding the determinants of father involvement, particularly when he no longer or has never co-resided with his child is therefore important. The available literature suggests that fathers who co- reside with their children and fathers who are older are more involved with their children (Castillo, Welch and Sarver, 2011; McWayne et al, 2013). Fathers’ working hours are not strongly related to father involvement (McGill, 2014), and income was also less important than residential status and racial identity of fathers (McWayne et al. 2013). Taking four or more weeks of paternity leave did not predict higher levels of involvement in an Australian study, and any leave taking was associated with a higher chance of solo father care at the weekend but not during the week (Hosking et al. 2010). His educational levels however do correlate with fathers’ taking part in parenting activities, particularly activities that are important for child development (for example, for children aged 3 to 5, fathers’ education was correlated to his engagement in reading and other cognitively stimulating activities, Grarcia, 2014). Furthermore, determinants of father involvement appear to not only be associated with fathers’ characteristics, but also how these characteristics interact with the couple and family contexts. For example, father–child attachment was more closely linked to couple and family contextual variables, while mother–child attachment was more closely linked to mother involvement (Coyl-Shepherd and Newland, 2013). The rise of men’s involvement with their children, in addition to their traditional role as financial provider, has given rise to so-called “new fathers”, who are expected to be more equal partners in parenting, and performing both interactive and physical caregiving. The roles of provider and involved father may conflict (McGill 2014). In France, the double-role of “new fathers” appears to still not be a significant phenomenon: traditional, gendered conceptions of different tasks still play an important role, despite an increase in equality in the workplace. In particular, regular, time- consuming tasks continue to be seen as the preserve of women, while tasks involving children’s hobbies and other social activities see a more mixed participation of the two parents. Father involvement is mostly linked to household’s and couple’s characteristics such as their socio- economic profiles, their social and gender values, the mother’s working hours (most influential factor for changing patterns); the number, age and sex of children (Brugeilles & Sebille, 2013)
SLIDE 4
The French context
Similarly to other countries, France has seen an increase in the diversity of family forms and life course patterns in the last decades. This has been especially marked by an increase in cohabitation, in unmarried pregnancies, in divorce and separations (Bonnet et al., 2010). Non-traditional family forms have also emerged, for example, about 4% of all couples are classed as “Living Apart Together” (Buisson & Lapinte, 2013). As elsewhere in Europe, entry into parenthood is increasingly delayed (Toulemon et al. 2008). However, unlike other European countries, this postponement has had little effect on completed fertility and France has one of Europe’s highest fertility rates. This relatively high level of fertility is related to the strong two-child family norm (Régnier-Loilier 2006) and to a generous and diversified family policy, i.e., a combination of allowances, tax deductions and childcare facilities (Toulemon et al. 2008). Another specificity of the French context is that the socio-economic stratification of fertility behaviours is less important than in Anglophone countries (Rendall et al., 2009), possibly because universal childcare and generous maternity leave gives incentives to women to enter the labour market before having children. Within this context, the proportion of children not living with both biological parents is the relatively lower than other Western countries but not trivial: in 2011, 18% of children did not live with both parents (Lapinte, 2013). For these children, residence with the mother is the most common arrangement: for example, at 10 years of age, 18% of children live with their mother only, against
- nly 4% living only with the father in 2005 (Breton & Prioux, 2009). In spite of its regularisation in
2002, shared custody remains relatively rare: for example, a study shows that a year after divorce,
- nly 15 % of children are in shared custody, while 76 % live with their mother only, 9 % with only
their father (Bonnet, Garbinti, Solaz, 2015). Shared custody is very rare for infants and young children and concerns mainly children aged 3 years of age and over. In case of separation or divorce, the parent who does not have custody of the children can exercise parental authority jointly and has a right of access. In turn, he/she must contribute to the child’s maintenance and education by paying child maintenance to the person who has been granted custody, as fixed by the court. These rights and obligations depend on the condition for the recognition of maternity/paternity. Paternal affiliation is automatically established if the parents are married to each other. If they are not, the father must recognize the child to establish his paternity. Recognition can be done before birth, at the time of the declaration of birth (usually within the 5
SLIDE 5
days of birth) or afterwards through an act with a civil registry official. A father who does not recognize his child after its first birthday is deprived of his parental authority. He can regain this authority afterwards by declaration if the mother agrees or by court order. If the father refuses to recognize the child, the mother can take legal action against the father to establish his paternity.
Hypotheses
Our analyses are guided by three research hypotheses, described here: First, the heterogeneity of households without a permanent father will be important in a setting like France, in terms of the socio-economic and demographic characteristics of parents. However, we don’t further hypothesize on the type of classifications that might be present, as the latent class methods used, by definition, do not allow imposing a priori assumption. Second, given these differences in household characteristics, we also expect that the levels of father involvement and contact two months after birth will vary significantly in this sample. Given the literature, we anticipate that contact will be more frequent for groups characterised by higher average levels of social and financial resources. Third, we expect that the latent groups of non-resident father households will predict father contact at 1 year through a number of channels. In our setting, we expect that it is not only socio-economic factors, as classically used as explanations in the literature and usually conceptualised as fixed concepts, that will predict father involvement one year after birth. We hypothesize that father contact at 1 year of age will also be predicted by the parents’ relationship trajectories prior to the birth of the child; the degree and nature of father involvement at birth and 2 months; and the household’s changing financial means, employment status, and housing situation once the child is born.
Data, study population and methodology
The ELFE data
The Etude Longitudinale Française depuis l'Enfance (French longitudinal study of children Elfe) is France's first large generalist birth cohort study (Charles et al., 2011), using a multidisciplinary
SLIDE 6
approach to explore the relationship between environmental exposures and the socio-economic context on child health and development. It follows over 18,000 children born in 2011, and is representative of all children born in France (excluding Overseas Territories). Elfe takes a multidisciplinary approach to explore different aspects of children's lives (environmental exposures, family life, living conditions) to assess their influence on the children's development and health. The study includes repeated questionnaires with the parents, the collection of biological samples, and linkage to other data sources at different points in the child's life, including in the hospital shortly after birth, at two months of age, on their first birthday, etc. Children were recruited in a random sample of 349 maternities among the 544 registered in metropolitan France. A first inclusion survey was carried out with the mother face-to-face at the hospital shortly after birth. Mothers were invited to participate if they had delivered a baby in a selected maternity during four inclusion periods (25 days in total: 4 days in April, 6 in June/July, 7 in September/October, 8 in November/December), if she was over the age of 18 and could sign a consent form available in 5 languages. Babies born at 33 weeks’ gestation or later were included invited in a parallel cohort study. The initial sample consists
- f 18,228 newborns. An in-depth telephone interview was conducted two months later: 15,536
mothers and 12,504 fathers responded to the full interview, collecting in-depth data on socio- economic status, family living arrangements, nutrition, and the environment children grow up in. In this wave, mothers and fathers are interrogated. A new wave of interviews was carried out around the child’s first birthday, with both parents (SAMPLE SIZES), and includes a similar, in-depth questionnaire to the previous wave. These three data points are used here; all results shown are weighted to take account of the survey sampling design. We consider that the child lives with both parents when the mother has declared that she is in a relationship with the father and that they cohabit full-time. Parents are classed as being in a non- cohabiting relationship when the mother has declared that she is in a relationship with the father but that he resides elsewhere or cohabits with her part-time (for example, only at weekends and holidays,
- r a few months in the year or more rarely). Parents are considered not together when the mother has
declared that she is not in a relationship with the father.
Analytical sample
We focus on children not (always or permanently) living with both their biological parents at the 2 month interview. Observations (n=15) have been removed from the analytical sample if the child was not living with the mother at the 2 month interview; if the child or one of the twins was (still) in the hospital at the 2 month interview; if the child was fostered; if the mother does not know who the
SLIDE 7
father i; if the father passed away. 934 households were included at the 2 month wave, 690 were present at 1 year wave.
Measures
In this paper, we distinguish four groups of fathers: co-residential fathers who live full-time with their child; Living Apart Together (LAT) fathers who have a relationship with the mother but not (always) live with the child; non-residential fathers who recognize the child at birth; and non- residential fathers who did not recognize the child at birth. We include in our analyses the last 3 groups. We use several socio-demographic measures, including the international standard classification of education (ISCED) for describing mothers’ educational levels (divided into three categories (primary school qualifications, lower secondary school qualifications, upper secondary qualifications and over). We also consider the immigration background of the mother, distinguishing whether she is of migrant origin (defined as being born abroad with a foreign nationality), descendant of migrants (defined as being born in France to migrant parents), and “natives” (which includes all those born in France to non-immigrant parents, but also those born abroad with a French nationality). We construct a measure of equivalised annual household income based on the OECD equivalisation scale and categorized in tertiales, as well as an indicator of welfare receipt based on the reported receipt of three benefits, including unemployment benefit, housing benefit, and the active solidarity revenue (allocated to individuals whose regular income is too low to cover a basic cost of living). We measure the father-child involvement and contact at three points: birth, two months, and 1 year. For father involvement at birth, we use an indicator for whether the father has formally recognized the child and a measure of whether he was present at the birth. Here, we account for the fact that when the birth is a Caesarean the father may be asked to wait outside the delivery room and consider the percentage of Caesarean births amongst those births where the father is not present. At two months, we measure frequency of the contact between the father and the child categorized as: sees the child weekly or more, sees the child less often, never sees the child.1
Methodology
To classify households without a priori assumptions, we use Latent Class Analyses (LCA) to decrease subjectivity in choosing groups. LCA is a clustering method that uses a probabilistic model
1Weekly contact: we also include here LAT fathers who live with their child: mostly weekend and holidays; mostly during the week; some months of
the year. Less than weekly: we also include here fathers who live with their child some of the time (LAT fathers) but for whom the mother did not know how often they lived with them.
SLIDE 8
to describe the distribution of data, which can be used to identify subsets underlying the observed heterogeneity in a population (Collins & Lanza, 2008). To classify the 934 families without full-time resident fathers at the 2 month interview, we use a number of variables from the birth and two month interviews: residency of the biological father at 2 months (whether the father lives part-time in the household); demographic characteristics of the mother, including her age at birth of the child, whether the child is her first child, her migration status; socio-economic characteristics of the mother: her educational qualifications, her employment status, household equivalized income; father involvement at birth, including his physical presence at the birth and whether he recognized the
- child. The LCA is carried out with the Stata-LCA plugin (LCA Stata Plugin, 2014; Lanza et al.,
2014).2 Sample weights are applied in the LCA. Only complete observations are considered 3. In a final step, we carry out regression models to estimate whether the derived LCA groups predict father-child contact at 1 year of age, and explore which factors determine father contact one year after birth. To do so, we run two separate sets of logit models: a first model predicting weekly contact at 1 year; a second set predicts no father-child contact at 1 year. Alternative specifications were tested through a multinomial logistic regression model, simultaneously predicting father-child contact at 1 year classed as weekly contact, less regular contact, and no contact. The logit models are presented for ease of interpretation. Sets of covariates are added step by step in order to analyse how the specific effect of each set of covariates is related to the variables of interest. The sets of covariates are grouped in a temporal manner as follows: first, characteristics from birth (sex of the child, whether the pregnancy was wanted) are entered; second, characteristics relating to the parents’ relationship (whether the parents ever lived together, whether the parents have ever been in a relationship) and to the father-child contact (weekly, less often, never) as measured at the 2 months interview; third, we enter variables capturing the socio-economic and housing status of the household (whether they have difficulties making ends meet, home ownership, whether any grandparent lives in the child’s household, whether the household moved home between 2 months and 1 year) as measured at one year, as well as a dummy capturing whether the mother is in a relationship with the biological dad at one year. Sample weights are used throughout the analyses and take account of the sampling framework, non- response at inclusion, and non-response through the 3 waves of data collection used here (Juillard et al., 2015).
2 The LCA Stata Plugin uses baseline-category multinomial logistic regression to predict latent class membership. 3 Records with missing data on covariates are dropped from the analysis (Lanza et al., 2014).
SLIDE 9
Results
Descriptive analyses
Family situation of children shortly after birth
In the Elfe study, the large majority of the children (91.5%) live with both parents at two months (Table 1). About 8% live only with their mother. At this age, very few children live only with the father (1 child in our sample). Overall, at two months, the family environment of children that live
- nly with their mother is very diverse. About one fourth of these parents are in a relationship but do
not cohabit permanently. These parents are referred to as ‘living apart together’ (LAT) and 2% of families with a two-month old child fall into this category. Half of these parents do not cohabit due to professional reasons, one fourth as they want to remain independent. About 6% of the children have parents who are not in a relationship 2 months after birth. Amongst the households where the child’s father was classed as non-resident, only 2% of the mothers are in a relationship with a different partner, who could be referred to as the ‘social’ father present in the household.
- -- Table 1 about here --
In our analyses, we focus on the family situations 2 months after birth as this is when the main survey (which collected richer data on family situations and living arrangements) was conducted. However, the comparison between the 2 month survey with the more limited survey conducted at birth shows there is relatively little change between birth and the 2 month interview. About a quarter
- f parents not co-residing with a partner at birth were now doing so. Inversely, about 1.4% of parents
co-residing at birth were no longer living together at the two month interview. The birth of the child triggered the formalization of some unions: 4% of couples who were neither married nor in a civil union at birth had either married or celebrated a civil union by the child’s 2 months, and 3% of those in a civil union at birth were married two months later.
Involvement of the father around birth
SLIDE 10
Focusing on this group of children living with the mother only all or most of the time, non-resident father involvement around birth can be characterised looking at a number of variables. In about 38%
- f cases, non-resident fathers were present at their child’s birth, and 61% had legally recognized their
child by 2 months. Being present at the birth or inclusion of the father’s name on the birth certificate suggests some degree of involvement and is policy-relevant: recognizing a child obliges fathers to attend to their children’s material needs and raise them. Non-resident fathers who were present at birth were more likely to recognize their child (91% of them recognized the child) than those who were not present at the birth (47%). Their presence may be due to the fact that these births are rarely the result of a transient relationship. Only 0.1% of mothers declared not to know or have no contact with the father. Most non-cohabiting parents (three-quarters) were in a relationship for at least six months before birth and nearly one-third lived together for at least six months.
Main results
Latent Class groupings
The LCA analyses allow classifying the 934 households without full-time resident fathers at the 2 month interview. The best solution describes five groups4 which are reported in Table 2, along with the conditional probabilities of the variables used in the LCA analyses. These analyses show that non-residential fatherhood is a heterogeneous phenomenon (see table 2). The first 3 groups are of relatively similar sizes and make up three quarters of our sample, they are marked by different types of socio-economic disadvantages. The first group (25% of the sample) is mainly characterized by young maternal ages: 71% are under 25 years. We will therefore refer to as the “Young parents” group5. This group is disadvantaged in terms of socio-economic characteristics, including low educational level and the highest proportion of non-employment (84%). However, this group presents high levels of father involvement around birth: 91% recognize the child, and 60% were present at birth. The second group (24%) is characterized by a high proportion of migrant
- mothers. We refer to it as the “Migrant mothers” group. This group has a high proportion of lower
educated, out of the labour force mothers; mothers tend to be older than in other groups, and the
4 We have tested model specifications with 1 to 10 classes. To determine the optimal number of classes we consider the
Bayesian Information Criterion (BIC), which is lowest for the optimal number of classes. This suggests that a five-class model describes our data best, as the BIC is lowest for this specification, taking a value of 1675.54.
5 Annex 1 shows that the Young parents group also comprises the youngest fathers.
SLIDE 11
cohort child is not usually their first child. 98% of fathers have recognized the child, and 44% were present at birth. Finally, the third largest group (23%) comprises the most disadvantaged households, particularly for maternal education (81% of mothers did not achieve any end-of-secondary schooling qualifications), and with low levels of father involvement. Only 2% of fathers had recognized the child, and 5% were present at his or her birth. We name this group “Low educated mothers”.
- -- table 2 about here ---
The next two groups make up about a third of the sample, and stand out for their relatively advantaged socio-economic profiles. The largest of these two groups’ (19%) main characteristic is that 82% of fathers live at least part of the time with the child. We label this group “Living Apart Together (LAT)”. This group is the most advantaged in terms of socio-economic indicators, and virtually all fathers recognize their child and were present at birth (99 and 82% respectively). Finally, the smallest group (9%) is made up of relatively well educated, employed mothers, who are having their first child; father involvement around birth is very low: 17% recognized the child and 12% were present at birth. We name this group named “Solo mothers”. This group resembles the LAT-group in terms of demographic characteristics (such as maternal age and migrant status) and socio- economically conditions (although they appear to be slightly less well-off than the LAT group). The socio-economics profiles of the different groups highlighted above are reflected in other variables available in the survey which were not used in the LCA analyses, such as receiving welfare benefits, reporting having difficulty making ends meet, being home owner, or household income (see Annex 1). Other insights from these additional variables provide a more detailed picture of these
- groups. For instance, a third of the Young parents group lived with their own parent at two months
post-birth, suggesting that this group is at an entry into adulthood lifestage or are struggling to get past this stage. We do not know if they return to the parental home because of financial constraints or because they require parental support to combine their studies with child-rearing. In any case, this living arrangement may partly explain why they do not co-reside with the non-resident father, even though his level of involvement around birth is very high. The majority of mothers (80%) declared wanting a(nother) child before this pregnancy; this proportion is similar across the groups if slightly lower for the Migrant and Low educated mothers. Even if this is retrospective question, posed after the birth of the child and is therefore not ideal to establish fertility intention, the homogeneity across groups is striking. Similarly, the idea that these
SLIDE 12
births are not unwanted pregnancies is confirmed by the fact that two thirds of these parents had been in a relationship for at least 6 months. About a third of Solo mothers and Low educated mothers had lived for at least 6 months with the father, suggesting a separation had occurred, although we do not know if this separation is due to the pregnancy. Strikingly, only 4% of the LAT group had co-resided for at least 6 months, but the vast majority (90%) had been in a relationship for at least 6 months. The vast majority (83%) of non-resident fathers was in contact with their child two months after birth, and most of this contact was frequent but with some differences across the groups described above (see Table 3): while in total 65% non-resident fathers see their child at least weekly, this proportion is lower for Solo mothers (40%). Only 16% of mothers declare that the child never saw their non-resident father, this is highest for the low educated (37%) and the Solo mothers (52%) but virtually never the case for the LAT or Young parents groups (1 and 2% respectively).
- -- table 3 about here---
Evolutions in father involvement between 2 months and 1 year
30% of the sample described above is no longer present in the 1 year old survey. This is not surprising as these households are probably more likely to be mobile (for instance, 30% of the retained sample changed address between 2 months and 1 year, see Annex 1). Furthermore, the three more disadvantaged groups have the higher rates of attrition (between 36 to 28%) compared to the two more advantaged groups (10-18%). Of the 690 remaining households, at one year after birth, our latent class groups describe significant variation in father contact (see table 4). A third of LAT and Young parents fathers had now moved in with the child; for the other groups this proportion was 15% for the Migrant mothers, 8% for the Solo mothers and only 3% for the Low educated mothers. For about 38% of our sample, the father saw the child at least weekly and 15% saw the child but less often. This varied markedly by group: we note for example that while few Migrant mothers now co-resided with the father, over 50% of fathers saw the child at least weekly. On the other hand, regular contact was very low for the Low educated group, with only 18% of these fathers seeing the child at least weekly. In a third of our sample, at 1 year fathers never saw the child. This proportion is driven by the Low educated and Solo mothers groups, where respectively 68% and 54% of fathers never saw their child.
SLIDE 13
While the figures above relates to living and contact arrangements, a significant proportion of fathers was (still) in a relationship with the mother (table 4): this was particularly the case for the LAT group (80% of parents were in a relationship) and the Young parents and Migrant groups (39% and 38% respectively). However, the proportion of fathers in a relationship with the mother was low in the Solo mothers group (18%) and almost null for the Low educated mothers group (5%). While fathers can recognise their child later, there is very little change in the proportion of children who have been recognised by their father between 2 months and 1 year. This may be because for some groups (such as the LAT, Young parents, and Migrant mothers), the proportion of children recognised by their father at 2 months is already very high (above 90%) and therefore all fathers who would recognise their child in these groups have already done so by 2 months. For the Low educated and the Solo mothers, the proportions are very low at 2 months and do not change at 1 year. Given the low level of father involvement described for these groups at birth, 2 months and 1 year, it is perhaps unsurprising that fathers are not engaging in this process later on.
- -- table 4 about here ---
What predicts father contact after 1 year?
In a final step of analyses, we explore whether the LCA groups, built using variables relating to the time around birth and up to 2 months of age, predict father contact 1 year post-birth, and if so, through which variables. We first look at results of a logit model predicting no father contact at 1 year of age (table 5). Once everything is controlled for (Model 5), there is no significant difference in the probability of no contact for migrant mothers and young parents compared to the LAT group. Solo mother and low educated mothers have higher probability of no contact than the LAT group, even after characteristics relating to the relationship history of the parents, contact at 2 months, the socio-economic and housing status at 1 year, and the relationship status of the parents at 1 year are accounted for. If we look at the intermediate models, we can note that the initial increased risk of no father contact for the Migrant and Low educated groups (Model 1) is slightly decreased when the relationship history and the father contact at 2 months are included (Model 3, in particular, not knowing the father and not having contact at 2 months strongly predict no father-child contact at 1 year) and slightly decrease again when the housing status at 1 year is included, suggesting housing constraints.
SLIDE 14
However, the largest change in the estimates, and the loss of statistical significance, for these groups is observed when the relationship status at 1 year (whether the parents are in a relationship) is included in Model 5. While for the Low educated and Solo mothers groups estimates remain significant in Model 5, we also observe a strong decrease in the odds when relationship status at 1 year is included, suggesting that this is a key mechanism to explain the low levels of no contact in the LAT group. We note on effects associated to the baseline characteristics selected: the sex of the child or whether the mother wanted a(nother) child before this pregnancy. Therefore, these results suggest that it is not just disadvantaged groups that lose contact over time, and, vice versa, relatively advantaged groups can also be at risk of losing contact between fathers and non-resident children: the two groups at higher risk of no father contact at 1 year include both a relatively advantaged group (the Solo mother) and a disadvantaged group (the Low educated group). What characterizes these otherwise very different groups is the low father involvement at birth, with very low proportions of fathers recognizing the child and an even lower presence at birth.
- --table 5 about here---
The second set of logit models looks at the opposite potential outcome: regular father contact (at least weekly) at 1 year of age, using the same set of covariates (Table 6). Once everything is controlled for (Model 5), compared to LAT group there is no significant difference in the probability
- f weekly contact except for the Low educated mother groups who have a lower probability of
regular contact. A key difference with the previous outcome (no contact) is therefore that the Solo mothers group does not appear to be disadvantaged in terms of maintaining regular father-child contact compared to the LAT group, once all covariates are included. The intermediate models show similar patterns as described above, including a very strong link with the relationship status of the parents at one year, and an effect of father-child contact at 2 months (but not of relationship history). Socio-eco variables at 1 year, in particular whether a grandparent lived in the household, and having difficulties making ends meet, predicted a lower chance of regular factor contact. However, housing
- wnership did not predict regular father contact, unlike in the models for no contact.
- --table 6 about here---
The picture that emerges therefore shows that while socio-economic disadvantage may play some role in predicting father-child contact in early childhood, the relationship history between the parents
SLIDE 15
and the trajectories of father involvement and contact appear to play an equally important role. These relationships are defined from birth: father involvement from birth (for example, through legally recognizing the child or being present at the birth) is an important factor in understanding later father-child contact, and those early inequalities in father involvement appear to re-inforce over time. The LCA classes, capturing the early interplay of socio-economic and relational characteristics, are therefore particularly interesting when thinking about future trajectories of father-child involvement and contact.
SLIDE 16
Conclusions
Until recently, fathers have been largely ignored in the statistical portraits of families. Research and data has focused primarily on mothers. Yet an increasing body of research shows that the involvement of fathers, including when they no longer live with their child, is important for child well-being. The assumption often made is that, once fathers no longer reside with their children, they are likely to lose contact and therefore no longer “matter” in our understanding of child trajectories, especially as it is assumed that the disadvantage profiles of these households will be the key explanation for subsequent trajectories of child well-being. This is particularly the case when fathers are already not living with their child at birth, and in fact the little research available on non-resident father pertains to those who have lived for at least some time with their children. Using latent techniques and a recent, nationally representative sample of France, we show that, among households without a co-resident father shortly after birth, a heterogeneity of situations exists: non-residential fatherhood at birth is not always defined by disadvantaged socio-economic conditions nor by a complete lack of involvement of all non-resident fathers. In fact, two groups emerging from our data have a relatively advantaged status, and only two groups show very low levels of father engagement around birth. And socio-economic disadvantage and low father engagement did not always overlap. The resulting LCA groups were therefore picking up on a diverse mix of situations, describing both advantaged and low involvement groups, as well less disadvantaged but involved groups. These latent groups, built using characteristics shortly after birth, strongly predicted father-child contact subsequently at one year of age, and a number of channels emerged to explain these correlation: notably, father engagement around birth (for example, being present at the birth, or formally recognizing the child), whether they were currently in a relationship with the mother, and, to a smaller extent, housing and financial constraints. These results suggest that trajectories of father- child contact and of parental relationships are at least as important as socio-economic conditions to understand future father involvement.
SLIDE 17
Tables
Table 1: Family situation of 2-months old children Child living arrangement Weighted proportion N (non- weighted) Living permanently with two parents 91,5% 15207 Living permanently with the mother 8,4% 1079 Parents in a relationship (LAT) 2,0% 290 Father does not cohabit permanently 1,5% 227 Father non-cohabiting 0,5% 63 Parents not in relationship 6,4% 789 Parents separated, in contact 6,3% 722 Father unknown/refusion of contact 0,1% 67 Different situation (fostered child, child in hospital, …) 0,1% 14 Total 100,0% 16300
SLIDE 18
Table 2: Class sizes and conditional probabilities for the Latent Class Analysis Young parents Migrant mothers Low educated mothers LAT Solo mothers Maternal age Under 25 71% 0% 38% 2% 5% 25 to 34 29% 60% 39% 68% 56% Older than 34 0% 40% 24% 30% 40% Maternal education Primary/lower secondary 67% 60% 81% 6% 1% Upper secondary 22% 23% 17% 8% 30% Higher 11% 18% 2% 86% 69% Maternal occupational status Employed 16% 38% 32% 83% 79% Unemployed 31% 15% 29% 9% 4% Out of labour force 53% 47% 39% 8% 16% Maternal immigrant status Migrant 22% 56% 19% 11% 8% Descendent 20% 8% 12% 8% 14% Native 58% 36% 68% 81% 78% Maternal occupational class Low 93% 85% 97% 41% 29% Medium 7% 14% 3% 30% 51% High 0% 1% 0% 30% 19% Rank of child First 71% 14% 46% 52% 77% Second or more 29% 86% 54% 48% 23% Residential status father Non-residential 71% 71% 99% 18% 93% LAT 29% 29% 1% 82% 7% Father recognized the child No 9% 2% 98% 1% 83% Yes 91% 98% 2% 99% 17% Father present at birth Yes 60% 44% 5% 78% 12% No 27% 32% 75% 6% 65% Other (incl. Cesarean, no information) 13% 24% 19% 15% 23% N (unweighted) 234 227 213 175 85 Unweighted proportions 25% 24% 23% 19% 9%
SLIDE 19
Table 3: Father-child involvement over the first 2 months of life, weighted % Young parents Migrant mothers Low educated mothers LAT Solo mothers Weekly 71 69 61 69 40 Less often 24 21 2 29 7 Never 2 8 37 1 52 No information 3 2 1 1 N (unweighted) 233 226 214 175 84 Table 4: Father involvement at 1 year, by latent class groups, weighted % Young parents Migrant mothers Low educated mothers LAT Solo mothers Father-child contact at one year Child lives with biological father 29 15 3 31 8 Non-resident, child sees him at least weekly 36 54 18 56 28 Non-resident, child sees him less often 19 16 12 9 10 Child never sees him 13 13 68 2 54 Missing 2 1 2 Biological parents in a relationship 39 38 5 80 18 N (unweighted) 154 166 142 156 72
SLIDE 20
Table 5: Logit regression model predicting No father-child contact at 1 year Model 1 Model 2 Model 3 Model 4 Model 5 LCA group Migrant mothers 2.253*** 2.233*** 1.923*** 1.729** 1.002
(0.666) (0.670) (0.695) (0.703) (0.773)
Solo mothers 4.265*** 4.253*** 3.493*** 3.453*** 2.458***
(0.666) (0.667) (0.744) (0.745) (0.845)
LAT (ref.) (ref.) (ref.) (ref.) (ref.) Young parents 2.194*** 2.189*** 2.055*** 1.619** 0.842
(0.675) (0.674) (0.702) (0.705) (0.763)
Low educated mothers 4.830*** 4.816*** 4.366*** 4.074*** 2.838***
(0.635) (0.637) (0.676) (0.682) (0.764)
Characteristics Boy 0.086 0.051
- 0.027
0.269 from birth
(0.276) (0.292) (0.297) (0.317)
Mother wanted a child
- 0.100
- 0.174
- 0.192
- 0.307
(0.371) (0.382) (0.385) (0.399)
Relationship Parents lived together at least 6 0.595 0.547 0.381 status 2 months months
(0.372) (0.359) (0.380)
Parents not lived together but had been in relationship (ref.) (ref.) (ref.) Parents neither in relationship, 0.263 0.220
- 0.089
nor lived together
(0.348) (0.375) (0.383)
Mother does not know the 1.674*** 2.158*** 3.692*** father
(0.552) (0.600) (0.688)
Father-child contact, weekly (ref.) (ref.) (ref.) Father-child contact, less often 0.406 0.576 0.610
(0.423) (0.422) (0.477)
Father-child contact, never 1.195*** 1.375*** 1.586***
(0.461) (0.465) (0.478)
Socio-eco status Difficulties making ends meet
- 0.170
- 0.273
1 year
(0.356) (0.375)
Home owner
- 1.121**
- 1.152*
(0.559) (0.606)
Household moved between 0.243 0.567* 2 months and 1year
(0.317) (0.335)
Any grandparent in the 0.749* 0.819* household
(0.393) (0.421)
Relationship In relationship with bio
- 4.247***
Status 1 year dad
(0.838)
Constant
- 4.084***
- 4.036***
- 4.142***
- 3.928***
- 2.597***
(0.597) (0.701) (0.730) (0.763) (0.849)
N 635 635 635 635 635
* p<0.1; ** p<0.05; *** p<0.01
SLIDE 21
Table 6: Logit regression model predicting regular (at least weekly) father-child contact at 1 year Model 1 Model 2 Model 3 Model 4 Model 5 Cluster Migrant mothers
- 1.174***
- 1.112***
- 0.991**
- 0.713*
0.391
(0.357) (0.369) (0.412) (0.407) (0.482)
Solo mothers
- 2.669***
- 2.647***
- 2.141***
- 2.053***
- 0.790
(0.421) (0.420) (0.513) (0.497) (0.591)
LAT (ref.) (ref.) (ref.) (ref.) (ref.) Young parents
- 1.258***
- 1.236***
- 1.225***
- 0.908**
0.156
(0.363) (0.361) (0.390) (0.414) (0.470)
Low educated mothers
- 3.450***
- 3.405***
- 3.271***
- 3.006***
- 1.465***
(0.374) (0.374) (0.427) (0.440) (0.499)
Characteristics Boy 0.015 0.133 0.183
- 0.033
from birth
(0.240) (0.249) (0.253) (0.306)
Mother wanted a child 0.271 0.295 0.361 0.537
(0.339) (0.348) (0.359) (0.420)
Relationship Parents lived together at least 6
- 0.309
- 0.331
- 0.014
status 2 months months
(0.328) (0.325) (0.338)
Parents not lived together but had been in relationship (ref.) (ref.) (ref.) Parents neither in relationship,
- 0.318
- 0.349
- 0.036
nor lived together
(0.315) (0.326) (0.396)
Mother does not know the
- 0.882
- 1.009*
- 1.903***
father
(0.570) (0.584) (0.587)
Father-child contact, weekly (ref.) (ref.) (ref.) Father-child contact, less often
- 1.035***
- 1.082***
- 1.626***
(0.319) (0.316) (0.463)
Father-child contact, never
- 1.404***
- 1.518***
- 1.676***
(0.469) (0.466) (0.462)
Socio-eco status Difficulties making ends
- 0.648**
- 0.701**
1 year meet
(0.302) (0.340)
Home owner 0.241
- 0.047
(0.358) (0.427)
The household moved
- 0.116
- 0.564
between 2 months and 1year
(0.289) (0.347)
Any grandparent in the
- 0.667*
- 0.679
household
(0.354) (0.416)
Relationship In relationship with bio 3.349*** Status 1 year dad
(0.489)
Constant 2.088*** 1.825*** 2.188*** 2.188*** 0.456
(0.278) (0.451) (0.476) (0.524) (0.590)
N 635 635 635 635 635
* p<0.1; ** p<0.05; *** p<0.01
SLIDE 22
Bibliography
Amato, P.R. and B. Keith, 1991. "Parental Divorce and Adult Well-Being: A Meta-Analysis." Journal of Marriage and Family 53(1):43-58. Amato, P. R, 2001. "Children of Divorce in the 1990s: An Update of the Amato and Keith (1991) Meta- Analysis." Journal of Family Psychology 15(3):355-70. Amato, P.R, 2005. "The Impact of Family Formation Change on the Cognitive, Social, and Emotional Well- Being of the Next Generation." The Future of Children 15(2):75-92. Bourreau-Dubois, C. and Jeandidier, B., 2005. Pauvreté des enfants et structures familiales: familles nombreuses, familles monoparentales, enfants de parents divorcés. La famille, une affaire publique, Michel Godet et Evelyne Sullerot (rapporteurs), Rapport du Conseil d'Analyse Économique n° 57, pp.337-388. Breton D., Prioux F., 2009, "Observer la situation et l’histoire familiale des enfants", Portraits de famille : l’enquête Etude des relations familiales et intergénérationnelles, sous la direction de Arnaud Régnier- Loilier - Paris : Ined, 2009 - 1 vol. (543 p.), pp. 143-162. Buisson, G. and Lapinte, A., 2013. Le couple dans tous ses états. Insee première, 1435. Charles M.-A., Leridon H., Dargent P., Geay B. and the Elfe team, 2011. “Le devenir de 20 000 enfants. Lancement de l'étude de cohorte Elfe", Population et Sociétés, n° 475. Collins, L. M., & Lanza, S. T. (2010). Latent class and latent transition analysis: With applications in
the social, behavioral, and health sciences. New York: Wiley.
Kiernan, K., 2006. Non-resident fatherhood and child involvement: Evidence from the Millennium Cohort
- Study. Journal of Social Policy, 35, 651-669.
Juillard, H., Thierry, X., Razafindratsima, N., Bringe, A., Lanoë, J.L. (2015) Ponderations de l’ênquete Elfe en maternité. Retrieved from pandora.vjf.inserm.fr/public/ Lacey, R., Bartley, M., Pikhart, H., Stafford, M., Cable, N. and Coleman, L., 2012. "Parental Separation and Adult Psychological Distress: Evidence for The'reduced Effect'hypothesis?". Longitudinal and Life Course Studies 3(3):359-68. Lamb, M. E., 2004. The role of the father in child development (4th ed.). Hoboken, NJ: John Wiley and Sons. Lanza, S. T., Dziak, J. J., Huang, L., Wagner, A. T., & Collins, L. M. (2014). LCA Stata plugin users'
guide (Version 1.1). University Park: The Methodology Center, Penn State. Retrieved
from methodology.psu.edu Lapinte A., 2013, Un enfant sur dix vit dans une famille recomposée. Insee Première, N°1470 LCA Stata Plugin (Version 1.1) [Software]. (2014). University Park: The Methodology Center, Penn State. Retrieved from methodology.psu.edu McLanahan, S. and Carlson, M.S., 2004. Fathers in fragile families. The role of the father in child development, 4, pp.368-396. Panico, L., M. Bartley, Y. Kelly, A. McMunn and A. Sacker. 2010. "Changes in Family Structure in Early Childhood in the Millennium Cohort Study." Population Trends 142(1):78-92. Rendall, M.S., Ekert-Jaffé, O., Joshi, H., Lynch, K. and Mougin, R., 2009. Universal versus economically polarized change in age at first birth: A French-British comparison. Population and Development Review, 35(1), p.89. OECD, 2015. OECD Family Database, OECD, Paris (www.oecd.org/social/family/database.htm) Tamis-LeMonda, C. S., and Cabrera, N. (Eds.), 2002. Handbook of father involvement: Multidisciplinary
- perspectives. Hoboken, NJ: Lawrence Erlbaum Associates.
SLIDE 23
Annex 1: Characteristics of household at birth and 2 months, by LCA group
Young parents Migrant mothers Low educated mothers LAT Solo mothers Total Characteristics at birth Father's age at birth Under 25 37% 5% 8% 3% 3% 14% 25-34 43% 31% 14% 51% 17% 31% Older than 34 13% 46% 11% 42% 17% 26% Unknown 7% 17% 67% 3% 63% 30% Characteristics at 2 months Mother's matrimonial status Married couple 5% 8% 0% 25% 0% 6% Civil partnership 0% 1% 0% 12% 1% 2% Free couple 18% 18% 3% 27% 6% 14% Lone mother 77% 72% 97% 34% 93% 78% No information 0% 0% 0% 2% 0% 0% Parents were in relationship/lived together for 6 months Lived together and were in relationship 13% 19% 31% 4% 25% 19% Never lived together were in relationship 55% 51% 44% 83% 39% 52% Neither in relationship nor lived together 31% 30% 24% 13% 32% 27% Mother not sure who is father 0% 0% 2% 0% 4% 1% No information 1% 1% 0% 0% 0% 1% Mother wanted a child No 12% 25% 25% 7% 15% 19% Yes 87% 74% 75% 93% 86% 80% No information 1% 1% 0% 1% 0% 1% Any grandparent in the household No 69% 95% 83% 95% 89% 84% Yes 31% 5% 17% 5% 11% 16% Social welfare dependency Yes 81% 75% 85% 26% 46% 72% Difficulty making ends meet Yes 19% 37% 25% 10% 12% 25% Housing ownership status Owned 14% 9% 7% 43% 36% 15% Rented 39% 35% 41% 29% 35% 37% Publicly subsidised 36% 43% 42% 17% 18% 36% Family member 4% 6% 4% 4% 10% 5% Other 7% 9% 6% 8% 1% 7% Quantiles of revenue First 38% 48% 41% 2% 17% 37% Second 28% 32% 40% 19% 23% 31% Third 16% 13% 11% 75% 56% 22% Unknown 18% 7% 8% 4% 4% 10% Total 234 227 213 175 85 934 Characteristics at 1 year
SLIDE 24