The Association between Norms and Actions The Case of Men’s Participation in Housework Daniela Grunow, Nina Baur Abstract: Gender and family researchers are often puzzled by the differences be- tween gender norms and actual behaviour. For example, many previous studies have shown that since the 1970s, attitudes have changed toward more modern gen- der role beliefs. However, this is generally not refl ected in behaviour, this being especially true for men’s involvement in housework and care work. In this paper, we analyse to what extent gender norms correspond with men’s participation in housework. We examine in particular cooking, vacuum-cleaning and doing the laun- dry, because these activities refl ect pronounced variation in norms and actions with regard to modern and traditional gender role attitudes. Going beyond previous re- search, which assesses the general infl uence of norms on practices, we investigate the relative impact of different types of gender ideologies, namely the strength with which respondents endorse (1) the male provider and female homemaker model, (2) women’s employment, (3) employment as men’s primary role, and (4) men’s participation in unpaid work. Our empirical fi ndings indicate that positive attitudes towards male contributions to housework are a key predictor of men’s housework. Keywords: Germany · Gender · Women · Men · Housework · Division of labour · Gender ideology · Cooking · Vacuum-cleaning · Doing the laundry 1 Introduction: The relationship between gender roles and daily routine behaviour Nowadays, it is assumed that both men and women have increasingly developed a more critical attitude toward traditional gender roles. Young couples and fami- lies in particular are believed to seek more gender egalitarian ways of combining employment and household tasks. Indeed, empirical research has shown that atti- tudes toward gender-specifi c norms have shifted toward egalitarian norms over the past decades, especially in the areas of family, employment, and housework (Blohm 2002; Lück 2004, 2009). Comparative Population Studies Vol. 39, 3 (2014): 521-558 (Date of release: 29.09.2014) © Federal Institute for Population Research 2014 URL: www.comparativepopulationstudies.de DOI: 10.12765/CPoS-2014-10en URN: urn:nbn:de:bib-cpos-2014-10en7 • Daniela Grunow, Nina Baur522 Historically, it seems that society’s attitude toward male and female responsi- bilities in private and professional life has changed. Especially housework, which was generally considered a woman’s responsibility and therefore “unmanly”, is increasingly considered men’s responsibility as well. Panel studies analysing the gender-specifi c distribution of housework of married couples and families show that nearly half (45.5 percent) of recently married couples share domestic respon- sibilities equally (Grunow et al. 2007, 2012). During the fi rst year of marriage, men and women take care of traditional female housework equally. It is striking, though, that most men gradually withdraw from doing housework, especially after entering parenthood. Thus, for many men, there appears to be an increasing discrepancy between egalitarian gender norms and their actual daily behaviour, which tradition- alises during family formation (Oberndorfer/Rost 2005). This discrepancy of norms and action is sociologically interesting as it is not clear why men and women give up egalitarian divisions of labour in favour of a more traditional division of labour. Economic factors (i.e. higher education and a better earning potential of men compared to their partners) cannot suffi ciently ex- plain this relationship, as recent empirical studies have shown (Gupta et al. 2010). Instead, family context appears to play an important role in the ‘traditionalisation’ of gender roles (Grunow et al. 2007, 2012; Kühhirt 2011). As a consequence, inter- est in the infl uence of norms on gender-specifi c behaviour has increased in recent years (for a summary see Davis and Greenstein 2009). Unfortunately, there is a lack of studies that measure and represent male and female norms in a contemporary and adequate way (Grunow 2010). In particular, studies measuring attitudes toward male roles and responsibilities in the private and domestic sphere are lacking. It is also rarely examined whether the importance of men’s role as earners has changed. Recent empirical studies from the Netherlands and Sweden hint at gender-specif- ic attitudes toward housework being an important predictor of male and female participation in housework (Poortman/Van der Lippe 2009; Evertsson 2014). More precisely, the respondent’s own attitudes play a greater role than the partner’s atti- tudes, and men's attitudes correspond more strongly to their actual behaviour than the attitudes of their female partners. Hence, this paper addresses the urgent need for research on gender role change, especially in regard to men (Ostner 2005; Tölke/Hank 2005). Following Lück (Edito- rial on this special issue CPoS 39, 3), we analyse individual attitudes toward gender roles, i.e. specifi c normative beliefs regarding male and female roles in the family, as well as individual assumptions regarding the consequences of traditional and non- traditional gender-specifi c behaviour which often explain individual attitudes. We empirically show that both types of beliefs express common latent normative con- structs, which help to explain individual behaviour. Subsequently, we label these constructs norm systems. Subsequently, we introduce theories about the development and the infl uence of gender norms on housework. Building on recent fi ndings that gender roles change during the formation of households and families (Davis 2007; Schober 2011), we discuss the impact of household and family contexts on gender roles. Using data from the study “The image of men in society”, we examine both normative attitudes The Association between Norms and Actions • 523 and assumptions regarding gender roles held by men and women and the extent to which men participate in traditional female housework depending on their family context. One advantage of our study is that our data allows us to separate in our analyses norms regarding male and female behaviours which relate to different realms of life. We view these distinct norms as separate norm systems. Specifi cally, we investigate the infl uence of following norm systems on behaviour: 1) The extent to which men and women endorse the male provider and female homemaker model, 2) Men’s and women’s attitudes regarding women’s employment, 3) Men’s and women’s attitudes toward employment being a man’s central so- cial role, and 4) Men’s and women’s attitudes regarding men’s participation in unpaid work. Especially the third and fourth system have been neglected by earlier studies due to lack of data. Thus, we allow for a dimensionally differentiated view on male gender norms. Using multivariate analyses, we explore how much of the variation in men’s participation in cooking, doing the laundry and vacuum-cleaning these norm systems explain. Of particular interest is the amount of variation explained by normative beliefs compared to experience-based economic and demographic fac- tors, especially household and family contexts. Due to spatial restrictions, our paper focuses on men's correspondence of norms and behaviour. This decision is sup- ported by international research indicating a signifi cant relationship between men’s own attitudes and their actual behaviour (Poortman/Van Der Lippe 2009; Evertsson 2014). Our paper contributes to current research literature both by measuring the multi-dimensionality of gender norms as different male and female systems and by describing empirically how these norms correspond to daily routine behaviour of men in Germany. As our empirical analysis is based on cross-sectional data, our fi ndings have merely explorative value. 2 Theories and concepts on the relationship between norms and behaviour Although – across all social milieus – only a small share of Germans (5.7 percent) favour the traditional male provider and female homemaker model, it is still the dominant pattern of division of labour (52.3 percent) (Veil 2003: 14; see also Both- feld 2005). Most men still have higher earnings than their female partners, and even in families where women have a superior professional status, women still carry out the majority of the housework (Koppetsch/Maier 2001). It is yet unresolved which social mechanisms contribute to the development of both traditional and egalitarian behaviour and why the traditional division of la- bour between men and women predominates in couple households and families (Grunow et al. 2007, 2012). Normative beliefs are considered to contribute to this situation, as the associated theories predict that attitudes and behaviour of men and • Daniela Grunow, Nina Baur524 women change at different paces. These theories state that a traditional division of housework develops because men and women develop different normative beliefs due to their respective social context (Grunow 2007). In order to provide a conceptual and theoretical frame for our anticipated empiri- cal fi ndings, we describe briefl y how the terms “norms”, “beliefs” and “attitudes” relate to one another conceptually and why they are assumed to have an impact on the behaviour of individuals (cp. Lück, editorial on this special issue CPoS 39, 3). Historically, research on gender roles has produced a number of different ter- minologies, as is refl ected in the theoretical description which follows. We follow the assessment of Davis and Greenstein (2009: 88-89) that these different termi- nologies are only partly of conceptual signifi cance. For our paper, it is important to understand norms and behaviour as two separate processes which change over the life course. Following Bolzendahl and Myers (2004), a number of interest-based and experi- ence-based factors account for the development and shift of individual normative gender beliefs (also labelled “gender ideologies”). According to the interest per- spective it is argued that men hold less egalitarian beliefs than women, because they expect to gain less from gender equality than women due to cultural expec- tations. The experience-based factors that are expected to affect gender norms are socialisation, education, labour market experience, and the different stages of family formation such as marriage, divorce and the birth of children (cf. Davis and Greenstein 2009). The fact that gender norms as well as behaviours change over the life course implies that they can infl uence each other. For example, due to their stronger family orientation, traditionally-oriented women are more likely to start a family and have children than women with an egalitarian attitude (Kaufman 2000). Furthermore, a re- cent panel study shows that women receiving higher wages and holding egalitarian gender norms are less likely to return to a traditional division of labour after entering parenthood (Schober 2011). This literature largely argues that in the case of house- work, norms infl uence behaviour and not vice versa (for a summary, see Davis and Greenstein 2009). Lück (Editorial on this special issue CPoS 39, 3), for example, argues that while normative beliefs both directly and indirectly infl uence behaviour, comparatively explicit individual attitudes toward gender roles have a direct effect on behaviour. Attitudes are normative beliefs that act as a basis for behaviour and do not need to be further justifi ed. Individual assumptions regarding explicit gender roles of men and women are assumed to have an indirect effect on behaviour, but a direct effect on individual gender attitudes. Based on these conceptual considerations, we will discuss the two most infl u- ential theoretical approaches that explicitly deal with gender-specifi c infl uences of norms on behaviour: role theory and the “Doing Gender” approach (described in detail in Becker-Schmidt 2005; Grunow 2007, 2010). In a second step, we discuss the implications of both theories taking into account changing social conditions and household and family constellations. The Association between Norms and Actions • 525 2.1 Role theory Role theory is highly contested – especially in the fi eld of women’s studies (Beck- er-Schmidt 2005; Grunow 2010). However, it has a comparatively long tradition in family research going back to the 1940s to Ralph Linton and Talcott Parsons and – viewed historically – constitutes the starting point for current conceptions of gender identity. Role theory assumes that people generally behave compliant to their attitudes and thereby align themselves with social norms. Men's willingness to help with do- mestic and family responsibilities, and women's disposition to demand this com- mitment, should therefore relate strongly to their normative gender role beliefs. Attitudes toward gender roles are generally measured by an individual's acceptance or disapproval of traditional gender roles, and it is assumed that individuals with traditional attitudes practice a traditional division of roles whereas individuals with egalitarian gender-specifi c attitudes attempt to practice egalitarian principles in their daily life. Although empirical support can be found for this approach (Fthenakis et al. 2002a: 100-101), the number of theoretical arguments and empirical fi ndings contradicting this theory has grown in recent years. For example, fi ndings showing a discrepancy between increasingly egalitarian gender norms and gender-specif- ic behaviour falling short of these norms speak against the gender role approach. While 70 percent of West-German men and women agreed to the statement that a traditional division of labour is better for everyone involved in 1982, this agreement decreased to approximately 50 percent in 2000 (Blohm 2002: 537). In East Germany, the percentage of proponents of a traditional division of labour was even lower after the German reunifi cation (35 percent of men, 30 percent of women). Yet, longitudi- nal studies over a similar period of time show only small changes in the daily routine behaviour of men (Künzler et al. 2001; Grunow 2007). These results, however, have not been systematically linked to one and other; fi rst behaviour and norms have mostly been investigated separately, and, second research has so far primarily focused on the infl uence of changes in attitudes to- ward female roles. Changes in normative attitudes toward male roles – especially regarding housework – and their infl uence on men's participation in housework have been neglected. We argue that progressive attitudes and beliefs regarding women’s employment are of limited value to predict men's behaviour in the house- hold. Following the basic principles of role theory, the latter should rather be the result of normative beliefs regarding male participation in domestic responsibilities. Hence, in contrast to previous research, we assume that individual attitudes toward men’s unpaid work are primarily responsible for this relationship (in contrast to, for example, attitudes toward women’s employment), as only these attitudes directly impact men's participation in housework. 2.2 Doing and Undoing Gender Role theory offers an explanation for why norms have an infl uence on behaviour, but it has a fundamental defi cit in explaining which social mechanisms cause changes in • Daniela Grunow, Nina Baur526 gender norms. Drawing on socialisation theory, role theory assumes that divergent attitudes toward gender roles and behaviour already develop during adolescence and remain fairly stable through adulthood. This theoretical conjecture, however, does not correspond to empirical reality. Davis (2007), for example, shows that gen- der norms are infl uenced much more by individual experiences such as household formation, cohabitation and employment than by the gender norms held by parents. The assumption that gender norms do not change over the life course has also not been conclusively verifi ed. For example, a recent hypothesis of life course re- search states that men and women experience a change of normative frames dur- ing the transition to parenthood, which – together with prevalent institutional con- ditions – further traditional attitudes (Grunow et al. 2007, 2012). This hypothesis is consistent with psychological theories of cognitive dissonance (Festinger 1957; Jakobsen 2009). These theories assume that psychological stress, caused by dis- sonance between people’s individual beliefs and own actions, leads them to adapt their attitudes to their circumstances. In line with this view demographic research has found that individuals in changing household contexts adapt their norms and values based on their new experiences (Surkyn/Lesthaeghe 2004). The more recent approaches of “Doing” and “Undoing Gender” refl ect these fi ndings which suggest that norms and behaviours change over the life course. The “Doing Gender” theory suggests that the production and reproduction of male and female gender identities is continuously reinforced through symbolic be- haviours in daily life (West/Zimmermann 1987). Following this theory, it is assumed that housework is, on the one hand, a necessary task of household production, but, on the other hand, a legitimate form for women to affi rm their gender identity and to enact prevalent gender norms (Berk 1985). Conversely, men identify with their social role as provider and avoid housework as it is considered unmanly (Brines 1994). In contrast to the initial pessimism implied by this theoretical perspective on role changes, recent feminist approaches emphasise the possibility of deliberate breaches with prevalent gender norms through non-conforming behaviour (Lorber 2000, 2004). Both perspectives (“Doing Gender” and “Undoing Gender”) stress that social conditions defi ning and institutionally framing gender-specifi c behaviour are important points of reference for the development and shift of individual normative beliefs and actions regarding the role of men and women. 2.3 Social conditions for gender norms Social conditions infl uence how men and women interpret and reproduce gen- der roles in daily life. Classic “Doing Gender” theories base their assumptions on historically developed social and institutional conditions that reproduce a gender- specifi c division of labour. International comparative research shows that women in countries with high gender equality spend less time on housework than women in countries with low gender equality (Stier/Lewin-Epstein 2007; Gupta et al. 2010; Grunow 2013). Male participation in housework is also found to be higher in egali- tarian countries (Fuwa 2004). Social conditions in the German welfare state and The Association between Norms and Actions • 527 labour market are traditionally aligned with the male provider and female home- maker model (Ostner 1995; Pfau-Effi nger 1999; Baur 2007; Hofmeister et al. 2009). The German Civil Code dating from 01.01.1900 defi ned the husband and father as the superior patriarch to whom the wife and mother is legally subordinate (in West Germany until 1977). This political model was gradually relaxed – at least in regard to male and female roles and responsibilities within the family – through various leg- islative amendments (e.g. legal entitlement to a [half-day] place at a day nursery in 1993, fl exible parental leave in 2001, introduction of tax exemptions for household help and childcare in 2002, changes in maintenance obligations and parental leave benefi ts in 2007) (Leitner et al. 2004). Political changes that primarily affect traditional female gender roles are much more numerous and have a longer history than changes affecting traditionally male gender roles. For example, maternity leave regulations were introduced in West Germany in 1952; the right to paternity leave was only introduced through the en- acting of the child-raising benefi ts act in 1986 (Vaskovics/Rost 1999). As a conse- quence of these social developments, we expect heterogeneous attitudes toward gender norms that cannot fully be explained by role theory. In particular, we antici- pate differences between male and female attitudes toward gender norms regard- ing the approval and disapproval of different norm systems that are affected by these structural changes: alignment with the male provider and female homemaker model and approval of women’s employment. 2.4 Household and family context Household and family contexts play a central role with regard to the issues men- tioned above, since laws, structures, and regulations construe gender in connection with marital status and the presence of children. Further, both classic role theories as well as “Doing Gender” theories assume gender-specifi c socialisation within the family to be the starting point for the formation of male and female roles and identi- ties. While it is undisputed that boys and girls are still socialised in gender-specifi c ways (Zimmermann 2006), it is becoming increasingly likely that men gain basic profi ciencies of housework such as cooking, doing the laundry and vacuum-clean- ing during their adolescence and early adult years. Since young people are becom- ing more likely to live on their own and start a household well before partnering and starting a family (Burkart 1997), single men often have to do housework themselves regardless of their normative attitudes. In comparison, men living with a partner are more likely to be able to avoid housework. It can also be assumed that, when liv- ing on their own for a longer period of time, doing housework becomes a matter of course for men. How these experiences infl uence the relationship between gender norms and housework has unfortunately been empirically neglected. Therefore, we investigate the following hypothesis: Doing housework is not only dependent on the normative framing of gender roles, but also on structures offering men oppor- tunities to avoid housework. Accordingly, men living with a partner should do less housework than men living without a partner. • Daniela Grunow, Nina Baur528 Research has shown that women in relationships without children do more housework than single women; however, relationship status does not infl uence the weekly time men spend doing housework (Künzler et al. 2001: 141). The latter fi nd- ing is contrary to our expectation that men seek to withdraw from housework. It has further been shown that the transition to parenthood is paralleled by an increasingly traditional division of labour in the household (Grunow et al. 2007). It is yet unre- solved, though, whether normative attitudes are responsible for this. Care tasks such as child-raising are traditionally seen as maternal responsibilities (Brannen 2007; Dermott 2007) and in a broader sense include housework and other “caring” jobs. Nevertheless, a change in normative attitudes has occurred over the last cou- ple of years, especially regarding the role of fathers (Baur 2007), which is refl ected in recent amendments of parental leave regulations. Research from Sweden has shown that couples in which the father takes parental leave have a higher chance of having another child (Duvander/Andersson 2006). Therefore, it remains unresolved whether the effects of normative attitudes on daily behaviour affect the formation of households and families, e.g. regarding the number of children, or whether norms still have explanatory power regarding the participation in housework after control- ling for family contexts. We therefore investigate two opposing hypotheses: since fathers are expected to participate more regularly in unpaid work, the number of children positively infl uences the probability that men do housework. As institu- tional and normative contexts merely constitute frames for individual behaviour, individual normative attitudes remain signifi cant predictors of male participation in housework, even after controlling for the family context. 2.5 Gender-specifi c connotation of housework In the male provider and female homemaker model, men and women divide work in a traditional way, where the man invests more time in paid work and the wom- an spends more time doing unpaid work (Hewener 2004; Grunow 2007; Dechant/ Schulz 2013 in CPoS 39,3). Thus, there are traditionally “male” and “female” tasks, and housework is typically considered a female task. The concept of tasks being gendered not only applies to the division of em- ployment, housework and child-raising, but also to the division within these fi elds. Research within the sociology of work, for example, has shown that both typically “male professions” (e.g. construction worker or mechanic) and “female professions” (e.g. nurse or nursery-school teacher) exist. In modern times, women have not only increasingly sought employment, they have also – although with a delay – increas- ingly entered traditionally male professions (Krüger 2001; Körner 2006; Lundgreen/ Scheunemann 2006; Baur/Akremi 2011). A similar development can be observed for the typically female domain of “housework”. When men do housework, they undertake typical tasks: DIY (do it yourself), repairs, gardening, car washing, negotiations with authorities, and fi ling tax returns – tasks that are associated with traditionally male professions and rather concern the outside world. Other fi elds of responsibility such as household budg- eting and household organisation, doing the laundry, and taking care of relatives The Association between Norms and Actions • 529 continue to be female domains (Zulehner 2004; Döge/Volz 2004; Döge 2006; Baur/ Akrem 2011). However, since 1991 – parallel to the trend of women entering male profes- sions – a small equalisation of male and female tasks in the household can be ob- served (Döge 2006: 28-29). This equalisation is not evenly distributed over every household. Gender norms seem to have an infl uence on whether men participate in housework and on what tasks they do. Zulehner (2004) distinguishes between “old” men with traditional gender role beliefs and “new” men with modern gender role beliefs. “New” men take on additional tasks in the household that “old” men still leave to their partners such as child care and cooking (Zulehner 2004). Hence, similar to employment, there is not just “housework”, but rather a number of dif- ferent tasks, which have partly male and partly female connotations. It is therefore reasonable to distinguish different tasks when investigating whether and how much housework men undertake. In congruence with the current state of research indi- cating that “new men” also select rather “male” tasks in the household, we expect our multivariate analyses to show that individual normative attitudes of men best explain variations regarding tasks that are “legitimately male”, namely vacuum- cleaning and cooking. Conversely, we expect situational factors such as household contexts to explain differences between those men doing and those men not doing the laundry. 3 Data Analyses are based on data from the survey “The image of men in society”, a CATI survey from 2006 conducted by Siegfried Lamnek and Nina Baur from the depart- ment of sociology and empirical social research of the Catholic University Eichstätt- Ingolstadt. The questionnaire consists of 120 questions capturing behaviour as well as different norms and values. For our research question, it is important that the gender role beliefs of participants and the information concerning who manages each household task are captured. The respondents were male and female German citizens from 16 different mu- nicipalities in Bremen, North Rhine-Westphalia, Baden-Württemberg and Saxony- Anhalt that were selected using purposeful sampling. Detailed information regard- ing the municipalities and the generalisability of the sample can be found in Otte and Baur (2008). Detailed information on the sampling method and the validation of data can be found in Baur and Luedtke (2008). The full questionnaire and a de- tailed description of the sampling method are available on request from the second author. The sample raises concerns regarding the generalisability of our fi ndings. While respondents were selected randomly within the municipalities, the municipalities themselves were not selected randomly. Therefore, signifi cance tests only indicate whether distributions can be generalised at the level of the different municipalities, but not at a federal level. We are, however, not interested in generalising distribu- tions, but in testing theoretical assumptions about patterns and associations. To • Daniela Grunow, Nina Baur530 this end, the sampling design is adequate as contrasting municipalities were con- sciously selected, with the aim that their inhabitants have distinct daily routines and norms. Thus, the different combinations of norms and behaviours that occur empirically should be captured in the data. Due to the above mentioned inferential limitations, our interpretations focus on effect sizes that are relatively large, and we report signifi cance tests mainly for exploratory purposes. The municipalities were selected to maximise the regional spread of urbanity, employment situations and political situations. Within every municipality, a random sample accounting for age and gender was selected using data from the registry offi ce. Based on the ascer- tained telephone numbers, the response rate was 39.3 percent. In total, 709 men and women were interviewed, and 696 answered the questions relevant for our research. In order to distinguish differences between male and female attitudes and responses, we fi rst compare men and women. In a second step, we focus solely on men to explore the question why some men do housework and others do not. 4 Distribution of housework in the family context 4.1 Dependent Variables: Femininity of household tasks In order to investigate the question which normative attitudes and which other fac- tors affect housework, we follow the recommendations of current research and analyse specifi c tasks more closely instead of evaluating housework globally (cf. Schulz/Grunow 2011). Specifi cally, we focus on three tasks: vacuum-cleaning, cook- ing, and doing the laundry. Two criteria were relevant for selecting these activities: First, the selected tasks should produce different patterns regarding the infl u- ence of gender roles. Whereas doing the laundry is nearly always the responsibility of women – even in non-traditional households – the greatest differences between “new” and “old” men become evident in cooking (Zulehner 2004). According to Zul- ehner's research, although men vacuum frequently, it seems that this is not related to modern gender role beliefs. Second, it is unresolved whether certain household tasks are (arduous or cum- bersome) housework or (pleasing and relaxing) recreation (see Döge 2006 in com- parison to Pinl 2004). To account for this issue, two activities (vacuum-cleaning and doing the laundry) were selected that are unanimously seen as housework in recent studies, whereas one activity (cooking) should be somewhere in between work and leisure. For example, cooking for the family with children during the week is gener- ally tedious routine work. Cooking sophisticated or special meals on the weekend, in contrast, can be exciting and recreational leisure. This distinction highlights gen- der-specifi c typologies: whereas women are nearly always responsible for tedious routine work, men cook what they like whenever they enjoy doing it – with or with- out their partner (Lincke 2007: 155-161). The Association between Norms and Actions • 531 4.2 Variations in the division of housework depending on family and household context Most research operationalises the question of who does how much housework by letting participants state whether their partner or they themselves do it. However, looking at what couples actually do, it becomes apparent that a number of other constellations are possible in daily life. To begin with, activities can be undertaken together. Young couples with an egalitarian attitude, for example, often cook togeth- er (Lincke 2007: 155-161). Thus, it is not surprising that when having to choose in surveys between themselves and their partner carrying out an activity, participants generally pick themselves. In consequence, results on housework differ consider- ably depending on who was asked. For instance, Zulehner (2004) fi nds 80 percent of women with a traditional attitude stating to vacuum, whereas less than half of men with a traditional attitude state that their partner carries out this task. Further, other people may live in a household and also be capable of doing house- work, e.g. children, parents, or fl atmates. Housework can also be externalised, that is, carried out by other people not living in the household such as friends, acquaint- ances, relatives, neighbours, a cleaner or social services (e.g. meals on wheels). It can be expected that housework is more likely to be delegated to women, since private help is provided more often by women (Meulemann/Beckers 2004) and pro- fessions that are closely related to housework and social services are traditionally female. Therefore, the frequent conclusion that young couples and singles practice a modern division of labour may partly result from the way housework is measured. After all, daily life can be organised in such a way that certain tasks do not occur at all. Employed people often eat warm meals in the canteen and do not cook during the week (Lincke 2007: 155-161). To circumvent this problem, our study specifi cally asks who carries out each of the mentioned tasks. The introduction for the questions concerning housework was, “In every household, different tasks have to be taken care of. Please, tell us who carries out each of the following tasks: mainly you, mainly another person, or you and another person in equal parts? By another person we mean e.g. your part- ner, your parents, children, fl atmates, but also social services and paid workers”. We then asked, “Who does the laundry in your household? Do you mainly do it yourself, does another person mainly do it, or do you and another person share it in equal parts?” A response alternative that wasn’t specifi cally offered was, “This task does not occur in my household”. If a respondent answered “mainly me” or “me and another person in equal parts”, we asked, “Which other people do the laundry in your household?”. Response alter- natives were not specifi cally offered, but coded for gender. Following options were possible: my wife/partner/girlfriend – my husband/partner/boyfriend – a female friend or acquaintance – a male friend or acquaintance – my mother – my father – my daughter – my son – other female relatives – other male relatives – a female neighbour – a male neighbour – a female fl atmate – a male fl atmate – a cleaner – social services (e.g. meals on wheels) – other, namely: [open-ended] – I’m the only one doing it. For vacuum-cleaning, we asked, “Who does the vacuum-cleaning?”, • Daniela Grunow, Nina Baur532 for cooking, “Who cooks in your household?” (questions and answers were trans- lated by CPoS). The remaining questions and response alternatives were equivalent to the questions on doing the laundry. As Table 1 illustrates, our data confi rm previous research: overall, women carry out the mentioned household tasks more often than men, (irrespective of relation- ship status). At least two thirds of men receive help from their partner or other people, irrespective of the specifi c task, whereas women more often manage them on their own. As expected, there are clear differences between the various tasks. Participation by gender has almost equalised for vacuum-cleaning: seven out of ten men and eight out of ten women vacuum; 35 percent of men and 45 percent of women are solely responsible for this task in the household. The division of labour differs con- siderably, though, for cooking and doing the laundry: nine out of ten women cook and do laundry. Every second woman is solely responsible for cooking, and eight out of ten women never receive any assistance in doing the laundry. In contrast, only every second man cooks, and only three out of ten men do laundry at all (of these, only one out of fi ve does it on his own). Examining gender-specifi c differ- ences for these household tasks, it becomes apparent that doing the laundry most strongly polarises the genders whereas vacuum-cleaning is most equally distrib- uted. Cooking holds a position between the two. 4.2.1 The relevance of household and family contexts for the division of housework The household and family context affects the participation of men and women in housework. As Table 2 illustrates, 83 percent of single men vacuum, 57 percent cook, and 51 percent do laundry either on their own or together with another per- Tab. 1: Distribution of selected household tasks for men (n=361) and women (n=335), in percent Oneself Oneself with Only other Oneself Only partner other people people with partner Men Vacuum-cleaning 35 12 12 23 19 100 Cooking 21 7 19 20 34 100 Doing the laundry 17 4 22 10 47 100 Women Vacuum-cleaning 45 16 12 20 7 100 Cooking 56 13 9 19 4 100 Doing the laundry 76 9 7 7 1 100 Source: Own calculations based on the study “The image of men in society” The Association between Norms and Actions • 533 son. If they have a partner with which they do not cohabit, the share of men doing the laundry remains almost the same (49 percent), but the share vacuum-cleaning and cooking decreases considerably (75 percent and 51 percent, respectively). If men move in with their partner, participation in housework decreases further: only 65 percent vacuum, 44 percent cook, and 21 percent do laundry. Overall, men do considerably less housework when in a relationship. This result corresponds to our expectations. For women we do not fi nd these pronounced partnership effects. Independent of their relationship status, eight out of ten women vacuum. The only partner effect for women shows that the percentage of women vacuum-cleaning together with an- other person doubles from approximately 20 percent to approximately 40 percent. Tab. 2: Division of selected household tasks depending on relationship status Source: Own calculations based on the study “The image of men in society” rather oneself with another person rather another person does not occur N in % in % in % in % Vacuum-cleaning Men with partner, cohabiting 27 39 35 226 with partner, not cohabiting 49 26 23 2 53 singles 51 32 16 1 77 Women with partner, cohabiting 41 38 21 197 with partner, not cohabiting 39 44 17 59 singles 60 23 17 78 Cooking Men with partner, cohabiting 13 31 56 226 with partner, not cohabiting 38 13 47 2 53 singles 32 25 42 1 77 Women with partner, cohabiting 60 32 7 1 197 with partner, not cohabiting 31 39 31 59 singles 64 26 10 78 Doing the laundry Men with partner, cohabiting 6 15 79 226 with partner, not cohabiting 36 13 49 2 53 singles 36 14 49 77 Women with partner, cohabiting 84 13 4 197 with partner, not cohabiting 51 31 19 59 singles 78 12 10 78 • Daniela Grunow, Nina Baur534 If women have a partner, but are not cohabiting, the percentage cooking or doing the laundry decreases from 90 percent to approximately 70 percent and 80 percent, respectively, as compared to singles. Among those with a partner but not cohabit- ing approximately 40 percent and 30 percent, respectively, carry out these tasks together with another person. If women move in with their partner, the percentage exceeds the level for single women, with 92 percent and 97 percent respectively, and they receive assistance as often as when they were single. 4.2.2 Delegation of tasks to third parties Our results indicate that relationship context only partly explains the variation of male participation in specifi c household tasks. As Table 1 shows, 31 percent (doing the laundry) to 70 percent (vacuum-cleaning) of men participate in the examined household tasks – irrespective of their relationship status. One in four men is as- sisted by another person or his partner in each of these tasks. One in four women receives assistance with vacuum-cleaning, but only one in fi ve with cooking and one in six with laundry. These results indicate why women take on more house- work when moving in with a man: they replace the other people who previously carried out household tasks for their partner. Thus, when the man does not carry out these tasks or when he does not do them on his own, it depends on the family context who takes them on. Further analyses (not displayed) show that both men and women are most often assisted by their mothers, and that women receive help from third parties considerably less often than men. 5 Gender role beliefs When investigating the impact of norms and values on male participation in house- work the question arises which norms and values are most relevant for actions. Existing surveys have operationalised attitudes toward the division of labour within the family differently, e.g. in the ISSP (ZUMA/Infas 1994, 1997, 2002), the LBS-fami- ly-study (Fthenakis et al. 2002a, 2002b), or Zulehner’s (n.d.) research on men. Since the focus of this paper is on male participation in housework, and since we follow Poortman and Van der Lippe (2009) in assuming that men’s participation depends on their own normative attitudes (and not those of their partners), the fi ndings re- ported here focus on male gender role beliefs. In order to increase comparability to previous research on female gender role beliefs, we compare male and female normative attitudes, but do not relate these normative attitudes of women to actual behaviour of men. Following the theories on the impact of norms on behaviour described above, we focus on four normative systems that are assumed to infl uence male participa- tion in housework for different reasons: attitudes toward the male provider and female homemaker model (system 1) and attitudes toward employment as the cen- tral role for a man (system 3) are important, because both aspects normatively free men from the responsibility of participating in housework in order to be “good” The Association between Norms and Actions • 535 partners and fathers (cf. Bielby/Bielby 1989). The approval of women’s employment (system 2) has been selected as a predictor variable in previous research, as it was Fig. 1: Distribution of gender norms 0 100 Attitude toward the male provider and female homemaker model Approval of women's employment Endorsement of men's unpaid work Employment as the man's central role 6% 41% 43% 10% 23% 37% 29% 12% 12% 39% 49% 36% 49% 12% Men 0 100 Attitude toward the male provider and female homemaker model Approval of women's employment Endorsement of men's unpaid work Employment as the man's central role 9% 42% 41% 8% 16% 28% 33% 23% 10% 32% 58% 20% 56% 23% Strongly disagree Rather disagree Rather agree Strongly agree Women % % % % 354 354 354 353 N 332 333 333 332 N Source: Own calculations based on the study “The image of men in society” • Daniela Grunow, Nina Baur536 assumed to be logical that men have to participate more in housework if women are also expected to work. In contrast, we have argued in section 2 that women’s role as earners has changed institutionally much more than norms regarding men’s house- work. Therefore, the hypothesis of previous research that both norms are interre- lated may be empirically incorrect. Our fi nal indicator – the endorsement of men’s unpaid work (system 4) – seems more appropriate to determine male participation in housework, because this norm is directly related to the associated behaviour. The literature frequently discusses a further aspect regarding the attitude toward the division of labour within the family: the disposition of men to fatherhood. Research has confi rmed repeatedly that men desire to be more active fathers and increasing- ly want a stronger role in the upbringing of their children (Erler et al. 1988; Fthenakis et al. 1999; Matzner 2004; Zulehner 2004; Oberndorfer/Rost 2005; Cyprian 2005; Grunow 2007; Baur 2007; Hofmeister et al. 2009). However, we decided to omit this aspect as it primarily concerns how men raise their children. Figure 1 shows a summary of the different systems and allows a comparison of male and female rates of approval. The following sub-sections describe which items constitute each system and how men agree to each norm system. 5.1 Identifi cation with the male provider and female homemaker model When investigating “gender norms”, the fi eld of gender studies often examines the extent to which couples identify with the male provider and female homemaker model (Ostner 1995; Pfau-Effi nger 1999). This model conforms to the ideal of clas- sic role theory. We constructed a number of items capturing different aspects of the male provider – female homemaker model in order to identify our participants’ attitudes toward this model. The specifi c questions and the distribution of male responses can be found in Table 3. The comparatively large Cronbach’s α in Table 3 indicates that items correlate strongly with each other. Looking at the different items, it becomes apparent that the male provider and female homemaker model involves more than a complementary division of roles in daily life, in which men are assigned the task of employment and women are assigned household tasks and – as soon as there are children – child- raising tasks. Our empirical fi ndings further confi rm Keddis’ (2003) results indicating that the male provider – female homemaker model not only implies a pragmatic divi- sion of roles, but that the idea of partners representing a single unit plays a central role. This single unit is expressed in the belief that partners in a good relationship always spend their weekends together and always think and feel the same. Table 3 shows that specifi c normative aspects polarise men, since the rates of approval vary greatly: only one third of the interviewed men state that partners should always think and feel the same in a good relationship, but seven out of ten men think that partners should always spend their weekends together, and three quarters think that in a good relationship, both partners want children. Although only 15 percent think that the career of the man takes precedence over the family, four out of ten men agree to the statement that a man should make sure that his partner never needs to worry about fi nancial issues. At the same time, one in two The Association between Norms and Actions • 537 T a b . 3: A tt it u d e s o f m e n t o w a rd t h e m a le p ro v id e r an d f e m a le h o m e m a ke r m o d e l Le v e l o f si g n ifi c an ce : + α ≤ 0 .1 0 ; * α ≤ 0 .0 5 ; ** α ≤ 0 .0 1 ; ** * α ≤ 0 .0 0 1 S o u rc e : O w n c al cu la ti o n s b as e d o n t h e s tu d y “ T h e im ag e o f m e n in s o ci e ty ” D is tr ib u ti o n o f re sp o n se s in % C o rr e la ti o n o f n o rm s an d b e h av io u r (P e ar so n ’s r ) N S tr o n g ly d is ag re e R at h e r d is ag re e R at h e r ag re e S tr o n g ly ag re e M an v ac u u m s M an co o ks M an d o e s la u n d ry A tt it u d e t o w ar d t h e m al e p ro v id e r an d f e m al e h o m e m ak e r m o d e l 6 4 1 4 3 1 0 -0 .1 5 ** -0 .1 1 * -0 .1 3 * 3 5 4 In a g o o d r e la ti o n sh ip , p ar tn e rs a lw ay s sp e n d t h e ir w e e ke n d s to g e th e r. 5 2 6 3 9 3 0 -0 .0 4 -0 .1 1 * -0 .1 0 + 3 5 8 In a g o o d r e la ti o n sh ip , p ar tn e rs a lw ay s th in k an d f e e l t h e sa m e . 1 7 5 0 2 3 1 0 < 0 .0 1 0 .0 6 -0 .0 2 3 5 8 In a g o o d r e la ti o n sh ip , b o th p ar tn e rs w an t ch ild re n . 4 2 0 3 7 3 9 -0 .0 1 -0 .1 3 * -0 .1 6 * 3 5 6 T h e c ar e e r o f th e m an t ak e s p re ce d e n ce o v e r th e f am ily . 3 8 4 7 1 0 5 -0 .1 0 + < - 0 .0 1 < - 0 .0 1 3 5 2 In a g o o d r e la ti o n sh ip , th e m an m ak e s su re t h at t h e w o m an n e v e r h a s to w o rr y a b o u t fi n an ci al is su e s. 1 3 4 8 2 4 1 5 -0 .1 6 ** -0 .0 5 -0 .1 1 * 3 5 6 In a g o o d r e la ti o n sh ip , a w o m an m ak e s su re t h at t h e m an ca n f o cu s o n h is jo b a n d n e e d s n o t w o rr y a b o u t o th e r is su e s. 9 3 7 3 2 2 2 -0 .0 7 -0 .0 4 -0 .1 1 * 3 5 6 T h e m o st im p o rt an t ta sk o f th e w o m an is t h e f am ily . 1 8 3 6 2 6 2 0 -0 .1 7 ** -0 .1 2 * -0 .1 2 * 3 5 4 T h e w o m an c ar ri e s th e m ai n r e sp o n si b ili ty f o r th e h o u se h o ld . 2 3 4 2 2 3 1 3 -0 .2 8 ** * -0 .2 4 ** * -0 .1 7 ** * 3 5 7 A w o m an s h o u ld n o t w o rk , if h e r ch ild is y o u n g e r th an t h re e y e ar s o ld . 1 0 2 4 2 1 4 5 0 .0 2 -0 .1 3 ** -0 .0 9 + 3 5 4 C ro n b a ch ’s 0 .7 6 • Daniela Grunow, Nina Baur538 men agrees that the most important task of a woman is the family and that she should take care of things so that the man can focus on his job. The latter applies more to raising children than to housework: only one third thinks that the woman has to carry the main responsibility for the household, but two thirds think that a mother should not work if her child is younger than three years old. 45 percent of the interviewed men even oppose the idea of mothers with young children being employed. Although the distribution of responses varies, a pattern can be detected. Re- spondents that agree to a particular item tend to agree to the other items as well, whereas respondents that disagree with an item tend to disagree with the other items as well. We therefore assume that the items measure the latent system “at- titude toward the male provider and female homemaker model” and summarise them in one variable using factor analysis (Fromm 2004).1 As Figure 1 illustrates, 10 percent of the interviewed men strongly agree to the male provider and female homemaker model, 43 percent rather agree, 41 percent rather disagree, and 6 per- cent strongly disagree. As expected, male rates of approval are (slightly) higher than female rates of approval. 5.2 Approval of women’s employment The male provider – female homemaker model assigns the role of employment to the man and the task of taking care of the family to the woman. However, this does not say much about women’s employment: when supporting the male provider – female homemaker model it is both possible to oppose women’s employment gen- erally or in particular stages of life and to “allow” women to work while viewing this employment solely as a provision of extra income. Conversely, opposing the male provider – female homemaker model does not automatically imply seeing women’s employment as the central role of women. In a qualitative study, Keddi (2003) shows that some men and women neither perceive employment nor the household as the central point of their life, but consider personal self-fulfi lment as more important. Other people oppose the male provider – female homemaker model, but align their life completely with their partner or do not know how they want to live. Indeed, fac- tor analysis shows that approval of women’s employment is a separate normative system. Independent of whether respondents support or oppose the male provider – female homemaker model, some endorse women’s employment whereas others object to it. As Table 4 illustrates, 58 percent of interviewed men hold the view that both men and women should work. However, only 40 percent believe it is good for a child to have an employed mother, and 66 percent hold the view that a mother should not work if her child is younger than three years old. Summarising these items into a single variable, six out of ten men object to women’s employment. In contrast to the 1 We initially conducted factor analyses separately for men and women, achieving very similar results. This paper reports the results for the whole data set. The Association between Norms and Actions • 539 T a b . 4: M e n 's a p p ro v a l o f w o m e n 's e m p lo y m e n t Le v e l o f si g n ifi c an ce : + α ≤ 0 .1 0 ; * α ≤ 0 .0 5 ; ** α ≤ 0 .0 1 ; ** * α ≤ 0 .0 0 1 S o u rc e : O w n c al cu la ti o n s b as e d o n t h e s tu d y “ T h e im ag e o f m e n in s o ci e ty ” D is tr ib u ti o n o f re sp o n se s in % C o rr e la ti o n o f n o rm s an d b e h av io u r (P e ar so n ’s r ) N S tr o n g ly d is ag re e R at h e r d is ag re e R at h e r ag re e S tr o n g ly ag re e M an v ac u u m s M an co o ks M an d o e s la u n d ry A p p ro v al o f w o m e n ’s e m p lo y m e n t 2 3 3 7 2 9 1 2 0 .0 3 0 .1 5 ** 0 .1 0 + 3 5 4 A w o m an s h o u ld n o t w o rk , if h e r ch ild is y o u n g e r th an th re e y e ar s o ld . (n e g at iv e ) 1 0 2 4 2 1 4 5 0 .0 2 -0 .1 3 * -0 .0 9 + 3 5 4 M e n a n d w o m e n s h o u ld b o th b e e m p lo y e d . 1 0 3 2 3 3 2 5 0 .0 5 0 .0 5 0 .0 8 3 5 5 It is g o o d f o r a ch ild , if t h e m o th e r is e m p lo y e d . 1 3 4 7 3 1 9 0 .0 1 0 .1 6 ** 0 .1 0 + 3 5 0 C ro n b a ch ’s 0 .5 7 • Daniela Grunow, Nina Baur540 attitude toward the male provider – female homemaker model, considerable gen- der-specifi c differences can be observed: six out of ten women endorse women’s employment parallel to having young children in the household. Previous research, such as the ISSP (ZUMA/Infas 1994, 1997, 2002), the LBS- family-study (Fthenakis et al. 2002a, 2002b), as well as Zulehner’s (n.d.) research on men, and longitudinal analyses show that the share of proponents of women’s employment has continuously risen in industrialised countries over the last decades (Lück 2004, 2006, 2009; Hofäcker/Lück 2004; Lück/Hofäcker 2008). 5.3 Employment as men's central social role Whereas the question of how important employment is for women in comparison to her family is captured comparatively frequently in research, the inverse question – how important is employment for men – is rarely asked. Is employment – as the male provider – female homemaker model suggests – really considered the central social role of the man (Baur/Luedtke 2008)? When examining these questions, an inverse pattern in comparison to the at- titude toward women’s employment can be observed: while the majority of women endorse women’s employment and the majority of men object to it – and men there- fore hold a traditional view of gender roles – women are even more conservative with regard to male employment: eight out of ten women, but only six out of ten men view employment as the central social role of men (see the rates of approval “strongly agree” and “rather agree” in Fig. 1). Overall, the attitude toward the role of men is much less ambiguous than toward the role of women. Both attitudes are not necessarily related to each other. As displayed in Table 5, an overwhelming majority of men believe that men work in order to support their family (99 percent), because they then feel needed (77 per- cent), and because they enjoy working (72 percent). Roughly half think that employ- ment is part of being a man. Only four out of ten consider work the most important thing in life, and only every fourth man thinks that men can only gain recognition from others via their job. Only 15 percent believe that the career of the man takes precedence over the family. 5.4 Endorsement of men's unpaid work Are men able to and should they do housework? As Table 6 illustrates, the times in which men perceived housework as demeaning are over: only 4 percent see it that way, even if one in three thinks that women should carry the main responsibility for the household. 86 percent of the interviewed men believe that they are able to do housework as thoroughly as women, and 79 percent consider themselves just as competent in parenting. Overall, nine out of ten respondents endorse the participa- tion of men in housework, women even slightly more than men. The bivariate analyses in Tables 3 to 6 highlight a number of aspects: fi rst, norms and behaviour correlate substantially. Second, the direction of infl uence of particu- lar norm systems vary: the more men support the male provider – female home- The Association between Norms and Actions • 541 T a b . 5: M e n 's a tt it u d e s to w a rd m a le e m p lo y m e n t Le v e l o f si g n ifi c an ce : + α ≤ 0 .1 0 ; * α ≤ 0 .0 5 ; ** α ≤ 0 .0 1 ; ** * α ≤ 0 .0 0 1 S o u rc e : O w n c al cu la ti o n s b as e d o n t h e s tu d y “ T h e im ag e o f m e n in s o ci e ty ” D is tr ib u ti o n o f re sp o n se s in % C o rr e la ti o n o f n o rm s an d b e h av io u r (P e ar so n ’s r ) N S tr o n g ly d is ag re e R at h e r d is ag re e R at h e r ag re e S tr o n g ly ag re e M an v ac u u m s M an co o ks M an d o e s la u n d ry E m p lo y m e n t as m e n 's c e n tr al s o ci al r o le 3 3 6 4 9 1 2 -0 .1 9 ** * -0 .1 2 * -0 .0 7 3 5 3 M e n w o rk , b e ca u se w o rk in g is p ar t o f b e in g a m an . 1 5 3 2 2 9 2 4 -0 .1 4 ** -0 .1 2 * -0 .0 9 + 3 5 2 M e n w o rk , b e ca u se t h e jo b is t h e m o st im p o rt an t th in g in lif e f o r th e m . 1 5 4 4 2 7 1 4 -0 .0 8 -0 .0 9 + 0 .0 3 3 5 4 M e n w o rk , b e ca u se t h e y e n jo y w o rk in g . 5 2 4 4 8 2 4 0 .0 7 -0 .0 3 -0 .0 2 3 5 4 M e n w o rk , b e ca u se t h e y t h e n f e e l n e e d e d . 3 2 0 4 3 3 4 -0 .0 5 -0 .0 3 -0 .0 6 3 5 4 M e n w o rk in o rd e r to s u p p o rt t h e ir f am ily . 0 1 2 7 7 2 -0 .1 6 ** -0 .1 7 ** -0 .1 2 * 3 5 4 M e n w o rk , si n ce t h e y c an o n ly g ai n r e co g n it io n f ro m o th e rs v ia t h e ir jo b . 3 1 4 4 1 9 5 -0 .1 5 ** -0 .0 6 0 .0 5 3 5 4 M e n 's c ar e e r ta ke s p re ce d e n ce o v e r th e f a m ily . 3 8 4 7 1 0 5 -0 .1 0 + < - 0 .0 1 < - 0 .0 1 3 5 2 C ro n b a ch ’s 0 .7 1 • Daniela Grunow, Nina Baur542 T a b . 6: M e n 's e n d o rs e m e n t o f m e n 's u n p a id w o rk Le v e l o f si g n ifi c an ce : + α ≤ 0 .1 0 ; * α ≤ 0 .0 5 ; ** α ≤ 0 .0 1 ; ** * α ≤ 0 .0 0 1 S o u rc e : O w n c al cu la ti o n s b as e d o n t h e s tu d y “ T h e im ag e o f m e n in s o ci e ty ” D is tr ib u ti o n o f re sp o n se s in % C o rr e la ti o n o f n o rm s an d b e h av io u r (P e ar so n ’s r ) N S tr o n g ly d is ag re e R at h e r d is ag re e R at h e r ag re e S tr o n g ly ag re e M an v ac u u m s M an co o ks M an d o e s la u n d ry E n d o rs e m e n t o f m e n 's u n p ai d w o rk 0 1 2 3 9 4 9 0 .2 0 ** * 0 .1 6 ** 0 .1 1 * 3 5 4 M e n a re ju st a s co m p e te n t in p a re n ti n g a s w o m e n . 2 1 9 3 5 4 4 0 .1 3 * 0 .0 4 0 .0 7 3 5 4 M e n c an d o h o u se w o rk a s th o ro u g h ly a s w o m e n . 2 1 3 2 9 5 7 0 .1 4 ** 0 .1 2 * 0 .0 7 3 5 7 T h e w o m an c ar ri e s th e m ai n r e sp o n si b ili ty f o r th e h o u se h o ld . (n e g at iv e ) 2 3 4 2 2 3 1 3 -0 .2 8 ** * -0 .2 4 ** * -0 .1 7 ** * 3 5 7 It is d e m e an in g f o r a m an t o d o h o u se w o rk . (n e g a ti v e ) 6 9 2 7 3 1 -0 .1 3 * -0 .0 8 -0 .0 1 3 5 7 C ro n b a ch ’s 0 .5 4 The Association between Norms and Actions • 543 maker model and see employment as their central role, the less they participate in the selected housework tasks. In contrast, if they endorse women’s employment and men’s unpaid work, the probability of doing housework increases. Third, it is worth distinguishing and looking at different norm systems and particular tasks individually, since the direction of infl uence of different norm systems on particular tasks is similar, but unequally strong. For example, men endorsing women’s em- ployment cook and do laundry more often than men objecting to it (r = 0.15 and r = 0.10, respectively). The relationship between the approval of women’s employment and vacuum-cleaning, however, is insignifi cant. These relationships are different for employment as the central role of men: the impact on doing the laundry is insig- nifi cant, whereas men focussing on their career vacuum and cook signifi cantly less often (r = -0.19 and r = -0.12, respectively). Subsequently, we want to investigate the relative infl uence of the different norm systems on each other and whether the relationship between norms and behaviour can still be observed when controlling for the impact of experience-based factors that infl uence gender norms (Bolzendahl/Myers 2004). Specifi cally, we account for regional differences between West and East Germany, level of education, employ- ment situation, employment prestige and income as well as household and fam- ily context (accounting for partner’s employment status, number of adults in the household and number of children under the age of 18 in the household). 6 The association between gender norms and men’s housework controlling for experience-based factors Following sociological theories on the impact of norms on behaviour and previous research examining the impact of gender-specifi c attitudes toward housework on behaviour (e.g. Poortman/Van der Lippe 2009), we empirically analyse to what ex- tent the identifi ed norm systems can explain men’s participation in housework. We use each identifi ed norm system as an explanatory variable in multivariate analyses. The dependent variables are selected household tasks, namely vacuum-cleaning, cooking, and doing the laundry (cf. sub-section 4.2). Response alternatives “one- self”, “oneself with another person”, and “oneself with partner” each describe the participation of men in a household task (coded “1”). The alternatives “only other people” and “only partner” were coded “0”. For each of the three dependent vari- ables we calculate three logistic regression models.2 The fi rst model only includes experience-based factors and represents current practice in model-building regarding the division of labour in the household (for a 2 As a comparison of logit coeffi cients across different models is not appropriate (Mood 2010), we use the validation method by Karlson, Holm, and Breen (2010) to test for signifi cant changes between models. This procedure – also called KHB method – allows the differences between logit coeffi cients in different models to be decomposed into a scaling and a confounding com- ponent (calculations based on the KHB method are not presented in the tables and are available on request from the authors). • Daniela Grunow, Nina Baur544 summary, see Davis and Greenstein 2009). The second model contains the norma- tive attitudes of respondents described above. The third model joins the variables of the previous models in order to analyse the extent to which the predictive power of normative attitudes is due to variations in the socio-demographic, economic, household and family context. This third model accommodates Bolzendahl and My- ers’ (2004) argument that individual normative beliefs are infl uenced by experience- based socio-demographic factors as well as the household and family context. Thus, the socio-demographic factors and the household and family context serve as control variables for experience-based differences between men, which may confound the coeffi cients for normative attitudes in the second model. The aim of our analyses is not to develop a parsimonious regression model, but to examine the empirical relevance (or irrelevance) of the identifi ed norm systems for explaining men's behaviour. The following tables display the logit coeffi cients. 6.1 Vacuum-cleaning As Table 7 (model 1) shows, our data suggest that neither level of education nor employment status or employment prestige have a signifi cant impact on whether men vacuum. There are also no differences between East and West Germany. How- ever, men are less likely to vacuum the older they are and the higher their household income is. As further calculations with fully standardised coeffi cients show, the net equivalent household income is the most important explanatory variable in the fi rst model.3 Neither being partnered nor the partner’s employment status explains why some men vacuum and others do not. Among the family context variables, only the num- ber of adults in the household has a highly signifi cant and negative impact: men are less likely to participate in vacuum-cleaning the more adults live in the household. This result confi rms our assumption that opportunity structures matter when men withdraw from doing housework. The fi ndings on the household context in sub- section 4.2.1 illustrate that this relationship is not trivial, as single men and women generally do not have an alternative to doing housework themselves. Model 2 illustrates the relative impact of the four norm systems on the probabil- ity of men vacuum-cleaning. As expected, a signifi cant relationship between norms and behaviour can only be found regarding the endorsement of men’s unpaid work (cf. sub-section 2.1).4 The more men support men’s unpaid work, the greater the 3 To compare the explanatory power of different independent variables in logistic regression models, coeffi cients have to be standardised – similarly to OLS. Both independent and depend- ent variables are standardised with a mean value of 0 and a standard deviation of 1 in order to create fully standardised coeffi cients (Long/Freese 2006). Results for the full model (model 3 in Tables 7-9) using fully standardised coeffi cients are reported in Table A1 in the appendix. 4 The high value of the logit coeffi cient (b = 3.87) needs to be seen in light of the extreme skew- ness of the distribution of responses for this norm system (see Table 6). Further analyses ruled out multicollinearity. The correlation of the explanatory variable and the response variable is 0.5. As displayed in Table A1 in the appendix, the effect is strong but not implausibly high for the standardised coeffi cients. The Association between Norms and Actions • 545 probability that they at least participate in vacuum-cleaning. Further analyses indi- cate that this effect becomes even stronger if children live in the household. When we additionally control for the experience-based variables in the full mod- el (model 3), the impact of attitudes toward men’s unpaid work on the probability of a man vacuum-cleaning does not decrease signifi cantly, as calculations based on the KHB method confi rm. At the same time, the effect of age loses its signifi cance indicating that older men hold more traditional views. The probability that single men vacuum also signifi cantly increases in this model. A change of sign can be ob- served for education and region, but these values are below the signifi cance level of 0.10. The other variables remain fairly stable. Tab. 7: Explanatory Variables for men vacuum-cleaning, logistic regression Level of signifi cance: + α ≤ 0.10; * α ≤ 0.05; ** α ≤ 0.01; *** α ≤ 0.001 Source: Own calculations based on the study “The image of men in society” Model 1 Model 2 Model 3 Region, ref. West Germany East Germany -0.02 0.16 Age (in years) -0.03** -0.02 Education, ref. maximum “Lower School Leaving Certificate” (Hauptschule) “General School Leaving Certificate” (Realschule) 0.01 -0.28 Higher education entrance qualification 0.33 -0.09 Graduate degree 0.24 -0.09 Employment status, ref. employed Pensioner 0.37 0.49 In education -0.92 -0.66 Other 0.24 -0.18 Treiman prestige 0.02 0.02 Net equivalent income (in thousand euro) -0.67* -0.76** Partner’s employment status, ref. not employed No partner 0.66 0.91* Partner employed part-time 0.08 0.19 Partner employed full-time 0.23 0.35 Family context Number of adults in the household (apart from the respondent) -0.45** -0.50** Number of children in the household -0.19 -0.23 Normative attitudes Approval of the male provider - female homemaker model 0.44 0.28 Approval of women’s employment 0.11 -0.55 Employment as men's central social role -1.61 -1.27 Endorsement of men’s unpaid work 3.87** 4.35** Explained variance (pseudo R2) 7% 7% 15% N 303 303 303 • Daniela Grunow, Nina Baur546 The full model accounts for 15 percent of the variance. Thus, the experience- based factors mentioned in previous research only have limited explanatory power regarding male participation in vacuum-cleaning. A comparison of the different models suggests that the four norm systems in model 2 have as much explanatory power as the socio-demographic, economic, and family context variables in model 1. Both models are able to explain 7 percent of the variance in the participation of men in vacuum-cleaning. In the full model, the endorsement of men’s unpaid work is the strongest predictor of the participation of men in vacuum-cleaning, followed by household income and number of adults in the household (see Table A1 in the appendix). Thus, norms appear to be highly signifi cant in explaining male participa- tion in vacuum-cleaning. 6.2 Doing the laundry Table 8 summarises the explanatory variables of the multivariate logistic regres- sions examining men’s doing of laundry. In contrast to vacuum-cleaning, doing the laundry is the most feminine task, in which men participate the least – a fact previ- ous research has also shown. In contrast to vacuum-cleaning, none of the normative attitudes have a (signifi - cant) impact on the probability of men doing the laundry (model 2, α ≤ 0.10). The co- effi cients refl ecting the approval of the male provider – female homemaker model, the approval of women’s employment, and the endorsement of men’s unpaid work, however, point in the expected direction. The variable measuring the endorsement of men’s unpaid work becomes signifi cant in the full model (model 3). Thus, men who endorse men’s unpaid work are more likely to do laundry than men who object to men’s unpaid work. Again, in contrast to vacuum-cleaning, neither demographic nor economic fac- tors are able to explain why men (do not) do laundry (model 1). Relationship and family context, however, have explanatory power. The probability of single men and women doing the laundry is about two and a half times as high as it is for men with partners who are not in employment. If other adults or children live in the household, men also do laundry signifi cantly less frequently. This again indicates that men often make use of opportunity structures in order to avoid housework. The effects of relationship and family context remain signifi cant in the full model. The full model in Table 8 explains 17 percent of the observed variance. How- ever, normative attitudes only explain 3 percent of the observed variance (model 2), whereas relationship and family context explain 15 percent. The number of adults in the household, followed by the number of children in the household and having a partner produce the largest effects for male participation in doing the laundry (see Table A1 in the appendix). This confi rms our expectation that normative at- titudes explain mainly the participation of men in “legitimately male” household tasks whereas the more feminine task of doing the laundry is mainly explained by situational factors such as being single and the household context. The Association between Norms and Actions • 547 6.3 Cooking The results of the explanatory variables for men cooking are illustrated in Table 9. Men cook comparatively often; however, according to previous research, it is not clear whether cooking is actually a leisure activity. Model 1 shows that men with a general school leaving certifi fi cate (Realschule) and a higher education entrance qualifi cation cook more than twice as often as men with a maximum of lower secondary schooling (Hauptschule). The effect for gradu- ates points in the same direction, but it is smaller and not statistically signifi cant. In the full model, only the positive effect of a higher education entrance qualifi cation remains signifi cant in explaining the participation of men in cooking. The higher the Tab. 8: Explanatory variables for men doing the laundry, logistic regression Level of signifi cance: + α ≤ 0.10; * α ≤ 0.05; ** α ≤ 0.01; *** α ≤ 0.001 Source: Own calculations based on the study “The image of men in society” Model 1 Model 2 Model 3 Region, ref. West Germany East Germany 0.43 0.32 Age (in years) -0.02 -0.02 Education, ref. maximum “Lower School Leaving Certificate” (Hauptschule) “General School Leaving Certificate” (Realschule) 0.30 0.20 Higher education entrance qualification 0.21 0.09 Graduate degree -0.13 -0.30 Employment status, ref. employed Pensioner -0.27 -0.26 In education 0.74 0.77 Other 0.51 0.40 Treiman prestige 0.02 0.02 Net equivalent income (in thousand euro) -0.14 -0.10 Partner’s employment status, ref. not employed No partner 0.86* 1.03* Partner employed part-time 0.43 0.43 Partner employed full-time 0.43 0.42 Family context Number of adults in the household (apart from the respondent) -0.81** -0.78** Number of children in the household -0.61* -0.58* Normative attitudes Approval of the male provider - female homemaker model -1.59 -0.52 Approval of women’s employment 0.74 0.72 Employment as men's central social role 0.44 0.66 Endorsement of men’s unpaid work 1.25 1.71+ Explained variance (pseudo R2) 15% 3% 17% N 303 303 303 • Daniela Grunow, Nina Baur548 employment prestige, the more likely men are to participate in cooking. All of these factors indicate that particularly men from higher social classes cook. These result fi t the modern trend that cooking – in particular the preparation of elaborate meals – has increasingly gained social appreciation, partly mediated by cooking shows on television. Household income, however, has a negative effect. Relationship and family context also infl uence male involvement in cooking. Sin- gle men are more than twice as likely to cook in comparison to men whose partner is not employed. Men with a partner in full-time employment cook three times as often. This result indicates that cooking is not always a leisure activity for men, but often a necessity. Similarly to vacuum-cleaning and doing the laundry, men who do not live on their own make use of household structures that enable them to avoid Tab. 9: Explanatory variables for men cooking, logistic regression Level of signifi cance: + α ≤ 0.10; * α ≤ 0.05; ** α ≤ 0.01; *** α ≤ 0.001 Source: Own calculations based on the study “The image of men in society” Model 1 Model 2 Model 3 Region, ref. West Germany East Germany -0.52 -0.79* Age (in years) -0.01 0.00 Education, ref. maximum “Lower School Leaving Certificate” (Hauptschule) “General School Leaving Certificate” (Realschule) 0.76* 0.60 Higher education entrance qualification 1.00* 0.87* Graduate degree 0.44 0.31 Employment status, ref. employed Pensioner 0.18 0.26 In education 0.33 0.51 Other 0.49 0.30 Treiman prestige 0.02* 0.02+ Net equivalent income (in thousand euro) -0.66** -0.63* Partner’s employment status, ref. not employed No partner 0.74* 0.91* Partner employed part-time 0.37 0.39 Partner employed full-time 1.31** 1.35** Family context Number of adults in the household (apart from the respondent) -0.73** -0.72** Number of children in the household -0.08 -0.08 Normative attitudes Approval of the male provider - female homemaker model 0.52 1.43 Approval of women’s employment 1.38* 1.7* Employment as men's central social role -1.62 -1.78 Endorsement of men’s unpaid work 2.37** 2.66** Explained variance (pseudo R2) 12% 6% 16% N 303 303 303 The Association between Norms and Actions • 549 cooking: the number of adults in the household has a negative impact on the prob- ability of men cooking. Model 2 shows the infl uence of normative attitudes on the participation of men in cooking. The endorsement of men’s unpaid work has the strongest effect. The more men approve this norm system, the greater the probability that they cook. The approval of women’s employment also has a signifi cant positive effect. The full model (model 3) indicates that men from East Germany cook less fre- quently than men from West Germany when keeping norms, socio-demographic and economic factors constant. The other effects are largely unchanged compared to the previous models. The full model only explains 16 percent of the variance in cooking. Norms account for 6 percent of the variance (model 2), and the socio- demographic, economic, and family-related variables in model 1 explain 12 percent of the variance. 7 Conclusion and discussion of fi ndings What can be learned from these fi ndings about the apparent discrepancy between egalitarian gender norms on the one hand and behaviours lagging far behind these norms on the other? Our discussion follows the order of the outlined theoretical as- sumptions on the relationship between norms and behaviour in the household and family context. Building on the basic assumption of role theory, our fi rst expectation was that people adapt their behaviour to existing normative gender roles and that there is a relationship between normative attitudes and the participation of men in housework. We argue – and fi nd empirical evidence suggesting – that an issue of previous research is the inaccurate theoretical defi nition and empirical measure- ment of gender norms. Previous research has focussed especially on attitudes to- ward women’s employment (cf. sub-section 5.2). Our empirical fi ndings suggest in- stead that this normative system is only relevant for explaining male participation in cooking. Normative approval of the male provider – female homemaker model and attitudes toward employment as the central social role of men also appear only mar- ginally relevant for predicting whether men participate in household tasks such as vacuum-cleaning, doing the laundry, and cooking (see also Schober 2011). Attitudes toward men’s unpaid work play a much more decisive role for male participation in housework. The empirical results not only hint at a relationship between norms and behaviour in regard to men’s unpaid work, but also suggest that it is important to examine specifi c norms and tasks. The effect of a norm system can vary for dif- ferent tasks, and, depending on the task, various norm systems may be relevant predictors. Furthermore, based on more precise theoretical assumptions and data allowing a separate analysis of male and female roles regarding employment and housework, our fi ndings suggest that historical changes regarding women’s em- ployment do not automatically lead to changes in norms regarding men’s and wom- en’s participation in unpaid work. The asymmetry of gender role changes (Grunow et al. 2007) therefore does not only apply to behaviours, but also to attitudes. • Daniela Grunow, Nina Baur550 Based on recent theoretical contributions arguing that gender constructions re- fer to (normative and institutional) context, our second expectation was that men generally hold less egalitarian beliefs than women. Due to the asynchronous chang- es of male and female roles in Germany during the 20th and the early 21st century, our third expectation was to fi nd gender-specifi c differences regarding the endorse- ment of or objection to those norm systems that were mostly affected by structural changes. Our fi ndings indicate that there is still variation in the acceptance of, or ob- jection to, women’s employment, both among men and women, and that employ- ment is still seen as the central social role of men by a vast majority. Our fi ndings further point to men having more traditional attitudes toward women’s employment than women themselves. At the same time, women are more conservative regard- ing male employment: eight out of ten women, but only six out of ten men view employment as the central social role of men. A possible interpretation of these results is that normative stereotypes – such as those expressed here – are applied more readily to the opposite gender. Perhaps life contexts viewed from the outside appear less complex and ambivalent in comparison to one’s own life context. The fi ndings are also in line with the hypothesis of interest-based factors infl uencing in- dividual normative beliefs (Bolzendahl/Myers 2004). Furthermore, the evidence we have found suggests that individual attitudes toward men’s and women’s employ- ment are not necessarily related to each other, i.e. someone holding a progressive view on the role of women does not automatically hold a progressive view on the role of men. Our fourth expectation focused on the impact of household and family context on normative framing and on individual behaviour regarding participation in house- work. We assumed that both household structures offering men opportunities to withdraw from housework and the individual gender role beliefs infl uence whether men participate in housework or not. Our fi ndings support this hypothesis. We fi nd that single men vacuum, do laundry and cook signifi cantly more often than men with a partner. Men’s own normative endorsement of men’s unpaid work plays a decisive role in predicting whether they perform either of these tasks. The evidence suggesting that men withdraw from housework with an increasing number of adults in the household further emphasises the importance of opportunity structures. The more adults live in the household, the less likely men are to vacuum, do laundry, or cook. Our descriptive analyses further indicate that women take on more house- work when moving in with their partner, particularly because they replace other people (often mothers) who carried out these tasks before. Thus, the family context not only infl uences whether men do housework, it also infl uences who takes on certain tasks when the man does not do them himself. Our fi fth expectation was that the number of children in the household has a positive effect on the participation of men in vacuum-cleaning, doing the laundry, and cooking, since more commitment is demanded of fathers with young children in the private domain these days and research has shown that couples with com- mitted fathers have more children. Our multivariate analyses, however, indicate the opposite, yet the effect is only statistically signifi cant for doing the laundry: the more children living in the household, the less inclined fathers are to participate in The Association between Norms and Actions • 551 doing the laundry. This result complements earlier time-use studies (Künzler et al. 2001) and research on the relative participation of men in household tasks (Grunow et al. 2007) in one important aspect: parenthood not only leads to a more traditional gender-specifi c division of work because women take on the greater share of ad- ditional housework, some tasks – such as doing the laundry – are not performed by certain fathers at all. Since institutional and normative contexts merely frame individual behaviour, we moreover assumed that individual normative attitudes of men infl uence whether they do housework even after controlling for family context. This hypothesis is confi rmed in our analyses. Finally, our fi ndings on the gender-typing of specifi c household tasks are in line with the current state of research on the commitment of “new men” to housework. As even men with an egalitarian attitude tend to perform “male” rather than “fe- male” tasks in the household, we assumed that normative attitudes of men espe- cially explain male participation in “legitimately male” tasks (vacuum-cleaning and cooking). In the case of doing the laundry – still the most feminine task – we as- sumed a bigger impact of situational factors such as household and family context. This hypothesis is confi rmed in our multivariate analyses: in the case of vacuum- cleaning, the norms-based models explain as much of the variance as the models including socio-demographic, economic and family context variables. In the case of cooking, norms explain one third of the variance explained by the full model. However, in the case of doing the laundry, norms make up only one fi fth of the total explanatory power of the full model. Our fi ndings have both theoretical and methodological implications: fi rst, our norm systems help explain some of the variation in all three dependent variables in addition to socio-demographic, economic and family context variables. This implies that regression models not controlling for norms are incomplete, and results may be biased by spurious effects or suppressor effects. Second, our fi ndings suggest that future studies on the division of housework should not – in contrast to common practice so far – sum up domestic work hours using indices or time budgets. Such measures are likely to lead to ambiguity regard- ing both the contribution of men toward housework and the gender-specifi c con- notation of particular household tasks. Third, we want to point out the limitations of our fi ndings. It has been argued re- peatedly (Cooke 2004; Grunow et al. 2007) that it is necessary to analyse processes of gender-specifi c divisions of work with longitudinal data. Cause and effect of com- peting theoretical concepts can only be separated from mere composition effects or life course effects with panel data collecting norms that are differentiated over time and on an individual level (cf. Grunow 2007). Our multivariate cross-sectional analyses presume that the operationalised norm systems infl uence behaviour and not vice versa. However, this assumption cannot be confi rmed by our data. Thus, our regression models mainly have explorative value. Future research needs to col- lect and analyse differentiated longitudinal data on gender norms to verify these results. However, we interpret the relatively stable effects of norm systems after controlling for household and family context as an indication of our results not be- ing signifi cantly affected by the mixture of norms and family context as explanatory • Daniela Grunow, Nina Baur552 variables for the participation of men in housework, despite using cross-sectional data. In theory, it would have been appropriate to look at the impact of interaction ef- fects between family context and normative attitudes of men on the investigated household tasks. Although we conduct these analyses and report some results, the full potential of such analyses could not be exploited due to the small sample size. We fi nd some indication that household and family context re-emphasise the effect of certain attitudes. It would be interesting for future research to understand how normative attitudes and gender-specifi c behaviour infl uence each other and change over the course of a relationship. Acknowledgement We would like to thank Sandra Buchler for further editing of the translated English manuscript. Daniela Grunow would like to thank the European Research Council for fi nancial support in the context of the ERC Starting Grant Project APPARENT (Grant agreement no.: 263651). References Baur, Nina 2007: Der perfekte Vater. Vaterschaftsvorstellungen deutscher Männer. In: Freiburger Geschlechterstudien 21: 79-114. Baur, Nina; Luedtke, Jens 2008: Männlichkeit und Erwerbsarbeit bei westdeutschen Männern. In: Baur, Nina; Luedtke, Jens (Eds.): Die soziale Konstruktion von Männ- lichkeit. Hegemoniale und marginalisierte Männlichkeiten in Deutschland. Opladen/ Farmington Hills: Barbara Budrich: 76-99. Baur, Nina; Akremi, Leila 2011: Lebensstile und Geschlecht. In: Rössel, Jörg; Otte, Gunnar (Eds.): Lebensstilforschung. Sonderheft der Kölner Zeitschrift für Soziologie und Sozialpsychologie 51. Wiesbaden: VS-Verlag: 269-294 . Becker-Schmidt, Regina 2005: Von soziologischen Geschlechtsrollentheorien zur ge- sellschaftstheoretischen Erforschung des Geschlechterverhältnisses. In: Vogel, Ulrike (Ed.): Was ist weiblich, was ist männlich? Aktuelles zur Geschlechterforschung. Biele- feld: Kleine: 89-112. Berk, Sarah F. 1985: The Gender Factory. The Apportionment of Work in American Households. New York: Plenum Press. Bielby, William T.; Bielby, Denise D. 1989: Family Ties. Balancing Commitments to Work and Family in Dual-Earner Households. In: American Sociological Review 54: 776-789. Blohm, Michael 2002: Einstellungen zur Rolle der Frau. In: Statistisches Bundesamt (Ed.): Datenreport 2002. 2nd edition Bonn/Wiesbaden: Bundeszentrale für politische Bildung/Statistisches Bundesamt: 533-540. Bolzendahl, Catherine I.; Myers, Daniel J. 2004: Feminist Attitudes and Support for Gen- der Equality. Opinion Change in Women and Men, 1974-1998. In: Social Forces 83: 759-789 [doi: 10.1353/sof.2005.0005] . Bothfeld, Silke 2005: Vom Erziehungsurlaub zur Elternzeit. Frankfurt a.M./New York: Campus. The Association between Norms and Actions • 553 Brannen, Julia 2007: Childcare. In: Ritzer, George (Ed.): The Blackwell Encyclopedia of Sociology. Oxford: Blackwell: 457-459. Brines, Julie 1994: Economic Dependency, Gender, and the Division of Labor at Home. In: American Journal of Sociology 100,3: 652-688. Burkart, Günter 1997: Lebensphasen und Liebesphasen. Vom Paar zur Ehe zum Single und zurück? Opladen: Leske+Budrich. Cooke, Lynn P. 2004: The Gendered Division of Labor and Family Outcomes in Ger- many. In: Journal of Marriage and Family 66: 1246-1259 [doi: 10.1111/j.0022- 2445.2004.00090.x]. Cyprian, Gudrun 2005: Die weißen Flecken in der Diskussion zur „neuen Vaterrolle“. In: Zeitschrift für Familienforschung 1: 76-79. Davis, Shannon N. 2007: Gender Ideology Construction from Adolescence to Young Adulthood. In: Social Science Research 36: 1021-1241 [doi: 10.1016/j.ssre- search.2006.08.001]. Davis, Shannon N.; Greenstein, Theodore N. 2009: Gender Ideology. Components, Pre- dictors, and Consequences. In: Annual Review of Sociology 35: 87-105 [doi: 10.1146/ annurev-soc-070308-115920]. Dermott, Esther 2007: Fatherhood. In: Ritzer, George (Ed.): The Blackwell Encyclopedia of Sociology. Oxford: Blackwell: 1647-1650. Döge, Peter 2006: Männer – Paschas und Nestfl üchter? Zeitverwendung von Männern in der Bundesrepublik Deutschland. Opladen: Barbara Budrich. Döge, Peter; Volz, Rainer 2004: Männer – weder Paschas noch Nestfl üchter. Aspekte der Zeitverwendung von Männern nach den Daten der Zeitbudgetstudie 2001/2002 des Statistischen Bundesamtes. In: APuZ B46/2004: 13-23. Duvander, Ann-Zofi e; Andersson, Gunnar 2006: Gender Equality and Fertility in Sweden. A Study on the Impact of the Father‘s Uptake of Parental Leave on Continued Child- bearing. In: Marriage and Family Review 39: 121-142 [doi: 10.1300/J002v39n01_07]. Erler, Gisela; Jaeckel, Monika; Pettinger, Rudolf; Sass, Jürgen 1988: Kind? Beruf? Oder Beides? Brigitte Untersuchung 88. Hamburg/München: Brigitte/DJI. Evertsson, Marie 2014: Gender Ideology and the Sharing of Housework and Child Care in Sweden. In: Journal of Family Issues 35: 927-949 [doi:10.1177/0192513X14522239]. Festinger, Leon 1957: A Theory of Cognitive Dissonance. Stanford: Stanford University Press. Fromm, Sabine 2004: Faktorenanalyse. In: Baur, Nina; Fromm, Sabine (Eds.): Datenana- lyse mit SPSS für Fortgeschrittene. Wiesbaden: VS-Verlag: 226-256. Fthenakis, Wassilios E. et. al. 1999: Engagierte Vaterschaft. Opladen: Leske+Budrich. Fthenakis, Wassilios E.; Kalicki, Bernhard; Peitz, Gabriele 2002a: Paare werden Eltern. Ergebnisse der LBS-Familien-Studie. Opladen: Leske+Budrich. Fthenakis, Wassilios E.; Kalicki, Bernhard; Peitz, Gabriele 2002b: Übergang zur Eltern- schaft. Questionnaire. Fuwa, Makiko 2004: Macro-Level Gender Inequality and the Division of House- hold Labor in 22 countries. In: American Sociological Review 69: 751-767 [doi: 10.1177/000312240406900601]. Grunow, Daniela 2007: Wandel der Geschlechterrollen und Väterhandeln im Alltag. In: Mühling, Tanja; Rost, Harald (Eds.): Väter im Blickpunkt. Leverkusen: Barbara Budrich: 49-76. • Daniela Grunow, Nina Baur554 Grunow, Daniela 2010: Sex roles in the Family. Perspectives from the Field of Women’s Studies. In: Kapella, Olaf et al. (Ed.): Family Diversity. Collection of the 3rd European Congress of Family Science. Opladen/Farmington Hills: Barbara Budrich: 127-142. Grunow, Daniela 2013: Aufteilung von Erwerbs-, Haus- und Familienarbeit in Partner- schaften im Beziehungsverlauf. Der Einfl uss von Sozialpolitik in Europa. In: Lück, Detlev; Cornelißen, Waltraud (Eds.): Geschlechterunterschiede und Geschlechterun- terscheidungen in Europa. Stuttgart: Lucius & Lucius: 237-263. Grunow, Daniela; Schulz, Florian; Blossfeld, Hans-Peter 2007: Was erklärt die Tradi- tionalisierungsprozesse häuslicher Arbeitsteilung im Eheverlauf: soziale Normen oder ökonomische Ressourcen? In: Zeitschrift für Soziologie 36,3: 162-181. Grunow, Daniela; Schulz, Florian; Blossfeld, Hans-Peter 2012: What Determines Change in the Division of Housework over the Course of Marriage? In: International Sociology 27,3: 289-307 [doi: 10.1177/0268580911423056]. Gupta, Sanjiv et al. 2010: Economic Inequality and Housework. In: Drobnič, Sonja; Treas, Judith (Eds.): Dividing the Domestic. Stanford: Stanford University Press: 105- 122. Hewener, Vera 2004: Geschlechtsspezifi sche Unterschiede im Umgang mit Zeit. In: APuZ B31-32/2004: 26-32. Hofäcker, Dirk; Lück, Detlev 2004: Angleichung nationaler Einstellungsmuster in Rich- tung eines liberaleren Rollenmodells? Einstellungen von Frauen zur geschlechtsspezi- fi schen Arbeitsteilung im internationalen Vergleich. In: Informationssystem Soziale Indikatoren (ISI) 32: 12-15. Hofmeister, Heather; Baur, Nina; Röhler, Alexander 2009: Versorgen oder Fürsorgen? Vorstellungen der Deutschen von den Aufgaben eines guten Vaters. In: Villa, Paula- Irene; Thiessen, Barbara (Eds.): Mütter/Väter. Diskurse, Medien, Praxen. Münster: Westfälisches Dampfboot: 194-212. Jakobsen, Georg 2009: Public versus Private. The Conditional Effect of State Policy and Institutional Trust on Mass Opinion. In: European Sociological Review 26,3: 307-318 [doi: 10.1093/esr/jcp023]. Karlson, Kristian Bernt; Holm, Anders; Breen, Richard 2010: Comparing Regression Co- effi cients Between Models using Logit and Probit: A New Method. CSER Working Pa- per. Aarhus University: Centre for Strategic Educational Research. Kaufman, Gayle 2000: Do Gender Role Attitudes Matter? Family Formation and Dissolu- tion Among Traditional and Egalitarian Men and Women. In: Journal of Family Issues 21: 128-144 [doi: 10.1177/019251300021001006]. Keddi, Barbara 2003: Projekt Liebe. Opladen: Leske+Budrich. Koppetsch, Cornelia; Maier, Maja S. 2001: Männlichkeit im Milieuvergleich. In: Döge, Peter; Meuser, Michael (Eds.): Männlichkeit und soziale Ordnung. Opladen: Leske+Budrich: 27-48. Körner, Gabriele 2006: Berufswahl, Lebensentwurf und Geschlecht. In: Zeitschrift für Frauenforschung und Geschlechterstudien 24,4: 141-157. Krüger, Helga 2001: Ungleichheit und Lebenslauf. Wege aus den Sackgassen em- pirischer Traditionen. In: Heintz, Bettina (Ed.): Geschlechtersoziologie. Sonderheft 41 der Kölner Zeitschrift für Soziologie und Sozialpsychologie. Opladen/Wiesbaden: Westdeutscher Verlag: 512-537. Kühhirt, Michael 2011: Childbirth and the Long-Term Division of Labour within Couples: How do Substitution, Bargaining Power, and Norms affect Parents’ Time Allocation in West Germany? In: European Sociological Review [doi: 10.1093/esr/jcr026]. The Association between Norms and Actions • 555 Künzler, Jan; Walter, Wolfgang; Reichart, Elisabeth; Pfi ster, Gerd 2001: Gender Division of Labour in Unifi ed Germany. Tilburg: University Press. Leitner, Sigrid; Ostner, Ilona; Schratzenstaller, Margit 2004: Einleitung. In: Leitner, Si- grid; Ostner, Ilona; Schratzenstaller, Margit (Eds.): Wohlfahrtsstaat und Geschlech- terverhältnis im Umbruch. Was kommt nach dem Ernährermodell? Wiesbaden: VS- Verlag: 9-27. Lincke, Hans-Joachim 2007: Doing Time. Die zeitliche Ästhetik von Essen, Trinken und Lebensstilen. Bielefeld: transcript. Long, J. Scott; Freese, Jeremy 2006: Regression Models for Categorical Dependent Variables Using Stata. 2nd Edition. Verlag: Stata Press. Lorber, Judith 2000: Using Gender to Undo Gender. A Feminist Degendering Movement. In: Feminist Theory 11: 79-95 [doi: 10.1177/14647000022229074]. Lorber, Judith 2004: Man muss bei Gender ansetzen, um Gender zu demontieren. Feministische Theorie und Degendering. In: Zeitschrift für Frauenforschung und Ge- schlechterstudien 222,3: 9-24. Lück, Detlev 2004: Cultural Persistence of Traditional Gender Attitudes Over Time. Cross-national Comparison. In: National Council on Family Relations (Ed.): Vision 2004. What is the Future of Marriage? Minneapolis: NCFR: 24-32. Lück, Detlev 2006: Cross-national Comparison of Gender Role Attitudes and their Im- pact on Women’s Life Courses. In: Blossfeld, Hans-Peter; Hofmeister, Heather (Eds.): Globalization, Uncertainty and Women’s Careers. An International Comparison. Chel- tenham, UK/Northampton: Edward Elgar: 405-432. Lück, Detlev 2009: Der zögernde Abschied vom Patriarchat. Der Wandel von Geschlech- terrollen im internationalen Vergleich. Berlin: edition sigma. Lück, Detlev; Hofäcker, Dirk 2008: The Values of Work and Care among Women in Mod- ern Societies. In: van Oorschot, Wim; Opielka, Michael; Pfau-Effi nger, Birgit (Eds.): Culture and Welfare State. Values of Social Policy from a Comparative Perspective. Cheltenham, UK/Northampton, USA: Edward Elgar: 289-313. Lundgreen, Peter; Scheunemann, Jana 2006: Geschlechtsspezifi sche Berufsbildung und Arbeitsmarktchancen 1950-2000. In: Zeitschrift für Erziehungswissenschaft. Bei- heft 7-06: 127-149. Matzner, Michael 2004: Vaterschaft aus der Sicht von Vätern. Wiesbaden: VS-Verlag. Meulemann, Heiner; Beckers, Tilo 2004: Das sichtbare und das verborgene soziale Engagement. Häufi gkeiten und Hintergründe von Ehrenamt und privater Hilfe in Deutschland im Jahre 2002. In: Soziale Welt 55: 51-74. Mood, Carina 2010: Logistic Regression: Why we cannot do what we think we can do, and what we can do about it. European Sociological Review. First published online: March 9, 2009: 67-82 [doi: 10.1093/esr/jcp006]. Oberndorfer, Rotraut; Rost, Harald 2005: Neue Väter – Anspruch und Realität. In: Zeitschrift für Familienforschung 1: 50-65. Ostner, Ilona 1995: Arm ohne Ehemann? In: APuZ. B36-37/1995: 3-12. Ostner, Ilona 2005: Einführung: Wandel der Geschlechtsrollen. In: Zeitschrift für Famil- ienforschung. Sonderheft 4: 46-49. Otte, Gunnar; Baur, Nina 2008: Urbanism as a Way of Life? Räumliche Variationen der Lebensführung in Deutschland. In: Zeitschrift für Soziologie 37,2: 93-116. • Daniela Grunow, Nina Baur556 Pfau-Effi nger, Birgit 1999: Welfare Regimes and the Gender Division of Labour. In: Chris- tiansen, Jens; Koistinen, Pertti; Kovalainen, Anne (Eds.): Working Europe. Aldershot et al.: Ashgate: 69-96. Pinl, Claudia 2004: Wo bleibt die Zeit? Die Zeitbudgeterhebung 2001/02 des Statisti- schen Bundesamts. In: APuZ B31-32/2004: 19-25. Poortman, Anne-Rigt; Van der Lippe, Tanja 2009: Attitudes Toward Housework and Child Care and the Gendered Division of Labor. In: Journal of Marriage and Family 71,3: 526-541 [doi: doi:10.1111/j.1741-3737.2009.00617.x]. Schober, Pia S. 2011: The Parenthood Effect on Gender Inequality. Explaining the Change in Paid and Domestic Work When British Couples Become Parents. In: European So- ciological Review 29,1: 74-85 [doi: 10.1093/esr/jcr041]. Schulz, Florian; Grunow, Daniela 2011: Comparing Diary and Survey Estimates on Time Use. In: European Sociological Review 28,5: 622-632 [doi:10.1093/esr/jcr030]. Stier, Haya; Lewin-Epstein, Noah 2007: Policy Effects on the Division of Housework. In: Journal of Comparative Policy Analysis 9: 235-259 [doi:10.1080/13876980701494657]. Surkyn, Johan; Lesthaeghe, Ron 2004: Value Orientations and the Second Demograph- ic Transition (SDT) in Northern, Western and Southern Europe. In: Demographic Re- search. Special Collection 04/2004: 45-86 [doi: 10.4054/DemRes.2004.S3.3]. Tölke, Angelika; Hank, Karsten 2005: Das „vernachlässigte“ Geschlecht in der Familien- forschung: Untersuchungen zu Partnerschaft und Elternschaft bei Männern. In: Zeit- schrift für Familienforschung. Sonderheft 4: 7-17. Vaskovics, Lazlo R.; Rost, Harald 1999: Väter und Erziehungsurlaub. Stuttgart: Kohlham- mer. Veil, Mechthild 2003: Kinderbetreuungskulturen in Europa. Schweden, Frankreich, Deutschland. In: APuZ B44/2003: 12-22. West, Candace; Zimmermann, Don H. 1987: Doing Gender. In: Gender and Society 12: 125-151. Zimmermann, Peter 2006: Grundwissen Sozialisation. Einführung zur Sozialisation im Kindes- und Jugendalter. 3rd edition. Wiesbaden: VS Verlag. Zulehner, Paul M. 2004: Neue Männlichkeit – neue Wege der Selbstverwirklichung. In: APuZ B46/2004: 5-12. Zulehner, Paul M. (o.J.): Männer 2002. Questionnaire. ZUMA; Infas 2002: Internationale Sozialwissenschaftliche Umfrage: ISSP 2002 – Family and Changing Gender Roles III. Fragebogen für Deutschland [URL: www.gesis.org, 1.1.2006]. ZUMA; Infratest Burke 1994: Internationale Sozialwissenschaftliche Umfrage: ISSP 1994 – Family and Changing Gender Roles II. Fragebogen für Deutschland [URL: www. gesis.org, 1.1.2006]. ZUMA; Infratest Burke 1997: Thema „Arbeit“: Meinungen in Deutschland. ISSP 1997 – Work Orientations. Fragebogen für Deutschland [URL: www.gesis.org, 1.1.2006]. The Association between Norms and Actions • 557 Translated from the original text by the Federal Institute for Population Research, for information only. The reviewed and authors’ authorised original article in German is available under the title “Die Korrespondenz von normativen Vorstellungen und Handeln. Das Beispiel männlicher Hausar- beit”, DOI 10.12765/CPoS-2014-10de or URN urn:nbn:de:bib-cpos-2014-10de1, at http://www.com- parativepopulationstudies.de. Date of submission: 25.02.2011 Date of acceptance: 02.02.2012 Prof. Dr. Daniela Grunow (). Goethe-University Frankfurt am Main, Faculty of Social Science, Sociology Institute, Frankfurt am Main, Germany. E-Mail: grunow@soz.uni-frankfurt.de URL: http://www.uni-frankfurt.de/47217051/home? Prof. Dr. Nina Baur. Technische Universität Berlin, Department of Sociology, Berlin, Germany. E-Mail: nina.baur@ tu-berlin.de URL: http://www.mes.tu-berlin.de/Baur • Daniela Grunow, Nina Baur558 Appendix Tab. A1: Relative explanatory power of independent variables for men vacuum- cleaning, doing the laundry and cooking, logistic regression, fully standardised b-coeffi cients Level of signifi cance: + α ≤ 0.10; * α ≤ 0.05; ** α ≤ 0.01; *** α ≤ 0.001 Source: Own calculations based on the study “The image of men in society” Man vacuums Man does laundry Man cooks Region, ref. West Germany East Germany 0.05 0.09 -0.23+ Age (in years) -0.29** -0.17 -0.04 Education, ref. maximum “Lower School Leaving Certificate” (Hauptschule) “General School Leaving Certificate” (Realschule) -0.08 0.06 0.18 Higher education entrance qualification -0.03 0.03 0.25* Graduate degree -0.03 -0.09 0.09 Employment status, ref. employed Pensioner -0.14 -0.07 0.07 In education -0.15 0.17 0.11 Other -0.04 0.09 0.07 Treiman prestige 0.20 0.20 0.23+ Net equivalent income (in thousand euro) -0.35* -0.04 -0.28* Partner’s employment status, ref. not employed No partner 0.26* 0.28* 0.25* Partner employed part-time 0.05 0.12 0.11 Partner employed full-time 0.10 0.11 0.37** Family context Number of adults in the household (apart from the respondent) -0.31** -0.47** -0.44** Number of children in the household -0.12 -0.29* -0.04 Normative attitudes Approval of the male provider - female homemaker model 0.03 -0.06 0.17 Approval of women’s employment -0.08 0.11 0.25* Employment as men's central social role -0.14 0.07 -0.20 Endorsement of men’s unpaid work 0.52** 0.20+ 0.31** Explained variance (pseudo R2) 15% 17% 16% N 303 303 303 Published by / Herausgegeben von Prof. Dr. Norbert F. Schneider Federal Institute for Population Research D-65180 Wiesbaden / Germany Managing Editor / Verantwortlicher Redakteur Frank Swiaczny Assistant Managing Editor / Stellvertretende Redakteurin Katrin Schiefer Language & Copy Editor (English) / Lektorat & Übersetzungen (englisch) Amelie Franke Copy Editor (German) / Lektorat (deutsch) Dr. Evelyn Grünheid Layout / Satz Beatriz Feiler-Fuchs E-mail: cpos@bib.bund.de Comparative Population Studies www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet) Scientifi c Advisory Board / Wissenschaftlicher Beirat Paul Gans (Mannheim) Johannes Huinink (Bremen) Michaela Kreyenfeld (Rostock) Marc Luy (Wien) Clara H. Mulder (Groningen) Notburga Ott (Bochum) Peter Preisendörfer (Mainz) Zsolt Spéder (Budapest) Board of Reviewers / Gutachterbeirat Martin Abraham (Erlangen) Laura Bernardi (Lausanne) Hansjörg Bucher (Bonn) Claudia Diehl (Konstanz) Andreas Diekmann (Zürich) Gabriele Doblhammer-Reiter (Rostock) E.-Jürgen Flöthmann (Bielefeld) Alexia Fürnkranz-Prskawetz (Wien) Beat Fux (Salzburg) Joshua Goldstein (Berkeley) Karsten Hank (Köln) Sonja Haug (Regensburg) Aart C. Liefbroer (Den Haag) Kurt Lüscher (Konstanz) Dimiter Philipov (Wien) Tomáš Sobotka (Wien) Heike Trappe (Rostock) © Federal Institute for Population Research 2014 – All rights reserved