Literature on the impact of union dissolution on household tasks

9Studies on housework, the descendants of time-use studies, focus mainly on the division of tasks between spouses at different stages in family life. They aim to provide data on equality between spouses, household production, and the reconciliation of work and family life (Coltrane, 2000). From the early 1990s, the popularity of the “doing gender” theory (West and Zimmerman, 1987) in explaining the unequal division of household tasks between men and women doubtless contributed to orienting research in this direction. This theory, founded on symbolic interactionism and ethnomethodology, treats gender less as a collection of individual traits than as the result of social interactions. Performing certain activities is a way for men and women to affirm their gender identity and to show others and themselves that they are competent members of their sexual category who are capable of the corresponding expected behaviours. In western countries, this is supposedly illustrated by the fact that the gap between men and women in amounts of housework, already found in people living alone, increases with entry into cohabitation and, even more so, with parenthood (for international comparisons, see Finch, 2006; Gershuny and Sullivan, 2003; Sullivan and Gershuny, 2001). In these situations, women tend to work fewer hours in paid employment and increase the time they spend on household and family work, whereas the opposite tendency is found in men. As an illustration, in 2010 in Switzerland, women aged 15-65 living alone spent 32.8 hours per week on paid work and 16.8 hours on domestic and family work (versus 33.9 and 14.6 hours for men). When women live with a partner and without children, their domestic time increases at the expense of working hours: 28.8 hours of paid work and 20.9 hours of domestic and family work per week (versus 35.2 and 13.6 hours for men). The ratio is reversed when women live with a partner and have children under 6 years old: 11.8 hours per week of paid work versus 55.6 hours of domestic and family work (40.1 and 29.4 hours for men). [10] Although significant, these figures are based on cross-sectional data, which limits their explanatory power as they compare different individuals and households.

10Longitudinal studies, which can be used to track households and their members over time, overcome the biases inherent to cross-sectional studies (Gupta, 1999, p. 710), offering a better understanding of factors that influence the division of household tasks within couples. However, few such studies have been performed (Artis and Pavalko, 2003; Baxter et al., 2008; Cooke, 2004; Gupta, 1999; Henchoz and Wernli, 2010; Sanchez and Thomson, 1997), as data collection requires substantial infrastructure and processing. Furthermore, available longitudinal studies provide little information on the impact of union dissolution on hours spent performing household tasks.

11To our knowledge, no cross-sectional or longitudinal study has been specifically devoted to examining the impact of union dissolution on the time spent by men and women on household tasks in Switzerland or France. However, cross-sectional analyses including different types of households (Schön-Bühlmann, 2006b and 2009) show that women in Switzerland who live alone spend less time than women in couples without children. Single mothers spend less time on domestic and family tasks than women in couples with one or more children (Schön-Bühlmann, 2006b), which seems to confirm the results of an American study (Bianchi et al., 2006). This may be due to single mothers’ greater investment in paid work, and to their children’s higher average age – older children requiring less care (Gerfin et al., 2009; Schön-Bühlmann, 2006b).

12Conjugal life appears to have little influence on the hours that Swiss men spend on housework. According to the available cross-sectional analyses, men who live alone spend about as much time performing domestic and family tasks as men in couples without children. Single fathers apparently spend almost as many hours as fathers with a partner (Schön-Bühlmann, 2006b, p. 40).

13The few available longitudinal studies seem to confirm the trend indicated by cross-sectional surveys for women. The results for men are less clear, however. In Australia, married women perform more housework than widowed, separated, divorced or never-married women (Baxter et al., 2005). After separation, women spend less time on household tasks, while men spend more (Baxter et al., 2008; Gerner et al., 1990; Gupta, 1999). An Australian longitudinal study (Baxter et al., 2005, p. 598) shows that separated, divorced or widowed men perform more housework than married men.

The principal factors explaining the division of domestic work

14Several factors explain the time devoted to domestic tasks (for a review see: Coltrane, 2000; Gerfin and Strub, 2009; Schön-Bühlmann, 2009). These factors are generally cited to account for the division of tasks within the couple during a specific period in family life. They thus present two drawbacks for our study. First, they offer no predictions concerning variations, because they explain a state at a given time. Second, the division of household tasks within the couple is irrelevant to our study because the partners no longer live together. This article will therefore also examine the extent to which these factors are pertinent for explaining variations in time spent on household tasks following union dissolution. The three levels that we distinguish here for expository reasons are in reality intrinsically linked and interdependent.

15• Individual factors: these include the characteristics and resources specific to each individual. Sex, age, level of education or employment situation, and personal income are factors found to have an essential influence on time spent on domestic work (Schön-Bühlmann, 2009). A previous longitudinal study (Henchoz and Wernli, 2010) showed that in Switzerland, personal resources that are socially valued (such as income, level of education and professional status) have little effect on men’s domestic involvement over the course of the family building process. This is not the case for women, who draw on their income to pay for household help in order to reduce their own domestic workload. This partly explains the observation that the time women spend on such work varies widely.

16• Interactional and family factors: these are factors whereby the presence or absence of other people is assumed to modify individual behaviour. For example, the number of dependent children influences partners’ household involvement (Schön-Bühlmann, 2009). The presence of a partner also leads to interactions that can change the time devoted by each person to domestic work. According to the theories of human capital (Becker, 1981), economic dependence (Delphy and Leonard, 1986) or resources (Blood and Wolfe, 1960), domestic work is a source of drudgery to be avoided. Persons who possess socially valued resources (a high income, for example) will arrange to negotiate a lesser share of household tasks. These theories are based on the implicit logic that housework is divided according to the economic or quasi-economic rules of instrumental exchanges, and that these rules and principles of exchange are gender-neutral (Brines, 1993, p. 303). For others, such as the partisans of the doing gender theory (West and Zimmerman, 1987) briefly presented above, the daily construction of gender is, on the contrary, the central explanatory factor in the distribution of household tasks (Coltrane, 2000). According to this theory, the division of work within the couple reflects “the material embodiment of wifely and husbandly roles, and derivatively, of womanly and manly conduct” (West and Zimmerman, 1987, p. 144). There is thus a symbolic dimension to the performance of household tasks: they are a way for individuals to show that they have adopted the behaviour appropriate to their sexual category. Gender is “done” differently, therefore, depending on the people with whom an individual interacts, since expectations vary if the person is a man or a woman, single or a parent. Through differentiation in their performance of paid and unpaid work, men and women also differentiate themselves in gender terms. They express and demonstrate their love for and interest in their family members (Ferrand, 2004; Sanchez and Thomson, 1997; Zarca, 1990). Men’s professional engagement and the quality of women’s domestic work can also be seen as traditional symbolic reaffirmations, for men as good fathers and husbands, and for women as good mothers and wives (Berk, 1985).

17• Contextual or structural factors: these highlight the fact that the division of household tasks is also determined by opportunities and constraints resulting from the context in which families live. Individual choices depend not only on personal or family preferences, but also on the options available in society and the likelihood that their choices will be accepted (Krüger and Levy, 2001, p. 153). People must make compromises that allow them to reconcile their behaviours and desires with institutional constraints and opportunities that are external to the family (laws on equality, availability of child care, structure of the labour market, etc.).

18Contextual factors include, among others, the influence of culture, of public policy and of economic and social development (notably in terms of gender equality). [11] In a Norwegian time budget study designed to reveal the effect of structural factors, Grønmo and Lingsom (1986) highlighted the central role of ideological and cultural factors in determining the time spent on housework. Certain Swiss studies have also shown differences in the distribution of household tasks between linguistic regions (Bonoli and Gay-des-Combes, 2005; Sousa-Poza and Widmer, 1998). More generally, the structure of Swiss society, and notably the labour market, result in a model of the work-family balance that favours full-time employment for men and a female working schedule that is tailored to family needs (Krüger and Levy, 2001; Levy et al., 2006; Widmer et al., 2003a; Widmer et al., 2004; Widmer et al., 2003b). In Swiss families, as elsewhere in Europe, domestic work remains women’s work, independently of individual factors such as generation, income, or the respondents’ family situation (Aliaga, 2006).

19In this study, individual factors will be studied for each transition. Structural factors, notably ideological and cultural factors, will be taken into account through the notion of generation, which we define below. Interactional factors, such as the influence of the presence of a partner, will be examined via the different transitions involved in union dissolution.

Transitions examined

20The analysis of SHP data is based on all the transitions of waves 1 to 11 (1999 to 2009). As we have previously shown with regard to the family building process (Henchoz and Wernli, 2010), the time devoted to household tasks is influenced less by marital status than by living arrangements (entry into cohabitation or arrival of children), so we have chosen to look at the end of union from this angle. Furthermore, changes in marital status at the end of union are neither obligatory nor instantaneous. However, we do differentiate the causes of union dissolution – i.e. separation or death of one of the partners – in order to measure any possible impact of this factor.

21Death and separation can lead to different types of family reconfiguration that must be distinguished:

• the transition from living with a partner to living alone;
• the transition from family life to living alone;
• the transition from family life to single-parent family life.

The population

22As Table 1 shows, the above-mentioned transitions are not equally common among men and women of all ages. For example, most of those who undergo a transition from family life to living alone are fathers, whereas those who go from family life to single-parent family life are mostly mothers. Those who have experienced union dissolution due to the death of their partner are generally among the oldest respondents.

23For ease of presentation, we chose to distinguish three generations, based on the criteria of Attias-Donfut (1997). The “young generation” is composed of persons born in 1955 or later (under 45 years old at the start of the panel). The “pivot generation” includes people born between 1935 and 1954 (aged 45-64 in 1999). The “elder generation” includes respondents born before 1935 (aged 65 years or above in 1999). This choice reflects the notion of family generations, the pivots being those who can be both the children of the elder generation and the parents of the young generation. Furthermore, these different generations have undergone different experiences which “give them common social and historical references that will shape their conceptions of the world” (Attias-Donfut, 1997, p. 43). This division into three cohorts allows an analysis of groups that are relatively homogeneous in terms of labour market participation (active or otherwise), type of relationship (marriage/cohabitation), family organization, and conceptions of family and of gender relations. In this sense, this categorization provides a means to explore the impact of norms on the division of domestic and family work. These factors, which we have classified as structural or contextual factors, may explain the consequences of the partner’s departure on the organization of domestic work. As shown by our analysis of the division of household tasks over the course of the family building process (Henchoz and Wernli, 2010), these norms have a decisive impact on the performance of household tasks. For example, all other things being equal, it is the oldest women who invest the most time in such tasks.

24Out of the 455 transitions observed by the SHP between 1999 and 2009, we analyse only those affecting at least 20 people (cells with a grey background in Table 1), as the others cannot reliably be interpreted. Thus, in this analysis, the transition from family life to living alone will only be examined for men in the young generation, and the transition from family life to single-parent family life for women in the young generation.

Table 1

Number of men and women by generation for each type of union dissolution and transition

Generation  1955 onward 1935 to 1954 Before 1935 Separation Death Separation Death Separation Death Transition from living with a partner to living alone Women 70 1 24 30 6 41 Men 81 2 22 5 4 9 Transition from family life to living alone Women 4 0 2 0 0 0 Men 45 0 6 0 0 0 Transition from family life to single-parent family life Women 81 3 4 0 0 1 Men 12 1 1 0 0 0

Number of men and women by generation for each type of union dissolution and transition

25The only transition that can be examined for all generations is that from living with a partner to living alone. In this case, the distinction between three generations makes it possible to model a non-linear effect of birth cohort on our dependent variable, which our analyses will show to have been a judicious choice. It will thus be a matter less of comparing the different generations than of examining the transitions that most often affect these different generations.

Details of the method

26The effect of the above-mentioned transitions will be examined while controlling for other parameters generally recognized to have an influence (Henchoz and Wernli, 2010), namely working time, [12] level of education, [13] net personal income [14] and, where applicable, the number of dependent minors. Because the age of the youngest child could not be used due to problems of collinearity (Pearson’s r of 0.78 with the respondent’s age for all parents), we included a dichotomous indicator reflecting the presence in the household of one or more children below age 7, who generally generate a greater amount of domestic work for parents. While we do not look at marital status at the time of the transition, we do make use of a relative indicator of the type of union before transition, [15] in order to take investment in the relationship and family life into account.

27We also included the use of domestic help [16] in our analyses as well as an indicator of eating out in restaurants, [17] which are substitution strategies that could be used to reduce the time spent on household tasks.

28The duration of the union prior to transition is too highly correlated with age (Pearson’s r 0.81) to be included in the models. Furthermore, several other variables, such as linguistic region, use of paid home child care, or the fact of remaining in the same dwelling (or not) after the transition were also tested. They were not included in the models presented here, as they had no impact.

29These longitudinal analyses are based on the data accumulated from waves 1 to 11 (1999 to 2009) of the SHP, structured in the form of person-period or long format files, with the information on an individual represented in the form of several recordings.

30The sample for analysis is defined differently for each type of transition. The departure of the partner is studied, with different configurations, among persons initially living with a partner under the same roof. Once the event under study (separation or death of the partner) occurs, the relevant persons leave the field of observation, the goal being to determine the immediate effect of such events. They may return to the sample after going through the initial condition again (repartnering).

31We used a mixed linear model [18] (MLM), in which repeated observations of the same individual constitute a sub-level of the analysis that takes the form of a composite multilevel model for change (Singer and Willett, 2003). The continuous dependent variable is taken to be the linear sum of fixed and random effects. Fixed effects, or the structural component, have the same effect for all individuals, whereas random effects are randomly distributed among individuals, with a standard normal distribution (mean of 0) and unknown variance, but with a structure that can be modelled.

32In our models, these random effects make it possible to vary the y-intercepts of the fixed effects for each individual, allowing us to structure the residuals and correlate them in time. The chosen model is an autoregressive structural model [19] of the covariance of the residuals, a common choice for repeated measures, which means that for a given individual residuals are correlated from one observation to the next, but in a variable fashion, as a function of their proximity. The variables presented in the tables are chosen not in order to find the best model on the basis of technical criteria, but for purposes of analysis and comparison. Parameters that are not statistically significant but that are highly important for interpretive reasons were retained, in order to compare equivalent models for the different cohorts.

Transition from living with a partner to living alone

36We have a large enough sample to examine the transition from living with a partner to living alone for all generations. However, this transition is not generated by the same event in different generations. In our analysis, women and men of the young and pivot generations are led to living alone by a separation. Women of the pivot generation also experienced the death of a partner, as did women in the elder generation.

37The impact of the transitions studied in Table 3 confirms the bivariate analyses of Table 2: the death of a partner reduces women’s domestic workload. The impact is slightly greater for women in the elder generation (–7.0 h/week versus –6.1 h/week for women in the pivot generation).

Table 3

Impact of transition from living with a partner to living alone on domestic workload in Switzerland (hours per week)

Women Men Estimation Significance Estimation Significance Constant 23.025 0.00 8.784 0.00 Generation 1955 onward –6.683 0.00 +0.153 0.65 1935-1954 –2.018 0.00 +0.156 0.59 Before 1935 (Ref.) 0.000 0.000 Interaction between generation and type of union dissolution Death of the partner * 1955 onward –2.334 0.77 +2.128 0.54 Death of the partner * 1935-1954 –6.097 0.00 +6.247 0.00 Death of the partner * before 1935 –7.014 0.00 +7.512 0.00 Separation * 1955 onward –2.180 0.03 +0.851 0.13 Separation * 1935-1954 –4.697 0.01 +3.383 0.00 Separation * before 1935 –9.562 0.02 +2.181 0.37 Type of union before transition Married +2.340 0.00 –0.486 0.03 Not married (Ref.) 0.000 0.000 Year (wave) 1 to 11 (1999-2009) –0.064 0.02 +0.005 0.81 Working hours (0-100% full time) –0.068 0.00 –0.036 0.00 Level of education 0 to 10 –0.278 0.00 +0.045 0.19 Net personal annual income By multiples of 10,000 Swiss francs +0.007 0.36 –0.020 0.06 Paid household help Yes –0.729 0.02 0.273 0.17 No (Ref.) 0.729 0.000 Restaurant at least once per month Yes –0.481 0.02 –0.082 0.56 No (Ref.) 0.000 0.000 Number of observations 10,507 9,623 Number of different individuals 2,680 2,452 Fit of model -2 restricted log-likelihood 74,397 58,884 AIC 74,403 58,890 BIC 74,425 58,911 Individual random effect AR1 diagonal (covariance) 54.784 0.00 21.714 0.00 AR1 rho (correlation from one year to next) 0.076 0.00 0.171 0.00 Constant Variance of random effect 31.991 12.193 Estimation: Unstandardized fixed effects. Interpretation: Following a separation, the mean time spent on housework by women born in 1955 or later fell by 2.2 hours per week on average, all other things being equal.

Impact of transition from living with a partner to living alone on domestic workload in Switzerland (hours per week)

38The effects of separation vary by sex and cohort. This transition leads to a reduction in women’s domestic work, which is greater for older women (–4.7 h/week for the pivot generation, –2.2 h/week for young women; the event is not interpretable for women of the elder generation). By contrast, it does not increase men’s domestic workload, except in the pivot generation (+3.4 h/week). For men in the young generation, it does not have a statistically significant effect, thus confirming the results presented in Table 2.

39Here again we see that the hypothesis of substitution strategies for domestic tasks is confirmed only for women. These parameters are not significant for men, in general or more specifically just after a separation. [20]

40These results show that just as entry into cohabitation with a partner leads to an increase in women’s housework, the reverse can also occur when the union is dissolved. Women reduce their investment in housework and men of the pivot generation find themselves having to increase theirs. This is explained in part by the fact that, in this generation, women devote more time than men to household tasks. Their departure obliges men to invest more time in a minimum of domestic chores.

41Examining the impact of the various parameters defined in the theoretical section and used to control for the effect of transition also provides new insights. Gender and the type of union before transition have a central impact, although nuanced by generation. In general, women devote considerably more hours to housework than men (constant of 23.0 versus 8.8), and married women more than single women (+2.3 h/week). Note, however, that this parameter varies only for the youngest generation, which includes only 49% of married people. For the two older generations, marriage is the dominant configuration for couples (92-96%).

42Moreover, working hours and level of education also have a significant impact for women, that once again depends on generation. The younger the women, the longer their working hours, and the higher their level of education, the less time they devote to domestic work. Note also that recourse to external help reduces the time spent on household tasks (–0.7 h/week), as does eating out in a restaurant at least once a month (–0.5 h/week). The effect of income, however, is not significant. Finally, we found a slight effect of wave (–0.06 h/week) on these women, which implies that within the same birth cohort and controlling for other parameters, the time spent on household tasks falls slightly each year, which may reflect a lowering of the normative requirements in this domain.

43According to the model in Table 3, [21] a women in the young generation with a university degree, living with a partner, unmarried, and working full-time with an annual income of 100,000 Swiss francs, will spend 6.5 hours per week on household tasks: 6.5 = 23.025 (constant) – 6.683 (young generation) – 2.78 (university degree, education = 10) + 0 (cohabiting union, reference category) – 6.8 (working full time) + 0.069 (income) – 0.32 (survey wave). By comparison, the time spent by a woman of the elder generation, married, with a minimal level of education and not in employment is estimated at 25.0 hours per week. Women’s investment in household tasks both before and after a transition is thus highly variable and strongly dependent on individual factors.

44For men, on the other hand, individual factors [22] have a lesser impact. For example, being in employment has an impact that is significant (– 0.04 hours per week), but around half of that seen in women (– 0.07 h/week). The type of union before the transition also has a small influence, close to the threshold for significance (p = 0.03), and in the opposite direction from the effect in women. A married man thus spends around half an hour less per week (– 0.5 h/week) on household tasks than a man who is not married, which reveals a certain specialization in tasks after marriage. The other parameters discussed above, such as income, recourse to paid household help, and eating out in restaurants have no impact. The same applies to the effect of wave, which is not significant.

45The same simulation as the one performed for women can be used for men. This exercise confirmed our results: the difference in investment in household tasks between a young, unmarried man with a university degree and a full-time job and a man of the elder generation, married, not in employment and with little education is much smaller (5.6 hours per week versus 8.3) than we found in women with similar characteristics.

Transition from family life to living alone

46Table 4 looks at the transition from family life [23] to living alone after a separation, exclusively among men of the young generation, as this event was too rare in the other cohorts or among women, who most often have custody of their children in this situation. Controlling for the other parameters included in the model, this transition leads to an increase in the time spent on household tasks by men in the young generation (+1.5 h/week). This result can be linked to the preceding one, which showed that the transition from living with a partner to living alone does not have a significant effect on the time spent on household tasks by men in the young generation (Table 3). This suggests that after this transition, men once again take over the domestic tasks that they had left to their partners due to the specialization of tasks following the birth of children (Henchoz and Wernli, 2010). Despite the smaller household size, men of the young generation increase the time they spent on household tasks, albeit modestly so.

47Once again, the hypothesis of resorting to substitution strategies, such as recourse to domestic help or eating out in restaurants at least once a month, is not confirmed [24] among the men of the young generation. However, here we see that income does have a significant effect, being associated with reduced time spent on household tasks.

48The model presented here includes the same parameters as the previous model, along with an indicator of the number of children. [25] It confirms several results from Table 3: working hours have a negative impact on performance of household tasks, while level of education has no effect. Although the number of children does not have a significant effect, the presence of young children (below age 7) tends to slightly increase the time that men in the young generation spend on household tasks (+0.4 h/week). The fact of being or having been married loses its importance here and does not have a significant impact, given this parameter’s status as a quasi-constant (with 94% of men married) and its limited impact (Table 3).

Transition from family life to single-parent family life

49Table 5 concerns the transition from family life as a couple with (minor) dependent children [26] to single-parent family living, with no new partner in the household, following a separation. This event was studied only for women in the youngest generation. Indeed, after a separation, in Switzerland, women generally retain custody of their children. [27] As above, the number of dependent minors was taken into account, while a dichotomous indicator of the presence of young children was used.

50After controlling for all parameters, the transition to single-parent family life was found to have substantial effects on women in the young generation. Even though they no longer have a partner to perform some part of the household tasks, this event leads to a reduction in the time spent on housework (–1.9 h/week).

51Table 5 also confirms the results in Table 3, namely that for the women in the young generation, the time spent on housework increases with age. By contrast, the higher their level of education, the longer their working hours and the greater their tendency to employ domestic help and to eat out in restaurants, the less time they devote to cooking, dishwashing, cleaning and laundry. Moreover, as in Table 4, increased income is associated with a drop in time spent on household tasks. We observe as well that for a given age, and controlling for the other parameters included in the model, each passing year reduces the time that women spend on basic household tasks (–0.2 h/week), similarly to what was seen for women in Table 3. This may illustrate a progressive drop in normative expectations in this domain with the arrival of the younger generations.

52Finally, the presence of children leads to a net surplus of domestic work for women (respectively +2.5 hours/week per child), whereas that of young children (below 7 years) does not have a significant impact (p = 0.10).