Abstract
Discretionary time deficits are a result of a disproportionate amount of time spent in paid and unpaid work. Such time deficits are related to negative health outcomes. Discretionary time deficits may be influenced by family structure, because other adults can provide additional economic and time resources. Research to date does not examine differences in discretionary time deficits by family structure. Using the 2003-2010 American Time Use Surveys (ATUS), this study measures discretionary time deficits using a residual time poverty measure and investigates differences in time poverty rates by family structure for women in combination of two social roles—worker and parent. Never married women have significantly lower time poverty rates as compared with married women. There are no family structure differences in time poverty among married, cohabiting, and previously married employed mothers. Living with other employed adults significantly reduces the odds of time poverty for never married women.
Introduction
Major social and demographic changes over the past several decades have resulted in an increase in women’s labor force participation, higher rates of divorce, and the increased prevalence of single parenting. Given these changes, women face competing burdens of employment, child care, and household labor. Researchers and policy makers have generally focused their attention on the economic disadvantages for single-parent families (Casper & Bianchi, 2002); however, it has been argued by some scholars that certain women also experience severe discretionary time deficits, or insufficient discretionary time after considering time spent in paid work and unpaid tasks in their lives (Vickery, 1977). Very little research examines discretionary time shortages or constraints faced by some families, despite evidence showing that diminished leisure time is linked to negative health outcomes (Bittman, 2002) and discretionary time is important to form strong interpersonal relationships and social networks (Mattingly & Bianchi, 2003).
Discretionary time scarcity is particularly pertinent to women and is affected by their relationship status and the responsibilities associated with the roles of worker and parent. Women who are particularly prone to experience discretionary time shortages from day-to-day responsibilities are single mothers who have to juggle work and household commitments with half as many adults to provide economic and caregiving support (Bianchi, Robinson, & Milkie, 2006; Casper & Bianchi, 2002). Single mothers also lack the economic resources to purchase goods and services that may free up their time.
Literature on time use investigates discretionary time shortages by utilizing the concept of “time poverty.” Time poverty concerns itself with individuals who have insufficient time for rest and leisure, after time spent working (paid or unpaid) is taken into account (Vickery, 1977). The time poverty literature has used relative measures to consider the social context and make comparisons to other individuals in a given setting. Relative time poverty measures generally apply a relative threshold, which is defined as 50% or 60% of the median time in certain activities, in order to identify individuals who are time poor.
Existing evidence, using relative time poverty measures, generally finds single-parent families to experience more time poverty than dual-parent families (Harvey & Mukhopadhyay, 2007). A recent analysis using American data found that workers and parents experience the highest rates of time poverty (Kalenkoski, Hamrick, & Andrews, 2011), but their study did not consider differences by marital status or family structure. The current literature is limited by the lack of focus on family structure and on families in the American context. Consequently, our knowledge of how the distribution of social roles—worker, spouse or cohabiting partner, and parent—influences women’s time poverty is sparse. A more careful examination of family structure and the roles of worker and parent is needed in order to accurately document women’s time poverty.
In this study, I examine differences in women’s time poverty by family structure and the influence of the roles of worker and parent on women’s time poverty by family structure. The combination of some or all of these characteristics provides a useful framework for understanding time constraints and discretionary time inequality faced by women. I utilize the 2003-2010 waves from the American Time Use Survey (ATUS) to answer the following research questions:
Background
Time Poverty
Although conceptualizing time shortages is not a new concept (Bardasi & Wodon, 2006), there is limited research on formally operationalizing time poverty. Time use research has generally used informal definitions of time poverty, for example, long weekly employment hours, total unpaid and paid work time, and a handful of studies have examined free time (Bittman & Wajcman, 2000; Gershuny, 2009). This research has found leisure time inequalities to be gendered (Mattingly & Bianchi, 2003; Sayer, 2005) and educated and high-income earners in dual-income families to be leisure time poor (Robinson, 1997).
Time poverty research has since expanded to consider several relative time poverty measures. As opposed to making judgments or assumptions about required time in certain activities, relative measures consider the social context and make comparisons to other individuals in a given setting. This approach mimics the measurement of relative income poverty (Townsend, 1979), in that, those individuals whose discretionary time is considerably lower than the average or median person are deemed to be incapable of participating in civic, social, and leisure activities (Bittman, 2002). Relative time poverty measures generally apply a relative threshold, which is defined as 50% or 60% of the median time in certain activities, in order to identify individuals who are time poor.
Most relative time poverty measures are operationalized using one domain of time or combinations of time in certain activities (Bardasi & Wodon, 2006; McGinnity & Russell, 2007). Most techniques construct time poverty measures using “four kinds of time” (As, 1978). The first type of time is “contracted”: the amount of time explicitly indentured to paid work or educational pursuits, including related travel. The second kind is “committed” time: the amount of time dedicated to unpaid work such as housework, child care, shopping, or providing help to others. Third is “necessary” time: the amount of time vital for maintaining physiological health, such as time spent toward eating, sleeping, and hygiene. The fourth is “free” or “leisure” time, that is, time leftover after the other three kinds of time have been subtracted from a person’s day. For this reason, leisure has been described as the “residual” (Bittman, 2002). A threshold (e.g., 50% or 60% of the median), or cutoff, is calculated using one or a combination of these activities, and an individual is determined to be time poor relative to this threshold. In the current study, a residual measure is constructed by subtracting paid and unpaid work time and time in biological activities from 24 hours. This measure is comparable to recent work in the time poverty literature (Burchardt, 2008; Kalenkoski et al., 2011).
An individual’s decision to allocate time to these four activities is constrained by a set of resources and responsibilities (Burchardt, 2008). The most obvious resource is time, which can be used to generate income in order to purchase goods and services. Other forms of resources include human and social capital. Human capital encompasses skills and experience that determine job opportunities, and social capital, such as a network of family and friends, can provide support for household tasks and caregiving without any financial expenditure.
These available resources are used to meet an individual’s responsibilities. Responsibilities include personal care and time for biological needs, such as sleeping and eating. Child care and care for elderly relatives can be provided directly or purchased from the market. An individual may also have obligations to maintain resources. For example, a house requires an investment of time and/or financial resources to maintain its physical conditions (Burchardt, 2008). Investing in human capital means increasing education as well as maintaining good health. Spending time with friends and family to enhance relationships and expand social networks are all mechanisms to increase social capital.
Previous decisions about time use have important consequences on an individual’s available resources and responsibilities. Earlier decisions about getting married, having a child, attaining further education, and so on, have major implications on an individual’s time use. Time poverty research has considered these decisions to be fixed (Gershuny, 2007), and analyses are conducted on current circumstances. This practice is akin to income poverty research, which investigates how much income an individual currently has amassed, not what he could have earned.
In all, these resources, responsibilities, and earlier decisions on time use are captured in most time poverty measurements by determining which individuals are time poor because of insufficient time for rest and leisure (Vickery, 1977). More time spent in market work or unpaid productive work means less discretionary, or leisure time, and thus greater “time poverty.”
Theoretical Frameworks
The theoretical perspectives that explain the differences in women’s discretionary time emphasize marital, employment, and parental statuses. The time availability perspective posits levels of discretionary, or free time, are affected by one’s own and other household members’ supply of time to paid work and the demand for housework (Coverman, 1985). Family members who devote less time to market work have more time available for household labor and child care. Married or cohabiting mothers have the benefit of a partner who can share the burden of paid and unpaid work, while single mothers are at a disadvantage as they cannot share the demands of employment and housework. In addition, single mothers have fewer financial resources to pay for child care and other household responsibilities. The time availability perspective suggests single mothers suffer from greater time deficits and are more likely to be time poor than their partnered counterparts.
The gender perspective emphasizes that family work is not a gender neutral activity and that mothers do not perform unpaid work because they have a comparative advantage; rather, time use patterns are produced by unequal power relations between married men and women. Thus, women “do gender” within the household by taking on the burden of unpaid household labor. Marriage and childbirth are gendering activities that can solidify gender roles and the division of household labor (Thompson & Walker, 1995). Cohabiting mothers may experience discretionary time deficits similar to married mothers since they are primary caregivers and have a partner in relation to whom they can perform their “gender roles.” On the other hand, divorced mothers have experienced societal expectations of gendered behavior that may differentiate them from never married mothers, but their single status may distinguish them from married mothers. In contrast, single mothers’ lack of a romantic partner suggests that they do not participate in “gendered activities.” As such, married or cohabiting women may have higher time poverty rates as compared with single women.
Employment, Economic Resources
All else being equal, as market work hours increase, time available for other activities decreases. Nock and Kingston (1988) showed that additional hours of market work are related to a decline in discretionary time, but the decrease is not an hour for an hour. This is most likely because individuals can make tradeoffs in other activities, such as unpaid work. If women bear the burden of managing and coordinating family life, more so than men, and family responsibilities are given a higher priority than leisurely activities, an additional hour of paid work may crowd out discretionary time for women. Previous literature finds that women with paid jobs have more housework and child care responsibilities than men (Bianchi, Milkie, Sayer, & Robinson, 2000).
Married mothers are better educated and have higher incomes and thus lower rates of poverty than single mothers (Casper & Bianchi, 2002). Single mothers lack economic resources to purchase goods and services, such as child care and prepared food, that can lead to an increase in their discretionary time. Particularly, never married mothers have lower household incomes than divorced mothers. On average, divorced mothers have higher employment rates and are twice as likely to have full-time, better paying jobs as never married mothers. Thus, one can expect that time poverty rates are lower for never married mothers than divorced mothers, because they are less likely to have employment demands than divorced mothers.
Parenthood
Earlier studies on time use show that presence of children increases the demand for household work more than any other characteristic (Berk, 1985). The more children and the younger they are, the greater the child care demands, and the more likely women are to be time poor (Bittman & Wajcman, 2000). Although partnered women have the potential supply of time from fathers to contribute to family care, research finds that women bear the burden of managing family life, coordinating responsibilities for their partners and children, and supervising homework for school-aged children (Bianchi, 2000). Time poverty rates may be lower for partnered mothers over single mothers purely because partnered mothers have greater latitude because of a working partner or spouse and higher household incomes than single mothers, although the child care demands may be high for both groups of women and may result in similar time poverty rates.
Although children require a great deal of time investments, younger children need constant supervision. Most studies find that the presence of preschool-aged children in the household increases mothers’ child care time (Bryant & Zick, 1996; Milkie, Mattingly, Nomaguchi, Bianchi, & Robinson, 2004; Sandberg & Hofferth, 2001). The research on the relationship between number of children, gender of children, and maternal child care time is mixed (Milkie et al., 2004; Nock & Kingston, 1988; Sayer, Bianchi, & Robinson, 2004). Never married mothers tend to have younger children than divorced and married mothers. Divorced mothers generally have older children than married mothers (Casper & Bianchi, 2002). To the extent the role of parenting affects a mother’s discretionary time, never-married mothers may have more caregiving demands and higher time poverty rates than divorced mothers.
Empirical Evidence
In the literature on time poverty, researchers generally consider the roles of partner, parent, and worker together when discussing time shortages. Generally, time poverty studies have made comparisons across all men and women (McGinnity & Russell, 2007) or focused on specific groups, such as workers (Chatzitheochari & Arber, 2012). Studies find high levels of time deficit among employed, single-parent households in Canada, and also that time poverty exacerbates problems generated by income poverty (Douthitt, 2000; Harvey & Mukhopadhyay, 2007). Bittman (2002) used cross-sectional Australian Expenditure data and found the combination of gender, long work hours, and family responsibilities to predict leisure time poverty. In particular, full-time working mothers with young children are most at risk of being excluded from leisure. McGinnity and Russell (2007) found employment and caregiving (of adults and children) to be the strongest predictors of time poverty for women. Significantly, these studies do not distinguish between married and cohabiting couples, are beset with small sample sizes, and are either not in the American context or are outdated.
A more updated version of time poverty thresholds, using the ATUS, constructs a time poverty measure similar to Burchardt (2008) and to the present study (Kalenkoski et al., 2011). They found that employed individuals and households with children had higher time poverty rates. Despite showing rates for various household types, they did not consider marital status or family structure, which could explain the null findings of similar time poverty rates between households that differ by number of adults. The analyses in the current article build on their findings by assessing discretionary time disparities for women by family structure.
Another strand of literature examines women’s leisure time but does not use relative time poverty measures to investigate discretionary time inequality. Earlier studies that examine the relationship between marital status and mothers’ leisure suggest modest differences between single and married mothers’ leisure time. The gap is driven mostly by employment status and the age of the youngest child (Douthitt, Zick, & McCullough, 1990; Sanik & Mauldin, 1986). However, these findings are based on small, nonrepresentative samples from the 1970s. More recent time diary studies report less leisure among married mothers versus their single counterparts (Bittman & Wajcman, 2000; Mattingly & Bianchi, 2003). These studies tend to have very small samples and do not scrutinize the heterogeneity in family structure. Although these studies suggest family structure differences, they do not examine time shortages using a relative time poverty framework.
In sum, while the extant research provides provocative evidence that time use varies by marital status and suggests important mechanisms to explain a women’s likelihood of experiencing time poverty, previous time poverty research has been limited by outdated, nonrepresentative data, and a lack of emphasis on family structure. The present study builds on previous research that has operationalized time poverty as a relative construct to measure discretionary time shortages and contributes to the literature by investigating differences among women by their partner, work, and family roles.
Data and Method
Data
This article uses the ATUS, a nationally representative sample of the noninstitutionalized population, aged 15 years and older. Respondents are phone-interviewed about their time use 2 to 5 months after completing the Current Population Survey (CPS). Respondents report their “diary day,” starting at 4 a.m. the previous day and ending at 4 p.m. on the interview day. Respondents report on what they were doing, with whom, for how long, and the location of their activities. The sample consists of interviews on all days of the week and all months of the year. To analyze the relationship between time poverty and women’s social roles, I utilize the pooled 2003-2010 waves of the ATUS. In all, the ATUS sample includes 112,038 respondents for whom there is a time diary, reflecting an average response rate of 58% (Bureau of Labor Statistics, 2010). Previous research on these response rates has shown that busy people are no less likely to respond to the ATUS, but individuals who are weakly integrated within their communities are less likely to respond (Abraham, Maitland, & Bianchi, 2006). All analyses use sampling weights to account for sample stratification, distribution of diary days, and different response rates across demographic groups and days of the week.
The primary analysis is on adult women’s time use and their family responsibilities, and thus, I exclude persons under 18 or older than 64 years of age (N = 24,903), all males (N = 38,552), and those for whom information on family structure and analysis variables was missing or who had data quality issues (N = 6,943). This results in an analytical sample of 41,153 women.
Time Poverty Measure
I construct a residual time poverty measure by subtracting contracted, committed, and necessary time from 24 hours or 1,440 minutes. The table in the appendix outlines the ATUS major activity codes assigned to each of the aforementioned time domains. Contracted time includes work and work-related activities. Committed time is time in household activities and caring for and helping household members. Committed time that is given to children is exclusive to primary child care. Primary child care includes time in physical care, teaching, reading, and playing with children. The ATUS captures information on secondary child care, or child care done in conjunction with a primary activity. The time poverty literature does not make any considerations for secondary care, and thus the focus in my analysis is only incorporating primary child care. In doing so, no time is double counted, but my analyses may underestimate mothers’ time in child care. Necessary time is personal care, such as sleeping and grooming, and eating and drinking. The residual, after subtraction, is discretionary time. In the same vein as Bittman (2002), Bardasi and Wodon (2006), and Burchardt (2008), a time poverty cutoff, or threshold, is calculated by taking 60% of the median discretionary time distribution of the entire sample of men and women. This threshold is 342 minutes, or 5.7 hours, per day. After this measure and threshold is constructed, the sample selects on women for analyses. Women who have discretionary time below this cutoff are deemed discretionary time poor, and those at or above the cutoff are not discretionary time poor.
In observing women’s time poverty, one reason to caution against using an individual’s reported sleep minutes in measures of time poverty is because sleep may mask the incidence of time poverty. A woman may draw on her sleep time in order to allocate time to fulfill necessary tasks, such as paid and unpaid work. Previous studies are silent on this potential concern. I examine mean sleep, in minutes per day, by family structure and presence of children (results available on request), and find no clear evidence of sleep deprivation by family structure. For two reasons, I choose to impute a respondent’s sleep time with the recommended level of 8 hours. One, the “sleep” category in ATUS includes a range of terms that are not only exclusive to physiological sleep (i.e., falling asleep, dozing off, getting up, waking up, etc.) but also refer to a transition state. It is likely that ATUS respondents overestimate their physiological sleep time in a 24-hour period. There is evidence that observed sleep duration in this sample is on average longer as compared with self-reported sleep times in other population-based studies (Patel et al., 2004). Two, leveling the sleep time for all respondents decreases any possible measurement error.
Informed by the work of others, I construct two alternative relative time poverty measures (McGinnity & Russell, 2007; Spinney & Millward, 2010; results not shown). The findings in the present study are robust to these two additional measures. The residual measure proves to be intuitive and easily interpretable, because it considers discretionary time as the residual; the time left over after netting out the activities one has to do (personal care, unpaid work, and paid work).
Independent Variables
Family Structure
Questions asking about the respondent’s marital and cohabitation status are used to construct a family structure variable with four mutually exclusive categories. Women are classified into one of the following: (a) living with a married spouse, (b) living with an unmarried partner, (c) separated/widowed/divorced, (d) never married.
Individual Characteristics
Living arrangements are obtained from the household roster. A three-category variable indicates the presence of other adults in the household and their employment status. This variable excludes spouses and cohabiting partners, but can include adult children, parents, and other adult relatives. A respondent is categorized as residing with other nonemployed adults, living with other adults who are employed, or living without the presence of other adults (the reference group). This variable construction follows Kendig and Bianchi (2008) and captures the increased supply of economic and time resources that may come with residing with other adults. Employment status is identified outside the diary day by using the labor force status of the time diary respondent at the time of the interview. A three-category variable was created to indicate if, in the week preceding the ATUS interview, the respondent worked full-time, part-time, or was not employed. Full-time employment is considered to be 35 hours or more per week. In regression analyses, comparisons are to women employed full-time, the reference category. Education is coded into four binary variables: (a) less than a high school degree and no GED, (b) high school graduate or equivalent GED, (c) some college, and (d) college graduate or higher. The college plus category is excluded as the reference category in regressions. Family income is measured from 16 responses that range from (1) less than $5,000 to (16) $150,000 and over. Income is converted from these categorical responses to dollar amounts by assigning the midpoint of each category and representing income in thousands of dollars. The last category is topcoded to $200,000. Age is a continuous variable ranging from 18 to 64 years of age. Race and ethnicity is coded into five binary variables for non-Hispanic White, non-Hispanic Black, Asian, Hispanic, and Other. Non-Hispanic White is the reference category in the regressions.
Children’s Characteristics
The ATUS defines own household children as biological, adopted, and step-children who live in the household. The number of household children is measured using categories for none, one, and two or more; larger families increase the amount of care work and can affect one’s time poverty. The age of the youngest child in the household is measured using categories for 0 to 2 years, 3 to 5 years, 6 to 17 years, and no children; this is pertinent to test if mothers with younger children are more time poor, because younger children have the most binding time constraints, in that they require adult supervision 24 hours a day.
Additional controls for a weekend and summer diary response were included in the original regressions. Including these indicators did not substantively alter the point estimates. Furthermore, separate regressions were run for weekday and weekend diaries. The results of these analyses did not reveal any dramatic changes to warrant stratification by weekday and weekend diaries. The largest changes in point estimates were for employment variables, and these changes are meaningful because the vast majority of employment hours is during the weekday. All other correlates of time poverty remained similar in size and significance to estimates from regressions that did not stratify by weekend and weekday diaries.
Analytic Strategy
Using the residual measure of time poverty, I provide time poverty rates for subgroups defined by family structure and the role of worker and parent. T-tests are used to test for statistically significant differences in the rates. Logistic regression models are used to assess the relationship between an individual’s experience of time poverty and sociodemographic characteristics, including age, race, education, income, employment status, and child characteristics. Separate models are run for each family structure to explore how patterns of individual and household characteristics differ by family structure. Post hoc Chow tests are computed to compare coefficients on individual and child characteristics across family structure. All analyses are weighted to account for oversampling of certain demographic groups, sample stratification, and differential response rates.
Results
Table 1 presents descriptive statistics for all women in the analytic sample and by family structure. Little over half the sample is composed of married women. Less than 4% of women are cohabiting and about one fifth of the sample are each previously married or separated and never married. Cohabiting and unpartnered women spend significantly more time in contracted and less time in committed activities than their married counterparts. Previously married or separated and never married women spend significantly more time in personal care or necessary time than married women. Unpartnered women have nearly 30 to 90 minutes more of discretionary time than married women.
Percent Distributions and Means of Selected Characteristics by Relationship Status, 2003-2010.
Note. Percentages are weighted with sample weights. Sample sizes are unweighted. Standard deviations are presented in parentheses.
Cohabiting, previously married or separated, and never married women have significantly lower household income and are less well-educated than their married counterparts. Cohabiting women are significantly more likely to live without other adults in the household, to be employed full-time, and to be childless than their married counterparts. Separated or previously married and never married women are significantly more likely to live with other adults in the household, to be childless, and less likely to have their youngest child be less than 5 years old than married women. Never married women are significantly more likely to work part-time than married women. Unpartnered women are also more likely to be Black than their married counterparts.
Table 2 shows time poverty rates for all women in the sample and by family structure and presence of children in the household. The residual measure uses a 60% threshold defined for the total population, which is 342 minutes (5.7 hours, not shown). Women whose discretionary time falls below this level are labeled as time poor and those whose discretionary time is at or above this threshold are not time poor.
Time Poverty Rates by Family Structure and Number of Children in the Household.
Note. Percentages are weighted with sample weights. Sample sizes are unweighted. Significant results are in comparison to married women. Residual Measure’s time poverty threshold is 342 discretionary minutes.
p < .05. **p < .01. ***p < .001.
The first column of Table 2 examines time poverty rates of all women and by number of children in the household. Thirty percent of the sample is time poor and nearly one third of mothers are time poor. The first row examines time poverty rates by family structure. Never married women have significantly lower time poverty rates as compared with married women.
Among childless women, never married women have significantly lower time poverty rates than their married, childless counterparts. For women with children, there is no evidence that time poverty rates increase with more children in the household. Although, a priori, we would expect a heavier time burden with more children in the household, it is unclear if the results are conclusive without knowing the age and gender of the children. Among mothers with only one child, never married mothers have significantly lower time poverty than married mothers with one child. Previously married mothers with two or more children in the household have significantly higher rates of time poverty and never married mothers of two or more children have significantly lower time poverty relative to married mothers.
Table 3 examines time poverty and the nexus of the roles of worker and parent by family structure. Panel A focuses on employed women, and Panel B examines women who are not employed. In the first panel and first column, nearly two fifths of employed women are time poor. Forty-three percent of employed mothers are time poor.
Time Poverty Rates by Family Structure, Employment, and Age of Youngest Child.
Note. Percentages are weighted with sample weights. Sample sizes are unweighted. Significant results are in comparison to married women.
p < .05. **p < .01. ***p < .001.
Focusing on family structure differences, Panel A indicates that cohabiting and never married employed women have lower time poverty rates than their married counterparts. Nearly one third of never married employed mothers are time poor, and this rate is significantly lower than their married, employed counterparts. When examining the impact of age of the youngest child in the household on the rates of time poverty of employed mothers, there are no family structure differences when the youngest child is less than 3 years of age. Among older children in the household, never married mothers present with a discretionary time advantage: never married employed mothers have significantly lower time poverty than their married counterparts when the youngest child in the household is at least 3 years of age. There are no family structure differences in time poverty among married, cohabiting, and previously married employed mothers, regardless of age of the youngest child in the household.
In Panel B, unpartnered women who are not employed have significantly lower time poverty rates than their married counterparts. Among unemployed mothers, married mothers have significantly higher time poverty rates than cohabiting and unpartnered mothers. With respect to the age of the youngest child in the household, never married unemployed mothers have significantly lower time poverty rates than their married counterparts, regardless of the youngest child’s age. Cohabiting unemployed mothers of young children and school-aged children have significantly lower time poverty rates than their married counterparts.
In supplementary analyses not shown and focusing on differences within family structure, lower time poverty rates are found among childless women as compared with mothers. Among mothers, and within family structure, as age of the youngest child in the household decreases, the higher the time poverty rate. Although, this is only true for married and never married mothers.
I turn to multivariate models predicting time poverty to examine to what extent variation in sociodemographic characteristics is associated with the experience of time poverty. Table 4 presents the results of logistic regression models predicting time poverty stratified by family structure. Living with other adults in the household who are employed decreases the odds of time poverty for never married women. Similar to the descriptive results, part-time employment or not working is associated with significantly lower odds of time poverty, compared with full-time work. An increase in income is associated with higher odds of time poverty for previously married women. Last, with respect to children’s characteristics, childlessness is associated with lower odds of experiencing time poverty for married and previously married women. Children in the household, who are younger than 6 years of age, are related to higher odds of time poverty.
Odds Ratios From Logistic Regressions Predicting Time Poverty by Family Structure.
Note. Standard errors are provided in parentheses. All models are weighted.
Statistically significant differences between Married and Never Married women.
Statistically significant differences between Married and Separated/Widowed/Divorced women.
Statistically significant differences between Married and Cohabiting women.
p < .05. **p < .01. ***p < .001.
I conducted further analyses to examine whether certain individual and child characteristics are statistically different by family structure (differences are highlighted in Table 4 with superscripts). The association between age and time poverty is statistically different between married and cohabiting women. The coefficients on unemployment and Asian are statistically different between married and separated/widowed/divorced women. The differences in characteristics between married and never married women are similar to descriptive results in Table 1. Living with other employed adults significantly reduces the odds of time poverty for never married women and this estimate is significantly different from that of married women. Nearly two thirds of never married women live with other adults who are employed, while only half as much is true for married women (Table 1). Part-time employment and nonemployment associations with time poverty are also different between married and never married women. This is not surprising given that never married women have higher rates of part-time employment and lower rates of nonemployment than married women. Although only 10% of never married women have children in the household less than 2 years of age, the odds of experiencing time poverty with the presence of an infant in the household are significantly higher than that of married women.
Discussion
In this article, I draw on the ATUS, a time diary data set, to report on new evidence concerning the relationship between time poverty and key social roles borne by women. Current literature on time poverty does not examine how this metric discerns time disadvantages by family structure. The analyses in this study attempt to give greater scrutiny to the diversity in living arrangements in order to shed light on the differences in time shortages by women’s work–family roles.
Descriptives reveal significant discretionary time advantages for never married women, as compared with married women. Never married mothers and childless women have lower time poverty rates than their married counterparts. When examining the nexus of family structure, work, and parenthood, there are no family structure differences among employed mothers when the youngest child is less than 3 years of age. There is discretionary time equality among married, cohabiting, and previously married employed mothers of infants. Multivariate models reveal living with other employed adults significantly reduces the odds of time poverty for never married women.
I offer a few explanations to explain the minimal differences in time poverty rates between married and cohabiting and previously married women. One, it may be that partnered women are in traditional relationships and are “doing gender” by increasing their unpaid work time, such as housework. Two, bargaining theory posits an important relationship between household responsibilities and earnings share among partners. It is possible that women are unable to negotiate their housework time due to a lower relative income share. Previously married women are unable to display a gendered performance, because they lack partners. The gain in discretionary time for unpartnered women resulting from the absence of a husband or cohabiting partner may be offset precisely due to the absence of a partner with whom to share employment and household responsibilities. In the net, partnered and some single women are comparable with regard to their experience of time poverty. Third, the presence of other adults in the household may be disadvantageous if these other adults increase a woman’s committed time (e.g., extra housework, caretaking responsibilities, or household management). Due to data constraints, we cannot examine the time of other household members.
My findings are still somewhat puzzling. Why do never married mothers have significantly lower time poverty rates than their married counterparts? These findings are surprising, because they are counter to that of existing literature on family structure and its role in economic and social inequality (McLanahan & Percheski, 2008). In this regard, qualitative work by Blair-Loy (2003) may be instructive as far as the lack of inequality in time poverty among mothers with infants. She argues women experience a tension between two gendered schemas, the “family devotion schema” and the “work devotion schema.” The first represents the expectation that women are primary caregivers and family managers, and the second framework captures women’s complete dedication to work, leaving little time for family care. Blair-Loy argues that earlier cohorts of women may have opted out of the family devotion schema, while later cohorts of women have tried to negotiate the two schemas by pursuing a career while also trying to meet the expectations of household and family responsibilities. Mothers from later cohorts are juggling to satisfy their careers and to fulfill a family schema that requires time for household demands and family involvement. Although Blair-Loy’s work does not focus on time poverty and her conclusions are limited to women in executive occupations, it is highly suggestive with regard to time poverty among women, regardless of family structure.
Never married women are more likely to live with another adult and to live with another employed adult. Also, their time in committed activities is significantly less than married counterparts (nearly 2 hours less; see Table 1). This could be for many reasons. Perhaps the presence of more adults in the household increase the supply of time and economic resources, all of which help reduce the likelihood a never married woman is time poor. Due to data constraints, we do not know if the experience of time poverty caused other adults to move into the household and thereby increase the total discretionary time for these women. It could also be argued that the significantly less time in committed activities is due to less time in childcare as compared with married mothers. Using the same data, Kendig and Bianchi (2008) found single mothers to invest less time in their children, as compared with married mothers; however, these differences are explained by sociodemographic characteristics, which are the same covariates used in the current analyses. Kendig and Bianchi find single mothers who reside with another adult who is employed engage in more interactive care activities than single mothers who live alone. Thus, given economic constraints, single mothers may choose to make important tradeoffs with their time in order to protect time with their children.
Another perspective that may elucidate the underlying mechanisms between family structure and time poverty is the constraint of cultural and societal values and norms. In particular, ideologies about good mothering may influence caregiving time and time allocation in other activities. Despite the increasing acceptance of mothers in market work, they are still expected to be primary caregivers and not breadwinners (Riggs, 1997). Furthermore, as the amount of time needed to create a “good” childhood has increased (Arendell, 2001) mothers are expected to not only be experts on their children’s immediate emotional and physical needs but also to be child development specialists—that is, from a long-term perspective—and to manage the various and complex dimensions of their children’s development and overall well-being. It may very well be the case that mothers’ time allocation is increasingly governed by normative notions about motherhood.
It is crucial to consider how the design of the ATUS limits the comparisons made in this study. Only data at the individual level are available and only information on one household member’s time poverty, on a given day. It may be that other members of the household are not time poor and are not making difficult tradeoffs with their time vis-à-vis important activities. Therefore, at the household aggregate level, it is an open question if the household or family is indeed time poor. The data structure also limits our ability to consider day-to-day variation in time poverty.
Although the ATUS provides a very large and representative sample of individuals, the data are cross-sectional and do not allow for causal analyses. It is difficult to comment on how changes in social roles—such as entering employment or exiting from a cohabiting relationship—affect an individual’s time poverty. Furthermore, the ATUS provides a small set of observed covariates and does not provide information on physical or mental health conditions of women and their children. Such factors could in themselves explain higher or lower time poverty rates, or they could influence one’s selection into the role of worker, parent, or partner. The findings in this study are thus to be interpreted while bearing such limitations in mind.
In sum, the rich information on women’s time use in the ATUS allowed for the investigation of differences in time poverty rates by various social role combinations. This research has revealed the significance of time consumption resulting from employment and caregiving roles. It is of course difficult to make judgments on the quality of discretionary time afforded to women who are time poor because of long work hours and child care responsibilities. It may be the case that such individuals are choosing to dedicate these hours in order to achieve higher levels of income or work satisfaction, or they are choosing to focus their energies on spending time with their children. Despite being time poor—because they have less time available for leisure, rest, or time spent with friends—these women may not be worse off than women who are “time rich”—for time poverty is just one of the many measures for ascertaining an individual’s level of well-being.
Footnotes
Appendix
Contracted, Committed, and Necessary Time by Major Activity Group
| ATUS major activity code | |
|---|---|
| Contracted activities | |
| Work and work-related activities | 05 |
| Travel related to work | 18 |
| Committed activities | |
| Household activities (includes housework, food and drink prep, and household management) | 02 |
| Caring for and helping household children and adults | 03 |
| Travel related to household activities | 18 |
| Necessary activities | |
| Personal care (includes grooming and health-related self-care) | 01 |
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.