Time-use studies show that contemporary fathers are more engaged in child care (Bianchi et al. 2006; Monna and Gauthier 2008), significantly impacting the cognitive and social development of their children (Dermott 2008; Pleck 2004). Yet other scholars contend that compared with women, men’s lower involvement in childrearing reflects traditional family arrangements that impede women’s career commitments and reinforces gender inequality (Gornick and Meyers 2003; Hochschild 1989; Williams 2000). Because of the importance of this question, scholars still vigorously debate the proposition that men are more family-centered now than in the past (Coltrane 2000; Hochschild 1989; Townsend 2002).

Some research supports the notion that involved fatherhood is a recent phenomenon. Several qualitative studies show that contemporary men do not want to repeat the mistakes of their own workaholic fathers, and instead spend considerable time with their children at the expense of their careers (e.g., Coltrane 2000; Gerson 1993; Pleck 2004; Risman 1987). Survey findings generally support these narratives, but one noteworthy study found that men’s time in child care was much larger on the weekends (Yeung et al. 2001). Of course, some families may prefer that fathers focus on career pursuits, and many fathers contend that earning a paycheck is their most important contribution to their children’s well-being, even if work demands limit their shared time together (LaRossa 1997; Pleck 2004; Townsend 2002). On the other hand, in an age when dual-earners are the modal family type among couples (Jacobs and Gerson 2004), fathers who limit their child-care time on workdays obligate mothers to care for children at these times (either by providing direct care or by arranging and supervising third-party child care), which will in turn hinder women’s own career pursuits (Budig and England 2001; Gornick and Meyers 2003; Hochschild 1997; Williams 2000). A more accurate assessment of men’s role in family life may require examining temporal trends in men’s shared time with children, by workdays versus non-workdays.

Of course, to assess inter-cohort differences in the levels and antecedents of fathers’ child care time allocations requires accessing recent and past data on time use. In general, there is a paucity of research of this type, and the studies that have been done are limited in at least two ways. First, bearing on point the above, most studies examined trends in total weekly time with children and did not distinguish between men’s child-care time on workdays versus weekends. Second, most studies were hampered by data limitations that limited model specification, potentially affecting the conclusions drawn from these studies. This article addresses these limitations by comparing a more recent cohort of men with a prior generation, and by specifying more extensive analytic models of fathers’ time with children. In addition, men’s child-care time commitments will be measured on workdays and non-workdays, and the observed change in both measures will be decomposed into portions due to inter-cohort differences in the characteristics of fathers versus wholesale changes in men’s behavioral propensity to spend time with children.

Prior Research on Fathers’ Time with Children

Fathers’ involvement in the lives of their children can be characterized in many different ways. On one level, fathers may be involved in their children’s lives yet have comparatively little direct contact with their offspring. Indeed, the fathers in Townsend’s (2002) study who provided a “package deal” of support to their children (i.e., food, clothing, and a good home in a good neighborhood with good schools) vigorously disputed the notion that they were “uninvolved” in their children’s lives, even though they described their wives as the “default parent” to their children because they spent so much time working. Besides being accessible to or responsible for children, scholars place a great deal importance on direct interaction and engagement with children, such as providing personal care, reading to and playing with children, going places with and/or transporting children from place to place, and providing general supervision (Dermott 2008). Involvement with children that includes one-on-one interaction is an important part of fathering, facilitating children’s social, emotional, and cognitive development (Hall 2005; Pleck 2004). Although engagement with children may be only a subset of activities, exchanges, and support that characterizes the relationship between fathers and offspring, it is an important element of fatherhood that is deserving of attention in its own right.

When considering direct interaction and engagement with children, most studies show that women remain primarily responsible for these activities, irrespective of the intensity of their work efforts (Bianchi et al. 2006; Coltrane 2000; Craig 2007; Monna and Gauthier 2008). Two cross-sectional studies report men converging on women’s time with children, but on weekends, not on weekdays. Galinsky (1999) surveyed adolescents and found gender parity in shared parental time with children on weekends, but on workdays twice as many mothers as fathers spent five or more hours in the company of their children. Similarly, Yeung et al. (2001) examined children’s time diaries in 1997 and found that the father-to-mother ratio of time with children approached 0.9 on the weekends, but was 0.66 on weekdays.

Of course, temporal data are needed to determine if contemporary men exceed their ancestors in time spent with children. There are several well-designed studies of this type, but most were conducted before the emergence of the norm of involved fatherhood (e.g., Bryant and Zick 1996a, b; Nock and Kingston 1988). Three studies are exceptional, however, in comparing recent fathers’ reports of shared time with children compared with fathers of a prior generation. First, Sayer et al. (2004) analyzed adult time diaries collected in four national surveys between 1965 and 1998. In each survey, adults reported the flow of their activities in the previous 24 h, stating what they were doing (primary activity), who else was present, and what else they were doing at the same time (secondary activity). Of course, child care can be a primary activity, or a secondary activity performed while the parent does something else, like leisure or housework. But Sayer et al. (2004) focused only on primary activities, and showed that fathers’ time in primary child care increased from two to six hours per week between 1965 and 1998. Second, Sandberg and Hofferth (2001) examined children’s time diaries, finding that children’s time in primary and secondary activities increased from 18.5 to 22.7 h per week from 1981 to 1997. Third, Hall (2005) found that between 1977 and 1997, men’s self-reported estimates of shared time with children increased by approximately 26 and 70 min per day on workdays and non-workdays, respectively.

Sayer et al. (2004) provide the most detailed summary of the behavioral and compositional factors that explain fathers’ increased shared time with children, and their argument is summarized briefly here. In regards to change in men’s behavioral propensities, Sayer et al. (2004) first note that in the past, contraceptive technologies were less reliable, access to abortion was more restricted, and societal norms equating fatherhood with adulthood were stronger. As these factors have changed over time, the transition to fatherhood has become less spontaneous and more voluntary, and contemporary men invest more time in their children as a result. Second, rising crime rates and fear of crime increased parental concern for children’s safety, causing fathers to spend more time supervising their children. Third, societal norms of involved parenthood now dictate that adults must be intensively involved in all aspects of their children’s lives (Hays 1996). Although the expectations of intensive motherhood fall more heavily on women, contemporary fathers risk getting negative reactions from wives, relatives, and acquaintances if they are largely detached from their children’s upbringing (Gerson 1993; Hochschild 1997; Townsend 2002). Of course, this argument appears to suggest that contemporary fathers are constrained to spend more time with their children. An alternative argument is that fathers have always wanted to be involved in their children’s lives (LaRossa 1997; Pleck 2004; Risman 1987), and that as society increasingly viewed parenthood as a shared activity and altered institutions like the workplace to facilitate it (Williams 2000), fathers acted on their desire to be heavily involved in their children’s lives.

The arguments above give reasons why fathers have a greater propensity to spend more time with their children. But, it may also be true that over time structural change in the composition of American families affected fathers’ time investments in children, irrespective of men’s motivation to engage in childrearing. Sayer et al. (2004) and Sandberg and Hofferth (2001) considered such compositional changes as rising levels of educational attainment and maternal labor force participation rates, increasing age at first birth, smaller family sizes, and increased divorce. Some of these factors may be more relevant for understanding trends in mother’s time investments in children (e.g., maternal employment and single motherhood), and some may increase fathers’ time with children (e.g., smaller family sizes), while others may decrease fathers’ time with children (e.g., increasing divorce rates). Both Sayer et al. (2004) and Sandberg and Hofferth (2001) estimated multivariate regression models of men’s time with children, finding that none of the temporal increase in fathers’ time investments in children could be attributed to compositional changes in parents and family; by implication, then, all of the observed increase in men’s time with children was due to changes in men’s propensity to spend time with children.

Assessing Temporal Trends in Men’s Time with Children

Two methodological questions can be raised about findings from recent studies, suggesting the need for additional research. First, if men’s time in child care partly reflects their willingness to be an equal partner in raising children, (Dermott 2008; Gornick and Meyers 2003; LaRossa 1997; Townsend 2002), the claim that contemporary men aspire to co-parenthood may be overstated if much of the temporal increase in child-care time occurred on weekends rather than on workdays. Sandberg and Hofferth (2001) did measure children’s shared time with their fathers on a school day and on weekends, but they summed these estimates and did not report their analyses of child-care time by work- versus non-workday status of their fathers. Sayer et al. (2004) did a multivariate analysis of child-care time in a single 24-h cycle and controlled for whether or not fathers reported their activities on a weekday or a weekend. Type of diary day had no effect on fathers’ time with children, yet relatively small sample sizes (e.g., n = 198 fathers in 1998) may account for the insignificance of this effect. Hall (2005) distinguished between workdays and non-workdays paternal interaction with children, but his analytic models employed a limited set of controls (an important point further discussed below). In sum, the literature is still relatively silent on the determinants of fathers’ time investments in children, by workday status.

Second, two of the three temporal studies decomposed the weekly increase in men’s time with children, finding that changes in men’s behavioral propensities, not compositional changes in families, accounted for the increase. Of course, the conclusions drawn from a decomposition analysis depend heavily on model specification. Thus, it is noteworthy that Sandberg and Hofferth’s (2001) analysis of children’s diaries controlled only for family structure (i.e., single parent versus married), and the education and employment of mothers. In their analysis of parental diaries, Sayer et al. (2004) controlled for number of children, presence of a preschooler, employment status, hours worked, age, education, and type of diary day. Hall (2005) estimated pooled models of men’s time with children controlling for sample year and other covariates, but he did not perform a decomposition analysis.Footnote 1 It may be true that men’s increased propensity to interact with children accounts for all of the temporal increase in shared time with children, but this conclusion is based on analytic models employing relatively few controls.

This point is not merely academic, but rather points to the sources and durability of change in men’s family behaviors. Change that stems from cohort replacement processes will manifest in change in the types of families in which fathers live (and the kinds of places where they work), and as such, this type of change is less likely to reverse course in the future (Brewster and Padavic 2000; Ryder 1960). An example of such a compositional change in family life is wives’ increasing contribution to family earnings, which studies of housework show is positively related to men’s contributions to household labor. The usual explanation offered for this finding is that women’s higher pay translates into greater decision-making power within the family, prompting fathers to engage more intensively in family life (for reviews, see Coltrane 2000; Greenstein 1996; Williams 2000). Given that increases in women’s proportionate contributions to household earnings is the historical product of increases in women’s educational attainment, employment rates, and relative pay, few scholars expect these markers of women’s status (and thus their relative power in family life) to reverse course in the future (e.g., Gornick and Meyers 2003; Williams 2000). Estimates of the compositional contribution to a temporal trend (estimated at zero in the two studies of men’s shared time using diaries) may be suspect when analytic models employ few predictors, and it should be noted that none of analyses of fathers’ child care time controlled for women’s relative earnings within the family.

Similarly, estimates of the propensity contribution to a temporal trend (estimated at 100% in the time diary studies cited above) are also suspect in analytic models with limited predictors. A rapid change in men’s behavioral proclivity to spend time with children may be welcomed by family scholars and the lay public alike, but such change could be produced by unique period effects that were absent in the past and may disappear in the future. For example, the decomposition analyses cited above tracked men’s behavior into the late 1990s, when tight labor markets and a growing corporate concern to appear “family friendly” may have emboldened men to work less and spend more time with children (Golden 2008). But when labor markets slacken, corporations tend to cut back on work-life balance programs (Fried 1998; Hochschild 1997; Jacobs and Gerson 2004), and concerns about job security may cause men to prioritize their roles as family providers over their desires to nurture children (Maume and Bellas 2001; Townsend 2002). In the studies reviewed above, most used demographic and family covariates to predict trends in men’s shared time with children, with few controls for the types of jobs and firms in which fathers worked.

In sum, a re-analysis of men’s shared time with children that draws on more fully-specified models and which decomposes temporal increases in child care time into its compositional and propensity components would add to the body of literature that seeks to determine the source and durability of change in men’s family behaviors. Before describing the data and measures for such an analysis, the next section briefly reviews the determinants of men’s time in child care.

Theoretical Perspectives on Men’s Time with Children

Scholars draw on several overlapping perspectives to explain variation in men’s time investments in children (for reviews see Coltrane 2000; Monna and Gauthier 2008; Sayer et al. 2004), and these perspectives are reviewed below. In addition, I offer some tentative research propositions as to how the covariates identified by these perspectives will predict men’s child care time allocations depending on whether compared with fathers in the past, contemporary fathers are more family-centered or remain largely focused on work and cede child care responsibilities to their wives.

First, the time availability perspective posits that demographic and economic factors affect the amount of time men can give to their children. Despite some contrary findings, most studies report that men’s time with children is negatively related to their age and work schedules (Greenstein 1996; Hall 2005; Yeung et al. 2001), and positively related to family size and the presence of young children (Bryant and Zick 1996a, b; Sandberg and Hofferth 2001). If fathers are more committed to their children now than in the past, we may expect that demographic traits such as father’s age, family size, and presence of young children will have stronger effects (i.e., significantly larger slopes) on men’s child care time allocations (especially on work days) in contemporary samples of fathers. Yet, if contemporary fathers largely resemble traditional fathers of the past, we may expect that these covariates have small and stable effects on child care time allocations across time. Furthermore, if the results from the decompositions performed by Sandberg and Hofferth (2001) and Sayer et al. (2004) are robust, then temporal changes in these structural features of family life (i.e., temporal differences in means) should account for little of the temporal increase in fathers’ shared time with children.

According to the time availability perspective, when wives work more hours husbands may be obligated to provide more child care, yet there is no empirical support for this proposition in the literature (e.g., Bryant and Zick 1996a; Coltrane 2000; Nock and Kingston 1988). This non-finding is anticipated by the gender perspective, which posits that care giving responsibilities reflects cultural prescriptions about who should perform these tasks. Since women are largely responsible for child care irrespective of whether and how much they work (Bianchi et al. 2006, Coltrane 2000; Hays 1996; Monna and Gauther 2008), men’s child care time is often determined by their rejection of traditional role responsibilities in family life. This proposition is more often tested in studies of the division of household labor, but two studies reported that egalitarian men spend more time with children (Bulanda 2004; Hofferth 2003). However, since the objective of this study is to determine how much of the change in men’s time investments in children is due to changes in men’s propensity to parent, controlling for gender ideology in the analytic models is tantamount to controlling on the phenomenon to be explained. Instead, changes in men’s propensity to spend time with children will be inferred from a decomposition of the increase in men’s child-care time into its compositional versus propensity components. If, compared with fathers in the past, contemporary men are more family-centered then the majority of the observed increase in child care time allocations should be due to an increased propensity on the part of men to spend time with children.

Although fatherhood is fulfilling most of the time, raising children can also be challenging and stressful at other times. In these instances the relative power of spouses will determine who provides more child care. Studies have conceptualized relative power in many ways (e.g., by spousal differences in age, education, and occupation), but most tests of this perspective found that men did more housework as their wives contributed a larger share of the couple’s earnings (for reviews, see Coltrane 2000; Greenstein 1996; Monna and Gauthier 2008). The effect of wife’s relative earnings on men’s time allocations to children has drawn much less attention from researchers, and as mentioned above, prior studies failed to control for wife’s relative earnings in models of men’s shared time with children. Nevertheless, one might expect this covariate to positively predict men’s time in child care, and in the decomposition analysis, changes in the mean on this covariate will account for a substantial portion of men’s increased shared time with children.

Finally, recent studies have paid increasing attention to how the work-family nexus affects family behaviors. In the 1990s employers increasingly implemented work-life balance policies (e.g., flexible schedules, working at home; see Golden 2008), yet there is little empirical support for the proposition that men with access to these policies perform more household labor (e.g., Estes et al. 2007; Jacobs and Gerson 2004). In part, this may be because in this decade firms also sought to convert many high-wage and full-benefit workers to contingent and contract workers in order to lower their wage bills; the “survivors” of these restructurings included skilled workers, professionals, and managers, who worked longer hours to ensure their job security and increase their chances of promotion (Jacobs and Gerson 2004; Maume and Bellas 2001). A concurrent trend in the economy was an expansion of work schedules into evenings and weekends, increasingly affecting parental efforts to spend time with children (Presser 2003). Although research on the work-family nexus was largely conducted with male professionals employed in large firms, research with a general sample of men may show stronger employment-related effects on family behaviors. This discussion suggests at the very least that analytic models of men’s time investments in children should control for employment conditions, including occupation, promotion chances, job autonomy, work schedules, and concerns about job security. In general, if men are persistently work-oriented (increasingly family-centered) over time, we may expect stable (weaker) effects of work-related covariates on men’s time investments in children. Of course it is also possible that as the economy changes over time, some covariates (e.g., more large firms with “family-friendly” policies) will enhance men’s ability to spend time with children, while other covariates (e.g., more non-standard work schedules) will constrain men’s shared time with children.

Data

Two data sources were used to assess temporal change in men’s time investments in children. The U.S. Department of Labor funded the 1977 Quality of Employment Survey (QES), with the aim of studying the work-family nexus in a representative sample of the labor force. The Families and Work Institute conducted the 1997 National Study of the Changing Workforce (NSCW), intending to replicate the 1977 QES and assess change over time (the NSCW was also administered in 1992 and 2002, but these surveys do not replicate the 1977 QES). The 1977 QES conducted face-to-face interviews with 1,515 respondents who were at least 16 years old, spoke English, and worked 20 or more hours per week for pay. The 1997 NSCW was a telephone survey of adults (age 18 or older) with 3,552 respondents. In pooling the two surveys the sample was limited to English-speaking respondents working for pay for 20 or more hours per week (to match the 1977 QES eligibility criteria), and who were at least 18 years old (to match the 1997 NSCW eligibility criteria). Both samples are representative of the employed labor force in their respective years (for the QES, see Staines and Quinn 1979; for the NSCW, see Bond et al. 1998). Both surveys had minimal design effects, such that applying weights had little bearing on the pattern of results; thus, all results reported below were derived from unweighted samples.

Sample

Total sample size in the pooled data set was 4,823 cases, but to investigate men’s time investments in children the sample was initially limited to 1,296 men who lived with dependent children (n = 513 in 1977; n = 783 in 1997).Footnote 2 Some readers may prefer that the analytic sample include mothers, and make gender comparisons in parental time investments in children. But, the literature on time in mothering is more extensive than is research on fathering (Bianchi 2000; Bianchi et al. 2006; Monna and Gauthier 2008). If women were incorporated into the analysis, most analysts would expect that mothers significantly exceeded fathers in shared time with children in both 1977 and 1997, and that women’s workday time investments in children declined slightly over the interval (supplemental analyses showed this to be the case; these results are available to readers on request). Thus, the analyses below will focus on fathers’ time investments in children, about which less is known for certain.

As a measure of men’s commitment to parenting, the estimates of time investments in children may be biased by limiting the sample to men working at least 20 h per week. Although this limitation was imposed by the original administrators of these surveys, it potentially excludes fathers who withdrew from the labor force to stay at home with their children. Although stay-at-home men have been profiled in qualitative studies (e.g., Gerson 1993; Risman 1987), they are atypical of most men, constituting only a handful of respondents in survey data, if they are present at all (Hakim 2002). Of course, other family-centered men may choose to work between 1 and 19 h per week, and they would similarly be excluded from the sample for this study. It was not possible to determine how many men were affected by this exclusion criterion in 1977, but in 1997, only 2% of men were excluded from the analytic sample because they worked between 1 and 19 h per week (data not shown).

Finally, estimates of fathers’ time with children may be affected by the growing phenomenon of single fatherhood (Dermott 2008; Sayer et al. 2004). In the 1997 sample, approximately one in 11 fathers were single, yet in 1977 all but a handful of fathers were married. Thus, to make a consistent comparison over time, the sample was further limited to married fathers. Although practical considerations motivated this selection, a sample of married fathers is appropriate for addressing a timely issue often raised in the literature. That is, if men’s time with children aids in freeing wives to pursue careers and is an indicant of equality in contemporary family life (Budig and England 2001; Gornick and Meyers 2003; Hochschild 1989), the day-specific measures of fathers’ child care time are most salient in a sample of married men. After taking missing data on predictors into account, the final analytic sample sizes were n = 428 in 1977 and n = 654 in 1997.

Measures

Fathers’ Time with Children

To assess time investments in children, the QES/NSCW surveys asked, “On average, on days when you’re [working/not working], about how much time do you spend taking care of or doing things with your (child/children)?” Responses were converted to minutes per day and top-coded at the 95th percentile to reduce the influence of outliers (logging child-care time resulted in analytic models with weaker explanatory power; results not shown). Between 1977 and 1997, married fathers increased their daily shared time with children by approximately 26 and 78 min on workdays and non-workdays, respectively (see bottom row, Table 1). As these figures show, the increase in non-workday time with children was three times larger than the increase on workdays, a finding consistent with Yeung et al.’s (2001) conclusion that “involved fatherhood” is much more likely to be observed on non-workdays.

Table 1 Means and OLS metric determinants of married fathers’ minutes per day with children, by type of day and year
Full size table

These estimates of fathers’ child care time may be inflated by men’s desire to conform to cultural norms of involved fatherhood, suggesting that child care time estimates taken from time diaries may be more accurate. By way of comparison, Sandberg and Hofferth (2001) reported that between 1981 and 1997, weekly father-child shared time increased by 4.2 h (22.7–18.5) in children’s time diaries. The QES/NSCW surveys provided information on typical number of days worked per week, which when multiplied by the average daily child care times and summed, show a similar temporal increase of 4.4 h (22.5–18.5 = 4.4) in fathers’ reported weekly shared time with children. Thus, in these data, there is no reason to think that self-report measures of men’s time with children diverge widely from estimates reported in children’s diaries.

Predictor Variables

Research on family behaviors typically control for the demographic and familial characteristics of respondents, work effort and employment context, and the relative incomes of partners. Although it is possible that work efforts and earnings may be endogenous to parental time investments in children, this is less likely to be true for men. Only a few qualitative studies suggest that family priorities determine men’s work efforts (Gerson 1993; Risman 1987), and in survey data, Hakim (2002) could not identify any men whose family priorities determined their work efforts. Consistent with prior research on household labor time (Bianchi 2000; Coltrane 2000; Monna and Gauthier 2008), this study will treat demographic, familial, and work-related characteristics as exogenous to men’s time in child care.

Given these considerations, the analytic models controlled for age (captured by a vector of dummy variables, with age 25–34 as the reference category), nonwhite (1 = Black or other; 0 = white), and years of education. As well, high incomes give families the resources to purchase third-party child-care (formal day care for young children, and enrollment in programs, lessons, sports, etc for older children), which may partially relieve fathers of childrearing duties (Lareau 2002; Monna and Gauthier 2008). To account for this, the models controlled for the couple’s combined income in logged 1997 dollars.Footnote 3 To measure the relative incomes of spouses, the quotient of wife’s income to the couple’s combined incomes was entered as a covariate in models of men’s time with children (adding squared and cubed terms for wife’s relative income did not show any non-linearity in this measure’s effect on father’s child care time).Footnote 4 Finally, the models included controls for number of children (capped at four to reduce the influence of outliers), the presence of a preschool child (i.e., younger than 6 years old), and the presence of other adults in the household (who may assume child care responsibilities in place of fathers).

The analytic models also control for work-related covariates. For example, large firms typically offer benefits that allow employees to schedule their work so that they may spend time with their families (Estes et al. 2007; Golden 2008); thus, the analytic models will include a binary control for working in a firm of 500 or more employees. At the same time, many firms use the threat of layoffs to get more work out of their employees, especially among professionals and managers (Jacobs and Gerson 2004; Maume and Bellas 2001), necessitating controls for workers’ perceived likelihood of losing their jobs (1 = “somewhat” or “very likely” to lose job in next 2 years; 0 = otherwise), hours worked per week, and professional/managerial occupation (1 = Census-defined manager or professional; 0 otherwise). In addition, the models included a binary control for expecting a promotion (1 = “good” or “excellent” chance for promotion; 0 otherwise) and a four-item index of job autonomy.Footnote 5 The flexibility and timing of men’s work schedules was also captured by a measure of the frequency of working at home (with responses ranging from 0 = never to 5 = more than once a week), working nights (1 = works most hours after 4 pm and before 8 am; 0 otherwise),Footnote 6 and a binary measure for typically working more than 5 days per week. Finally, the models included a binary control for self-employment for theoretical (self-employed men may set their own schedules) and practical (to prevent the loss of cases) reasons.Footnote 7

Results

Table 1 presents the year-specific means on the predictor variables (in the two left-most columns), as well as the results of an ordinary least squares regression analysis of married men’s time with children, by workday status and year. The right-most column in Table 1 shows the results of t-tests for temporal differences in means on predictors (denoted by the “a” symbol), and slope effects on workday and non-workday time with children (symbolized by “b” and “c” symbols, respectively). All t-tests were conducted with one-tailed tests since the literature review above clearly implied temporal change in men’s family behaviors and its antecedents; the significance of individual slope effects were evaluated in both one- and two-tailed tests.

Before examining the determinants of men’s time with children, it is noteworthy that on most covariates the characteristics the 1997 sample differed significantly from the 1977 sample. For example, relative to the reference category of men ages 25–34, the 1997 sample had more men in the 35–44 cohort than the 1977 sample (42% in 1997 versus 32% in 1977), and the level of education was higher in 1997 as well. These comparisons likely reflect trends toward having children later in life and rising levels of educational attainment (Sayer et al. 2004). As well, the 1997 sample had more than twice as many nonwhite fathers as the 1977 sample (18 versus 7%), which may be due to period effects and selection criteria in drawing the earlier sample. The 1977 survey was conducted during a severe recession and limited to those working 20+ hours per week; since non-white workers suffer higher unemployment than whites, the 1977 survey likely excluded many non-whites. In regards to income, after taking the anti-log of the year-specific means, the average couple’s combined real income grew by $5,312 over the interval. In part, income growth may be attributable to wives’ work efforts, with their percentage contributions to the couple’s income doubling between 1977 and 1997 (the means of approximately 9 and 18% in 1977 and 1997, respectively, were depressed by large proportions of men who were married to non-working spouses in both years; see note 4). Finally, number of children declined over time, although in both years about one-half of fathers had a pre-school age child.

As for employment covariates, over the interval fewer men worked in the largest firms and/or worked as professionals and managers, and twice as many (24 versus 12%) perceived they would likely lose their jobs. These findings likely resulted from corporate restructurings during the 1980s that trimmed middle managers from firms’ payrolls, increased the use of temporary workers, and resulted in many survivors feeling vulnerable to layoffs. This argument may also explain why work intensity increased over the interval, in the form of longer work weeks and more frequently working at home (Jacobs and Gerson 2004; Maume and Bellas 2001). Yet, at the same time, more men expected to be promoted over the interval. Finally, a surprisingly low (and declining) percentage of men work nights, which conflicts with Presser’s (2003) reports of rising rates of working non-standard schedules, yet her analysis showed that it was young singles and women with children who were the most likely to work these schedules, not married fathers.

Determinants of Fathers’ Time with Children

Turning now to the determinants of men’s time with children, the next-to-last row of Table 1 shows that the regression models explain between 15 and 19% of the variation in time with children, depending on the type of day and survey year. The right-most column shows that unlike nearly all of the covariate means differing significantly over time, only seven of 42 slope contrasts (denoted by the “b” and “c” symbols) significantly varied over time. Nevertheless, Table 1 does reveal some interesting patterns in predicting married men’s time investments in children.

In many studies age is viewed as a proxy for socialization experiences, in which older cohorts were more likely to grow up in traditional households, and were more likely to have traditional households themselves (Coltrane 2000; Hall 2005). But, in this study there are necessarily few baby-boomer and later fathers in the 1977 sample, and there are few 1950s and earlier fathers in the 1997 sample. Thus, the negative slopes for older age cohorts likely reflect aging effects on men’s time with children, in which, compared with men ages 25–34, older fathers spend less time with their children. These effects are stable over time with the exception of the youngest fathers (ages 18–24) who significantly increased their time in child care, but on non-workdays (b = 112.66 in 1977 versus b = −9.02 in 1977), not on workdays.

Some variable effects seem to indicate that contemporary fathers with higher social class standing spend less time with children. For example, the couple’s combined income is negatively associated with men’s time with children on workdays in 1997 (b = −12.77). If family earnings reflect social class standing (at least in part), Lareau (2002) found sharp differences in childrearing practices between working- and middle-class families. That is, middle-class families use their resources to cultivate their children’s talents by enrolling them in activities organized by other adults (e.g., scouting, piano lessons, sports participation, etc.) which would decrease fathers’ own shared time with children; by contrast, children in working-class families, spent more time “hanging out” with family members (Lareau 2002; see also Monna and Gauthier 2008). If this argument is true, it would account for the negative effect of family income on men’s workday time investments in children in the 1997 survey year. Similarly, if education taps class-specific childrearing practices in a similar manner, this may explain its negative effect on non-workday time investments in 1997. And, of course, if non-whites are more likely to be lower class, this may also account for why non-white fathers report more non-workday time with their children in 1997 than do white fathers in that survey year.

To the extent that men view childrearing as primarily women’s work (Hochschild 1989; Williams 2000), it is noteworthy that as women bring more income into the family, in both sample years fathers increase their workday time investments in children. Yet, the impact of this covariate on 1997 non-workday time with children is insignificant. This suggests that over time, on non-workdays men have less need to be constrained to spend time with their children, and instead choose to become more involved with their children.

This proposition gains additional support when considering the positive effect of having a pre-schooler on fathers’ time in child care. In 1997, fathers of a child younger than six spend 49 and 138 more minutes per day in child care on workdays and non-workdays, respectively, than do fathers of older children. Moreover, the non-workday 1997 slope is significantly larger than its corresponding slope in 1977 (the temporal contrast in workday slopes is marginally insignificant). These patterns indicate that among fathers who have young children at home, contemporary men are more family-centered (especially on non-workdays) than were fathers in the past.

There is evidence that supports and disputes the notion that fathers continue to prioritize their work efforts over parenting concerns. On one had, men with longer work weeks spend less time in childcare on workdays, an effect that is relatively stable over time (b = −1.60 in 1977; b = −0.97 in 1997). Similarly, working more than 5 days a week reduced fathers’ shared time with children on non-workdays in both years, and also limited workday child care time among 1997 fathers. In sum, as the length of work schedules increased over time, men necessarily limited their time in child care. On the other hand, some variable effects suggest that when given the opportunity to control their work, men spend more time with their children. For example, men who exercise some autonomy over the conditions of their work spend additional time with their children on workdays (b = 10.33 in 1977; b = 6.93 in 1997). Also, men who work in large firms (reputed to be more family friendly) spend more time with children on workdays in 1997 (b = 20.68). Of course, the late 1990s was a time when cultural norms of involved fatherhood were strongest (Estes et al. 2007; Williams 2000). This may partially account for why men who worked nights in 1997 (and were presumably at home during the days) spent 30 min more a day in child care, a significant increase over the negative (but insignificant) effect of working nights in 1977.Footnote 8

To summarize the results in Table 1, the determinants of men’s time investments in children are varied, and do not always accord with the perspectives typically invoked to explain variation in men’s household labor. For example, the time availability perspective suggests that men’s engagement in family life depends on the time they have available for these responsibilities; similarly, men’s available time often coheres with the work-family perspective which suggests that variation in job demands affects men’s ability to meet their family responsibilities. Table 1 shows some support for these perspectives in that the length of men’s work schedules (in terms of days and hours worked per week) negatively affects their time allocations to children, whereas men who exercise autonomy on the job and those who work in large firms are able to spend more time with their children. The gender perspective emphasizes that child care is primarily the responsibility of women, and generally unresponsive to family characteristics. Yet, fathers do allocate more time to pre-schoolers, and because this effect strengthened with time, it suggests that contemporary fathers of younger children are more invested in their children’s welfare than were their past counterparts. Finally, the relative power perspective enjoyed strong support; i.e., as wives’ contributions to the couple’s income increased, men generally spend more time in child care. What is also notable about these results is that although the means on most covariates significantly changed over time (denoted by numerous “a” symbols in the t-tests column of Table 1), the effects of these covariates were generally stable over time (denoted by the general absence of “b” and “c” symbols). These results suggest that the changing composition of men’s families and work situations may have a greater impact on men’s time investments in children than what might be expected from prior research findings, a proposition that is empirically addressed below.

Decomposition Analysis

Table 2 shows the results of a decomposition analysis, the bottom line of which shows the total increase in men’s time investments in children by type of day; i.e., a difference of 26 min on workdays, and 78 min on non-workdays. The cells under the “minutes” columns in Table 3 were calculated by multiplying the temporal differences in the means on predictors (calculated from Table 1) by their corresponding 1977 slopes (chosen on the grounds that it is desirable that 1977 men resemble their 1997 counterparts by increasing their shared time with children; see Jones and Kelley 1984). These values can be interpreted as the change in time with children (in magnitude and as a percentage of the total temporal gap) that would be expected if the 1977 sample resembled the 1997 sample. Summing down the columns shows the total amount of change in men’s time investments in children attributable to compositional changes; any remaining difference is attributable to changes in men’s behavioral propensities (assuming fully specified analytic models).Footnote 9

Table 2 Decomposition of the temporal increase in married men’s minutes per day with children, by type of day
Full size table

Focusing on the summed compositional component first, the results do conflict with prior research findings on the sources of the increase in men’s time with children. That is, analyses of time diaries found that all of the increase in men’s time with children was due to changes in men’s behavioral propensity to care for children, and none of it could be attributed to compositional change in men’s work and family lives (Sandberg and Hofferth 2001; Sayer et al. 2004). Yet, after estimating more extensive analytic models of men’s child care time on workdays versus non-workdays, this finds study finds that 27 and 31% of men’s workday and non-workday time with children, respectively, is attributable to temporal changes in the composition of fathers.

And, in particular, it appears that women’s increased contribution to family earnings is the primary compositional factor that accounts for men’s increased time investments in child care. If fathers in 1977 were married to women who contributed as much to the couple’s income as their 1997 counterparts, 1977 fathers would spend 7 and 15 min more per day with children on workdays and non-workdays, respectively, accounting for 26 and 19% of the increase in men’s day-specific time investments. This finding suggests that compared with their past counterparts, recent cohorts of men are increasingly accommodating their parenting obligations to their wives’ increased earnings in the labor market. To the extent that wives’ contribution to family earnings is unlikely to reverse course in the future (unless one expects future declines in women’s educational attainment and employment, accompanied by a worsening of the gender wage gap), we might expect that this determinant of change in men’s time investments in children is also unlikely to reverse course.

Compositional changes in the workplace also partially account for the increase in men’s time with children, although the effects are smaller and countervailing. For example, the temporal increase in job autonomy explains 12% of the increase in men’s workday time with children. Since blue-collar work provides less autonomy than white-collar work (Jacobs and Gerson 2004), the growth of service jobs and the loss of manufacturing jobs likely accounts for the temporal increase in job autonomy between 1977 and 1997. Nevertheless, these results suggest that when fathers exercise some control over the demands and conduct of their jobs, they spend more time with their children (perhaps, not surprisingly, job autonomy accounts for a smaller portion of the increase in non-workday time with children). On the other hand, as men’s work hours lengthened over the interval, this reduced men’s workday time with children by a non-trivial 9%. Similarly, as the frequency of working at home increased, men reduced their time with children by 8 and 9% on workdays and non-workdays, respectively. And, to the extent that these trends reflect durable changes in the organization of work that require men to work longer and more frequently at home (Fried 1998; Jacobs and Gerson 2004), these factors are likely to continue to inhibit men’s time with children in the future.

Summary

There is a dearth of research on long-term temporal change in men’s time commitment to children. The few studies that are exceptions to this statement specified limited analytic models, and attributed all of the increase in men’s time in child care to an increased behavioral propensity among fathers to be involved in their children’s lives. This study drew two nationally representative samples of fathers in 1977 and 1997 to estimate more fully-specified models of men’s time investments in children, and further distinguished between time in child care on workdays versus non-workdays. Indeed, the temporal increase in involvement with children was three times higher on non-workdays than on workdays, suggesting that men may persist in their traditional orientation to cede routine child-care responsibilities to their spouses on days when they must work. This conclusion is reinforced by the finding that men’s time in child care was highly responsive to an increase in women’s power within the family (conceptualized by the relative incomes of spouses). On the other hand, there is some evidence that younger fathers, and fathers of young children, are more involved in their children’s lives, and that when they exercise some control over their jobs they spend more time with their children on workdays.

A decomposition analysis showed that approximately 70% of men’s increased time with children could be attributed to wholesale behavioral changes among all men, and that 30% could be attributed to inter-cohort changes in the composition of fathers. Foremost among these cohort-replacement changes is that more recent cohorts of men are more dependent on their wives’ earnings than were men in the past, and they exercise more autonomy at work. Change in men’s parenting behaviors that stem from cohort replacement processes, such as women’s increased power in family life, are likely to be more durable and long lasting (Brewster and Padavic 2000; Ryder 1960). Thus, this paper’s finding that approximately 30% of men’s temporal increase in child-care time could be attributed to structural changes in the characteristics of men’s lives differs from the findings of prior research, suggesting that these sources of increases in men’s parenting time are unlikely to reverse course in the future.

Discussion

Despite significantly improving on past analyses of temporal trends in levels and determinants of fathers’ time investments in children, this study has not definitively settled the question as to whether contemporary fathers resemble their past counterparts in prioritizing their work obligations over their parental responsibilities, or whether contemporary men are more involved in family life than were their ancestors. On one hand, men resemble their work-prioritizing ancestors in that the increase in workday time in child care was only one-third as large as the corresponding increase in non-workday shared time, demanding work schedules substantially limited men’s shared time with children, and a substantial financial contribution from wives to family incomes often compelled men to increase their time in child care. On the other hand, more findings support the notion that involved fatherhood is an emergent phenomenon, including the fact in that younger fathers in the most recent cohort significantly increased their time in child care (as did fathers of young children), when they had job autonomy or when they worked in large firms (reputed to be family-friendly), fathers used this freedom to attend to their children, and the largest contributor to increases in men’s child care time was a greater behavioral propensity among all men to spend time with children. Given these mixed findings, then, it appears safe to conclude that many, or perhaps most, men want to be intimately involved in their children’s live, but work demands often prevent men from realizing their goals of becoming involved fathers, certainly on work days (Jacobs and Gerson 2004; Williams 2000).

Of course, the limitations of this study should be noted, in hopes of stimulating further research that will assess the extent of men’s commitment to their children and by extension, egalitarianism in family life. First, this study could not examine parental time investments when fathers lived apart from their children. Although this paper’s sample of married fathers was appropriate for examining the larger issue of men’s parenting in the face of work-family commitments, many fatherhood studies ignore nonresidential fathers (LaRossa 1997; Sayer et al. 2004; but see Hofferth et al. 2010). To fully understand men’s involvement in the lives of their children, future studies should draw on samples that include both nonresidential and residential fathers.

Second, this study could only examine the quantity of fathers’ time with children, not the nature of shared time. One major advantage time diaries in examining family behaviors is their greater detail on the time fathers spend feeding, grooming, teaching, playing with, or doing household tasks with their children. If fathers are more apt to play with their children while mothers spend more time in personal care and instructing them, this may reflect a persistent traditionalism that holds mother largely responsible for the routine care of children (Hays 1996; Hochschild 1989; Hochschild 1989; Townsend 2002). On a related point, this study could not take into account the gender of the child, and this may converge with the content of shared time to further reflect a persistent traditionalism in parental time investments in children. Time diaries from the 1970s showed that mothers spent more time with daughters doing household tasks, while fathers spent more time with sons in leisure tasks (Bryant and Zick 1996b; Yeung et al. 2001). What is not clear at this point is whether recent data would show gender convergence (by sex of parent and sex of the child) in the nature of parent–child shared time.

Finally, if scholars are going to collect detailed diary data on the gendered activities in contemporary family life, there are other important uses of time that may reflect men’s commitment to equality in family life. For example, conventional wisdom and scholarly research suggests that working women must sacrifice sleep and leisure to avoid neglecting their children (Craig 2007; Williams 2000; Maume et al. 2009). Research has long shown that stress levels are higher in women, and their greater responsibility for family life partially explains women’s lower psychological well-being (e.g., Waldron et al. 1998). Yet, getting enough rest, maintaining an exercise program, and pursuing hobbies, reduces stress and improves health. Examining temporal change in gender differences in the amount and quality of time in sleep, exercise, and leisure should be added to any research agenda seeking to assess commitment to equality in family life among contemporary adults.