Attitudes towards sharing housework in couple context: An empirical, factorial survey approach

Objective: This study measured and compared the attitudes of German women and men towards sharing total housework, routine housework and non-routine housework in couples. Background: Although attitudes towards gender roles and the notion of separate spheres are important for understanding many aspects of family life, knowledge about situational variations of women’s and men’s attitudes towards housework sharing is limited. Method: Original data from a factorial survey of 1,120 German women and men from 2016 were used to describe variations in the attitudes of women and men towards three sets of housework using multilevel regression models. Results: Women and men expressed their attitudes towards equal sharing of total and routine housework, but non-routine housework was assigned to male partners in couples. Attitudes differed widely according to the context of the couple: In couples with similar economic resources, respondents favored equal sharing of housework, and in couples with unequal arrangements, the partner with fewer resources was tied to more housework and vice versa. Conclusion: When evaluating housework responsibilities, women and men in this study seemed to follow the principles of equity and balanced exchange.


Introduction
Attitudes towards gender roles are a fundamental part of social reality and help us to better understand many aspects of family life. Most notably, the traditional family model with its notion of the female homemaker is widely used to explain the division of labor between women and men. Within this framework, a high degree of agreement with the model of separate spheres points to 'traditional' gender role attitudes that assign the man the sole role of breadwinner and the female partner of a couple the sole role of homemaker and householder. The most popular counter-position implies maximum agreement with the idea of gender egalitarianism in all spheres of labor and proposes a completely equal division of housework in couples. The possibility of reversing the traditional roles of women and men seems to be of little theoretical, empirical, and normative significance.
By and large, research on attitudes towards gender roles has documented a growing acceptance of the model of gender egalitarian in practically all Western societies in recent decades and across cohorts (Bolzendahl & Myers 2004;Braun & Scott 2009;Brooks & Bolzendahl 2004;Davis & Greenstein 2009;Ebner, Kühhirt & Lersch 2020;Pampel 2011). In this increasingly liberalized environment, women seem to be somewhat more egalitarian than men, and people with a higher level of education seem to be more egalitarian than people with a lower level of education (Braun & Scott 2009;Kane 1995;Pampel 2011). However, there are signs of a 'stalled revolution', indicating a stability in attitudes towards gender roles since about the 1990s and evidence of a (re-)traditionalization of the life course, especially during the transition to parenthood (Baxter, Buchler, Perales & Western 2015;Cotter, Hermsen & Vanneman 2011;van Egmond, Baxter, Buchler & Western 2010). Nevertheless, taking into account the structural constraints that favor gender inequalities in couple households and recognizing the multidimensional nature of gender ideologies (Grunow, Begall & Büchler 2018), studies have shown positive effects of the commitment to gender egalitarianism on the propensity of women, men, and couples to share housework equally (for reviews, cf. Coltrane 2000; Lachance-Grzela & Bouchard 2010).
Since Pedulla & Thébaut (2015) have shown that preferences are sensitive to context variations, the main question of this article is: How do attitudes towards housework sharing vary depending on different couple situations that individuals in the process of forming attitude evaluate? An empirical approach to this issue using original data from a factorial survey of 1,120 German women and men is the main contribution of this paper. In this original survey, respondents were asked to assign arrangements of sharing housework to vignettes with randomized descriptions of couples. This method is particularly suitable for the investigation of attitude judgments between different scenarios under the assumption of particularly high internal and construct validity (Auspurg & Hinz 2015).
Furthermore, this paper offers two additional, rather small supplements to the current literature. First, I analyze in particular the attitudes of women and men towards housework sharing, since these attitudes should reflect the personal opinions of women and men about who should do housework, rather than general gender ideologies (Poortman & van der Lippe 2009). To achieve this goal, I use information from a direct survey question about housework sharing from the original factorial survey. This approach is driven by methodological concerns about previous approaches to measuring attitudes towards gender roles. Compared to general single items about the attitudes towards division of labor (Walter 2018), the approach of this paper points directly to the preferred arrangement of sharing in one explicit question (see Figure 1 below). In short: a clear interpretation of "agree-disagreeitems" has been problematic in previous research (Braun 2008;Walter 2018). For example, the rejection of a common item such as "A man's job is to earn money; a woman's job is to look after the home and family" (ISSP Research Group, 2016, US questionnaire) says nothing about the division of labor that women and men positively advocate: Is it an equal relationship (which is usually assumed, but rather arbitrarily), a female breadwinner model, or even something completely different (Braun 2008)?
Second, I compare the attitude of women and men towards sharing total housework with sharing routine and non-routine housework activities. The gender construction perspective has emphasized that housework consists of different tasks, which in turn are deeply gendered. As Barnett and Shen (1978) pointed out early on, routine housework activities, such as cleaning, cooking, or doing the laundry, are considered necessary and less optional, and usually cannot be postponed for long periods of time. These "low-schedulecontrol tasks" are regarded as "female tasks". Non-routine or "high-schedule-control tasks" such as doing repairs, caring for the car or doing administrative paperwork are regarded as "male tasks". From the perspective of gender construction, the doing gender of women is more related to routine chores and the doing gender of men to non-routine chores. Both genders should continue to dispense with the respective other bundle of chores in order to avoid dissonances in their gender identity (Berk 1985). Taking this into account, I assume that the attitudes towards sharing routine and non-routine housework activities are gendered accordingly. Thus, the investigation of attitudes towards total housework conceals a gendered heterogeneity in the more specific attitudes of women and men, which could be caused by the defining housework as target variable. Although several empirical studies have indeed analyzed different dimensions of the actual housework sharing (e.g., Barnett & Shen 1978;Craig & Powell 2018, Kan, Sullivan & Gershuny 2011, an empirical comparison of attitudes towards sharing routine housework, non-routine housework and total housework is still pending.

Background
Previous research has shown that attitudes are sensitive to contextual variations that may be independent of the actual socio-economic profiles of women and men (Pedulla & Thébaut 2015). Previous studies have emphasized the importance of equity (Auspurg, Iacovou & Nicoletti 2017;Carriero & Todesco 2016 and relative resources, as well as gender, (Doan & Quadlin 2019) in evaluating arrangements for housework sharing in couples. In addition, it is plausible that other signs of couples, such as marriage, influence the attitudes of the individual.
In the past, numerous studies have discussed the situation-related predictors for the division of housework (most recent overviews: Coltrane 2000; Lachance-Grzela & Bouchard 2010), which form the basis for perception and formation of attitudes towards housework sharing in couples. Overall, the gendered division of household labor is one of the most consistent empirical regularities observed in the social sciences. Current estimates suggest that women in Western economies perform about two thirds of routine housework (Bianchi, Sayer, Milkie & Robinson 2012).
Arguably, the transition to parenthood and the presence of children stand out as main constraints to an equal division of housework (Baxter, Hewitt & Haynes 2008;Kühhirt 2012;Sanchez & Thomson 1997). When children are born, the responsibility for housework usually shifts to women, regardless of other developments in terms of time availability, the autonomy of partners, or the relative resources of couples. The birth of a first child seems to pave the way for the model of separate spheres, which has proven to be difficult to reverse, even as the children grow older. Similarly, research theoretically based on different models of symbolic exchange, such as doing gender (Berk 1985;West & Zimmerman 1987), has provided evidence that marriage could also trigger a more traditional division of housework compared to all other forms of cohabitation (Gupta 1999, Baxter 2005, but: Baxter et al. 2008. Theoretically based on different economic exchange or bargaining, autonomy, or dependency models (overviews in Brines 1993;Gupta 2007), the absolute or relative employment and income of couples are decisive predictors for the division of housework. Research has shown that women or men with higher incomes or employment hours tend to do less housework than their partners. Similar resources yield rather egalitarian housework arrangements. However, as several studies have shown, the negative correlation between resources and housework performance is not necessarily linear (Baxter & Hewitt 2013;Gough & Killewald 2011;Killewald & Gough 2010). This is especially true if men or both partners are unemployed, or women earn more than their male partners. In cases like these, gender seems to be more important than resources, activating mental maps of gender or gendered identities that refer to the model of 'traditional' separate spheres, even in contexts that are highly gender-liberalized.
Assuming that there is a positive correlation between the actual division of unpaid work and the attitudes towards housework sharing (Poortman & van der Lippe 2009), couple situations provide clues as to what can be expected as empirical outcomes in the question of who should do the housework. Thus, parenthood and the age of children, marital status, the employment of women and men, and the absolute and relative income of women are used as factors to describe the situations that form the framework for the normative judgments of housework division in the factorial survey (see 3.1 and Figure 1 below).
Research on perceptions of housework sharing has offered several mechanisms of exchange that shape assessment of the fairness or satisfaction of women and men with the housework sharing arrangement, with ideology, power, and equity being the most important (Baxter 2000;DeMaris & Longmore 1996;Layte 1998;Lennon & Rosenfield 1994;Kane & Sanchez 1994). These mechanisms have been successfully used to explain the "apparent paradox" (DeMaris & Longmore 1996: 1043) that the apparent gender inequality in participation in housework is not perceived as unfair to wives as expected.
The research has argued and provided empirical evidence that processes of social or economic exchange influence women's and men's perceptions of fairness and satisfaction with certain arrangements of the division of housework in couples (DeMaris & Longmore 1996; Lennon & Rosenfield 1994;Robinson & Milkie 1998;Kane & Sanchez 1994). Economic resources, which are seen as fundamental sources of power, or resource differences between partners set the tone for bargaining on participation in housework and affect the influence of partners to bring about change. Material dependency, less power or few possible alternatives outside a partnership then lower partners' expectations of justice in terms of gender inequality in housework (DeMaris & Longmore 1996). This, in turn, reduces the degree of fairness and consequently increases the likelihood that individuals will perceive unequal divisions of labor as less unjust (Baxter 2000).
Within the notion of equity, however, women and men evaluate their outcomes from an exchange in relation to their inputs. In all relevant areas, i.e., not necessarily within a single sphere, women and men perceive equity in a couple's exchange relationship if their inputs do not significantly outweigh their outcomes, but are balanced (DeMaris & Longmore 1996).
Applying the exchange mechanisms to the study of attitudes towards housework sharing, women and men can be expected to opt for specific housework arrangements that they feel are fair or with which they would be satisfied given a couple's specific economic or time availability situation. Under otherwise equal conditions, one partner can be expected to need more time, measured by labor market participation, to change attitudes towards greater participation by the other partner in a couple. The same should apply to the autonomy of women, measured by women's absolute income (Gupta 2007). Looking at the relative resources of couples, one should not expect to find declared attitudes that assign greater relative amounts of housework to the partner with fewer employment or income resources, and vice versa. Couples in which both partners are similarly employed or contribute in a similar way to the total household income should encourage attitudes towards equal housework sharing.
These expectations are based on the idea of resource exchange as a gender-neutral model, which leads to symmetrical expectations regardless of the direction of dependence. The introduction of gender asymmetries according to the traditional model of separate spheres, gender ideology, family ties (Kane 1998), gendered interests in maintaining advantages at home (Kane and Sanchez 1994), or processes of neutralizing gender bias (Greenstein 2000) can change expectations of attitudes in general towards greater housework participation of women, which is even stronger in cases of male dependence.
Three recent studies examined the perceptions of fairness and the satisfaction of women and men with housework sharing against the background of these mechanisms using factorial survey designs: Auspurg, Iacovou & Nicoletti (2017) analyzed the preferences of British women and men for the division of paid and unpaid labor in couples, using satisfaction with different work arrangements as an evaluation indicator. They stated that their "main finding is that both men and women display a marked preference for equity, in terms of both the allocation of housework and the total allocation of paid and unpaid work" (Auspurg, Iacovou & Nicoletti 2017: 17). They also found little evidence for the separate spheres model of female homemakers and male breadwinners, selfish preferences, or differences in the preferences of women and men. In summary, this study emphasizes "that it is ... unlikely that gender differences in housework shares can be explained by systematic differences in the utility that men and women derive from doing housework" (Auspurg, Iacovou, and Nicoletti 2017: 18). Carriero & Todesco (2014) examined the perceptions of fairness and legitimacy of renegotiations of housework arrangements, using an Italian sample. They showed that equity principles are to a certain extent shaped by gender ideologies, because compromises between paid and unpaid labor are not evaluated by the respondents in a gender-neutral way, as the theory of equity predicts. In a follow-up study based in part on the same data, Carriero & Todesco (2016) further highlighted the role of comparison referents for the perceived fairness of housework arrangements. Following Thompson's (1991) framework of distributive justice, comparison referents, outcome values, and justifications shape the perception of justice in couples by women. When women -and possibly men as wellcompare themselves not only with their partners but also with their mothers or other members of their circle of acquaintances (Carriero & Todesco 2016), they adjust their sense of fairness and can thus 'deviate' from the symmetrical notion of exchange.

Empirical design and methods
My analysis was based on the original data from a factorial survey of 476 women and 644 men from Germany aged 18 years or older in the German "WiSo-Panel" (https://www.wisopanel.net/), an online panel conducted at the University of Freiburg under the direction of Professor Anja Göritz as part of her research in the field of psychology (Göritz 2014). Upon request, the "WiSo-Panel" offers external researchers the opportunity to conduct their own surveys in its pool of respondents. The online panel includes several thousand panelists, who have voluntarily chosen to participate in occasional scientific surveys. Essentially, the "WiSo-Panel" represents the German "online population", which is why the sample is, on average, younger and more highly educated than the entire German population.
For the survey of the present study, all panel members that met the selection criteria of German nationality and minimum age were invited by mail in spring 2016 and then voluntarily decided to participate in the survey (response rate of 19 %, but participants are similar to the total population of the panel). The average age of women was 45.3 years and that of men was 50.7 years. About half of all participating women and men have a university degree (.47 and .50, respectively). Most of both genders lived with a partner (.68 and .75) and slightly more than half of the respondents had children (.54 and .56).
Certainly, the respondent sample of the study is not a random sample of the German population. Most importantly, male respondents are older and more highly educated, and female respondents are younger and more highly educated than in the overall German population. This may limit the possibility of generalizing the results of the study to the entire German society. First, however, the sample is large enough to include a sufficient number of representatives of each age and educational group and their cross tabulation for women and men. Second, the factorial survey is based on a high internal and construct validity, which has proven to be advantageous compared to traditional surveys (Auspurg & Hinz 2015). Third, the testing of associations and "causal" hypotheses does not necessarily require strict statistical representativeness compared to estimations of population parameters.
Data from factorial surveys are well suited to investigate attitudes to social reality and its variations (Auspurg & Hinz 2015;Rossi & Anderson 1982). In a factorial survey, respondents evaluate certain outcomes (in this case, attitudes towards sharing housework) that are conditioned by vignettes describing conceivably relevant situations (couple characteristics). The stimuli of these situations are randomly manipulated, and the vignettes are randomly assigned to the respondents according to the experimental concepts of factor orthogonality and random placement (Rossi & Anderson 1982). Ideally, this promotes a high internal validity of factorial surveys, which should neutralize the unobserved heterogeneity of the respondents, avoid problems of multicollinearity and design confusion, and easily interpret treatment effects and higher order interactions (Auspurg & Hinz 2015). Furthermore, the vignette approach should be advantageous compared to item-based opinion surveys when it comes to studying the attitudes of the respondents. Human judgments and their determinants can be addressed more directly, even in cases of sensitive issues, without having to indirectly derive from the evaluation of statements, and with a lower risk of distorting social desirability (Alexander & Becker 1978;Auspurg & Hinz 2015).

The vignette study
Each vignette of the survey ( Figure 1) described the situation of a couple based on six factors whose categories were randomly manipulated within the scenarios: (1) marital statusunmarried, or married; (2) Parenthood and age of child -no children, one child aged six months (newborn age), one child aged between three and six years (kindergarden age), or one child aged 6 years or older (school age); (3 and 4) men's and women's employment situation -not employed, working part-time, or working full-time; (5) women's monthly income -no income, 800, 1,800, or 3,500 euros; and (6) women's relative income -less than partner, same as partner, or more than partner. These factors were derived from the empirical literature on the division of labor, as they proved to be relevant predictors of housework arrangements for couples (e.g., Baxter 2005;Baxter and Hewitt 2013;Baxter, Hewitt & Haynes 2008;Bianchi, Sayer, Milkie & Robinson 2012;Brines 1993;Gough and Killewald 2011;Gupta 1999Gupta , 2007Killewald and Gough 2010;Kühhirt 2012;Sanchez & Thomson 1997;West and Zimmerman 1987).

Figure 1:
Vignette example An unmarried couple is living together. They don't have children.
The man is working full-time, the woman is not employed. From the total number of possible vignettes (full factorial = 864 vignettes), some implausible combinations of employment and income (factors 3-6) were removed from the full factorial, especially being employed and having zero income, as well as some others (see Appendix A). From the remaining 648 vignettes, 120 scenarios were selected, which were included in the survey using d-efficient sampling (Su & Steiner 2020). This ensured the balance of the levels (all levels of each dimension occur equally frequently) and the factor orthogonality (the minimal correlation between the factors and the maximal variance within the questionnaire). The proposed design (d-efficiency = 92.3) complies with resolution IV, which means that in addition to the main effects, specified two-way interactions could be estimated without aliasing problems (Auspurg & Hinz 2015).
The 120 vignettes were blocked on 30 decks with 4 randomly arranged vignettes each. Each respondent was randomly assigned one of these blocks. Each respondent went through the assigned vignette block three times. After answering the first four vignettes about the attitudes towards total housework, a text-only page changed the focus to the set of routine housework tasks ("cooking, cleaning, doing the laundry"), followed by four vignettes about these particular attitudes. Then another text-only page shifted the focus to the set of non-routine housework tasks ("repairs, car, administrative paperwork"), again followed by four vignettes about these attitudes (see Appendix A). As the answers to a specific question in the survey indicated, only 3 % of all respondents rated the scenarios as "(very) difficult".
The question that triggered the respondents' attitudes was "Who do you think should be doing the housework [OR cooking, cleaning and laundry OR repairs, paperwork, and caring for the car] in this couple?" Response categories for the attitudes towards sharing housework ranged from 1 = "woman alone", 2 = "woman most", 3 = "woman more", 4 = "both equally", 5 = "man more", 6 = "man most", to 7 = "man alone". This scale outlined the range of possible alternatives for housework sharing within a couple, ignoring of course forms of domestic outsourcing for reasons of complexity.

Measures
The vignette factors marital status, parenthood and age of children, and absolute and relative income of women were operationalized as sets of binary variables (Table 1). The employment status of women and men was combined into a measure of relative employment with three categories: "woman<man", "woman=man", and "woman>man". The gender of respondents was measured as a binary variable (Table 1).

Estimation
The attitudes towards sharing housework, routine and non-routine tasks, which the respondents expressed for each vignette in the questionnaire, result from the combination of a respondent-specific basic attitude towards housework sharing and a reaction to the specific situation of the vignettes. Since four measurements (level 1) for each respondent (level 2) represent hierarchical data, I used regression techniques for multi-level data to analyze the individual and contextual variation of attitudes. I treated the seven-point-rating outcome of each vignette as a linearly dependent variable.
I applied a linear random intercept model, specified as where Yij indicates the rating of a vignette i by respondent j, γ00 denotes the average intercept, xpij is a vector for all vignette factors, γp0 represents the corresponding regression coefficients, zqj is a vector for respondent characteristics with γ0q as corresponding slope parameters, Uoj is the respondent-dependent deviation from the average intercept, and Rij is the residual (Snijders & Bosker 2012).

Sensitivity analyses
Several sensitivity analyses were performed to check if the results change due to different settings of the estimation. First, all models were run separately for the sex of the respondents and with multiple covariates (Appendix B; all other results of the sensitivity analyses are documented in the replication files). Second, all analyses were run for different subgroups of the sample, e.g. for married or childless respondents. Third, all predictions of the situational characteristics (Figure 3-7) were estimated as interactions with marriage and parenthood, utilizing the resolution IV-design of the vignette study (see 3.1 above). Fourth, all analyses were run with different models, most notably with a random-effects-panelmodel. Fifth, all analyses were checked for differences between respondents who lived in the eastern or western part of Germany. The code for the replication of all sensitivity analyses will be provided together with the data and replication files to article (see Acknowledgments below). None of these changes in the estimation setup altered the final conclusions.

Results
Figure 2 provides a first and basic illustration of the attitudes of women and men towards sharing total housework (green bars), routine housework (red bars) and non-routine housework (blue bars), as reported by the respondents of the factorial survey for all vignettes.
The purely descriptive results in Figure 2 showed that for all three outcomes, both women and men stated "both partners equally" as their preferred sharing arrangement. The attitudes towards sharing total and routine housework were symmetrically distributed around this modal category. Both women and men assigned about 25-30 % of all vignettes for total or routine housework to either to the female or the male partner of a couple. However, non-routine chores, however, were more likely to be attributed to men, as 45 % of these particular vignettes assessed by women and 50 % of those evaluated by men were rated "man more," "man most," or "man alone."

Figure 2:
Women's and men's attitudes towards housework in general, routine housework, and non-routine housework Notes: Relative frequencies how women and man responded to the vignettes in the factorial survey; the 95 %confidence intervals may be used to account for differences in gendered rating behavior and do not refer to any "super-population".
Figure 2 shows no relevant differences between women's and men's distributions of the attitudes towards the three outcomes. Most notably, there were no differences in the vignette ratings between attitudes directed towards total housework and towards routine housework. The proportion of vignettes rated as "women's work" was significantly lower for non-routine housework than for routine and total housework. Conversely, the shares of vignettes rated as "men's work" was higher for non-routine housework than for the other two outcomes. Figures 3-7 illustrate, how attitudes towards sharing housework differ according to the situational characteristics of the couples depicted by the vignettes. All figures show predicted margins and 95 %-confidence intervals for the attitudes of women (hollow symbols) and men (solid symbols) stated attitudes towards sharing total housework (green squares), routine housework (red triangles), and non-routine housework (blue circles). The predictions were calculated using the models in Table 2, with all other covariates set to their mean values.
Two minor findings from Figures 3-7 should be mentioned in advance: First, there were only minor differences between the attitudes of women and men towards each of the three definitions of housework (as in Figure 2). However, women slightly tended to assign total housework duties as well as non-routine tasks to themselves compared to men (negative main coefficients for respondents' gender in Table 2). Notes: Each model is displayed in two columns, left column: main coefficients, right column: interaction coefficients with respondent's sex; reference categories: a "Male respondent", b "Not married", c "No child", d "Relative employment: W<M", e "Income woman: 0", f "Relative income: W<M"; regression coefficients, standard errors in parentheses; levels of statistical significance: * p < .05, ** p < .01, *** p < .001. Figures 3-7 were derived from this table.
Second, there seemed to be no differences in the attitudes towards total housework and routine housework. At the same time, attitudes towards non-routine housework differed markedly from the other two sets of housework, and tended to greater responsibility of the male partner in a couple (predicted attitudes much higher than the value of 4). Now focusing on context variations, marital status did not yield any attitude differentials for women and men (Figure 3).

Figure 3:
Predicted attitudes towards housework by vignette couples' marital status Notes: Predicted mean vignette ratings and 95 %-confidence intervals. The predictions were derived as average margins from the models of Table 2 in the article. Potential range of attitudes: 1 = "woman alone" to 7 = "man alone", 4 = "both partners equally".
In couples with a newborn child, women expected a higher contribution of men for all three outcomes compared to no children or children in kindergarten or school age (Figure 4).

Figure 4:
Predicted attitudes towards housework by vignette couples' parental situation Notes: Predicted mean vignette ratings and 95 %-confidence intervals. The predictions were derived as average margins from the models of Table 2 in the article. Potential range of attitudes: 1 = "woman alone" to 7 = "man alone", 4 = "both partners equally".
Couples' relative employment produced the most striking attitude differentials ( Figure 5). In couples in which women and men are equally employed, both partners were said to be equally responsible for housework in general and routine chores (predicted attitudes around the value of 4, i.e. "both partners equally"). In cases of employment inequality, the partner who is working more was associated with less housework and vice versa. Again, both women and men assigned non-routine tasks to a greater extent to the male partner of a couple.

Figure 5:
Predicted attitudes towards housework by vignette couples' relative employment Notes: Predicted mean vignette ratings and 95 %-confidence intervals. The predictions were derived as average margins from the models of Table 2. Potential range of attitudes: 1 = "woman alone" to 7 = "man alone", 4 = "both partners equally".
The symmetrical pattern of relative employment that is resource-dependent and gender-neutral in its direction was also visible for relative income, albeit not that pronounced ( Figure 6).

Figure 6:
Predicted attitudes towards housework by vignette women's absolute income Notes: Predicted mean vignette ratings and 95 %-confidence intervals. The predictions were derived as average margins from the models of Table 2 in the article. Potential range of attitudes: 1 = "woman alone" to 7 = "man alone", 4 = "both partners equally".
Regarding women's absolute income, there are no significant attitude differences in those cases in which women have any positive income (Figure 7).

Conclusion
The present study provided an empirical approach to the attitudes towards sharing housework, using an empirical, factorial survey approach. This study, which focuses explicitly on attitudes towards housework sharing, analyzed the context-dependent variations in the attitudes of women and men towards sharing housework and compared three different bundles of housework tasks, i.e., total housework, routine housework (cooking, cleaning, laundry), and non-routine housework (repairs, car, paperwork).
With regard to the important question of contextual variations, the study provided one main finding: Couple contexts -presented with factors of marital status, parenthood and age of children, employment, and absolute and relative income of women in the factorial survey -did indeed change the attitudes of women and men towards sharing housework.
The distribution of couples' economic resources -especially employment and, to a lesser extent, the absolute and relative income of women -led to significant differences in attitudes in all three outcomes. (a) Equal sharing is highly associated with equal resources. (b) High relative employment or income resources of a partner are systematically associated with lower expected housework responsibilities, and vice versa. (c) These clear resourcedependent differences in attitude are again independent of the gender of the respondents (similar for satisfaction with housework sharing: Auspurg, Iacovou & Nicoletti 2017). This suggests that attitudes towards housework are conditioned to those situational characteristics that are theoretically focused on by exchange or equity theory with its notions of gender-neutrality.

Figure 7:
Predicted attitudes towards housework by vignette couples' relative income Notes: Predicted mean vignette ratings and 95 %-confidence intervals. The predictions were derived as average margins from the models of Table 2 in the article. Potential range of attitudes: 1 = "woman alone" to 7 = "man alone", 4 = "both partners equally".
On the one hand, this could indicate the existence of a kind of 'bargaining family' that functions according to the power mechanism. On the other hand, this kind of resource dependency could be the result of time-availability considerations. According to the arguments of the social exchange or equity theory, when assigning housework responsibilities to one or the other partner, women and men could consider an efficient allocation of total working time -rather than equality in each area of work -and, thus, shape their attitudes accordingly.
Two aspects stood out among the two "minor" topics of the study: First, attitudes towards sharing total housework and routine housework clearly tended towards equal sharing arrangements in couples. In contrast, non-routine tasks were attributed more to male partners in couples. This applied to both female and male respondents in the survey. Furthermore, traditional attitudes towards housework sharing were less pronounced among higher educated or less traditionally oriented women and men. This is in line with earlier research on changes in gender ideology (Braun & Scott, 2009;Pampel, 2011). In contrast, non-routine housework still stuck with men, even to a far greater extent than routine housework to women. I interpret this finding as evidence for a different change in the gendered meaning of housework. Routine housework seems to be losing importance for women's roles and has even become part of men's roles, at least on the level of attitudes. At the same time, the model of men doing non-routine chores is still a stable part of the gender construction.
Second, the attitudes of women and men towards sharing total housework and routine housework, as well as their individual and contextual variations, were virtually identical and differed significantly from attitudes towards sharing non-routine housework. I interpret this finding as evidence that women and men have routine chores in mind when they are surveyed about housework in general. This has important methodological implications, as it relativizes the importance and interpretation of housework indicators in existing studies. It seems likely that general items on attitudes towards housework or unspecified questions about housework could be limited to routine housework. Thus, the results of the present study require a more detailed measurement and a more careful interpretation of the activities around "housework".
At least five limitations of my study warrant future investigation. First, this study was based only on data from Germany and should thus be replicated elsewhere to check if the findings are stable in other contexts. However, since the processes of family change, demographic behavior, and the 'gender revolution', i.e., the convergence of gender in the use of time, proceeded in a similar fashion in Germany as in other Western societies (Goldscheider, Bernhardt & Lappegård 2015), the findings of comparable studies from the United States (Doan & Quadlin 2019), Great Britain (Auspurg, Iacovou & Nicoletti 2017), and Italy (Carriero & Todesco 2016 are consistent with the findings presented in this paper. Second, since the sample of the study was not a statistically representative random sample of the entire German population, the external validity of the study could be questioned. At least, a generalization of the findings could be somewhat problematic for Germany as a whole. Since the panelists and respondents within the "WiSo-Panel" at least represent the German "online-population", higher educated respondents were overrepresented in the sample. Thus, the results of the study may possibly overestimate the tendency of respondents to express egalitarian values and follow principles of equity. All interpretations are bound by this restriction. Third, there should be a systematic evaluation of the different approaches to measuring attitudes towards housework sharing for different individual activities and task bundles. As Poortman and van der Lippe (2009) have pointed out, including other attitudinal aspects of housework, such as enjoyment and perceived competences, but also unravelling the relationship between attitudes, satisfaction, and the perception of fairness, could broaden our understanding of the predictive power of attitudes in the process of gendered division of labor. Fourth, and directly related to this, there is need for research on the predictive power and interrelationship of attitudes towards housework sharing and the actual sharing behavior of women and men, on housework in general and on different housework activities, in cross-sectional and longitudinal settings. Fifth, the introduction of domestic outsourcing in the process of attitude formation should be a further step to come closer to the complexity of the everyday practices, beliefs, and processes of gender construction of women and men.

Data availability statement
The datasets generated during and/or analyzed during the current study are available as supplementary files to the article (doi:10.20377/jfr-419-506).

Full factorial
%mktex (2 4 3 3 4 3, restrictions=impl, interact=x1*x2 x1*x3 x1*x4 x1*x5 x1*x6 x2*x3 x2*x4 x2*x5 x2*x6 x3*x4, n=120, options=resrep, seed=2702); %mktblock(nblocks=30, seed=2702); Appendix B: Analyses of the paper with more controls Notes: Proportions, means and standard deviations (where applicable); rounding differences to 1 may apply. Respondent characteristics were operationalized as binaries for university degree, partnership (living together with a partner either married or unmarried), parenthood, a continuous variable of age, a score that represents the disapproval of traditional family roles and perceived parental division of housework. Parental housework division was operationalized with a five-point scale from 1 = "woman alone" to 5 = "man alone". The disapproval of traditional family roles was scaled (Cronbach's α = .81 for all respondents) from four of seven items that are commonly used in social science research, for example in the most recent ISSP on gender roles of 2012 (doi:10.4232/1.12339): (1) "A pre-school child is likely to suffer if his or her mother works"; (2) "All in all, family life suffers when the woman has a full-time job"; (3) "A job is all right, but what most women really want is a home and children"; (4) "A man's job is to earn money; a woman's job is to look after the home and family". Each item was rated from 1 = "strongly agree" to 5 = "strongly disagree", lower values depict consent to the 'traditional separate spheres model', and higher values represent disapproval of the traditional family model. Notes: Reference categories: a "Not married", b "No child", c "Relative employment: W<M", d "Income woman: 0", e "Relative income: W<M"; regression coefficients, standard errors in parentheses; levels of statistical significance: * p < .05, ** p < .01, *** p < .001.