Happy Moms, Happier Dads: Gendered Caregiving and Parents’ Affect

Abstract

Parenting is emotionally demanding and highly gendered. We use data from the American Time Use Survey to examine mothers’ and fathers’ momentary affect during childcare activities. We observe a gender imbalance in the emotional rewards of childcare: fathers report more happiness, less stress, and less tiredness than mothers. We introduce the “care context”—defined as the type of childcare activity, when and where it takes place, who is present, and how much care is involved—as an explanation for these gender differences in parents’ affect. The analysis reveals that most dimensions of the care context vary between mothers and fathers. We also find that the care context fully accounts for differences in mothers’ and fathers’ happiness, partially explains differences in stress, and does little to explain differences in tiredness. Thus, the gender imbalance in the emotional rewards of childcare is partially due to parents’ highly gendered engagement with their children.

Introduction

Parenting is highly gendered. A substantial body of research shows that, on average, mothers spend more time than fathers caring for children (Bianchi, 2000; Raley, Bianchi, & Wang, 2012), and that the nature and circumstances of the care that mothers and fathers provide differ significantly (Craig, 2006; Raley, et al., 2012). For example, fathers do proportionally more recreational activities with children, spend less time caring for very young children, and do less “solo” parenting than mothers (Raley, et al., 2012; Yeung, Sandberg, Davis‐Kean, & Hofferth, 2001). A separate, smaller body of research looks at the gender imbalance in parents’ feelings while caring for children. This research finds that mothers enjoy childcare slightly less than fathers do; when asked how they feel during childcare, fathers are more likely than mothers to report positive emotions, such as happiness, whereas mothers are more likely to report negative emotions, such as stress and fatigue (Connelly & Kimmel, 2015; Offer, 2014; Roeters & Gracia, 2016). Our study bridges these two bodies of research and argues that a key explanation for differences in mothers’ and fathers’ feelings during childcare is the gendered nature of caregiving activities. To evaluate this hypothesis, we introduce the “care context”—a multidimensional concept that includes the type of childcare activity, when and where it takes place, who is present, and the amount of care—and examine whether it can account for gender differences in parents’ reported feelings, or affect, during childcare.

Affect is a core component of subjective well-being that measures how people evaluate their own lives (Busseri, 2011; Diener, 2000). Affect is related to emotions, moods, and feelings and is distinct from cognitive dimensions of subjective well-being, such as life satisfaction. Momentary affect, our focus, emphasizes affect during specific times or activities, such as childcare (Kahneman, Diener, & Schwarz, 1999). We use data from the American Time Use Survey (ATUS), which is unique in that it collects information on momentary affect during specific activities for a large, nationally representative sample of women and men. Our analysis examines four domains of momentary affect during childcare activities—happiness, stress, tiredness, and meaning—allowing us to assess both positive and negative emotional experiences of caregiving as well as parents’ sense of how meaningful or purposeful the activity is for them.

Our study focuses on the emotions parents experience while actively engaged in caring for their children rather than how they feel spending time with their children in general. Given the growing share of parents’ time spent in active childcare and the persistently gendered nature of family life, it is important to consider how gendered engagement in childcare activities relates to parents’ well-being.

Background

Although family size in the U.S. has declined over the last century, the time and energy demands of parenting have not. In recent decades, the amount of time that parents spend with their children has grown steadily (Bianchi, 2000). An increasing share of that time is now spent in active childcare activities, such as teaching and playing, that enhance children’s learning and development (Bianchi, 2000; Sayer, Bianchi, & Robinson, 2004). These behavioral changes reflect a growing conviction—among many parents, educators, and developmental scientists—that for children to become successful adults, their social and human capital must be cultivated through intensive investments of parental time and resources (Lareau, 2011; Wall, 2010).

The demands of parenting do not fall equally on mothers and fathers, however. Although fathers now spend more time with children than in the past—including active engagement in childcare—mothers continue to spend substantially more time than fathers with children (Bianchi, 2011; Raley, et al., 2012). Moreover, mothers are more likely to reorganize other areas of their lives, such as their work schedules, to facilitate childrearing (Kaufman & Uhlenberg, 2000; Sanchez & Thomson, 1997).

Childcare and Momentary Affect

The demanding nature of childrearing has consequences for parents’ well-being. An emerging area of research focusses on parents’ momentary affect—an aspect of subjective well-being related to moods and emotions at particular moments in time. This research examines, first, how parents’ affect during childcare compares to affect during other activities, and second, how affect during childcare differs for mothers and fathers.

Studies comparing parents’ affect during caregiving to their affect during other activities have generally found that caring for children is emotionally rewarding. For example, a U.S. study based on a convenience sample of middle-class parents determined that childcare activities are more enjoyable and meaningful to parents than other daily activities (Offer, 2014). Research using nationally representative samples of activities from the ATUS has shown that parents report greater happiness and meaning during childcare activities than they do during paid work, housework or leisure (Connelly & Kimmel, 2015; Wang, 2013).¹ In contrast with these findings, data from a convenience sample of employed women in Texas found that childcare was less enjoyable than most other activities (Kahneman, Krueger, Schkade, Schwarz, & Stone, 2004).

Although parents appear to enjoy childcare more than other daily activities, only a few studies have considered affect during childcare activities exclusively and have found that fathers generally report more positive emotions than mothers do. One study based on a convenience sample of White, middle-class U.S. families used a single, continuous measure of affect and found that mothers reported less favorable emotional states during childcare than fathers (Larson, Richards, & Perry-Jenkins, 1994). A more recent study of middle-class parents determined that fathers felt more engaged and less stressed than mothers during physical childcare activities, such as feeding and bathing children (Offer, 2014).

Several recent studies have examined gender differences in parents’ momentary affect using nationally representative ATUS data. For example, Wang (2013) presented descriptive statistics from the 2010 ATUS data showing that mothers were more happy but also more tired and stressed than fathers during childcare activities. Two additional studies used the same data and controlled for sociodemographic characteristics of parents. The first of these studies found that mothers reported significantly higher levels of stress and tiredness than fathers during the most common childcare activities, whereas there were very few differences in mothers’ and fathers’ levels of happiness and meaning (Connelly & Kimmel, 2015). The second study focused on gender differences in stress and meaning and found that mothers reported more stress and slightly less meaning than fathers during childcare activities (Roeters & Gracia 2016).

The Care Context

Taken together, the emerging research on affect during childcare suggests a gender imbalance in the emotional rewards of parenting: mothers are more stressed, more tired, and less happy than fathers when caring for their children. Furthermore, some of these gender differences in affect during childcare are not explained by parents’ sociodemographic characteristics alone. We argue that differences in mothers’ and fathers’ affect could be due to their gendered engagement with children. We introduce the “care context,” a multidimensional concept encompassing the local, transitory factors that differentiate one act of caregiving from the next. The care context encompasses five dimensions that relate to the type of childcare activity, when it took place, where it took place, who was present, and how much care was involved. Several studies have included controls for one or more of these characteristics in their analyses of gender differences in parents’ affect during childcare activities (e.g., Offer, 2014; Roeters & Gracia, 2016). However, we cannot discern from these studies whether and to what extent the overall context of childcare explains gender differences in parents’ affect during childcare activities.

The explanatory power of the care context is supported by two bodies of research: one showing that each of its five dimensions is related to individuals’ affect and another showing that they vary between mothers and fathers. We outline the evidence for each dimension in turn.

The first dimension of the care context relates to the type of childcare. Childcare includes activities from changing diapers to ferrying children to school to playing soccer in the yard. The existing research on parents’ affect indicates that some activities are happier, more stressful or more meaningful than others. For example, parents are happier and less stressed in recreational activities than they are in most other types of childcare (Offer, 2014). In addition, some research suggests that managerial activities, such as researching and planning children’s schedules, may be particularly stressful for parents (Fox, 2009; Walzer, 1996). A separate body of research has documented substantial differences in the activities mothers and fathers undertake in caring for their children. Mothers spend a greater share of their childcare time in routine physical care, such as bathing, feeding, and dressing children (Craig, 2006; Raley, et al., 2012; Sayer, 2005). Mothers also take more responsibility for potentially stressful managerial activities (Fox, 2009; Raley, et al., 2012; Walzer, 1996). Fathers, on the other hand, spend proportionally more of their childcare time in enjoyable supervision and recreational activities, such as playing, reading stories, and sports (Craig, 2006; Raley, et al., 2012; Sayer, 2005; Yeung, et al., 2001).

The second dimension of the care context is when the activity takes place. This includes both the day of the week and the time of day. Affect levels vary throughout the week, with people reporting more positive emotions on weekends than on weekdays (Ryan, Bernstein, & Brown, 2010). Affect levels also fluctuate throughout the day. For example, negative emotions are most common in the morning and early afternoon, and tiredness is most common in the morning and at night (Kahneman, et al., 2004). The timing of mothers’ and fathers’ interactions with their children differ in ways that could impact their affect. Mothers spend more time with children on weekdays and fathers more time on weekends (Sayer, et al., 2004; Yeung, et al., 2001). Differences in parents’ labor force participation (Kramer, Kelly, & McCulloch, 2013) could lead fathers to spend less time doing childcare during typical daytime work hours than mothers do.

The third dimension of the care context is where the activity takes place. Research indicates that caring for children at home is associated with increased fatigue compared to caring for children in public places (Meier, et al., 2016). Conversely, caregiving in public may bring the added stress of scrutiny from others; parents are often subject to judgement and criticism when parenting in public (Trussell & Shaw, 2012). Differences in where mothers and fathers spend their caregiving time could influence their affect during childcare. However, we are not aware of research investigating gender differences in the location of caregiving activities to date.

The fourth dimension of the care context considers who was present during the activity. This dimension consists of the ages of children and whether the parent was engaged in “solo” parenting. Certain stages of child development, such as infancy and toddlerhood, are more demanding and stressful for parents than others (Brazelton, 2013), and childcare time tends to be most gendered when children are young. Mothers spend more of their total childcare time with very young children and fathers with children in middle childhood (Yeung, et al., 2001). Parents report that solo parenting—defined as caring for children without a partner present—is more demanding and less rewarding than time spent caregiving with a partner (Hattery, 2001). Multiple studies of parents’ time use have concluded that fathers spend much less time in solo parenting than mothers (Folbre, Yoon, Finnoff, & Fuligni, 2005; Raley, et al., 2012).

Finally, the fifth dimension of the care context captures two aspects of the amount of childcare performed by parents: the cumulative amount of time spent in childcare on a given day and the duration of the activity. Spending a lot of time caregiving on a single day could lead to increased stress or fatigue. Shorter activity duration could indicate time fragmentation or multitasking, which is related to negative affect and stress (Offer & Schneider, 2011). Research indicates that mothers spend more time caring for their children than fathers (Bianchi, 2011; Raley, et al., 2012), and mothers also experience higher levels of leisure time fragmentation (Mattingly & Bianchi, 2003). Thus, differences in mothers’ and fathers’ time spent in caregiving and activity duration could contribute to gender differences in their reported momentary affect.

We have introduced the concept of the care context and reviewed research supporting the idea that the context of childcare influences individuals’ affect and differs by parental gender. The goal of our analysis is to use activity-level data from the ATUS to document differences in mothers’ and fathers’ reports of happiness, stress, tiredness, and meaning during childcare activities and assess whether the five dimensions of the care context explain these differences.

Method

Data

We used pooled activity-level data from the 2010, 2012, and 2013 American Time Use Survey, collected by the U.S. Census Bureau on behalf of the Bureau of Labor Statistics (Hofferth, Flood, & Sobek, 2013). ATUS respondents are aged 15 years or older and are randomly drawn from households that have completed eight rounds of the Current Population Survey (CPS). Respondents are representative of the civilian, noninstitutionalized, U.S. population living in households. ATUS is a telephone survey. Respondents report each activity they engaged in during a 24-hour window from 4 a.m. the previous day until 4 a.m. on the day they make their report. Reports include when the activity began, how long it lasted, and where it took place. Where applicable, respondents also report who else was present during the activity. These reports produce a detailed record of how each respondent spent their time on the diary day. Data on many household and individual characteristics are available through the CPS.

Occasionally, topic-specific modules are included in the ATUS interview. This study uses information from the Well-Being Module, fielded in 2010, 2012, and 2013. As part of this module, respondents reported how they felt during three random activities from the previous day. We base our work on these activities, and thus our unit of analysis is a “person-day-activity.”

Our analytic sample included activities classified as primary childcare and accompanied by measures of momentary affect. Primary childcare is defined as childcare that is the respondent’s primary activity and includes activities where a parent is directly engaged in caregiving or other activities that promote a child’s well-being. Primary childcare is different than secondary childcare, which is defined in the ATUS as caring for a child under 13 while engaged in another primary activity, such as cooking.² It is also different than general time spent with a child, which occurs when a child is present during a parent’s main activity but is not being cared for as secondary childcare. We did not include these non-primary childcare activities in the analysis for two reasons. First, affect during these activities could be related to the main activity rather than the respondent’s interactions with their child. Second, information on some dimensions of the care context, such as the type and duration of care, was not available for these activities.

Our study was concerned with momentary affect during primary childcare activities directed at parents’ own children aged 17 and under (own children in the ATUS includes biological children, adopted children, and stepchildren). In constructing the analytic sample, we began with 4,391 primary childcare activities where respondents reported that an own child aged 17 and under was present, whether the child was coresident or not. We added n=158 additional primary childcare activities where no own child was reported present if we considered that the activity could plausibly have taken place without a child.³ For example, activities such as attending school events or waiting for children were included if they were undertaken by a parent with an own child age 17 or under. This gave us 4,549 cases. We then excluded 49 cases (1.1%) due to missing data on household income for the respondent, leaving us with 4,500 cases. There were no missing values for the remaining independent variables. We excluded between two and 14 activities (from 0.04% to 0.3%) from each of the analyses because of missing data on the dependent variables, the four measures of affect. The final analytic samples included 4,486 to 4,498 cases.

Response rates for the ATUS in 2010, 2012, and 2013 were 56.9%, 53.2%, and 49.9%, respectively. Abraham and colleagues (2006) determined that nonresponse in the ATUS has minimal effect on estimates of time use, and that this effect tends to be centered among the unemployed and those who are weakly attached to households. In line with their suggestions, we used person-level weights for all analyses to take account of nonresponse (Abraham, Maitland, & Bianchi, 2006). We also used activity-level weights to account for each activity’s probability of selection into the Well-Being Module.

Variables

Our dependent variables were four measures of affect during childcare. For each activity that was randomly selected for inclusion in the Well-Being Module, respondents were asked to report on six domains of affect pertaining to that activity. Research indicates that assessments of affect that are tied to specific activities, such as those in the ATUS, are generally more reliable than overall measures of life satisfaction and well-being (Kahneman, et al., 2004). The domains used in this study relate to how much happiness, tiredness, and stress respondents felt during the activity and how meaningful they found the activity to be.⁴ Whereas the first three domains relate to emotions, the fourth domain, meaning, captures a sense of how purposeful the activity was.

In the Well-Being Module respondents retrospectively reported affect levels for activities in the previous day. This method is sometimes called the day reconstruction method (DRM).⁵ Respondents reported how intensely they felt each emotion using a seven-point scale, from 0 – 6, where higher values indicated greater feelings of happiness, stress, tiredness, or meaning.

The key independent variable in the analysis is the gender of the parent performing the childcare activity (mother =1; father=0).⁶

We included person-level and household-level characteristics that we expected to be associated with both parents’ gender and their affect during childcare, including measures for age, race and ethnicity, education, household income, single-parent status, work hours, and survey year.

Earlier research indicates that age, race and ethnicity, education, and household income are associated with subjective well-being (Charles, Reynolds, & Gatz, 2001; Kahneman & Deaton, 2010; Mroczek & Kolarz, 1998; Yang, 2008). These characteristics are also likely to vary between mothers and fathers. Respondents’ age and education were measured in years. Respondents’ household income was measured in the ATUS on a one (less than $5000) to 16 (more than $150,000) scale. Income data were collected in the final CPS interview, which took place two to five months before the ATUS diary day. We coded respondents’ race/ethnicity into four categories: non-Hispanic White; non-Hispanic Black; Hispanic; and other, generally comprising of Asian, Native American, and mixed-race respondents.

With respect to partnership status, single parents have fewer social supports, more psychological strain, and are more likely to suffer from depression than partnered parents (Amato, 1993; Cairney, Boyle, Offord, & Racine, 2003). In addition, previous research using ATUS data concluded that single mothers’ experiences of childcare activities were less rewarding than those of partnered mothers (Meier, et al., 2016). The dummy variable, single parent, indicated respondents who did not have a coresident spouse or partner. Working for pay is associated with how parents feel about the time they spend with their children (Johnston & Swanson, 2007; Milkie, Mattingly, Nomaguchi, Bianchi, & Robinson, 2004). We used a continuous variable, usual weekly work hours, to account for respondents’ work hours. A small proportion of activities (n = 138) were performed by respondents who reported that their work hours varied from week to week. A follow-up question asked these respondents whether they usually work full or part time, i.e., more or less than 35 hours per week. In these cases, we recoded the variable to the mean work hours for respondents who work full or part time, as appropriate. Mean work hours for full-time workers differed significantly by gender, so for those activities performed by respondents who work full time and whose work hours vary, we used gender-differentiated means. Finally, we included dummies to control for survey year.

The care context was key to our analysis, and we included measures for five dimensions. To measure the first dimension of the care context we created a categorical variable, activity type, to group childcare activities into four categories. Three of these categories were developed by Raley and colleagues (2012) in their study of childcare using the ATUS data. We added a fourth for education-related childcare activities: 1) Physical childcare consisted of basic caregiving activities that parents undertake to meet their children’s physical needs, including tasks such as feeding, bathing, and dressing children, and putting children to bed. 2) Recreational childcare consisted of activities such as playing with children (including sports), talking with children, doing arts and crafts with children, reading to or with children, and attending children’s events other than those related to their education. 3) Educational childcare activities were those activities that are directly related to children’s formal education, including helping with homework, attending school conferences, and homeschooling children (but not reading or play). 4) Managerial childcare activities involved parental activities such as organizing and planning for children as well as picking up and dropping off children from school or other events.

To measure the second dimension of the care context—when the activity took place—we used four variables. The first of these, time of day, was a categorical variable that distinguished between activities that took place in the morning (activities that began between 6 a.m. and 10 a.m.), midday (activities that began between 10 a.m. and 2 a.m.), afternoon (activities that began between 2 a.m. and 6 a.m.), evening (activities that began between 6 p.m. and 10 p.m.), and night (activities that began between 10 p.m. and 6 p.m.). Three additional dichotomous variables accounted for whether the diary day was a weekend day, a public holiday, or took place in the summer, when children are likely to be on vacation from school.

To measure the third dimension of the care context—where the activity took place—we used a single dummy variable, activity at home, to distinguish between activities that took place in the respondent’s home or yard and those that took place elsewhere.

To measure the fourth dimension of the care context—who was present during the activity—we used a series of dummy variables. Nonresident child accounted for childcare activities that took place with a respondent’s own nonresident child (recall that ATUS defines own children as biological children, adopted children, and stepchildren). Solo parenting accounted for childcare activities where no coresident spouse or partner was present. In addition, we constructed a set of dummy variables to account for the presence of the respondent’s own children of various ages: infant (aged under 1), toddler (aged 1–2), preschooler (aged 3–5), older child (aged 6–11), and adolescent (aged 12 –17). Children were grouped into age categories because parents’ experience of childcare was expected to differ according to the developmental stage of the child(ren).

Finally, to measure the fifth dimension of the care context—how much care was involved—we used two different measures. First, activity duration measured the duration of the activity in minutes. Second, to account for the possibility that parents’ affect could vary by the amount of time already spent in childcare in the day, an additional variable was constructed to measure how much time the respondent had spent doing childcare activities between 4 a.m. (when the diary day begins) and the start of the focal activity. This variable, called earlier childcare, is measured in minutes. We confirmed that these two measures were not highly correlated.

Analysis

First, we examined descriptive statistics by parental gender. We were particularly interested in how the four domains of affect and the five dimensions of the care context differed for mothers’ and fathers’ childcare activities. Second, we estimated OLS regression models to assess how the four domains of parents’ affect were associated with parental gender in bivariate models and models including sociodemographic characteristics. The regression equation is:

$${\text{Y}}_{\text{ij}}={\beta }_0+{\beta }_1{\text{MOTHER}}_{\text{i}}+{\beta }_2{\text{X}}_{\text{ij}}+{\epsilon }_{\text{ij}}$$ (1)

where Y_ij is a measure of affect for person i during activity j, MOTHER_i indicates if the activity was performed by a mother (compared to a father), and X_ij is a set of sociodemographic characteristics.

Third, to assess whether the care context explained the relationship between gender and parents’ affect during childcare activities, we expanded the regression models to incorporate measures of the care context. The expanded regression equation is:

$${\text{Y}}_{\text{ij}}={\beta }_0+{\beta }_1{\text{MOTHER}}_{\text{i}}+{\beta }_2{\text{X}}_{\text{ij}}+{\beta }_3{\text{CARE}}_{\text{ij}}+{\epsilon }_{\text{ij}}$$ (2)

where terms were similar to equation (1) above with the addition of CARE_ij, a set of activity-level characteristics measuring the care context for person i during activity j. We used simultaneous estimation and Wald tests to determine whether the coefficients for gender changed significantly when measures of the care context were introduced.

Some ATUS respondents had more than one childcare activity randomly selected for inclusion in the Well-Being Module. As a result, the 4,500 activities in our sample were contributed by 3,683 respondents, of which 2,936 respondents contributed a single activity, 677 respondents contributed two activities, and 70 respondents contributed three activities. To account for the fact that some respondents contributed more than one observation to the sample, we adjusted the standard errors for clustering at the person-level using the cluster command in Stata 14.⁷ We also used person-level and activity-level weights.

Following the previous literature examining affect using the ATUS activity-level data (e.g., Connelly & Kimmel, 2015; Musick, et al, 2016; Offer, 2014; Roeters & Gracia, 2016), we treat our dependent variables as though they were continuous. However, the affect measures are based on a scale of 0 to 6 and thus are bounded, ordinal variables. Given the concerns associated with using OLS for this type of data, we also estimated all of our models using ordered logit and ordered probit models. In all cases, the results were substantively similar to those produced using OLS regression in terms of the direction of the associations and significance levels. We present OLS models for their ease of interpretation.

Results

Descriptive Statistics

Table 1 reports weighted means and standard deviations for all analysis variables and indicates where there were significant differences by respondents’ gender. Overall, parents reported high levels of meaning and happiness, moderate levels of tiredness, and low levels of stress during childcare activities. (See Appendix A for distributions of the affect variables.) However, mothers and father reported significantly different levels of affect in three of the four domains: mothers reported less happiness, more stress and more tiredness than fathers. It is notable that mothers and fathers did not differ significantly in the level of meaning they attributed to their childcare activities. These results are broadly consistent with the existing literature on gender differences in parents’ affect during childcare (Connelly & Kimmel, 2015; Larson, et al., 1994; Offer, 2014; Roeters & Gracia, 2016; Wang, 2013).

Table 1.

Weighted descriptive statistics for the study of mothers’ activities (n = 3, 074) and fathers’ activities (n = 1,426)

	Mothers’ activities			Fathers’ activities
Variable	M.	S.D.		M.		S.D.
DEPENDENT VARIABLES
During activity respondent felt:
Happy (scale 0–6)	4.71	1.39		4.90	***	1.33
Stressed (scale 0–6)	1.48	1.73		1.16	***	1.52
Tired (scale 0–6)	2.76	1.94		2.37	***	1.84
Activity was meaningful (scale 0–6)	5.18	1.41		5.22		1.31
INDEPENDENT VARIABLES
Age	34.42	7.64		37.73	***	7.94
Race/ethnicity:
White, non-Hispanic	.65	-		.67		-
Black, non-Hispanic	.09	-		.08		-
Hispanic	.19	-		.17		-
Other	.07	-		.08		-
Household income (1–16)	11.15	4.16		12.18	***	3.59
Education (years)	14.25	3.07		14.41		3.11
Single parent	.21	-		.06	***	-
Usual weekly work hours	19.37	19.09		37.05	***	19.47
Year of survey:
2010	.34	-		.33		-
2012	.34	-		.32		-
2013	.35	-		.32		-
Type of childcare activity:
Physical	.56	-		.53		-
Recreational	.21	-		.26	**	-
Educational	.05	-		.06		-
Managerial	.18	-		.15	†	-
Time of day
Morning (6am-10am)	.26	-		.23		-
Midday (10am-2pm)	.14	-		.09	***	-
Afternoon (2pm-6pm)	.23	-		.23		-
Evening (6pm-10pm)	.31	-		.40	***	-
Night (10pm-6am)	.06	-		.05		-
Weekend	.19	-		.24	***	-
Public holiday	.01	-		.01		-
Summer	.20	-		.19		-
Activity at home	.78	-		.79		-
Nonresident child	.00	-		.03	**	-
Solo parenting	.85	-		.67	***	-
Infant present (under 1)	.17	-		.13	**	-
Toddler present (1–2)	.28	-		.29		-
Preschool child present (3–5)	.34	-		.35		-
Older child present (6–11)	.39	-		.43	*	-
Adolescent present (12 –17)	.16	-		.14		-
Duration (minutes)	31.73	38.32		39.10	***	51.57
Earlier childcare (minutes)	90.63	104.25		64.22	***	90.07

Note. Asterisks signify significant differences between mothers and fathers within waves. We present weighted means for continuous variables and weighted percentages for categorical variables. We use Wald tests to test differences between mothers’ and fathers’ activities.

^†p < 0.1,

^*p < 0.05,

^**p < 0.005,

^***p < 0.001 (two-tailed tests)

Mothers and fathers who contributed activities to the sample differed significantly on several demographic characteristics. Mothers were younger, had lower household income, and worked fewer hours than fathers did. Mothers were also more likely than fathers to be single parents.

Supporting the view that parental engagement with children is gendered, most measures of the care context were significantly different for mothers’ and fathers’ activities (all descriptive findings discussed here were significant at the 0.05 level). First, the types of childcare activities mothers and fathers undertook varied: a greater share of fathers’ activities were recreational in nature (26% compared to 21% of mothers’ activities). Second, mothers’ and fathers’ childcare activities took place at different times of day: mothers’ activities were more likely to occur in the middle of the day, whereas fathers’ activities were more likely to occur in the evening. Fathers’ activities were also more likely to occur on weekend days (25% compared to 19% of mothers’ activities).

Third, who was present during childcare activities varied by parental gender. Fathers’ childcare activities were more likely to include an older child (aged 6–11) or a nonresident child, whereas mothers’ activities were more likely to take place with an infant. Fathers’ activities were also more likely to take place with a partner: just 61% of fathers’ activities were classed as solo parenting compared to 85% of mothers’ activities. Finally, mothers and fathers performed substantially different amounts of childcare. The sample included 3,074 childcare activities performed by mothers and 1,426 activities performed by fathers. Given that the ATUS is a nationally representative sample, this indicates that mothers in the U.S. undertake substantially more primary childcare activities than fathers do. In addition, fathers who contributed activities to our sample had spent less cumulative time in childcare activities at the beginning of the focal activity than mothers had (64 minutes for fathers’ activities compared to 90 minutes for mothers’ activities). On average, fathers’ activities were longer than mothers’ activities (39 minutes compared to 32 minutes). This finding, in conjunction with the finding that mothers had higher amounts of cumulative childcare, suggests that mothers’ childcare time is more dispersed throughout the day than fathers’ and may indicate fragmentation of mothers’ childcare time.

Regression Results

Table 2 displays the regression results for all four domains of affect: happiness, stress, tiredness, and meaning. The bivariate results in Models 1 reflect the descriptive findings. Models 2 include sociodemographic characteristics. Models 3 include the measures of the five dimensions of the care context in addition to sociodemographic characteristics.⁸

Table 2.

Parents’ subjective assessment of childcare activities by parental gender, selected sociodemographic control variables, and contextual variables

	Happiness									Stress
Variable	(1)			(2)			(3)			(1)			(2)			(3)
Mother	−.190	**	(.063)	−.176	*	(.077)	−.110		(.073)	.324	***	(.074)	.334	***	(.086)	.277	**	(.085)
Age				−.013	**	(.004)	−.004		(.005)				.017	**	(.005)	.011		(.006)
Race/ethnicitya
Black, non-Hispanic				.253	*	(.098)	.288	**	(.098)				−.374	**	(.130)	−.416	**	(.128)
Hispanic				.121		(.102)	.188	†	(.100)				.004		(.125)	−.051		(.126)
Other				.195	†	(.104)	.174	†	(.102)				−.186		(.130)	−.176		(.128)
Household income				−.002		(.011)	−.007		(.011)				−.023	†	(.014)	−.018		(.014)
Education				−.040	**	(.012)	−.043	***	(.012)				.038	*	(.016)	.043	**	(.016)
Single parent				−.226	*	(.096)	−.189	†	(.096)				.226	†	(.131)	.201		(.136)
Usual weekly work hours				.002		(.002)	.004	*	(.002)				.0002		(.002)	−.001		(.002)
Yearb
2012				.198	**	(.072)	.211	**	(.070)				−.002		(.087)	−.019		(.087)
2013				.166	*	(.075)	.175	*	(.072)				.113		(.094)	.093		(.093)
Type of childcare activityc
Recreational							.623	***	(.060)							−.522	***	(.081)
Educational							−.335	**	(.129)							.205		(.153)
Managerial							.029		(.112)							.019		(.150)
Time of dayd
Midday (10am-2pm)							.310	**	(.100)							−.263	*	(.128)
Afternoon (2pm-6pm)							.271	**	(.080)							−.250	*	(.103)
Evening (6pm-10pm)							.181	*	(.076)							−.150		(.102)
Night (10pm-6am)							−.092		(.148)							.123		(.173)
Weekend							.001		(.050)							−.010		(.071)
Public holiday							−.036		(.269)							−.018		(.233)
Summer							−.076		(.070)							−.016		(.079)
Activity at home							.153	†	(.092)							−.031		(.133)
Nonresident child							.152		(.309)							−.247		(.280)
Solo parenting							−.006		(.059)							.073		(.080)
Infant present (under 1)							.170	*	(.084)							.057		(.112)
Toddler present (1–2)							−.009		(.072)							.098		(.095)
Preschool child present (3–5)							−.088		(.060)							.087		(.082)
Older child present (6–11)							−.113	†	(.065)							.146	†	(.085)
Adolescent present (12–17)							−.144		(.094)							.333	**	(.119)
Duration (minutes)							.001		(.001)							.001		(.001)
Earlier childcare (minutes)							.000		(.000)							.000		(.000)
Constant	4.903	***	(.050)	5.775	***	(.205)	5.059	***	(.271)	1.161	***	(.055)	.237		(.279)	.446		(.332)
R2	.004			.029			.098			.008			.023			.053
n =	4,493			4,493			4,493			4,498			4,498			4,498
	Tiredness									Meaning
Variable	(1)			(2)			(3)			(1)			(2)			(3)
Mother	.391	***	(.082)	.483	***	(.092)	.493	***	(.092)	−.042		(.067)	−.059		(.077)	−.021		(.076)
Age				−.015	**	(.005)	−.003		(.006)				−.005		(.007)	.005		(.008)
Race/ethnicitya
Black, non-Hispanic				−.301	*	(.160)	−.324	*	(.152)				.062		(.114)	.094		(.121)
Hispanic				−.371	**	(.125)	−.291	*	(.121)				−.162		(.144)	−.080		(.135)
Other				−.503	**	(.153)	−.512	**	(.152)				−.070		(.113)	−.105		(.111)
Household income				−.030	*	(.014)	−.026	†	(.013)				−.025		(.017)	−.027		(.016)
Education				.024		(.017)	.022		(.017)				.000		(.016)	−.003		(.016)
Single parent				−.041		(.123)	.058		(.123)				.066		(.080)	.126		(.084)
Usual weekly work hours				.009	***	(.002)	.005	*	(.002)				.002		(.002)	.003		(.002)
Yearb
2012				.124		(.095)	.097		(.091)				.125		(.087)	.119		(.082)
2013				.148		(.097)	.097		(.093)				.111		(.095)	.095		(.086)
Type of childcare activityc
Recreational							−.306	**	(.090)							.419	***	(.059)
Educational							−.327	†	(.168)							.173		(.118)
Managerial							−.098		(.142)							−.315	*	(.125)
Time of dayd
Midday (10am-2pm)							−.148		(.138)							.083		(.118)
Afternoon (2pm-6pm)							.336	**	(.108)							−.071		(.104)
Evening (6pm-10pm)							1.159	***	(.103)							−.004		(.101)
Night (10pm-6am)							1.359	***	(.162)							.141		(.142)
Weekend							−.072		(.077)							−.130	*	(.054)
Public holiday							.239		(.209)							−.347		(.268)
Summer							−.054		(.091)							−.027		(.072)
Activity at home							.235	†	(.122)							−.069		(.086)
Nonresident child							−1.031	*	(.431)							−.123		(.290)
Solo parenting							−.044		(.090)							−.009		(.063)
Infant present (under 1)							.306	*	(.129)							.309	**	(.110)
Toddler present (1–2)							−.028		(.101)							.087		(.099)
Preschool child present (3–5)							−.059		(.087)							−.190	**	(.068)
Older child present (6–11)							−.055		(.086)							−.025		(.082)
Adolescent present (12–17)							.021		(.120)							−.093		(.104)
Duration (minutes)							.000		(.001)							.001	*	(.000)
Earlier childcare (minutes)							.000		(.000)							.001	†	(.000)
Constant	2.372	***	(.063)	2.680	***	(.295)	1.738	***	(.364)	5.221	***	(.048)	5.587	***	(.240)	5.210	***	(.362)
R2	.009			.034			.138			.000			.011			.061
n =	4,496			4,496			4,496			4,486			4,486			4,486

Note. Unstandardized regression coefficients reported. All models include person- and activity-level weights. Results are clustered at the person level.

^aThe reference category is White, non-Hispanic

^bThe reference category is 2010

^cThe reference category is physical childcare

^dThe reference category is morning (6am-10am)

^†p<0.1,

^*p<0.05,

^**p<0.01,

^***p<0.001 (two-tailed tests).

Happiness.

In the bivariate regression in Model 1, mothers reported significantly less happiness than fathers did during childcare activities. The magnitude of the gender difference was reasonably modest in size; on average, mothers’ happiness score was 0.190 points lower on the scale than fathers’. The gender imbalance in happiness remained significant when sociodemographic characteristics were accounted for in Model 2, though the coefficient was somewhat reduced in size; on average mothers scored 0.176 lower than fathers in this model. Several sociodemographic characteristics were significantly related to happiness at the 0.05 level in Model 2. Increases in parental age and education were associated with less happiness. Single parents reported less happiness, and Black parents reported more happiness than White parents.

When measures of the care context were included in the model for happiness in Model 3, the coefficient on parental gender was smaller, compared to Model 2, and no longer significant. Additional analyses determined that this result was not attributable to any one dimension of the care context but rather to a combination of dimensions. Using simultaneous estimation and a Wald test, we determined that the difference in coefficients on parental gender across Models 2 and 3 was statistically significant. Thus, the care context appears to fully explain the significant difference in mothers’ and fathers’ reported happiness during childcare.

Several domains of the care context were significantly associated with happiness. Parents reported more happiness during recreational activities and less happiness during educational activities, compared to physical childcare. Activities involving infants were associated with more happiness. The time of day was also important: parents reported more happiness during activities that took place in middle of the day, afternoon, and evening compared to those that took place in the morning. Results for the sociodemographic variables were generally similar in Model 3 to those reported for Model 2 with the exception of parental age, which was no longer significant, and of parents’ work hours, which was positively associated with happiness in Model 3.

Stress.

In the bivariate regression in Model 1, mothers reported significantly more stress than fathers did during childcare activities. The magnitude of the gender difference was modest in size; on average, mothers’ stress score was 0.324 points higher on the scale than fathers’. The gender imbalance in stress remained significant when sociodemographic characteristics were accounted for in Model 2, and the size of the coefficient was virtually unchanged. Two sociodemographic characteristics were significantly related to stress at the 0.05 level in Model 2: increases in parental age and education were both associated with more stress during childcare.

When measures of the care context were included in the model for stress in Model 3, the coefficient on parental gender reduced in size but remained significant. In this model, mothers’ score was 0.277 points higher than fathers’. Additional testing determined that the coefficient in Model 3 was significantly different to that in Model 2. Thus, the difference in reported stress was partially explained by the care context.

Several dimensions of the care context were significantly related to stress during childcare. Parents reported less stress during recreational activities, compared to physical childcare activities. Parents also reported more stress during activities with adolescents (ages 12–17). Activities that took place in the middle of the day and afternoon were less stressful for parents than those that took place in the morning. Results for the sociodemographic variables in Model 3 were similar to those reported for Model 2 with the exception that parental age was no longer significantly associated with stress.

Tiredness.

In the bivariate regression in Model 1, mothers reported significantly more tiredness than fathers during childcare activities. The magnitude of the gender difference was modest in size; on average, mothers’ tiredness score was 0.391 points higher on the scale than fathers’. The gender imbalance in tiredness remained significant and the coefficient on parental gender increased in size when sociodemographic characteristics were accounted for in Model 2. In this model, mothers’ score was 0.483 points higher than fathers’. Multiple sociodemographic characteristics were significantly related to tiredness in Model 2. Increases in age and household income were associated with less tiredness during childcare activities. White parents reported more tiredness than parents of all other racial and ethnic groups. Work hours were associated with increased parental reports of tiredness, although additional analyses (not shown) indicated that the association between work hours and tiredness was only significant for mothers. This is likely a consequence of the “double burden” of paid and unpaid labor that is common among working mothers (Frisco & Williams, 2003; Hochschild & Machung, 2012; Johnston & Swanson, 2007).

When measures of the care context were included in the model for tiredness in Model 3, the coefficient on parental gender remained significant and virtually unchanged in size. Additional testing determined that the coefficient in Model 3 was not significantly different to that in Model 2. Nevertheless, multiple measures of the care context were significantly associated with parents’ reported tiredness. Recreational activities were associated with less tiredness than physical childcare activities. Parents reported less tiredness during activities that involved nonresident children and more tiredness during activities that involved infants. Not surprisingly, the time of day was associated with tiredness; parents reported more tiredness during childcare activities that took place in the afternoon, in the evening, and at night than they did in the mornings. Results for the sociodemographic variables were generally similar in Model 3 to those reported for Model 2 with the exception of parental age, which was no longer significant.

Meaning.

The level of meaning that parents reported did not differ significantly by parents’ gender under any model specification. No sociodemographic characteristics were associated with parents’ reports of meaning. Conversely, several aspects of the care context were significantly associated with meaning in Model 3. Parents rated recreational childcare activities more meaningful and managerial activities less meaningful than physical care activities. Activities were more meaningful when an infant was present and less meaningful when a preschooler (age 3–5) was present. Parents reported that weekend activities were less meaningful than activities that occurred on weekdays. Longer activities were associated with more meaning, although the size of the coefficient was very small.

Discussion

This study used data from the American Time Use Survey to examine differences in mothers’ and fathers’ momentary affect, including feelings of happiness, stress, tiredness, and meaning, during childcare activities. Our main goal was to explain the gender imbalance in parents’ affect during this time. As such, we introduced the concept of the “care context,” which refers to five dimensions of parents’ engagement in childcare, including the type of childcare activity, when and where it takes place, who is present, and the amount of care involved. Previous analyses of gender differences in parents’ affect have incorporated limited aspects of the care context as controls. We took a more comprehensive approach and tested whether these multiple dimensions of parents’ gendered engagement with children accounted for differences in mothers’ and fathers’ feelings during childcare activities.

Our study deepens understanding of gender and parents’ affect during childcare with several important findings. First, we found that parents enjoy childcare and report high levels of meaning and happiness during childcare activities. However, there is an imbalance in the emotional rewards of caring for children. Mothers are less happy, more stressed and more tired than fathers, but mothers and fathers do not differ in their reports of how meaningful childcare activities are.

Second, our analysis of how the care context varies by gender demonstrated that there are substantial differences in the nature of the childcare activities that mothers and fathers perform. This supports the view that parental engagement with children is highly gendered. We observed significant differences in mothers’ and fathers’ childcare activities across four of the five dimensions of the care context. Fathers’ childcare activities are more likely to be recreational, to take place in the evening or on weekends, and to include a partner, a non-resident child and/or children in middle childhood. By contrast, mothers’ childcare activities are more likely to be managerial, to take place in the middle of the day or on a weekday, and to include an infant. Although mothers’ childcare activities are typically shorter than fathers’ activities, mothers tend to have spent more cumulative time in childcare each day.

Third, models that included measures of the care context revealed that the emotional experience of caring for children is shaped by multiple contextual factors beyond individuals’ sociodemographic characteristics. Indeed, all five dimensions of the care context were associated with affect. Some contexts of care are consistently associated with improved affect across a range of domains. For example, recreational activities are associated with more happiness, less stress, less tiredness, and more meaning than other types of childcare activities. Other contexts of care are associated with positive affect in some domains but negative affect in others. For example, childcare activities that include an infant are associated with more happiness and more meaning, but also more tiredness. This suggests that feelings of happiness, stress, tiredness, and meaning are distinct and confirms our decision to consider the four domains of affect separately in our analysis.

Fourth, our major aim was to examine the role of the care context, and we found that variation in the context of care explains some—but not all—of the gender imbalance in parents’ affect during childcare activities. In particular, we found that the care context fully accounts for parental gender differences in happiness, partially explains gender differences in stress, and does little to explain gender differences in tiredness.

With respect to our analyses of how the care context relates to gender differences in each domain of affect, we found that measures of the care context appeared to fully explain fathers’ greater happiness during childcare. These results imply that happiness—particularly when caring for children—is a relatively transitory emotion that changes from one context to the next. In addition, our results suggest that the contexts in which fathers care for children increase happiness compared to the contexts in which mothers care for children. This suggests that fathers do not inherently enjoy caring for children more than mothers do, but their activities tend to be concentrated in the most enjoyable contexts of care.

In our analysis of parental stress, dimensions of the care context partially explained mothers’ greater stress during childcare activities, suggesting that fathers’ childcare activities take place in less stressful contexts of care. The remaining unexplained gender imbalance in stress during childcare could result from unobserved contextual factors. For example, mothers might tend to take over when children are tired, unwell, or otherwise upset, and these times are particularly taxing for parents. Consistent with this explanation, qualitative research has shown that parents’ perceptions of time spent with their children are strongly influenced by children’s moods (Shaw, 2008). Unfortunately, the ATUS data do not include detailed information on children’s moods and behaviors during childcare activities. An alternative explanation for the remaining gender imbalance in stress is that mothers are generally more stressed than fathers regardless of the activity. Support for this explanation in the existing literature is mixed. On the one hand, research has reported that mothers report higher feelings of stress than fathers in a small number of non-childcare activities (Connelly & Kimmel, 2015), and that mothers report slightly more stress than fathers during childfree time (Musick, et al., 2016). On the other hand, a study in which half of participants were parents found no differences in men’s and women’s reports of stress at work (Damaske, et al., 2014), and another study on parents’ affect uncovered no gender differences in parents’ reported stress during leisure time (Offer, 2016).

The care context did not explain any of the observed gender difference in parents’ tiredness during childcare activities. It is likely that mothers are generally more tired than fathers and that reported tiredness is not specific to any particular activity. This interpretation is consistent with research that finds that mothers report more tiredness than fathers during many activities not involving childcare—such working, cooking, and eating—and during childfree time (Connelly & Kimmel, 2015; Musick, et al., 2016; Wang, 2013). Thus, tiredness appears to accumulate and spill over from one activity to the next. Research by Musick and colleagues (2016) identifies deficits in mothers’ overall sleep and leisure time as factors that contribute to this accumulation.

The ATUS is the only nationally representative data that allows for detailed exploration of the associations between parents’ momentary affect and specific childcare activities. However, there are limitations to the ATUS data that affected our analyses. First, we could not examine the importance of the care context in determining gender differences in affect during secondary childcare activities or other non-childcare activities when a child is present (Allard, Bianchi, Stewart, & Wright, 2007). As noted, the ATUS does not contain information on all dimensions of the care context for secondary childcare and non-childcare activities. Other research has revealed important gender differences in parents’ experiences of secondary childcare and other activities undertaken when children are present (Connelly & Kimmel, 2015; Musick, et al., 2016). More detailed measures of the care context of secondary childcare and non-childcare activities would provide valuable insights into gendered caregiving behaviors, parents’ experiences of multitasking, and gender imbalances in the emotional rewards of parenting.

A second limitation relates to the sampling procedures used in the Well-Being Module. Data on each ATUS respondent’s affect was collected for three randomly selected activities on the diary day. To form our analytic sample, we selected all the childcare activities in the Well-Being Module that were directed at respondents’ own children. As such, our data set was not a random sample of mothers and fathers, so we cannot infer how the average mother or father experiences childcare activities. Our unit of analysis was the person-day-activity, so we can infer, instead, how mothers and fathers feel during the average childcare activity. We also note that the design of our study results in only one childcare activity per person for the majority of respondents, which leads to little within-person variation and makes it difficult to distinguish within-person and between-person variations in affect. Although the results apply to the person-day-activity, we think they can and should inform future analyses of person-level patterns. Importantly, this limitation to inference is not unique to our analyses, but common to other analyses of affect and parenting behaviors using the ATUS (e.g., Connelly & Kimmel, 2015; Roeters & Gracia, 2016).

Third, although we consider endogeneity as it relates to the context of care activities, noted above, it is possible that we did not account for important individual-level traits that are direct predictors affect. For example, respondents with mental health conditions could be less likely to work or be in a partnership and also report lower levels of happiness. Thus, the observed associations we uncovered between work, single parenthood, and happiness would be spurious. Unfortunately, we were limited in the availability of detailed person-level information in the ATUS and associated CPS data. Ideally, future studies will include more comprehensive individual- and activity-level information and incorporate longitudinal or quasi-experimental designs to better isolate the direct determinants of affect and time use.

Finally, our analysis does not account for various complexities of parents and parenting experiences, particularly those related to race, ethnicity, and class. Previous research has demonstrated that there are considerable race and class differences in both parents’ and children’s time use (Hofferth, et al., 2013; Kalil, Ryan, & Corey, 2012), so we included these sociodemographic characteristics as covariates in our analyses rather than treating all parents as a homogenous group. It is clearly important to undertake a more detailed, intersectional analysis of variation in parents’ affect during childcare activities across race, ethnicity, class, and gender, in order to shed light on varied parenting experiences and their relationship to affect.

In summary, the care context helps to capture the myriad ways that one act of caregiving can differ from the next and to explain gender differences in parents’ affect during childcare. Nevertheless, the question remains as to why mothers and fathers undertake childcare activities in different contexts of care. Numerous structural and cultural factors are likely to contribute. For example, mothers are systematically disadvantaged in the labor market—in terms of pay, perceived competence, and benefits—relative to fathers (Blair-Loy, 2009; Budig & England, 2001; Correll & Benard, 2007). These realities may prompt women to concentrate on caregiving and men to concentrate on breadwinning when they become parents (Fox, 2009; Kotila, Schoppe‐Sullivan, & Kamp Dush, 2013; Sanchez & Thomson, 1997). Furthermore, the motivations, meanings, and rewards that men and women attach to parenting can influence the amount and types of childcare that they do. Individuals reaffirm their ideologies and identities through domestic labor, including caregiving (Berk, 1985; Bittman, England, Sayer, Folbre, & Matheson, 2003; Schneider, 2011). For many mothers, being the primary parent remains central to their parenting ideologies and to their sense of identity as women and mothers (Gaunt, 2008; Hays, 1996). For many men, successful parenthood may be less tightly tied to being a primary parent, and more centered on taking significant responsibility for children’s financial needs, and other supports (Shirani & Henwood, 2011; Townsend, 2010; Wall & Arnold, 2007). Factors such as these motivate and constrain parents’ engagement with their children contributing to distinct childcare involvements. Future efforts to promote optimal caregiving experiences for all parents will need to attend to building supportive contexts of care and to the gender systems that shape these contexts and give them meaning.

Appendix A. Figures

Figure 1.

Distribution of reported affect scores (scale 0–6) during childcare activities with higher scores indicating more happiness, stress, tiredness, or meaning