Couple Bargaining or Socio-Economic Status?: Why some parents spend more time with their children than others Open Access

Time use research consistently finds that fathers' and mothers' time with children has gradually increased over the past few decades. However, beneath this trend, there is great heterogeneity among countries, and within countries. Instead, parent's time with children and the types of activities they do vary greatly by social class and by gender. Even among dual-earner couples, mothers systematically spend more time than their male partners on child care due possibly to gender norms and expectations or relative resources. Differences in the quantity and quality of parental time investment across socio-economic groups, however, are more difficult to explain.

Since the organisation of child care is at a complex intersection of values, social structures and constraints, this paper aims to assess which parents are more prone to investing time in their children. One group of explanations focuses on couple bargaining or the relative resources of each parent. As is the case for household chores, parent's time with children is then hypothesised to be a matter of relative bargaining power between the members of a couple. This hypothesis links child care behaviour to the persistent gender inequalities existing in Western societies. Another group of explanations focuses on the reproduction of social stratification or how socio-economic status is related to differences in social norms, lifestyle or leisure activities, parental ability and parental expectations of building their child's human capital. We expect that socio-economic status influences parental time spent with children and types of activity done. Here, we focus on parental capacity to stimulate their child's learning, and link parental child care to the theoretical and social debate on social class and education. This debate considers parental child care to be a key mechanism in class reproduction.

In order to disentangle these two explanations, relative resources and class reproduction, it is important to distinguish among participation in a variety of child care activities, since each activity may reflect different parental motivations and also result in different consequences for children. As a result, in this paper, we go beyond the often-used dichotomy between primary/non-primary care in the literature in order to develop a classification of parental time with children that represents more broadly the ‘quality’ of child care.¹ Through this analysis, we attempt to capture two main dimensions of child care: the intensity of the parent/child relationship and the extent to which the activities can potentially stimulate the intellectual, emotional and social capabilities of the child.

This research contributes to the current literature by adapting theories of the household division of labour and the reproduction of social stratification to the explanation of patterns of the quantity and quality of parental time devoted to children. Using data from the Spanish Time Use Survey 2002–2003, we investigate whether the amount of parental time with their children and the type of child care performed (developmental, high intensity, low intensity or supervisory care² ) is explained by the two principal hypotheses: the relative resources of each partner, and parental socio-economic status. Our sample is restricted to two-parent households (heterosexual couples) with children younger than 10 years old. Given the purposes of the analysis, we do not include one-parent households. Spain is of particular relevance for family studies because of the rapid and profound changes that it has undergone in the last 40 years. As a result, many young Spanish families nowadays are dual-earner families and have to confront a ‘familistic welfare state’ to solve their reconciliation problems. Because these new roles have only recently emerged, very little is known about their consequences for children. In this way, Spain provides a unique and dynamic setting for assessing competing hypotheses about parental time-use and child care.

Family change and women's increased participation in the labour market have not produced a reduction in parental time. On the contrary, countries with long time series data, such as the USA, show an upward trend in the quantity of parental time devoted to children since the 1960s³ (Bittman 1999; Sandberg and Hofferth 2001; Gauthier et al. 2004; Sayer et al. 2004). Fathers have increased their participation in child care activities, particularly during weekends, although mothers remain the main providers of care, are also more often the sole caregiver and assume more routine tasks than fathers (Pleck 1997; Bianchi 2000; Craig 2006). Trends in parental time investment involve a combination of factors, such as the reduction of routine domestic work and caring tasks, and the increase of developmental and more rewarding parent–child interactions (Sayer et al. 2004). Parallel changes have also taken place in the way parents socialise their children, with an increasing emphasis on their development and autonomy (Alwin 1996). Parents nowadays reduce routine domestic tasks as much as possible while more often incorporating their children into their leisure time (Bianchi et al. 2006). Therefore, the transformation of advanced industrialised economies does not necessarily entail ‘less caring parents’.

Nevertheless, the total time parents spend with children does not account for the nature of the interaction between the parent and the child. The scientifically and socially important issue is the ‘quality’ of time that parents spend on children (and not so much the ‘quantity’) and its unequal distribution across children from different social backgrounds. This issue has been relatively under-researched, but evidence shows that children receiving ‘high quality’ time of child care at early ages have more chances of good performance in formal education (Danzinger and Waldfogel 2000; Meyers et al. 2003). Along a similar vein, Hsin (2009) shows that close parent–child relationships during the first year of life or during the pre-school period have positive effects on the educational performance and cognitive abilities developed by children once they enter the educational system. In short, evidence suggests that the quality of the time devoted to child care has potential positive effects on children's life chances. However, the social determinants of parental time allocation to child care remain largely unexplored, particularly in contexts undergoing rapid family change such as Spain.

Most empirical studies reveal a strong correlation between parents' education and the type of activities that parents do with their children. Mothers with higher educational attainment in the USA, for instance, spend more time in ‘high-intensity care’ activities (i.e. where a direct child–mother interaction is implied, such as reading to the child) and less time in ‘passive care’ such as watching television (Bianchi et al. 2004). In contrast, educational attainment does not substantially alter the time spent by fathers in routine child care, although it is significantly associated with an increased participation in particular tasks such as playing, reading or leisure. In general, improved population-level educational attainment has indirectly reduced the gender gap in child care dedication within couples, as educated fathers are often more inclined to participate in child care (Gauthier et al. 2004; Chalasani 2007). In addition, educated parents value the time spent on children more, although there are some differences in the relationship between parents' education and the total amount of time spent on children across Western countries (Sayer et al. 2004).

It must be noted that the causal mechanisms behind these differences remain unclear and are yet to be explored in the literature. Education could reflect differences in parenting styles or values about child care organisation (Brines 1994; Bianchi 2000; Bittman et al. 2003); or education may simply capture indirect family income effects. For instance, since educated parents can more easily afford non-parental child care, such as child-minders or day care centres, they may display different behaviour than their low-income equivalents. We also propose an alternative explanation, rooted in the literature of social differences in educational attainment. Parents' social origins strongly influence children's human capital attainment, leading to a clearly positive association between these variables (Shavit and Blossfeld 1993; Breen and Goldthorpe 1997; Erikson et al. 2005). Most social class analysts' assume that parents know in advance that their dedication to child care, in terms of time and intensity, has some effect on their child's development, specifically on the educational performance of their offspring. Furthermore, it has been shown that parental dedication to child care has a significant and direct impact on school readiness and academic performance of children and a more indirect impact via the associated social and cognitive development of children (Chase-Lansdale et al. 1995; Yeung et al. 2000; Brooks-Gunn et al. 2002). Therefore, one can expect that, relative to parents of lower social class, more privileged parents will spend more time with their children in activities that build their human capital. In this sense, parental time dedication could be considered an investment in the (future) educational performance of their offspring.⁴ It would be rational for parents who desire high(er) levels of education for their children to provide a higher amount of ‘quality’ child care than parents who have lower educational expectations for their children. Educational research has shown that parental educational expectations for their children are determined socially, in the sense that they follow a social class pattern, related to the resources available to the parents. First, educational expectations are shaped by the costs of education, including opportunity costs, and the differential ability by social class to meet them. And second, parents may desire that their children obtain a social class position that is at least equivalent to their own, i.e. try to avoid downward social mobility (Breen and Goldthorpe 1997; Goldthorpe 2007). This logic would translate into more ambitious educational goals for children of higher classes and, following our assumptions, also implies greater parental investments in child care.

Other explanations for child care time allocation come from two perspectives rooted in human capital theory – parents' relative resources and comparative availability of parents' time. This first perspective assumes that the time allocated to children depends on the preferences and relative resources of members of a given couple. Thus, the partner with the higher resources (i.e. human capital or work experience) may be better prepared to bargain the share and distribution of time spent with children (Manser and Brown 1980; Bittman et al. 2003; Stancanelli 2003; Geist 2005), or may even decide whether or not to outsource certain family responsibilities for the sake of their career. This approach assigns preferences to individual family members – contrary to the neo-classical view that assumes a ‘consensus’ utility function for the family as a whole (Becker 1981) – and explains, for instance, the fact that wives tend to reduce their housework time and husbands tend to increase theirs as the proportion of family income contributed by the wife increases. For the same reason, when the husband is employed fewer hours and earns less than the wife, he has less power to avoid tasks and, therefore, performs more domestic duties. In both cases, it is more efficient in terms of household utility for men to increase involvement in domestic and child care activities. More recently, some authors have suggested that couples may not follow the same logic for organising child care as domestic work, since child care is a valued activity, rather than something to be avoided (Coltrane 2000; Gershuny 2000).

It should also be acknowledged that income might influence couples' division of care and parents' time availability in different ways. Thus, individuals living in high-income households have more options to acquire paid child care while reducing the couple's care burden and, quite probably, reducing their obligations for routine tasks. Consequently, they may have less time to share, more chances to equally distribute time and more opportunities to spend time on high quality activities. On the contrary, in low-income households couples may have to cope with child care by adopting a rigid division of labour or acquiring informal care. In either case, they have more time left to bargain, more chances of unequal sharing and more barriers to an active fathering or mothering. Using data for 2214 couples from the British General Household Surveys, Henz (2010) found that men's share of parent care increases with household income, while low-income households tend to follow a more traditional pattern of caregiving.

The second perspective, the comparative availability of parents' time, assumes that partners bargain the time that is left outside their work hours (South and Spitze 1994). Therefore, time with children is the result of both parents' relative resources and working time. Using data of two-parent families with at least one child younger than 5 in the USA, Aldous et al. (1998) documented that as the mothers' number of hours of paid work outside the home increased, the fathers provided more physical care. Likewise, the number of hours that fathers spent on the job was closely related to their parenting.

We build upon current literature by exploring empirically the quantity and type of activities of Spanish parents' time with children – from the couple's perspective. First, we test the role of couples' bargaining to explain the quantity of time that mothers and fathers invest in child care. Bargaining models have mainly been applied to housework time allocation (e.g. Lundberg and Pollack 1993; Bittman et al. 2003; Evertsson and Nermo 2004; Breen and Cooke 2005⁵ ). Fewer studies have used bargaining to explain parental time (Aldous et al. 1998) or parent care (Henz 2010). This is probably due to the fact that most analyses focus on individuals, either the mother or the father, and neglect joint decision-making processes within couples. Second, we test the role of parents' social position in predicting the amount of time dedicated to different child care activities. The working hypotheses are summarised as follows.

Based on human capital theory, we hypothesise that the partner with a greater share of a couple's total resources – measured either by their income contribution to the household economy or by their educational attainment in relation to their partner's education – has more ‘marital bargaining power’ to negotiate less time with children (Relative Resources Hypothesis). On the one hand, greater bargaining power by one member of the couple should lead to a lower share of supervisory and routine activities with children, because these activities are less rewarding to parents. Furthermore, supervisory and routine care often involves a large amount of time that could conflict with other activities. On the other hand, parents may seek activities that involve closer interactions between parent and child or have a greater potential to impact child development. As a consequence, the ‘relative resources hypothesis’ may not be applicable to this last type of child care. These effects may be further mediated by gender, as women and men may tend to allocate time to child care following different patterns due to their socialisation in traditional gender values (e.g. women doing more physical care or routine caring activities than men). Finally, the ‘relative resources hypothesis’ implies an inverse correlation between time spent by one parent relative to the other parent, since the partners substitute each other in the total amount of child care.

Based on educational and social class theory, we hypothesise that there is a positive relationship between parents' social position and the time they spend with children (Social Status Hypothesis). This thesis also predicts that couples with higher social positions or educational levels will allocate more ‘high quality’ time to their children (i.e. parental time devoted to activities that enhance the cognitive and emotional development of the child). Conversely, this hypothesis may be less relevant to explain supervisory and routine forms of child care. It can be noted that the ‘Social Status Hypothesis’ does not imply any particular division of labour between the partners. Furthermore, a positive correlation between the time spent by one parent relative to the other parent can be expected, since the behaviour of each couple member reinforces the other couple member. Contrary to the logic of the Relative Resources hypothesis, for which parents bargain to minimise their own involvement in child care, while trying to increase their partner's share of care, in the Social Status Hypothesis consistency in the sense of both partners' behaviour should be expected. Finally, it should be emphasised the Relative Resources Hypothesis and the Social Status Hypothesis are not mutually exclusive, but may, in fact, be complementary to a certain extent. Our empirical analysis will seek to determine under what conditions each hypothesis is relevant in explaining parental time devoted to child care.

We use data from the Spanish Time Use Survey, 2002–2003 (INE 2004), a cross-sectional survey based on diaries of time allocation from 23,880 households. All members of the household aged 10 or above are interviewed and are instructed to complete a diary with their daily activities, which provides data for both members of a couple. Each day is divided into 144 periods of 10 minutes, and respondents have to note down what they were doing during each period. A main activity (‘What were you doing’) must be specified, but there is also space for a simultaneous (secondary) activity (‘What else were you doing’). In addition to these, respondents report on the presence of other household members during each period. This methodology has been widely tested and ensures high-quality data. For our analytic sample, we select adult parents living in a married or consensual partnership (age 18–50) with at least one child younger than 10, resulting in 7384 individuals.

The dependent variable is the daily amount of time devoted to child care by each parent. We measure time in minutes and take into account time devoted to different types of care, as well as total time. In order to test the Relative Resources Hypothesis, we also perform some complementary analyses in which the dependent variable is the proportion of child care devoted by one couple member (the man) relative to the other (the women). Mothers and fathers are analysed separately because there are theoretical reasons to believe that women and men have distinct time use patterns and are influenced differently by our control variables (Bittman et al. 2003), which control for partner's characteristics when a couple perspective is considered. In the previous sections, we have emphasised the importance of the quality of time spent with children. We now introduce a typology of activities considering on the one hand, the intensity of the interaction between parent and child; and on the other, the potential intellectual, social and emotional stimulation of the child. This typology is rooted in work by Bittman et al. (2004) who classify child care activities into four groups: developmental care, high-intensity care, travel and communication, and low-intensity care. Here we modify this classification in order to consider passive care activities separately and to adapt this to the Spanish setting, where time spent on travel is much lower. The specific activities grouped under each type of care are the following:

• Developmental activities: face-to-face parent–child interaction, including activities considered to be critical for the linguistic, cognitive and social development of the child. These activities comprise teaching something to the child or helping with homework, reading, playing or talking with children. These are primary activities.

• High-intensity activities: face-to-face parent–child interaction related to physical care of the child, such as feeding, bathing, putting to sleep or taking care of the child when sick. This type of care is especially beneficial for the child's health and emotional wellbeing. Again, these activities are primary activities.

• Low-intensity activities: activities where parents play a secondary role or do not explicitly interact with the child, but which require more attention than mere supervision. Child care as a secondary activity is included here, as well as several activities where the child is present and a certain interaction is inferred, such as meals, being with the family, playing sport, bringing the child to school or to the doctor, etc.

• Supervision or passive care: includes any activity where the child plays no role but is present. There is no explicit parent–child interaction but the parent is available to the child. This includes sleeping, personal care of adults, paid work, housework, studying, shopping, voluntary work, informal help to other households (except child care), gatherings and religious fests, reading, watching TV, commuting (except if due to child care) and other non-specified activities.

We include controls at two distinct levels – the individual and the couple. At the individual level, we control for age and age squared; age at first childbirth; educational attainment (primary, secondary I (compulsory), secondary II ((general and vocational), and college); income (four categories); labour market activity (employed, unemployed, not active). For those who are employed, we take into account their working hours, distinguishing among those who work less than 20 hours a week; those who work between 20 and 45 hours; and those who work more than 45 hours a week. We also consider whether workers have some flexibility in their schedule. The survey also includes information on whether the diary was filled on a weekend day, and we include it as a control because we expect time use patterns to differ and parents to devote more time to the family on leisure days. At the couple level, we include the number of children and age of the youngest child, as well as the level of educational homogamy, the absolute and relative income (we have a high proportion of women who are out of the labour market and report no income; they are therefore included in a specific category within each model), and the occupational status of the household, using an adaptation of Erikson and Goldthorpe's (1993) classification. We measure occupational status of the household by the highest level of each member of the couple. Our data do not allow us to control for domestic help as would be ideal. We resort to including a dummy variable for the households which had some type of paid domestic help. Spanish households often get unpaid help from family members, and we do not have reliable information on that. As a result, we are probably underestimating external help, and this is a caveat to our analysis.

We are interested in testing mothers' and fathers' individual behaviour, and therefore our unit of analysis is the individual, with separate analysis for men and women. In order to test the hypotheses, we also introduce partner characteristics in the models.

Like results from other contexts, the time devoted to child care, as in the case of housework, shows a strong gender asymmetry. In 2003, Spanish mothers with young (under 10) children devoted an average of 452 minutes per day to child care, in contrast to the 274 devoted by men. In our data, women devote more time to both high- and low-intensity care, as well as to supervision tasks. However, developmental activities are gender neutral with both mothers and fathers spend about 20 minutes per day on them. Nevertheless, parental time with children depends greatly on the children's age. Younger children need more attention to carry out tasks, and they often require more direct supervision at home. Gender differences are even more significant among parents of very young children (3 or younger) with mothers devoting a much more time than fathers. As child age increases, overall care time by mothers decreases. This is particularly true for children older than 3 and then again when most children enter the school system (in Spain children 3–6 years old attend Infant education which is voluntary, cost-free and constitutes the first level of School Education). Mother time is much more intense during the early years and greater overall across all ages of children. However, father time does not decrease, with men devoting a similar amount of time to children regardless of children age.

In order to control for the variables that may have an effect on care time, we have used a regression analysis. Tables 1 and 2 present the results for complete models for women and men respectively that allow for a first evaluation of our hypotheses. However, a more in-depth exploration of the ‘relative resources’ hypothesis involves an analysis of the educational level of one member of the couple relative to the other member of the couple (Tables 3 and 4). Furthermore, in order to provide an additional test of the ‘social status’ hypothesis, we include a specific analysis of the impact of the household's occupational status on time spent on child care (Table 5 and 6). In each of the tables, we include models for total time devoted to child care and to each of the four types of parental activities (developmental, high-intensity, low-intensity and supervisory care). In the presentation that follows, we shall explain the results obtained for each of the main indicators of our hypotheses by combining the analyses presented in different tables. These results are easy to interpret. The constant terms in the last row of Tables 1 and 2 represent time spent, in minutes per day, on the specified activity for an individual with the baseline characteristics.⁶ The coefficients for the remaining rows represent the effect of each of the independent variables on the minutes devoted to child care.

One of the key indicators of the study is educational level. According to the Relative Resources hypothesis, what matters is the educational level of one partner vis à vis the other, while according to our Social Status hypothesis, the higher the level of education, the stronger the involvement in child care will be (in particular, if the combined level of education of the couple is high) and the higher the quality of the child care provided. As can be observed in Tables 1 and 2, the level of education has a strong positive and significant effect on the total time devoted to child care, for women and even more clearly for men. For instance, a college-educated woman spends 31 more minutes in child care than a woman with a lower secondary education. This result is in line with many other studies that show a positive relationship of education with parental involvement (e.g. Bianchi 2000; Gauthier 2004). For men, it is particularly the low-educated (primary level) who show a lower involvement in child care (20 minutes less than a man with a lower secondary education).

For men, the positive gradient of education can be found for all types of child care studied, although effects are not statistically significant for low-intensity care and are only weakly significant for supervisory care. For women, a positive relationship is found between higher education and all types of care except supervisory care, where women with a primary education show a significantly positive coefficient. As we have shown earlier, supervisory activities are, by far, the most time-consuming activities. This type of child care is also the more likely to be subject to bargaining between couple members, as it more seriously constrains other parental activities (such as leisure or paid work) and is then more likely to be considered an obligation.

These results are both underscored and nuanced when we analyse the degree of educational homogamy between partners (Tables 3 and 4). Here, we see that homogamous couples show a clear gradient according to the level of education, with college-educated couples showing the highest coefficients with respect to low-educated couples. For men, these results hold for all types of activities, although this is especially true for the share of developmental and high-intensity care. When she has a higher educational level than he, the coefficients show a similar order of magnitude and significance level as in the case of homogamous couples (Table 3). However, when the level of education is higher for the man than for his wife, the results are much less clear-cut, suggesting that women's educational level is a stronger determinant than men's for men's care activities. With the exception of supervisory activities, the positive relationship of education and time spent on child care is also found for women, and this is consistent with the Social Status Hypothesis (Table 4). It should be noted that few results are significant when the level of education is different between partners, indicating that differences in bargaining power are not the main drivers of time spent in child care.

A complementary view is obtained when we turn to the results concerning the share of time devoted to children within each couple. In Tables 3 and 4, when the father has a college degree and the mother primary schooling, his share of low intensity care diminishes significantly (but not other types of care). When she has a college degree and he a lower degree, men increase their share of supervisory, as well as low and high intensity activities (but not developmental activities). However, when the educational resources of each partner are different, the results are most often not statistically significant (Tables 3 and 4). The use of a relative measure appears to better explain fathers' involvement than mothers' involvement, since women already take on a large part of caring responsibilities, pointing to mechanisms related to gender norms instead of bargaining. Furthermore, one should keep in mind that there is an increase in the time devoted to child care by educational level, especially for men. Overall, these results clearly provide support for the Social Status hypothesis, while they are (weakly) consistent with the Relative Resources hypothesis for supervisory and low intensity care.

A second and perhaps more direct indicator of the economic resources available to individuals is their income (see Tables 1 and 2). A possible shortcoming of this indicator is that it is measured at the time of interview and, as a consequence, it may reflect the result of a bargaining process among the couple's members, rather than the resources available for each of them. In addition, we suspect it is severely under-reported, although there are no reasons to think that the relative situation of households (or individuals) is modified.⁷ Keeping these clues in mind, it can be seen in Table 1 that men's involvement in all types of care is positively related to their income, a result consistent with the Social Status hypothesis. While in the case of women, a U-shaped relationship between income and care seems to exist, since women in the lowest and the highest income groups spend more time on child care. This last result could be related to the use of non-parental care by higher income groups, which could affect women's time to a greater extent than men's time with children, given her generally larger implication in child care. However, we should keep in mind that these models control for the individual's labour market involvement and that incomes of less than 1000 euros involves households where nobody has a job (73% of those cases), together with households with just one job-holder (an additional 12%), thus making it difficult to interpret the results of this very low income group, where at least one of the parents has a high time availability to take care of children. We have also run models for relative income (results available upon request), and the results were not significant for women. For men, having a partner with higher income increased their time spent on developmental and high intensity care.

The results presented so far concerning educational level and income are basically consistent with the Social Status hypothesis, with the exception of supervisory care activities. In Tables 5 and 6, for men and women, respectively, we included an analysis of the household's occupational status as an additional indicator of social class. The class scheme used to classify individuals is based on the one proposed by Eriksson and Goldthorpe (1993). Two comments should be kept in mind. The first is that the actual class into which a particular household is classified is the highest between the partner's job classes; individuals are therefore supposed to share the social status of the couple member with the highest job level.⁸ The second comment has to do with the control for women who do not work versus double-income couples (Table 5). This is necessary in order to properly isolate the effects of occupational status, since home-maker women devote a much higher amount of time to child care than women with a job. Turning first to the results for women, a clear positive and significant association is observed between a household's occupational class and all types of care time, except for supervisory activities, where no significant relationship is found. Here we can also see that this positive relationship is stronger for the most demanding types of care. Thus, in the case of developmental care, high-grade professionals and managers spent on average 12.9 more minutes than skilled manual workers (the reference category). The results for men similarly show a positive relationship between child care time and social class, particularly for ‘high-intensity’ types of care. However, this relationship does not show up for ‘developmental’ care, which is a puzzling result.

We now comment on a key control variable, the total time that the partner spends on child care, which assesses the impact of the partners' involvement in child care on the individual's own involvement (Tables 1 and 2). The coefficients show a clear-cut and very significant result for all types of care and for both genders: the higher the involvement of one partner, the higher the child care dedication of the other partner. Furthermore, the relative impact of partners' behaviour increases with the quality of care. This finding unambiguously indicates that partners do not substitute each other in child care dedication; rather, the behaviour of each couple member complements and reinforces the other couple member. This behaviour is at odds with the logic of the Relative Resources hypothesis, for which parents bargain to minimise their own involvement in child care, while trying to increase their partner's share of care. However, these results are in line with the Social Status hypothesis, where the behaviour of partners is inter-related according to the theoretical arguments presented above.

All the models presented above also include other relevant variables that control for several determinants of parental care. The results obtained for these variables are as expected and are consistent with those found by many other studies. Thus, for instance, the positive effect of an additional child on parent's dedication to child care is strong and significant for all types of activities. This effect mainly applies to the second child, not the third, possibly indicating scale economies in care time. The same applies to the age of the youngest child, which clearly reduces parental involvement as age increases. The individual's age has a negative effect on child care time, although this impact is not linear, as captured by the age-squared variable. And finally, the activity and number of hours of paid work are consistent with the expectation that individuals with less involvement in paid work (inactive, unemployed or part-time workers) do have a greater involvement in child care. This last result is worth noting in the case of men, as it contradicts traditional gender norms about child care.

The main objective of this paper was to analyse the factors that influence the time that parents spend with their children, focusing on the case of Spain. We identified two main hypotheses that concerned Relative Resources and Social Status. According to the first hypothesis, couples would negotiate the time devoted to child care depending on their bargaining power. For the second hypothesis, parental time with children would depend on parents' social class position. Overall, the results obtained are more consistent with the Social Status hypothesis than with the Relative Resources hypothesis. Higher educational attainment and higher social status position is associated with more time spent with children for both men and women, although women's position seems to have a stronger influence on child care.

In addition to the total time spent with children, we have considered whether these hypotheses may help us explain the quality of the time invested in child care. We have differentiated four types of activities: developmental care, high-intensity activities, low-intensity activities and supervision or passive care, depending on the degree of parent–child interaction and on the abilities implied. As expected, the relative resources approach has some explanatory power when applied to supervisory or routine care, but certainly it is not consistent with the results of our analysis for more demanding types of care. For these types of care, which are less time-consuming but more crucial for the life chances of children, the Social Status hypotheses is of particular relevance. This pattern of child care time allocation points at the polarisation of parental involvement with children: parents from more advantaged social backgrounds invest more time of higher quality with their children than parents from less advantaged backgrounds. The institutional setting, e.g. the educational system, may modify the potential negative effects of such polarisation. In the Spanish case, most children over 3 years of age attend school, but the coverage of the public sector for Preschool education (0–3 years) is still under-developed. As a result, families which cannot afford to externalise child care usually receive external help from other family members, a strategy which solves the work-child care conciliation problem but does not help to correct potential inequalities. Another source of inequality that we found in this analysis concerns gender: even when we controlled for employment and time availability, women spend more time on child care, especially during weekdays, whereas men concentrate their implication on weekends and do less routine activities. However, given that our data are cross-sectional, we could not measure the change in men's involvement in child care. And there has certainly been a change, as shown by the literature in other countries, and as illustrated by unemployed or inactive men's involvement in child care.

Further analysis is needed to investigate more closely the relationships between parents' social status and child care, as well as the factors inducing parents to spend more time in developmental activities, which the scientific literature deems to be most influential for children cognitive development. Our theoretical models, based on cross-sectional time-use data, explained around 35% of the variance of parental time spent on activities considered high-intensity, low-intensity and supervisory, but only 9% of the variance of developmental activities. Nevertheless, it is worth mentioning that the total number of minutes devoted to developmental activities is very small.

The authors would like to thank Spain's Ministry of Science and Innovation [CSO2010-17811/SOCI] for financing this research.

Pau Baizán is ICREA research professor at the Universitat Pompeu Fabra (Barcelona, Spain). His research interests are in the study of family dynamics, time use, and on international migration, particularly focusing on the effects of institutional factors and on comparative research.

Marta Domínguez is associate professor at the Observatoire sociologique du changement – OSC (Sciences Po, Paris, France). Her work focuses on the sociology of family and gender, and she contributes to TransParent, a European research program which studies the division of domestic duties among couples. She also teaches first and second year courses at Sciences Po, including the Introduction to Sociology and the Sociology of Family.

M. José González is associate professor at the Department of Political and Social Sciences, Universitat Pompeu Fabra (Barcelona, Spain). Her research interests are family formation, time use, child well-being and gender inequalities. She also participates in the TransParent research project (Ref. CSO2010-17811/SOCI) which analyses changes in the division of work around the time of first birth, and its impact on gender inequalities in the household and the labour market.