The extreme gendering of COVID−19: Household tasks and division of labour satisfaction during the pandemic

Abstract

For many years, scholars have directed our attention to the gender gap in domestic labour. Even when women engage in paid employment, they nevertheless perform the majority of the household labour in most wealthy countries. At the same time, disasters and crises both expose and exacerbate existing social inequalities. In this paper, we ask: in what ways has the COVID−19 pandemic contributed to the gender gap in household labour, including childcare? How do women and men feel about this gap? Using data from the Canadian Perspectives survey series (Wave 3), conducted by Statistics Canada three months into the pandemic, our analyses consider the task distribution that made household labour intensely unequal during COVID−19, with women ten times more likely than men to say childcare fell mostly on them, for example. Yet, in nearly all of our models, women did not ubiquitously report being more dissatisfied with the division of domestic tasks within the house, nor were they more likely than men to say that the household division of labour “got worse” during COVID; however, parents did feel that it got worse. We discuss what these findings mean for women's mental health, long‐term paid labour, and interpersonal power, and raise questions about why it is we are not seeing a decrease in women's reported satisfaction with this division of labour. These findings spotlight gender inequality and the family as ongoing pillars of capitalism, and how the structural and interpersonal weathering of the pandemic comes at a particularly great expense to women.

INTRODUCTION

At the onset of the COVID−19 pandemic, existing household chores intensified while new sorts of routine tasks emerged. This is because people were home together around the clock. Lockdowns abruptly required many people employed outside the home to tele‐commute (Belzunegui‐Eraso & Erro‐Garcés, 2020), with a 12,000% increase in work‐from‐home arrangements in some countries (Bloom, 2020)—and other people were furloughed or lost their jobs entirely. Schools and day cares closed. These closures and shifts home transformed how families functioned and ultimately how cohabiting couples divided household tasks like shopping and cleaning (Evans et al., 2020). People complained that the dishes never seemed to end, and they managed exposure to COVID−19 by sanitizing their groceries. Parents supervised their kids’ online schooling and had few breaks from the demands of home (Collins et al., 2021b; Garbe et al., 2020). We know that the division of household labour, including childcare, has been historically gendered and divided by sex (Bianchi et al., 2012; Craig & Powell, 2011; Davis & Greenstein, 2013), but the drastic changes in both paid and unpaid labour in the wake of COVID−19 brought families the opportunity to reimagine who does what at home—and scholars the opportunity to re‐evaluate the politics of household labour.

In this paper, we consider how households divided domestic tasks three months into COVID−19, as well as women's and men's ¹ satisfaction with the division of these tasks. Drawing from data gathered by Statistics Canada during the third wave of the Canadian Perspectives Survey Series—focused specifically on social experiences during COVID−19—we consider the impact gender and living with kids has on the division of household tasks. We find that sex identification often mattered more—in both magnitude and in consistency—than other relevant variables such as employment and parental status for explaining who performed which tasks. These findings reveal an extreme gendering of household and family labour not reflected in previous pre‐COVID scholarship, and in a country that has more social safety nets than most. Yet, women were no more likely than men to say that the household division of labour “got worse” during COVID, and they did not report feeling consistently less satisfied with that division of labour. Understanding who does what tasks contextualizes differences in rates of satisfaction with household division of labour during COVID−19, which together might ossify the gender regression of work‐family labour imbalances for years to come.

This analysis spotlights just how important the legacy of patriarchy is to supporting capitalism in times of crisis. Given what we know about recent shifts in gender relations (Connell, 1987; Gerson, 2010; Risman, 1998), we might be forgiven for expecting men to have taken on more household tasks. Instead, the data suggest that the pillars of unpaid and devalued domestic labour have been scaffolded during this “magnified moment” (Hochschild, 1994)—when the intensification of domestic responsibilities and the loss of or disruption to paid work provide insights into the need for cultural and legislative change to better supports economic and human life.

LITERATURE REVIEW

Gender and domestic labour

Women consistently shoulder a disproportionate amount of the domestic labour in heterosexual, cis coupled households (Baxter & Hewitt, 2013; Craig & Mullan, 2010; Erickson, 2005), an imbalance compounded with the addition of kids (Cowan & Cowan, 1992; Gjerdingen & Center, 2005; Offer & Schneider, 2011). This is despite whether these women work outside of the home and leads to what Hochschild (1989) coined “the second shift,” where women come home from the paid workplace to manage the household. This second shift shows that women's increasing workforce participation is not enough to pressure egalitarian divisions of household labour (Gerson, 2010; Shelton & John, 1999). At the same time, research shows growth in men's overall participation in domestic responsibilities across countries (Fisher et al., 2007; Hook, 2006; Sullivan & Coltrane, 2008). And while some scholars note that the needle has moved in a more egalitarian direction (depending on contextual expectations of manhood, see Adams & Coltrane, 2005), others argue that the growth rate in men's domestic labour gives too‐optimistic an impression of change—with women still largely “doing it all” (Robinson & Godbey, 1997).

Pre‐COVID−19 research in both Canada and the U.S. shows that the gap between men's and women's hours of domestic labour had been modestly closing, though this change was largely attributable to women doing less rather than men doing more (Bianchi et al., 2000; Guppy et al., 2019). This might mean women outsource domestic labour to in‐home service workers or the house just remains messy and the family orders‐in dinner. Men who hold more egalitarian views do not necessarily pick up more of the housework, unless their partners prioritize egalitarian arrangements (Greenstein, 1996b). This suggests that a more equal household comes at the cost of women holding their partners accountable—at least for “egalitarian wives” who associate marital quality with the equal division of household chores (Greenstein, 1996a). And many women are unwilling to do this accountability work (Stone, 2007), since white, middle‐class definitions of motherhood are conflated with sacrifice and a seemingly bottomless altruism.

Testing key theories on inequality in household labour, scholars have considered time‐availability, relative resources, and gender explanations (Bianchi et al., 2000; Carlson et al., 2020). Of course, these are not mutually exclusive, with gender explaining time‐availability as more women than men hold part‐time jobs or stay at home (Carlson et al., 2020); and with women generally having less authority and fewer resources within the home and in public life. And so, gender is a primary variable in explaining work‐home inequities. This makes sense if we consider gender a structure itself that shapes social relations (Risman, 2004). These various theories all point to how shifts in women's work circumstances during COVID−19 is a power issue, with the gains women have made in both public roles and household negotiating power at risk of eroding as family and paid work meld to a degree previously unseen.

Scholars considered the impact of the second shift on women's mental health pre‐pandemic, with a Korean study showing that women dissatisfied with their men partners’ participation in household labour report a 2.65 times higher rate of suicidal ideation (Lee et al., 2018). This is especially true for those women who are gender progressive—expecting egalitarian household arrangements. In China, women who experience high work and family stress are over five times more likely to think about suicide than other women (Lin et al., 2020). In Canada, a recent study shows that contributing more hours than men to unpaid household labour increases women's stress (MacDonald et al., 2005), and another suggests that they consequently become more dissatisfied in their partnered relationships (Staland‐Nyman et al., 2008). These findings raise questions about women's (dis)satisfaction with the intensification of domestic labour demands during COVID−19, and its residual interpersonal and political effects. One study finds that men and women might see mothers staying at home with the kids during the pandemic as “practical” and “natural” (Calarco et al., 2020). But this justification does not account for satisfaction, or an otherwise feeling of ease with the pressures of home and with the often‐competing pressures of home and paid work.

We also know that after having a kid, highly‐educated professional women are pushed out of the workforce (from lack of institutional support) and pulled home (from guilt‐inducing expectations for mothers to ‘put baby first’) (Stone, 2007). While women struggle to manage home and work with one kid, a second is often the proverbial nail in the coffin for professional women's jobs, which are sacrificed so men can both engage in paid work and have kids (ibid). Further, mothers often find it difficult to re‐enter the paid economy with a gap in their workforce participation (Baum, 2002a; Cahusac & Kanji, 2014), taking a significant hit to their lifetime earnings and retirement, and thus to their financial independence (Aisenbrey et al., 2009; Baum, 2002b).

In her work on “intensive mothering,” Hays (1998) examines the vast amount of time and energy working moms are socially expected to invest in their children—creating strategic feeding plans (Brenton, 2017; Elliott & Bowen, 2018), directing school searches (e.g. Brown, 2022), and cultivating their kids’ cultural and physical capital (e.g. Stirrup et al., 2015). While much research shows that this is a white, upper‐middle class conceptualization of motherhood, the Black women in Elliott et al.’s (2015) research also felt the need to meet these demands but with little to no community support, ultimately sacrificing their mental and physical health. Indeed, an enormous amount of stress is associated with intensive mothering (e.g. Nomaguchi & Milkie, 2017; Rizzo et al., 2013). This raises questions about how conflating “good moms” with self‐sacrifice makes it difficult, if not impossible, for mothers to invest in paid employment or feel they are successfully juggling ‘work and home.’ Even employed mothers who seek out jobs with flexible hours to better meet the demands of the household find themselves stressed as they manage increased workloads during non‐standard working hours (Wharton, 1994; Wight et al., 2008). With more people working from home during COVID−19, employed mothers are also “stay at home” mothers; and these mothers reported lower work production and decreased job satisfaction (Feng & Savani, 2020).

How COVID−19 Has Affected Work and Home

Most research on gender and COVID−19 focuses on concerns around the potential growth of gender inequality, with early evidence suggesting that pandemic related inequities in paid employment and unpaid care work may prompt regressive shifts in gender role attitudes (Reichelt et al., 2021). Public debate currently centers around whether COVID−19 related changes in work and family have either improved gender equality—since dads are home in historically high numbers—or negatively affected gender relations as women shoulder the acute increase in domestic work (Hipp & Bünning, 2021). Carlson et al. (2020) argue that both mothers and fathers in heterosexually coupled families have risen to meet the increased responsibilities at home. And while the authors report that fathers’ stepping‐up to care for the kids could mean good things for families long‐term, they also note that fathers are more likely than mothers to report sharing equally in household work during COVID−19. Social expectations conflating womanhood and motherhood with domesticity shape how people evaluate division of labour “fairness” and likely impacted the authors’ findings. Indeed, Calarco et al. (2021) find that working mothers have found themselves the “default” caretaker in the family during COVID−19—there is no conversation amongst spouses about who will care for the kids and mothers end up relying on readily available gendered narratives that naturalize their roles as primary caregivers and justify their lack of options and support received.

Recent U.S. research indicates that COVID−19 increased the gender gap in paid work hours by 20−50% (Collins et al., 2020a), as job losses disproportionately affected women (Kesler & Bash, 2021; Landivar et al., 2020). Qian and Fuller (2020) also show that the employment gap among heterosexually coupled parents of young kids in Canada widened considerably from February (pre‐pandemic) to May (near the height of the “first wave”). This widening was particularly pronounced for parents of elementary school‐aged kids. As causal factors, the authors contend that women's occupations, for instance education and the service sector, are often at higher risk of job loss, and point to women's greater participation in part‐time employment and the loss of stable childcare arrangements. In the United States, women experienced a decline in employment about four percentage points higher than men's (Cajner et al., 2020), the greatest gap we have seen in 36 years and effectively setting back women's employment parity an entire generation (World Economic Forum, 2021). Canadian workers between the ages of 15 and 64 saw a 12.4% job loss rate from February to April of 2020 (Chan et al., 2020).

With regard to how employment loss and reduced hours have affected household labour, we see that in Spain, women were more likely than men to lose jobs; and within the home, men increased their domestic work, but only slightly (Farré et al., 2020). Shafer et al. (2020) use a Canadian survey to look at division of household labour and satisfaction with it, much like this paper. They find that Canadian families experienced “small shifts” toward a more equal division of labour during the early weeks of the pandemic, with increased participation in childcare tasks by fathers (see also Carlson et al., 2020). And although research shows that men are doing more, it is still less than women (Andrew et al., 2020a; Hipp & Bünning, 2021). For example, in England, research shows that job loss does not affect men and women the same, with unemployed women picking up more domestic work, but not men who similarly find themselves unemployed (Andrew et al., 2020c). This may be because fathers who stay at home with the kids see domestic work and unemployment as temporary (Doucet, 2016), or because men tend to pick up more domestic work when their women partners are employed (Raley et al., 2012).

Even amongst mothers who did not lose jobs, early research suggests women were uniquely affected by increased childcare demands (Andrew et al., 2020c; Kallitsoglou & Topalli, 2021; Minello et al., 2020). These demands give new meaning to the idea that “home becomes work” from Hochschild's (1997) study of the “time bind.” While she was speaking specifically to the way people spent more time at work to avoid the stresses of home, COVID−19 has pushed work home and forced families to combine work and childcare. Parents are often multi‐tasking with the move of work to home, with 47% of mothers and 30% of fathers caring for kids while simultaneously doing paid work (Andrew et al., 2020b).

Mothers and non‐parents who have moved to “short‐time” work (reduced work hours and reduced wages) report a decline in family and work satisfaction, whereas fathers’ well‐being is less affected in the wake of COVID−19 (Möhring et al., 2021). Considering just how greedy and ill‐defined both paid employment and housework can be during a crisis, recent research shows that COVID−19 has had a negative impact on mental health due to isolation, the lack of novel experiences, the stresses of job and income loss, economic insecurity, and overwork (Cheng et al., 2021; Low & Mounts, 2022; Ollivier et al., 2021; Witteveen & Velthorst, 2020). But for parents specifically—who have experienced a reduction to their already little leisure time (Andrew et al., 2020a)—we see married mothers are experiencing greater parental burnout than fathers (Prikhidko et al., 2020), while men—who are still experiencing stress—report greater marital satisfaction and mental well‐being than women more generally (Mousavi, 2020; Mousavi et al., 2020).

Perhaps as a result of this parental burnout, mental health consequences of COVID−19 are most pronounced among women. Research in Japan finds that parents have faced an increased level of stress during the pandemic, much of it stemming from school closures (Hiraoka & Tomoda, 2020). Single mothers in particular have faced swelling stress, though social networks and a community of care can help to attenuate feelings of being overwhelmed (Hertz et al., 2020). This increased stress manifests specifically in terms of elevated rates of anxiety and depression among parents of small children (Brown et al., 2020). In Germany, Czymara et al. (2021) found that women are worrying considerably about the well‐being of home and family (Kallitsoglou & Topalli, 2021), whereas men are worried about employment and the economy far more. Mothers in Iceland also reported feeling “overwhelmed, frustrated, tired, annoyed, and angry” (Hjálmsdóttir & Bjarnadóttir, 2021, p. 274) with having to manage others’ fears and anxieties during the pandemic; and for a country at the top of the gender equity scale, mothers were frustrated with the lack of support. In Canada, mothers reported feelings of guilt and distress as they worked from home while also being full‐time caregivers (Smith, 2022). These feelings of burnout stem from the inequalities discussed above and point to a critical need for mental healthcare support during crises.

The literature cited above reveals that a substantial gap in domestic work amongst heterosexual partners existed prior to the COVID−19 pandemic, despite near‐parity in men's and women's market labour. This gap has narrowed somewhat in recent years but remains pronounced. At the same time, crises normally exacerbate rather than ameliorate inequalities; and early evidence from the COVID−19 pandemic suggests just that; women have experienced greater job losses than men, are performing more hours of unpaid labour than men, and are reporting feelings of burnout, exhaustion, and a lack of relationship satisfaction. Though studies using time‐use data show the overall gap in domestic labour hours, we still know little about who is performing which tasks and how many tasks, and about satisfaction with the household division of labour during the pandemic. Our analyses that follow help to fill in these gaps.

DATA AND METHODS

Beginning in March 2020, Statistics Canada periodically surveyed Canadians about the impacts of the COVID−19 pandemic, resulting in what they call the “Canadian Perspectives Survey Series.” They began with a sample of 4600 respondents. Their third wave of data collection (CPSS3), conducted between June 15 and June 21, 2020 (three months after the informal “start” of the pandemic in early March), focuses on “Resuming Economic and Social Activities During COVID−19.” Statistics Canada reached out to the same participants in each wave of the survey, but because different questions were asked in each wave, the data are more akin to cross‐sectional research. They note that “all of these surveys are statistically representative of the Canadian population” (Statistics Canada, 2020b).

The target population for CPSS3 is residents of the 10 provinces (excluding the three territories), 15 years of age and older. Excluded from the survey's coverage are persons living on reserves and other Aboriginal settlements, the institutionalized population, and households in extremely remote areas with low population density. These are hard to reach groups that together constitute less than 2% of the Canadian population (Statistics Canada, 2020a). The exclusion of Nunavut, Yukon, and the Northwest Territories results in the exclusion of much of Canada's Inuit population. These exclusions are problematic, and organizations like the Assembly of First Nations (2009), the Social Sciences and Humanities Research Council of Canada (2019), and the Truth and Reconciliation Commission of Canada (2015) problematize the exclusion of indigenous peoples from research, calling attention to a colonial history that continues to shape scientific data and policy in Canada, and call for an incorporation of indigenous methodologies and ways of knowing into western, colonial research practices (Morton Ninomiya & Pollock, 2017; Snow et al., 2016). At the same time, the three territories were not hit by COVID−19 until later the in summer, and so the data would have looked much different if they had been included in the study at this time. This results in an under‐sampling of the indigenous population, as 45% of all indigenous people live on reserves. The CPSS3 public‐use microdata file (PUMF) utilized a sampling frame of 7242 Canadians, and their efforts derived a 58% response rate for a sample size of n = 4209. For this analysis, we use the n = 2746 participants who lived with a partner and exclude both the n = 1459 who did not and the n = 4 who did not answer the cohabitation question.

The tables that follow include both Ordinary Least Squares (OLS) regression models in cases where the dependent variable is ratio‐level and Logistic Regression in cases where the dependent variable is binary (Yes/No) We make use of several key dependent variables in the analysis. First, the survey asked participants to reflect upon who did 11 specific household and family tasks. These tasks included preparing daily meals, doing housework, doing the dishes, doing the laundry, grocery shopping, taking care of household finances and paying bills, playing with children, putting children to bed, home‐schooling or helping children with homework, staying home with the children, and taking children to or from school/daycare. We have chosen not to use taking children to or from school/daycare, as there are many missing cases (presumably because most schools and daycares were closed at this point in the pandemic). For each of these questions, participants could answer: “Mostly you” [indicating the participant], “Mostly your spouse or partner,” “Shared equally between spouse or partner,” or “Always or usually someone else.” From these, we created a binary variable, representing whether the participant answered Mostly Me (1) or Someone Else (0). We chose to code “shared” along with “someone else” so that “Mostly Me” might remain alone and would allow us to hone in on participants who did a disproportionate share of the labour, relative to those who did not. We also created two additive scales with these variables. For those cohabiting participants with kids (n = 827), the scale ranges from 0 to 10 (since all of the items are valid for parent participants). For all cohabiting participants (n = 1919), we have a scale ranging from 0 to 6, which excludes the household tasks involving kids.

The survey then asked: “How satisfied are you [at month 3 of COVID−19] with the way household tasks are divided between you and your spouse or partner?” Participants could answer: “Very Dissatisfied,” “Dissatisfied,” “Neither Dissatisfied nor Satisfied,” “Satisfied,” or “Very Satisfied.” For analyses in Logistic Regression, we recorded this variable into Satisfied (including very satisfied and satisfied) and Not Satisfied (including the other three options). Participants were then asked: “How does your satisfaction with this division of household labour between you and your spouse or partner compare with how it was prior to the COVID−19 pandemic?” Participants could answer: “Much better than before,” “Somewhat better than before,” “About the same,” “Somewhat worse than before,” or “Much worse than before.” We recoded this as a binary variable, with Yes (including much worse than before and somewhat worse than before) and No (the other three options).

Our models use a combination of independent variables, including Female (1 = Yes; 0 = No; Male). We wish we had access to a more inclusive and useful measure of gender than the inadequate sex identification of “male/female” (see Footnote 1), and there is work in the field helping survey methodologists to devise such measures (Compton et al., 2018; Magliozzi et al., 2016). We urge Statistics Canada to make use of such measures. In this paper, we utilize the variable available in the survey; and while we are limited to the measure of “sex,” we focus the paper on a gender analysis, given that household and family labour are culturally gendered and gendered in the division by sex. The survey asked neither the sexual orientation of the participant nor the gender identification and sexual orientation of the participant's spouse. However, in Canada, 99% of cohabiting couples are opposite‐sex, according to Statistics Canada (2017) so assuming heterosexual pairings would be incorrect about 1% of the time—too little to affect the quantitative regression results below in a noticeable way.

The models also include Age (an ordinal variable with seven categories), whether the participant has a child under 18 residing in their dwelling (Yes = 1; No = 0), Immigrant (1 = Yes; 0 = No, Born in Canada), Married (1 = Yes; 0 = No), Household size (number of residents ranging from 1 to 5+), and whether participants have a University Degree or higher (1 = Yes; 0 = No). We also use current employment status (1 = Employed; 0 = Not Employed), though the survey did not differentiate between full‐time and part‐time employment. The “employed” category would include those who are both working in‐person and who are working remotely. Race, ethnicity, and indigeneity were not recorded by the survey, which reflects Canada's long history of colonization and the default whiteness of many survey instruments. The survey also did not ask numerous potentially useful questions about the respondent's spouse, including the employment situation of a respondent's spouse.

The PUMF contains sampling weights, as the design itself was not self‐weighted, and the analyses in this paper utilize these weights. Although the sample is highly representative of the off‐reserve Canadian population, we nevertheless employ the person‐weight, as recommended by Statistics Canada to correct for sampling error. Both the OLS and the Logistic Regression models utilize robust standard errors to correct for heteroskedasticity, lending a conservative bias to model estimates.

ANALYSIS

Inequalities in household tasks performed during COVID

Table 1 presents three ordinary least squares (OLS) regression models. The first two utilize an additive scale of the number of household tasks performed during COVID by “mostly me.” Here the survey asked participants who in their household did 10 different tasks during the first three months of COVID−19. If a person answered “mostly me,” then we coded it as 1. If it was shared or performed mostly by a partner, we coded it as “0” for “someone else.” The resultant scale variable ranges from 0 to 10, reflecting how many of the tasks were “mostly me.” We included only parents here because the childcare tasks (home‐schooling, putting children to bed, etc.) are solely valid for parents. In the third model, though, we included everyone to analyze non‐parenting tasks that were potentially available to all participants.

TABLE 1.

OLS regression models of number of tasks done by “mostly me”

	Model 1 – Total Tasks That Were “Mostly Me” – Parents Only	Model 2 – Total Tasks That Were “Mostly Me” – Parents Only	Model 3 – Total Tasks That Were “Mostly Me” – All Participants
	B (robust s.e.)	B (robust s.e.)	B (robust s.e.)
Female	2.63***	2.59***	1.18***
	(0.282)	(0.703)	(0.107)
Employed	−1.20***	−1.23*	−0.247*
	(0.372)	(0.596)	(0.120)
Age	0.460**	.460**	−0.019
	(0.160)	(0.160)	(0.042)
Married	1.21	1.21	0.474
	(0.627)	(0.630)	(0.270)
University Degree	−0.102	−0.102	−0.161
	(0.285)	(0.285)	(0.102)
Household Size	−1.08***	−0.108***	0.032
	(0.254)	(0.260)	(0.092)
Immigrant	−0.282	−0.284	−0.256*
	(0.313)	(0.307)	(0.129)
Urban	0.441	0.442	0.158
	(0.333)	(0.338)	(0.129)
Female*employed	–	0.052	–
		(0.767)
Parent	–	–	−0.052
			(0.125)
Constant	2.88**	2.91**	1.51***
	(1.01)	(1.13)	(0.337)
R 2	0.357	0.357	0.130
N (listwise deletion)	555	555	2660

***p < 0.001; **p < 0.01; *p < 0.05.

Model 1 demonstrates that women reported being primarily responsible for significantly more of these 10 household tasks than men, all else equal. The unstandardized regression coefficient (B) indicates that women perform on average 2.63 more of these tasks than men (p < 0.0001). And employed participants do 1.20 fewer household tasks than those who are not employed (p < 0.001), keeping in mind there could also be differences between those employed part‐time and full‐time, had we been able to account for it. Age is also significant, with each step up in age category associated with performing about half of an additional task. Household size is also significant, with larger households associated with fewer tasks that are done mostly by the participant.

Because of the gendered nature of both paid employment and household work, we ran an interaction term for female*employed, and inserted it into the model, which should theoretically help to discern if the gender effect found in Model 1 relates to the interaction (i.e. the competing demands) of gendered domestic work and employment, in ways that are unique or specific for employed women. Model 2 provides results from that analysis. This interaction leaves the original model largely unchanged; gender and employment remain significant, but the interaction term is not significant. This indicates that both gender and employment exert independent effects on the number of tasks for which participants were primarily responsible; and ultimately that gender matters for these tasks, even controlling for employment status and, in turn, employment status matters even taking into account gender. The standardized coefficients do indicate that gender (0.457) contributes relatively more than employment (−0.192) in explaining the number of tasks performed.

Because some of the tasks are valid only for parents (e.g. putting children to bed), Model 3 uses a different dependent variable, which is a scale including only the six tasks valid for all participants (grocery shopping, dishes, household finances, etc.) with none of the five childcare variables. We can therefore add in a variable for whether participants were parents (1) or not (0).

Results of Model 3 indicate that, once again, gender is significant, with women reporting primary responsibility for 1.18 more of these six tasks than men, all else equal (p < 0.0001). Parenthood is not significant, suggesting that parents are not collectively performing more of these household tasks by themselves than non‐parents. However, this finding was not equally true for men and women, as the bivariate data shows that mothers reported performing a mean of 3.02 of these six tasks mostly themselves, while fathers mostly performed 1.59, on average, a significant difference (p < 0.0001). In Model 3, employment is again significant, with employed participants primarily responsible for about one‐quarter—or about 2.5—fewer tasks than non‐employed (p < 0.04), all else equal.

The overall results from Table 1 tell us that, among participants with kids and those without kids, women reported performing more everyday household tasks on their own than men three months into the COVID−19 pandemic, controlling for all other variables in the models. Likewise, participants with paid employment performed fewer tasks on average than those without, with 53% of the women employed, relative to 58% of men. Importantly, gender mattered for household labour even while controlling for employment. That is, women are taking on the lion's share of household work whether or not they are employed. Consistent with the emerging body of literature on the social consequences of COVID−19, our findings suggest that the pandemic is a gendered event that burdens women—particularly mothers—in a familiar but acutely problematic way (Collins et al., 2020a; Landivar et al., 2020; Power, 2020).

As with any analysis, the non‐significant coefficients are just as instructive as the significant ones. In these models, having a bachelor's degree or higher is not predictive of the number of household tasks performed by “mostly me.” Though the literature points toward the liberalizing effect of a university education on gender roles (Shu & Meagher, 2018), this did not show up in the data. University educated participants were just as likely as those without a university education to pick‐up more tasks at home while living with a partner, perhaps because those with a college degree are more likely to work remotely (Dey et al., 2020). Likewise, marital status and urban/rural location did not matter for household tasks at this time. Despite a relatively small number of predictors included in the model, the explanatory power is notable, with 36% of the variation in the dependent variable explained (R ² = 0.357) in the first two models, and 13% (R ² = 0.130) in the third model.

Inequalities in specific household tasks during COVID−19

Here, we break out the household tasks individually to look at what factors help to explain who performed each. We focus specifically on housework (for all participants living with a spouse/partner), meal preparation (for all participants living with a spouse/partner), playing with the kids (for parents), managing the kids’ homeschooling (for parents), and staying home with the kids (for parents). We chose these specific variables to ensure parsimony, and because these five variables represent some of the most time‐consuming and labour‐intensive household work. Table 2 presents a regression model for each of these variables, using the same independent variables as above.

TABLE 2.

Logistic regression models of whether five individual household tasks were done by “mostly me” or by someone else

	Model 1 – Housework was “Mostly me” – All Participants	Model 2 – Meal Prep was “Mostly Me” – All Participants	Model 3 – Playing with Kids was “Mostly Me”—Parents Only	Model 4 – Homeschool was “Mostly Me” – Parents Only	Model 5 – Staying Home w/ Kids was “Mostly Me”—Parents Only
	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)
Female	4.92*** (0.723)	4.61*** (0.633)	3.16*** (0.950)	6.64*** (1.94)	10.12*** (2.82)
Employed	0.654** (0.101)	0.811 (0.124)	0.452*** (0.123)	0.454* (0.151)	0.276*** (0.085)
Age	0.928 (0.058)	0.996 (0.058)	1.15 (0.178)	1.04 (0.173)	1.15 (0.174)
Married	1.37 (0.503)	1.69 (0.525)	1.02 (1.03)	0.491 (0.623)	3.44* (2.15)
University Degree	0.858 (0.117)	0.956 (0.127)	1.15 (0.266)	1.25 (0.358)	0.872 (0.238)
Household Size	1.11 (0.123)	0.992 (0.105)	0.883 (0.221)	1.08 (0.318)	0.466** (0.146)
Immigrant	0.893 (0.162)	0.822 (0.147)	0.723 (0.208)	0.628 (0.226)	1.03 (0.318)
Urban	1.28 (0.202)	1.17 (0.181)	1.30 (0.431)	1.17 (0.378)	2.00* (0.637)
Parent	0.851 (0.156)	0.790 (0.135)	–	–	–
Constant	0.240*** (0.110)	0.206*** (0.088)	0.281 (0.360)	0.628 (0.944)	0.254 (0.268)
Pesudo R 2	0.115	0.104	0.091	0.183	0.267
N (listwise deletion)	2686	2695	771	645	717

***p < 0.001; **p < 0.01; *p < 0.05.

Table 2 provides five logistic regression models looking at these five household tasks. Model 1 looks at who performed the housework three months into COVID−19, and more specifically which respondents answered: “mostly me.” The odds ratio for “female” (4.92) indicates that women were about five times more likely than men to report that housework fell mainly on them (p < 0.0001), all else equal. Employment is once again significant and decreases the odds of answering that housework was done by “mostly me.”

Model 2 of the table looks at who did meal preparation during the pandemic and, once again, women are some 4.6 times more likely than men (e^b = 4.61, p < 0.0001) to say that this task fell mainly on them. For these two variables, valid for both parents and non‐parents, there is a clear pattern whereby gender matters, and whereby household tasks are much more likely to be performed by women than men.

The last three columns, Models 3 through 5, are valid for only the parents in the sample. Here, the gender effect is amplified, with women being three times more likely to play with kids than men (e^b = 3.16, p < 0.0001), but being nearly seven times more likely than men to manage kids’ schooling (e^b = 6.64, p < 0.0001) and ten times more likely than men to say that staying home with children fell mainly on them (e^b = 10.12, p < 0.0001). This highlights just how drastically parenting has differed for mothers and fathers during COVID−19, in terms of both time and kind. In short, we find that during the pandemic, domestic work was highly unequal for all the women, but amplified for those managing online schooling and staying home to provide kids with care and entertainment. These findings reveal an extreme gendering of household and family labour not reflected in previous pre‐COVID scholarship. They signal a mothering effect of managing the consequences of COVID−19 on home and family, with significant risks to mothers’ employment and later work opportunities, which also severely impacts their negotiating power within the household.

Satisfaction with Division of Household Work During COVID−19

Were women dissatisfied with this highly unequal division of labour? The results are surprising given just how much women reported being mostly responsible for the bulk of domestic work and what we know about how an unequal division of labour impacts women's mental health. Simple descriptive statistics show that 64% of women report being satisfied with their household division of labour three months into the pandemic, compared to 72% of men (p < 0.001). Along similar lines, 63% of parents are satisfied with this division of labour, compared to 70% of non‐parents (p < 0.01).

In terms of changes, we see that only 6.66% of women said their division of labour got worse during the early months of the pandemic, compared to 4.07% of men (p < 0.001). The major difference here is between parents and non‐parents, whereby 10% of parents said it got worse, compared to 3.52% of non‐parents. This finding is higher for mothers than fathers: 11.79% of mothers saying this division of labour got worse, compared to 7.36% of fathers (p < 0.032). For non‐parents the difference is smaller (4.35% for women and 2.68% for men) and marginally insignificant (p < 0.050) These percentages are small, with few women directly reporting dissatisfaction with the highly unequal division of household labour.

Table 3 includes six Logistic Regression models looking at both satisfaction with the household division of labour at the time of the survey, as well as a variable measuring whether the respondent felt the household division of labour got worse or not during the first three months of COVID−19.

TABLE 3.

Logistic regression models of satisfaction with household division of labour and perception of whether division of labour got worse during COVID

	Model 1 – Satisfied with HH Division of Labor	Model 2 – Satisfied with HH Division of Labor	Model 3 – Satisfied with HH Division of Labor	Model 4 – HH Division of Labor Got Worse	Model 5 – HH Division of Labor Got Worse	Model 6 – HH Division of Labor Got Worse
	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)	Exp(b) (robust s.e.)
Female	0.689*** (0.088)	0.741 (0.116)	1.06 (0.152)	1.622 (0.454)	0.791 (0.317)	0.938 (0.300)
Employed	0.919 (0.135)	0.910 (0.134)	0.829 (0.127)	0.980 (0.313)	1.04 (0.331)	1.095 (0.356)
Parent	0.812 (0.133)	0.901 (0.200)	0.817 (0.135)	1.91* (0.601)	0.822 (0.384)	1.985* (0.600)
Age	1.09 (0.062)	1.09 (0.062)	1.11 (0.064)	0.827* (0.081)	0.826* (0.081)	0.829 (0.088)
Married	0.718 (0.239)	0.710 (0.236)	0.818 (0.281)	1.05 (0.561)	1.11 (0.607)	0.772 (0.418)
University Degree	1.22 (0.162)	1.22 (0.162)	1.14 (0.154)	0.980 (0.272)	0.975 (0.272)	1.09 (0.300)
Household Size	0.942 (0.100)	0.942 (0.101)	0.949 (0.105)	1.301 (0.246)	1.31 (0.255)	1.30 (0.242)
Immigrant	1.38 (0.328)	1.38 (0.239)	1.20 (0.211)	1.05 (0.369)	1.04 (0.90)	1.30 (0.453)
Urban	1.33 (0.208)	1.330 (0.206)	1.45 (0.224)	1.03 (0.325)	1.07 (0.334)	1.00 (0.333)
Female*parent	–	.820 (0.223)	–	–	4.08** (2.25)	–
Total Jobs “Mostly Me”	–	–	.681*** (0.028)	–	–	1.530*** (0.155)
Constant	2.65** (1.14)	2.60** (1.13)	4.96*** (2.23)	.042*** (0.031)	.056*** (0.042)	.019*** (0.016)
Pesudo R 2	0.019	0.020	0.086	0.047	0.061	0.105
N (listwise deletion)	2693	2693	2652	2650	2650	2610

***p < 0.001; **p < 0.01; *p < 0.05.

Results in Model 1 indicate that women are 31% less likely than men to say they are satisfied with this division of labour (e^b = 0.689, p < 0.001), controlling for other variables in the model. In Model 2, we add an interaction term for female*parenthood, and this erases the significant effect of gender, but the interaction term is not significant either. This suggests that satisfaction is best explained by gender (in Model 1), not by the inclusion of an interaction for gender and parenthood. Though gender matters for satisfaction in Model 1 (controlling for parenthood), Model 2 shows us that it is not specifically the mothers who are less satisfied, but women more generally – keeping in mind the low overall levels of dissatisfaction discussed above. Model 3 adds in the total number of household tasks performed by “mainly me” during COVID, ranging from 0 to 6 (valid for parents and non‐parents). This variable is significant, with each additional task primarily performed, decreasing the odds of satisfaction by about 32% (e^b = 0.681). This indicates that the number of tasks performed is driving dissatisfaction, not gender itself. However, as we saw in Table 1, the total number of tasks performed is significantly affected by gender, indicating that a multi‐step casual process is at play and key to understanding the gender of satisfaction with the division of household tasks during COVID−19. In other words, women report performing more of these household tasks, and the number of tasks performed, in turn, has a significant effect on satisfaction.

Model 4 provides a Logistic Regression model of whether this division of household labour got worse during COVID−19, or not. Sex is not significant in this model, nor any of the models predicting whether the household division of labour “got worse” during COVID. In Model 5, we again add the female*parent interaction term, and this time the term is significant (p < 0.01), meaning that the unique interaction of gender and parenthood—not sex alone—explains for whom the household labour got worse, and this perception was held disproportionately by mothers. The non‐significant sex coefficient, coupled with the significant interaction term, indicate that while women did not necessarily feel that the household division of labour got worse during the pandemic, vis‐à‐vis men, mothers are a group that did feel that the division of household labour deteriorated during the first three months of COVID. This means there is a more nuanced story regarding motherhood here, specifically, and the care work that mothers performed in the pandemic, relative to both fathers and women without children.

Finally, Model 5 adds in a variable for the total number of household tasks that fell on “mainly me,” ranging from 0 to 6. This variable is significant, and the odds ratio (e^b = 1.530) indicates that each additional household task increases the odds of saying the division of labour got worse during the pandemic by 53%. Importantly, parents in this model were twice as likely as non‐parents to report that the division of labour became increasingly unequal with each additional household task (e^b = 1.985, p < 0.05). Overall, it is important to note that Models 4 through 6 are the weakest explanatory models in the table—and in the study—largely because such a small segment of women (6.66%) and men (4.07%) felt that the division of labour had worsened during COVID−19.

There is much to unpack from Table 3, but these findings tell us several things. First, women report less satisfaction with the division of domestic work than men, at month three of the pandemic, though this finding depends heavily upon model specifications and disappears in some of our models. The remainder of models in the table show that perceptions that the household division of labour “got worse” depend much more on the interaction of gender and parenthood, revealing the unique burden placed upon mothers vis‐à‐vis fathers and women without children. Second, parents generally are more likely than non‐parents to perceive that the domestic division of labour got worse from the start of the pandemic to the third month. The perception that the division of unpaid household labour worsened is explained by the interaction of gender and parenthood, whereby mothers in particular felt the burdens of a worsening division of household tasks. And lastly, the table demonstrates that being mainly responsible for more household tasks is related to higher odds of perceiving that the household division of labour got worse.

CONCLUSION

It is striking to see that in our study, 64% of women reported being satisfied with the highly unequal division of household tasks and family care work early on in COVID−19. Men and women in post‐industrial societies often feel the freedom to “indulge” in essentializing gender expectations, as well as stereotypical and structurally constraining experiences (Charles & Bradley, 2009). Gender here is seen as an expression of self, with clear impressions that we experience freedom of choice. There is also the problem of women serving as default caregivers, who pick up the burdens of hearth and home without much consideration from their partners (Calarco et al., 2021). And those mothers who experience the pressures to intensively mother (e.g. Hays, 1998; Nomaguchi & Milkie, 2017; Rizzo et al., 2013) might invest deeply in their kids at this time, feeling their kids need them now more than ever. At the same time, intensive mothering becomes compulsive or completely involuntary when daycares and schools close, so that moms have to work from home with kids on their laps or kids in their cars if they are in the growing gig economy. All of this can lead to a sort of resignation that we will occupy gender segregated roles and work; and so, while it's disappointing, it's also not all that surprising to see this rate of satisfaction. We might expect to see lower satisfaction amongst employed women, especially since they often struggle to hold together the household while also in paid work. Of course, since the pandemic has continued for much longer than many people expected, it would be worth following‐up with participants to see if this finding has endured.

Regarding feeling that the household division of labour got worse during early COVID−19, here again we find the female*parent interaction significant, suggesting a unique effect of motherhood on perceptions of the division of household labour, keeping in mind that the raw percentages of both men and women who said that the division of labour got worse were very low. This corresponds with pre‐pandemic findings that mothers may not only pay a “tax” in terms of a wage penalty (Anderson et al., 2003; Budig & England, 2001; Waldfogel, 1997, 1998), but also in terms of overall life satisfaction (Donath, 2015; Giesselmann et al., 2018). There are implications here for women's happiness in the home, with their partners, and with their paid work performances. Though Hochschild's (1989; 1997) “second shift” and “time bind” theories provide a useful starting point for explaining the unequal division of household labour between employed men and women, they fall short of helping us to understand how crises such as the COVID−19 pandemic further widen, not narrow, this gap through the simultaneous performance of paid work and childcare.

The above analyses here have limitations, of course. As we mention in the methods section, a more inclusive measure of gender would aid in teasing out a more nuanced analysis of labour during Covid−19, as would a measure of race/ethnicity. Likewise, the models above would benefit from a variable for the age(s) of children, for spousal employment, as well as for respondent's occupation, as particular occupations were deemed “essential” during the pandemic, perhaps exacerbating strains on families and making it more difficult to balance the demands of paid work and household responsibilities. Likewise, a pre‐pandemic wave of survey data would have allowed for a clearer argument that the gendered division of household labour had actually “gotten worse” for women, but as is the case with many disaster/crises analyses, no pre‐event data exist. We also wish that Statistics Canada would have asked this same question in future waves. Three months into the pandemic, many people still viewed the pandemic as time‐delimited and temporary, expecting a quick return to normal. This expectation may have made them less likely to report dissatisfaction with their household division of labour, as it might be more palatable to perform highly unequal work if there is an end in sight. However, as the pandemic approached one year in duration, and then two, we may have seen satisfaction fall considerably; however, the June 2020 data do not capture this.

Even in Canada, a country with more progressive social policies and a political climate embracing somewhat more social democratic traditions than the U.S., women found themselves burdened by an intensified labour compression, with consequences likely to reverberate for years negatively affecting their employment, financial independence, and mental well‐being. Shared pressures that have come with a loss of institutional support highlight the need for more or better social safety nets and attention to these remaining issues after the crisis has passed. This entails asking ongoing questions about the heightened division of household labour and why it is that women still must perform so much unpaid labour for the family as a capitalist institution and thus for the economy to function. Our findings related to satisfaction indicate some degree of naturalization of patriarchal arrangements and understandings of motherhood as self‐sacrificing. This reflects cultural ideas that women's opportunities and careers come second to men's and spotlights the resiliency of patriarchal dividends.

But what policies are in place to support mothers in times of crises? And how do we urge cultural shifts away from one where women are held mainly responsible for the home? Progressive policies around parental leave allow for a more equal balance of labour between mothers and fathers during normal times (Collins, 2019), and this can also be true in crises. Existing work in the field often rightly emphasizes policy interventions to create more egalitarian recoveries along lines of race and class (Birkland, 2006; Kousky, 2017), but too little work has focused on policies that protect women's paid employment during and after crises, and that prompt men to take on more domestic labour. Examples include paid parental leave for fathers (which Canada has), more robust financial and childcare disaster and emergency programs, free or subsidized mental healthcare, pay equity legislation, additional paid sick leave, and the creation and maintenance of governmental committees or task‐forces dedicated to the status of women and parents—and which are led by women and parents.

Haney, T.J. & Barber, K. (2022) The extreme gendering of COVID−19: Household tasks and division of labour satisfaction during the pandemic. Canadian Review of Sociology/Revue canadienne de sociologie, 59, 26–47. 10.1111/cars.12391