Electronic Journal of Sociology (2002)

ISSN: 1198 3655

Modern family or modernized family traditionalism?:
Master status and the gender order in Switzerland

René Levy
Université de Lausanne

Eric Widmer
Universités de Genève et de Lausanne

Jean Kellerhals
Université de Genève


Are present-day families still like those described by sociology in the middle of the 20th century or have they changed beyond resemblance? The media and much of the popular perceptions along with them have long ago opted for an « all-change » perspective on contemporary patterns of living together; the sociological literature is divided between diagnostics of change and diagnostics of resilience. A recent study of intra-couple dynamics in Switzerland shows the existence of a large array of couples’ internal structures of regulation, but with a persistent core of traditionalism. The concept of master status is used to interpret the fact that major changes point less towards clear-cut egalitarianism, which could be an alternative to traditional gender structures, than towards a traditionally biased « synthesis » that could be paradoxically called modernized family traditionalism.

1. « The Parsonian Family is Dead » - is It?

Modern family sociology mainly developed in the middle of the 20th century, in a period of stability (for countries like the USA or even Switzerland) or of social reconstruction after the Second world war (as for most other European countries). This development took place under the aegis of the then dominating functionalist paradigm (Parsons & Bales 1955, Goode 1963). Its basic model was that of the nuclear family, a couple of two adult partners living together with their children and forming an irreducible group securing fundamental tasks for social and hence societal integration, especially through socialization, a group that functioned in a relatively autonomous way, with little intimate contacts beyond its borders, which made it particularly attuned to the flexibility required by the industrial society. The internal structure of this family model was mainly organized around two ascribed criteria, sex and age. The role attribution according to the sex of the adult partners - internal tasks for the wife, external tasks for the husband - was said to correspond to expressive vs. instrumental orientations typical of sexual identities and was interpreted to be a highly functional way of performing all the necessary contributions to family and societal functioning.

This endemic view of what many family ideologists liked to call the « basic cell of society » was increasingly faced by two main challenges. The one was the early feminist critique of the functionalist model’s veiling and justifying of women’s exploitation by men, and pseudoscientifically glorifying discriminative practices by presenting them as functional necessities for the whole society as well as optimal solutions for the intrinsic identities of the individuals (Friedan 1963). The other challenge was the increasing evidence of variations and changes that became important enough to express themselves in demographical time series in all highly industrialized countries. It is true that this evidence did not really attest to the intrinsic stability of the family model, supposed to be dominant, and to its particularly good adjustment to the needs of the industrial society (Goode 1963). Since the 60ies, divorce rates rose, marriage rates plunged, fertility rates fell as well, often far below the level of demographic reproduction, and atypical family situations appeared increasingly or were increasingly identified as such (single parent, especially single mother families, children born out of wedlock and often to teenager mothers, non-marital cohabitation, homosexual couples, recomposed families, couples living apart together, not to speak of more basically different domestic units such as communes; Roussel 1989, Sardon 2000, specifically for Switzerland Calot et al. 1998).2 A complementary long-term change is the increasing participation in the labor-force by women after the Second World War (during the first half of the 20th century, the rate of female employment decreased). Demographic data are not the only indicators of family-relevant change, norms and values are also concerned. To mention but one example : the values mobilized in order to find partners show a pattern of change similar to the general change of values identified by Inglehart in various publications (Inglehart 1971, 1977, 1990), as Buchmann and Eisner (1997) have nicely shown in a content analysis of contact ads in two relatively class-specific (one lower, one middle class) Swiss-German newspapers over almost the entire 20th century (1900 through 1992).

Following these signs of recent temporal variation, discourses about family change have developed vigorously both in and outside sociology, especially since the 80ies (Cherlin 1981, 1992; Popenoe 1993; Beck 1986; Beck-Gernsheim 1998, 2000, and many others). A typical summary statement is, e.g., Cherlin’s remark that the title of his re-edited book on « Marriage, Divorce, Remarriage » should rather read « Cohabitation, Marriage, Divorce, More Cohabitation, and Probably Remarriage » (Cherlin’s introduction, 1992). Teachman et al. (2000) write squarely that « nuclear families consisting of two heterosexual parents, biologically related children, fathers who are in the paid labor force, and mothers who are homemakers have declined to the point of being a minority circumstance in our society ». While it seems at least plausible that some of the changes have been more pronounced in the USA, most of them appear also in European societies. However, European sociologists have probably reacted more vigorously to this discourse of « Family Diversity as the Norm » (subtitle of Peterson & Steinmetz 1999) than their American colleagues, possibly because traditional family constellations show more resilience in Europe (Bertram 1991; de Singly 1991; Hettlage 1992; Vaskovics 1994; a more change-convinced stance for European countries can be found in Crompton 1999).

Without questioning the reality and relevance of demographic indications of change in family-related behavior, we have to admit that they can not directly prove even the major part of the far-reaching conclusions they often motivate. Although aggregated time series say certainly something about the cumulated stability or variation of individual arrangements in general, they say very little about individual trajectories, forms of cohabitation, and their significance. To mention only one specific example: in several European countries, the alleged « new » family form of non-married cohabitation can be shown to constitute a new and by now widely practiced early phase in individual life courses, preceding more classical phases of family formation, rather than an lasting alternative to these (for Switzerland, see Levy et al. 1997; Kellerhals et al. 2003). A more substantial argument must be added: the personal composition of a social group, including the family, may perfectly change without the group’s structure changing. Changes in a group’s composition may strongly affect personal relationships, especially in such an emotionally loaded primary group as a couple or a family, but recomposition may very well take place under the same structural arrangement, if by this term we mean role definitions, power relationships, territorial and systemic boundaries, and the like. A direct consequence of this consideration is that rising divorce rates say nothing about structural change in couples - they might even say more about the absence of such change while personal norms and aspirations evolve.3

So what is really happening to couples and families? More specifically: what changes can be demonstrated with respect to their sexual order? And even more specifically, thinking of the traditional model of family organization: what has changed with respect to the sexual division of work for the family?

The descriptive exploration called for by these questions implies a series of theoretical queries: how do couples organize, and as a function of what? There are series of specific theories frequently used to explain the division of what Coltrane (2000) calls « routine family work », or in an even larger perspective the division of all work for the family, paid and unpaid. At least five theoretical approaches can be distinguished in the actual research literature: human capital or new home economics (starting with Becker’s 1981 classic), postulating that men and women allocate their time to different kinds of tasks or activities according to the principle of optimizing their returns; time constraints or time availability (sometimes also called demand/response capability or situational approach, Coverman 1985, England & Farkas 1986) postulating – put somewhat bluntly - that the volume of time needed for the family depends on the number and the age of the children present in the household while the number of hours effectively spent in this area depends on what employed work leaves available; gender-role attitudes with socialization and sex-role ideology as a background,4 postulating that gender differences in task allocation result from the socialized gender norms of both partners and/or from their responding to known role expectations in their wider social environment; gender performance or doing gender (West & Zimmerman 1987), postulating that gender differences, including family tasks, are actively performed constructions of identity differentiation, including the possibility that the very scriptedness of identities varies between the sexes (as, e.g., Chodorow 1978 postulates for mothering and fathering); and the very classical resource theory (Blood & Wolfe 1960 and many later authors), postulating that not only power differentials between the partners, but also task distribution between them are conditioned by their relative resource contribution (implying that family work is less satisfying than employed work and is therefore left to the partner with less resources and hence less power for internal negotiations).

There is neither room nor need here to discuss the merits, drawbacks and empirical relevance of these approaches. Various excellent overviews and discussions already exist, at least for English publications (Shelton & John 1996, Shelton 1999, Bielby 1999, Coltrane 2000). They all have their raison d’être, as well as all of them are known to be only moderately supported – but nevertheless supported – by empirical data, which makes them all look partial and possibly complementary (as most recent overviews underline; e.g., Coltrane 2000, Shelton 1999). Moreover, all of them are more or less clearly limited to a micro-social level (interpersonal interactions, exchanges or negotiations) as they focus on various mechanisms of individual or interpersonal behavior. This behavior is seen as motivated by such factors as norm conformism, identity expression or interest, and – at least implicitly - as being acted out in a given context that sets different incentives for various behavioral alternatives. The context itself, i.e., the social-structural setup of the environment in which people and couples live, appears as a « given » outside the explanatory models, as these barely focus on the context’s structure and its sex-discriminating features. Against this backdrop, we wish to relate our analysis to a larger theoretical perspective, starting with more general questions and approaches, four of which we spell out, without deriving hypotheses, however, since our method, although basically quantitative, is inspired by an exploratory approach.

Do couples and families constitute autonomous systems, unfolding their proper systemic logic, as the theory of autopoïetic systems (Luhmann 1995) would have it? Do the partners rather construct their own private universe on the basis of their personal norms and convictions (that may of course be influenced by their cultural background), as a constructivist perspective like that of Berger & Kellner (1964) would postulate? Does their internal organization reflect a balance between the resources each of the partners contributes, as resource theory hypothesizes (Blood and Wolfe 1960)? Or are they adapting to external conditions, to the constraints, options and rhythms of their socioinstitutional environment, as recent research leads one to believe (Krüger & Levy 2001)?

If the self-organizing process prevails, we should expect all families to take the same shape without respect to any other factor, since in this perspective « a family is a family », i.e., there is no a priori internal source of variation that could differentiate autopoïetic processes between individual families, at least if they have the same basic composition. If the constructionist view is correct – which postulates in fact another, agency-centered variant of self-organization - we should expect that the partners’ family norms are strongly related to their family’s actual organization, much more so than more structural variables. If external factors are important, there should be associations with these factors, i.e., with partners’ external status positions in the case of the Blood & Wolfe hypothesis, and with more institutional elements, possibly hard to identify, in the case of the institutional adaptation hypothesis, personal norms notwithstanding.

While these hypotheses concern the internal functioning of the family and its relation to external, societal factors, European sociology has also been the arena of much debate about a so-called second wave of modernization supposed to take place in postindustrial societies (Beck 1986, Giddens 1984). This modernization is said to have revolutionized the existing forms of living, alone, in partnerships or as single parents. This kind of hypothesis concerns temporal variation, predicting increasing diversification of these forms, dissolving pre-existing patterns of standardization.

We shall approach these hypotheses mainly on the basis of a recent survey on family functioning in Switzerland and do so in an exploratory perspective. Even if it is a part of the problem under study, the distinction of areas internal and external to the family is also convenient for the organization of our argument.5 Consequently, we shall first examine intrafamilial aspects, i.e., the gendered attribution of internal tasks and decisions on the levels of practice and of norms, and secondly extrafamilial participation, where we shall concentrate on the most important extrafamilial area, i.e., employment, and then look at the interconnections between male and female participations and their variations through the life course, focusing our attention on the importance and distribution of work in- and outside the family. We shall start with a short examination of the modernization hypothesis, for which we use supplementary data, and then proceed to the more analytical hypotheses.

2. An Empirical Exploration in Switzerland: Concepts and Data

The main data we shall use for our purpose stem from a large survey of couples in Switzerland. Both senior authors of this study, having examined family and gender relationships in Switzerland since the 70ies (Held and Levy 1974, Kellerhals 1982), were relatively skeptical about postmodernist interpretations of family change. One of our objectives was therefore to explore to what degree certain forms of couple or family organization could be identified that did clearly not correspond to the traditional model. Even if the precise definition of this traditional model may be less than consensual and clearly defined in the literature, it seems safe to say that social history (Mitterauer 1979, 1984) and functionalist theory (Parsons & Bales 1955) have identified a specific organizational model with respect to the dimensions we call regulation rather than those we call cohesion.6 Regulation is an umbrella term for the various aspects of how a couple organizes and coordinates its members’ actions or what we may call the everyday production of family life. We distinguish three basic aspects of regulation: role definitions, hierarchy, and routinization. The notion of role definitions can be further differentiated into relational aspects (who behaves how towards the other and the group) and instrumental ones (who does what). Hierarchy relates to the aspect of dominance or equality between the partners and need not be differentiated here on theoretical grounds. However, we may distinguish different operational approaches to identify the power position of a person, following the classical methodological distinction of decisional, positional and reputational power.7 By routinization we mean the degree to which a couple has established fixed rules concerning the territories and rhythms of its daily life. The following analysis will concentrate on the most usual regulative aspects of family functioning: task attribution and power differentiation.8 The various ways in which men and women combine labor market and family participation will be called their specific participation profile.

As already stated, the data to be used stem mainly from our recent nationwide study of couple dynamics in Switzerland that included 1534 couples, with both partners interviewed separately by telephone (CATI) in 1998 (Kellerhals et al. 2003). Various dimensions of couple functioning (cohesion and regulation) were analyzed as such and in relation to social status, social (extrafamilial) participation, and social time or life-course positioning, as well as their connection to problems, conflicts, coping and the quality of the couple’s life. The sample definition included all couples in Switzerland living together since at least one year whose younger partner was aged between 20 and 70; they could be married or not, nationals or foreigners. The response rate of 31% of the couples was moderately satisfactory as such, but we could hardly expect more given the necessity to interview both partners.9 Our checks of distributional representativeness yielded very satisfying results, with only a slight (and common) indication of overrepresentation of middle and upper middle social strata. Some other data, stemming from older studies, will be used only for an introductory temporal comparison at the beginning of the next section.

As already stated, the aim of the following analyses is rather an exploration than a strict testing of hypotheses. Accordingly, we refrain from using advanced methods of statistical analysis and rely on more exploratory techniques.

3. The Family: Still a Gendered Work Place

3.1 Traditional Task Attributions

As a first approach to our topic, taking up the modernization hypothesis, let us look at the recent evolution of the factual division of family work10 and then at norms concerning it. Unfortunately, there exists no single, homogenous data series in Switzerland about factual aspects of (internal) family regulation.11 On a very descriptive and spotty level however, four studies that have used identical or similar questions concerning task allocation and decisional attribution in couples offer an opportunity for a trend estimate. Table 1 shows those task items from the four surveys that are more or less comparable between 1971 and 1998.12

A quick look at the frequencies reveals some moderate tendencies towards less sex typing of task allocation in such items as administrative contacts, gifts, holidays, cleaning, but there seems to be a hard core of tasks showing very little change (cooking meals, washing); overall, the observed changes over almost three decades are not spectacular and certainly do not corroborate current, highly mediatised ideas about modernization and widespread egalitarianism. Moreover, they may be partly due to the generalization of the norm of gender equality, belonging to the realm of declared political correctness, whatever the personal practice may be. Nevertheless, it seems safe to say that overall, the practical sex-typing of family tasks according to the interviewed partners remains through the past thirty years, despite some weak signs of decrease, especially if we take into account that even granting for a less than « representative selection « of masculine and feminine items in our questionnaire, the latter concern much more regular and time-consuming tasks than the former (such as administrative contacts or repairs) and thus weigh heavier in the partners’ time and energy budgets.

Table 1: Items of gendered task allocation in couples (who does what), 1971-1998 (Switzerland; line percent)13

Obviously, there is a dimension of inequality hidden behind the apparently « horizontal » aspect of labor division between the partners in a couple. A common way of quantifying possible inequalities of male and female engagement in family work is to count the weekly hours spent at it. While being purely « quantitative » and possibly submitted to subjective distortions, this indicator has various advantages: unlike the domain specific task items, it is synthetic, as respondents have to intuitively add up the whole relevant work if asked to give an overall estimate. It is probably also less exposed to social desirability than specific tasks.14 Unfortunately, we found no longitudinally comparable data about hours of domestic work in Switzerland for a substantial period,15 but it is interesting to look at the present situation. An estimation of hours of household work done by men and women with the data from our 1998 survey shows an overall difference, measured by way of median values, of 29.3 for women and 10.0 for men. We find consistently high correlations of the hourly predominance of female family work and the sex-typing of the four female tasks considered (g between .35 and .47) and zero correlations with the two male tasks (g = .06 and .02).16 Although we cannot use the hours indicator for the assessment of historical change, we shall concentrate on this measure in our later analyses since it synthesizes at a time the degree of sex-typing of family tasks and its inegalitarian component.

Concerning norms about task allocation, the empirical situation is even more sketchy and does not warrant a table. Some existing results warn us not to suppose too simple trends. Held & Levy (1983) showed for a five-year period in the 70ies (1971-1976) inverse changes when comparing two types of contexts in Switzerland: the acceptance of traditional sex role definitions (norms of strict role segregation) increased in urban contexts (where their acceptance was moderate) whereas it decreased in rural contexts (where the acceptance was high). It may well be that both changes, adding up to a movement of ideological convergence between the two types of contexts during the 70ies, can be interpreted as early reactions to the rise of the second wave of militant feminism in the first half of the seventies and the public debates it spurred; it is more than plausible that since then, people’s norms have evolved towards more egalitarianism.17 As of 1998, our data point to the existence of a generalized norm of gender equality that belongs to the realm of well-anchored political correctness, but does not necessarily correspond to actual practice. Faced with the question of whether a mother of school-children should remain completely at home, work part-time in favor of her family duties, share equally part-time work and family work with the man, or whether he should be more at home, less than one percent of both the men and the women opted for the « man more at home » variant – ostensibly, it appears to be simply outlandish. Women’s answers were equally distributed between the other three possibilities (which means that one third was in favor of complete segregation in this phase of family life, another third for reduction of her employment to part-time) while an even greater majority of the men preferred the mother at home (41.0% completely, 27.1% in part). All these observations point consistently to two conclusions: first, that traditional norms of sex-specific assignment of employment and family tasks did loose some of their consensus, but are far from having disappeared, and second, that the distinction between a normatively rejected inequality and a more acceptable « difference » should not be neglected.

Our analyses published elsewhere (Kellerhals et al. 2003) show a strong component of normative conformism in both the normative and descriptive items about sexual task attribution (but even more so for items directly expressing inequality), backing the statement about their expressing political correctness as much as action-oriented personal convictions.18 Even though our evidence remains an indirect proof, it indicates that in the area of discriminative gender differences, the (declared) norms have become more egalitarian than the practice.

3.2 Traditional Decision Patterns

The other classical area of regulation, power distribution between the partners, is ideologically more touchy as it focuses more explicitly on the question of inequality; for this and other reasons, its operationalization is more problematic. Here again, only a sketchy comparison of different points in time is possible (table 2). All the items used by these studies can be seen in the operational tradition initiated by Blood & Wolfe (1960). With respect to the classical distinction of the community power methodology, they correspond to the decisional approach to power assessment.

Overall, the tendencies point towards slightly more joint decisions over the last thirty years (items: husband’s job and insurance, but the latter mainly in men’s perception; for car/important purchase it is doubtful whether the difference is real since the items are semantically not quite equivalent; the difference is inversed for holidays and there is no clear difference for wife’s job and eating expenses). The only indication that explicitly shows a case of stable sex typing is related to insurance with a slight decline of masculine preponderance. In our 1998 survey, the tendency towards equality in the appreciation of seven decision items was manifest, even in those with clearly recognizable sex typing; e.g., to arrange or furnish one’s home is the woman’s decision for 53% of the men and 45% of the women, but for 40%/49% of them it is a joint decision. This could be seen as a fair description of these couples’ practice, but our tests show again that the extent of declared equality is quite strongly influenced by the respondents’ normative conformism (see the explanation in relation with the next table). All this sums up to a situation where a clear change in favor of less sex typing of important couple or family decisions or a tipping of the balance in favor of the woman cannot be asserted (see also Höpflinger 1986 for a more rigorous comparison over the period between 1971 and 1980, yielding similar conclusions).

Table 2: Items of decisional power distribution in couples (who decides what), 1971-1998 (Switzerland)19

The measurement of decisional power has been highly controversial in the literature; several theoretical and technical questions are almost impossible to solve. Inspired by an idea of Eichler (1973), we have developed a complementary measure that tries to tap a more intricate meaning of family power: to what extent are there forms of dependence between the partners? This idea is an attempt to grapple with conceptual problems that plague measures of power used by research in the resource theory perspective. First, at least in principle, there is no reason intrinsic to Blood & Wolfe’s thesis to rely heavily or even exclusively on income or financial resources as a possible reason for internal asymmetries; other types of resources could obey to the same theoretical logic, provided they are central enough to the couple’s functioning or « common weal ». But it is difficult to ascertain and quantify emotional, ideological or relational resources of each partner in a way that allows to compare them according to the objectivist logic of « who contributes more » implied by resource theory. Second, there may be an important, but variable difference between a partner’s amount of a given resource (status position, cultural capital or whatever may be considered), and his or her effective contribution of this resource to the couple’s life; what is more important, potential or effective resource contribution ? Third, it is even more difficult to determine in a non-normative way which of the partners’ « goods » or « services » or « capacities » constitute relevant resources that should theoretically influence the couple’s power balance. Fourth, even supposing they are measured correctly, it is questionable whether an objectivist comparison of the respective resources corresponds to the couples’ reality as partners may define their proper, private equilibrium in a way that is not adequately indicated by equal « quantities » of a given resource. Asking in terms of (inter-)dependence could then be closer to the reality of a couple’s everyday life and mutual evaluation; this strategy can be seen as an adaptation of the positional approach to power assessment in the context of the family.20 Our question was asked for six areas and ran as follows: « To live as a couple means often to depend on the other, and this can be different for various areas. Would you say you depend on your partner in one of the follow areas? ». For each area (see table 3), respondents had to choose between four answers, « You depend and that’s OK for you », « You depend and resent it », « Partner depends and that’s OK for you », « Partner depends and you resent it » . Two more answers were coded separately when given spontaneously without having been explicitly offered to the respondents: « No one depends » and « Each depends on the other».21

Even more than in the previous cases, the answer distributions point to a strong norm of equilibrium or independence: in four of the six areas, the two very answers that have not been explicitly mentioned and express two different forms of equality occur most often ! Only in the two most factual and instrumental areas (housework and finances) do we find relatively high rates of asymmetric, sex-typed answers - and these two are also the most directly relevant ones for the traditional stereotype of the family, i.e., a strict role segregation along the public/private or external/internal divide. The tendency to avoid sex typification of equality-sensitive areas also appears in the fact that in the vast majority of those cases where one or the other direction of dependence is admitted, the no-problem version of the answer comes up much more frequently than the problem version. A comparison of the partners’ answers (the gamma coefficients in table 3) shows little divergence, especially no clear dissension. In fact, it is extremely rare that one partner states one direction of dependence and the other the opposite; if one kind of dependence is mentioned by one partner only, the other will rather tend to claim equality. A further indication of the strong normative component of these answers comes from our attempt at controlling the respondents’ normative conformism or sensibility to social desirability: strongly conformist respondents give more answers of equality (independence or mutual dependence) than weakly conformist respondents.

Table 3: Perception of (inter-)dependence by men and women, six areas (Switzerland 1998)

So we find a generalized picture of (declared) equilibrium by independence or interdependence between the partners. However, two significant exceptions appear; they concern the two instrumental items of housework and finances where the answer distributions are skewed towards the traditional model: men tend to depend on their female partners for housework (more so according to themselves than according to their wives), and women tend to depend financially on their male partners. If we combine these two aspects, one third of the couples declared symmetry in both areas (no unilateral dependence), a fifth sees the man depend for family work, but not the woman financially, another fifth sees inversely the woman depend financially but not the man for housework, and a fourth declares complementary dependence, in conformity with the traditional model.

We do not have detailed normative questions at our disposal that explicitly refer to the power distribution; such questions, implying open acceptance or rejection of male dominance or privilege (e.g., « When the man returns from work, he has to be able to get a rest, whatever the problems in the household may be «, or « A certain equality in marriage may be all right, but in family matters, the man should have the casting voice « ) were included in the 1971 survey, where they yielded distributions that did not seem ideologically biased. Thirty years later, in our 1998 survey, we found it impossible to use such items as they were too openly sexist to be accepted even by people who would personally agree with them. We nevertheless included an overall question about who should be the boss in the family. About two thirds of the respondents prefer the no-boss or equality answer, one fifth prefers the man, and less than a tenth (7.9% of the men and 4.2% of the women) the woman. Two thirds of the couples completely agree in their answers (g = .32). Again, almost all of those who don’t, do not give contradictory answers (only 2.7% of the couples), but one partner gives an equality answer while the other does assign the boss position (with two thirds of these couples assigning it to the man, one third to the woman). When compared to the couples’ description of their actual situation, the answer distributions of the normative variant are identical for couples saying they have no boss and those who identify the man as the boss: practically all (98% and 97%) prefer equality and only 2% the man. In the rare couples who identify the wife as the actual boss, the normative opinion that the man should have this role rises to 37%, which may appear as an expression of uneasiness with the actual situation; in the very small category of couples with directly opposite descriptions of who is actually the boss, possibly expressing something like a competitive power situation, this percentage is even higher (48%). Thus, in situations that do not correspond to the traditional model, the man appears as a desirable boss to a substantial part of respondents while in the other situations, both partners prefer equality, whether the traditional model of male dominance is admitted or equa-lity is declared. This may be interpreted in at least two ways. Either the norm of equality is strong and practically realized in the couples who recognize they have a (male) boss, or the man holds de facto the same dominant position in the two cases, but for norma-tive reasons, part of those couples prefer to declare the situation as equality. A third interpretation could be in terms of a couple’s private, subcultural definition of actual inequality as corresponding to its norms of equity and thus « equal », at variance to what it seems from the outside. Probably, only qualitative probing could give unequivocal clues to these competing interpretations’ adequacy.

We shall not try to decide that question here, but stick to more straightforward conclusions. Our findings on task attribution and decisional power show that the coincidence between norms and practice in this subject matter is not given by itself. There is some evidence confirming the persistence of a consi-de-rable amount of normative acceptance of male dominance, but, even more important-ly, the norm of equality seems to have been accepted much more generally than actual equality in contemporary Swiss couples.

4. Working Outside the Family: Gendered Employment Participation

4.1 Who is in the Labor Market ?

The traditional model of family organization, based on sexual task segregation, ruled out female occupation even though at all times a substantial, but variable and sometimes minor part of married women did not stay entirely out of the labor market (not to mention unmarried women). The quite inequitable complementarity between a wife looking after the internal affairs of the family and a husband assuring its survival by earning the necessary income outside was culturally – and for a long time also sociologically – defined as balanced (Held 1978). While this part of the traditional model is no longer generally dominant, neither on the level of norms nor on that of practical action, it has far from turned into its opposite, except for a specific set of countries. According to Maruani’s (1993) comparative analyses, Switzerland, like (West) Germany, belongs to the group of countries where employment goes currently along with marriage, but not with motherhood. Let us first look into the question of female labor-force participation in a static perspective and then bring in its variation through the life course.

The labor-force participation of men and women living in couples follows quite different patterns. Whereas very few men do not work full-time before they are retired, women work in relatively equal proportions on all levels of involvement: not employed, employed at less than 50%, 50%-89%, and full-time (i.e., 90%-100%). If we combine these four categories for men and women living together, we find, significantly, in the ensuing 16-cell table all of women’s participation degrees with sizeable frequencies if the man is working full-time, but the eleven cells with the man working less than full-time, taken toge-ther (under the hea-ding « atypical ») except for retirees, sum up to hardly more than the weakest single cell frequency of the more « typical » situations (13% as against 11% - 27%). Thus, in table 4, the 11 atypical cells have been collapsed into one.22

Given this background information, it is safe to say that in Switzerland, the occupa-tio-nal participation of men living in couples follows quite extensively a binary model of em-ployment: it is either full-time or none at all, as opposed to women’s occupational par-ticipation which is « flexible » and shows a large variability. This picture testifies to an en-during strength of the male breadwinner model; men who live in a couple and have finished their initial education work full-time until retirement except in cases of acci-dents like forced unemployment.23

Table 4: Partners’ combined rates of employment (%)

Thus, women’s activities are much less exclusively confined to the family than the Parsons-Bales (or historical bourgeois) model would have it. There seems to be considerable room for women’s employment while they live in couples. Nevertheless, our com-pa-risons of men’s and women’s employment ratios and also of their contri-bu-tions to the household income show that except for the atypical cases, the women’s contributions remain supplementary to the men’s income: in no more than 10.6% of our couples does the female income equal or even exceed the male one, and more than half of these couples are in the aty-pical situation as the man is less than full-time employed.24 A telling expression of the subsidiary status of most wo-men’s in-come is the median value which amounts to 5477 Swiss franks for the men and 1201 franks for the women (ratio 4.6: 1, which of course should not suggest that in a substantial number of couples, the wo-man’s income is not vital for the household).

A more systematic appreciation of this interpretation can be made by comparing the correlations between the partners’ individual incomes and their common household income (graph 1), for which they tend to give extremely similar estimates (g = .85).25 The household income correlates highly with the man’s individual income, much less with the wo-man’s whose contribution appears, thus, to be less determining for the overall common income than the man’s. The two individual incomes are independent of each other, suggesting that a probable tendency to compensation of one income by the other is overridden by the occupational corollaries of homogamy.26

Graph 1 Correlations (g) between partners’ individual and household incomes

In the sense of our general question we can bring home the result that em-ployment is clearly no longer in a fundamental contradiction with the status of the woman as a part-ner in a couple, but her contribution to the family’s finances re-mains subsidiary in a great majority of cases, whereas the male partner still appears as the major eco-nomic provider of the household. In this sense, we can indeed conclude that the male-bread-winner model has not died, at least not (yet) in Switzerland.

What about norms concerning this question? Our questionnaire allows us to confront the actual picture with a roughly corresponding norm formulated as follows: « In case your everyday life should be seriously disturbed - like having to move your house-hold, a serious illness, or an important problem in your kinship - who of you two would change durably his or her duties, time-table or projects in order to maintain your common life in the new situation? » This question is supposed to tap the partners’ norms about the dominant status area – employment or family - in a more general way than just referring to the presence of children: for whom are the family duties considered to be dominant, for whom the occupational duties? Again, the partners in the interviewed couples agree quite largely and in most cases in a sex-typed sense: 61% say it would be the woman, 12% the man, and 27% that it is not clear from the outset, that it would depend, or that it could be both or either of them. How does this norm relate to the partners’ actual employment situa-tion?

Table 5: Who would adapt his/her occupational enrolment in order to cope with urgent family needs, by effective rates of employment (%, women’s answers)

Table 5 shows that on this normative level, the adjustment of the employment to the family needs is considered to be the woman’s duty in a third to more than half of all the couples, the man’s in hardly 10% of them; another third to half of them do not distinguish between the partners.27 Whether it ought to be the woman is related to quite some degree – al-though not linearly - to her actual rate of employment, with a « low » 31% in couples where both partners are employed full-time, and over 50% in the other situations (ex-cept for the retiree couples and the atypically employed ones where part of the men are already retired). The correlation is significant and confirms a direct inverse relationship between the rate of occupational engagement and the (normative) availability for fa-mily ur-gencies, but considered overall, it is of only moderate strength (Cramer’s V = .14 for women’s answers, .16 for men’s). On closer scrutiny, the table reveals inte-res-ting par-tial relationships. The normative availability of men is very low, but it is slightly increased in atypical employment constellations when the man is occupied less than full-time. The normative availability of the woman decreases with her increasing degree of employ-ment, be-gin-ning with low-rate part-time occupation. One might say that to the extent that the weight of her employment increases, her exclusive availa-bi-lity decreases and the « both » answer becomes more acceptable. In parallel, but on a much lower level, the « don’t know » and « it depends » answers increase also. The fact that couples with full-time employed men and with women employed in a « small » part-time job (under 50%) have the most traditional answer distribution, even more so than couples with a complete segregation, is also interesting as it coincides with some other findings showing that small-rate female employment has no reducing effect on traditional forms of regu-la-tion. It may well be that this employment constellation expresses an especially strained compromise between both highly valued family activities and equally highly valued or necessary employment activities – this interpretation needs more refined analyses than have been conducted so far.

To sum up our findings concerning the normative priority of employment vs. the fa-mily, we can conclude that there is interdependence between the actual, objec-ti-vely defined employment situation of the couple and the normative dominance of the fa-mi-ly for the woman, but other factors must be of similar or higher importance, given the rather modest correlation we found. Only in couples with identical enrolment rates of the partners (both 0% or both 100%) the non-typed answers amount to a small majority.

In order to enlarge somewhat the horizon of the analysis, let us add an assessment of traditional male-female com-ple-mentarity, based on the two instrumental items of our question about (inter-) dependence. This complementary aspect is more internal than the actual employment and expresses a subjective perception to a higher degree (table 6).

Table 6: Who would adjust his or her occupational enrolment in order to cope with urgent family needs, by instrumental interdependence (%, women’s answers)

The main result of table 6 is a correlation in the sense that the more instru-men-tal interdependence there is according to the traditional model, the higher the propor-tion of couples where the adjustment to family needs at the expense of gainful employ-ment is considered to be the wife’s duty.28 This norm seems then to be quite logically re-la-ted to the traditional model of family organization. Here again, however, the correlation is signifi-cant, but moderately so, which can be interpreted as meaning that this norm is to a considerable extent autonomous, inde-pendent of situative elements.

Without going into the examination of other possible influence factors, let us conclude that there is some significant correspondence between the actual situation of employ-ment and the acceptance of the normative principle that the family ought to be the do-minant responsibility of the wife, but that another part of this norm seems to be either autonomous or related to other factors not identified yet.

4.2 How Are Men’s and Women’s Working Hours Related?

Let us come back to the factual level and exa-mine to what extent the employment configuration influences the effective family-work load of the woman. Table 7 has been established using an estimation of wo-men’s relative share of family-work hours; it shows quite a strong rela-tionship (Cramer’s V = .30).

Table 7 Women’s estimated share of total family-work hours, by rates of employment

We look first at the four situations with full-time employed men. There is an over-all re-lationship in the sense that with an increasing female employment rate, the man’s fa-mily-work engagement increases and the wo-man’s decreases, except for rates below 50% that seem to make no difference compared to no female employment at all. How-ever, couples with a substantial male share of family work remain rare even in the atypical situ-a-tion where the men’s employment intensity is less than full-time and often lower than their fe-male partner’s. We note more-over that retiree households show a work distribution si-milar to that of couples with an entirely traditional, segregated employment si-tuation. In all situ-ations, these va-riations notwithstanding, the woman bears the major part of family work (over 60%) in a majority of the couples.

In more technical terms, with men’s employment rates controlled, we find a substantial negative relationship between women’s employment rates and their share of family work. But this is only part of the possible interrelationships to which we finally turn.

How are employment hours and family-work hours related to each other for men and women? We do have approximate information on both partners’ working hours in both areas, in terms of employment rates and weekly hours spent for family work. To give a quick overview, graph 2 presents the zero-order intercorrelations of the four indicators.

Graph 2 Correlations (g) between men’s and women’s occupational and family work engagement, based on weekly work hours

The intercorrelations show an asymmetric pattern of interconnections. The strongest correlation is the negative one between women’s family and occupational engage-ment, indicating the presence of an either-or situation. Elimination of couples with retired partners reinforces this correlation. The moderate correlation be-tween male and female em-ployment rates cannot be interpreted in general terms because the male employment rate, unlike the other three variables, has shown to be mainly binary. The correlations implying this measure are largely constituted by the comparison between full-time employed and re-tired men,29 which makes them some-what trivial: the correlation mentioned mainly says that women living with non-retired men tend to have hi-gher employment rates than the others, which could be a simple age effect because most wives of retired men are retired themselves. Without retired couples, this correlation inverses its sign and becomes negative, showing a complementary relationship between non-retired partners’ employment rates, i.e., some tendency toward a compensatory link, underlining the fact that to some degree, the partners’ household unit functions also as an economic unit. The other two non-zero cor-relations are relatively weak. Female part-ners of em-ployed men have a slightly greater load of family work than those of retired men, pro-bably taking up part of the housework that could be done by the man were he not (or less) employed, which is interesting in that one could imagine a negative correlation meaning that non-employed (retired) men take up more family work than employed ones (this correlation is not influenced by retirement, .23 instead of .22). Female and male family work is somewhat positively correla-ted, possibly indicating a weak « solidarity » effect probably related to the total amount of fa-mi-ly work needed (again, we find no retirement effect: .21 instead of .22). Including and excluding retired couples, the correlation between male family work and female employment rates remains zero (.05 with and .06 without retired), expressing what we could subsume from earlier results, i.e., that men’s high employment rate and low family-work rate vary insignificantly, to the contrary of women’s, and consequently do not « respond » to women’s rates’ variations. The last correlation, between men’s occupational and family work, becomes reasonably negative (-.24) instead of zero (-.06) if only non-retired men are considered, indicating a slight tendency towards an either-or use of time, although with a coefficient that remains considerably more moderate than its female counterpart.

The whole situation strongly suggests the interpretation that when pro-blems about the distribution of work time appear in the couple, the real zero sum game about time use is on the side of the women, be-tween their employment and their family work; the va-riability of men’s occupational and family hours is weak, with a heavy pre-pon-de-rance of their occupational hours.30 There is no systematic relationship between men’s family-work hours and wo-men’s occupational hours, i.e., men’s family-work engage-ment does not lead to or allow higher female employment rates, nor is the former con-di-tioned by the latter, and there is no systematic overall relationship between men’s employment and the amount of their family work either (but a clear hint at a weak relationship when full-time and less-than-full-time employed men are being compared) while women’s family-work hours do have an « alleviating » relationship with men’s – and with men’s employment rate.31

To put it briefly, this first part of our exploration has shown that many women living in a couple today participate in the labor market and that tendencially, the partners’ share of family work varies with the intensity of women’s employ-ment according to a compensatory logic (whereas men’s does not vary). Even so, it clearly appears that women’s primary attachment re-mains, in a large majority of the couples, to the family and not to their employment, on both the factual and the normative level, and that the high, but variable female share of family work seems to be a condition for men’s high and steady occupational involvement. To put it squarely: when it comes to real hours of performed work and to real money earned on the labor market, there is a strong tendency for men to remain the main bread-winners, and for women to remain the main family workers.

5. Gendered Participation Profiles and Phases of Family Life

5.1 Towards a « Structural » Concept of Family Phases

After this structural and static analysis, let us add a more dynamic perspective although the synchronic data at hand impose an indirect approach that we call « metastatic ». This needs an introductory discussion. We can compare our responding couples by means of some simple information pertaining to the classical phases of family life, usually referred to in terms of the family life cycle, which correspond to the usual operationalization of this concept in sociology and social demography (Murphy 1987). We can distinguish five such phases and compare them with couples who have no children and will not have, according to statistical likelihood and personal projects.32 This « life course » variable allows us to compare groups of couples in specific phases of the typical family trajectory, but cannot tell us beyond doubt whether any one couple passing from one phase to the next will change or remain stable exactly according to the results of our comparison because we cannot follow them up. Several results give a strong credibility to a life-course interpretation of findings such as those in this section. First, the relatively concentrated age distributions within each of the five family phases and their differences between the phases support the interpretation of these phases in terms of a temporal succession. Second, most of our analyses that relate a family or participational variable both to age and to this life cycle variable produce more distinctive results with the latter than with the former. This is most clearly the case for the female employment patterns. This points to the fact that even if time is implied, the crux is less the personal age or the couple’s duration than the « structural » phase of family life as it is usually defined, i.e., the absence or presence of children and their situation in terms of schooling. Third, a relationship between female work load and family phase is more readily interpretable as an influence of the latter on the former than in the opposite sense since it is hard to see how a given work load could influence the phase in which the family actually is.

Before looking at the partners’ rates of employment in this metastatic life course perspective, it may be important to stress the « structural » aspect of the family phases. As this second remark underlines, we may not take the family phases to just represent time in an abstract sense. As they are usually defined, they represent different constellations with which the couple has to cope, depending mainly on the presence of children and their extrafamilial career with respect to school and occupation. Without necessarily taking up the perspective of developmental family tasks (Duvall 1957), we have to take into account the many feedbacks on internal family dynamics, and thus the parents’ everyday situation, provoked not only by the individual development, but also (or even more so) by the extrafamilial institutional career of the children. In this perspective, a life-course or family-life-cycle interpretation of differences between the phases of the family life must not necessarily assume that any individual person living in a given phase has reached this phase by the simple succession implied for the whole group. To illustrate this idea by an arbitrarily constructed example, take one mother grown-up children living in their own household, who has been living as a single for some years and rooms in with a man himself separated from another woman but with the custody of their pre-school twins. Her life course is all but linear: in terms of the classical succession of family phases she comes back to an earlier phase after having left a later family situation, but these precedents notwithstanding, there are good chances that she will find herself involved in the situational logic that corresponds to the classical phase of a family with pre-school children. In that sense, the family phases should rather be considered as structural configurations « normally » succeeding each other in a given order, but mainly defined by specific social situations, and not as a mere « timing of events » in individual life courses (Levy 1996, Giele & Elder 1998).

5.2 Explorative Analysis of the Impact of Family Phases

Graph 3 shows that the employment pattern of women, contrary to that of men, is clearly related to the phases of family life as we define them. This relationship is even more pronounced if the employment rates are weighted by the activity rates (not shown in the graph, but see table 8), with again no such relation for men. The arrival of a child is the main reason for the interruption of female employment; a later return to employment remains below the initial level (which is only slightly higher than that of couples without children: 70.7% for the women, 79.3% for the men – while this category contains a large range of ages, including retired persons).33

Graph 3: Women’s and men’s employment rates (% employed), by phases of family life

Since table 7 demonstrated the relationship between the couples’ employment configurations and the women’s share of family work, it is interesting to check whether there is a relationship between these configurations and the family phases. In accordance with graph 3, table 8 confirms that it is women’s, not men’s paid employment which is strongly related to the typical phases of family life. After the birth of the first child, almost half of the women quit their employment altogether and another third reduce their occupational time investment. The proportion of couples with continuing « double careers » (which are in fact often nothing more than a double fulltime employment without any career perspectives for the woman or both partners) drops dramatically and lastingly between the pre-child phase and the preschool-child phase while the proportion of traditionally segregated couples (man full-time, woman not employed) increases to almost half of all couples, to drop again in later phases where there is some tendency to resume or to increase again women’s paid employment (but to a rather limited extent). Let us insist once again that these rather strong family-bound variations hold for women whereas nothing of the sort can be observed for men.34

Table 8: Partners’ employment rates (%) and phases of family life

« Another side » of this relationship is confirmed by graph 4 that shows the association between family work hours and family phases for men and women. The median number of weekly hours rises in the presence of a small child and diminishes slowly in subsequent phases. The differential between female and male investment into family work remains consistent and substantial in all five phases. By visual inspection of the graph, we might not say that male investment in family work depends less on the « needs » of family life, but statistically speaking, the relationship is clearly stronger for women (Cramer’s V = .39) than for men (V = .21). Moreover, male investment remains inferior to female investment by ratios that rise considerably across the phases, from 2.2 to 4.9 (or 4.4 for the last phase).35 Let us add that the age differences between adjacent phases as well as the correlation of family-work hours with age are both too small to account for a substantial part of the relationship in graph 4.

Graph 4 Women’s and men’s median hours of family work per week and female/male ratios (right-hand axis), by phases of family life

A second « other» or rather a complementary side to the flexibility of women’s employment rate as a function of the family phases is its obverse: the relationship of the men’s family work hours as a function of their employment rate which we examined more globally in graph 2. We need no table to examine this relationship: the slight variations we find do not constitute a clearly interpretable association, even though it is significant (Cramer’s V = .07**). Men’s median weekly hours are practically the same for retirees (10.3) as for full-time employed (9.8) but higher (13.2) for the rare cases that are less than full-time employed.

5.3 Regression Analysis of Family Phase Effects

Although the main thrust of this paper is explorative, as we explained at the outset, we add a set of ordinary least squares regressions in order to quantify the effects of the family phases on the distribution of domestic work (female share) and on women’s employment rate, controlling the confounding effects of other variables (table 9). Model A considers the effects of family phases without any control variables. It shows that all family phases have a significant effect in making the women’s share bigger than in the prechild phase (taken as the reference category). When age and educational level of women as well as monthly income and employment rates of their male partners are included (Model B), the successive family phases still make the women’s share significantly bigger than in the prechild family. Only in families where the children have left the parental household (post-child phase) or in couples without any children do we find no difference in the women’s share of domestic work compared with couples in the prechild phase. Among the supplementary variables, both the woman’s age and her male partner’s involvement in the labor market have independent strengthening effects on the unequal distribution of family work whereas the woman’s educational level slightly redresses the gender balance. Age net of family phases, in this context, may well be a proxy for generational differences; the male partner’s occupational involvement expresses mainly the distinction between employed and retired men, as we have already argued. Thus our interpretation that family phases have an important influence on the organization of domestic tasks in families, net of possibly confounding variables among which the woman’s age is most prominent, is corroborated by these models. It is interesting to note, however, that inclusion of women’s employment rate as an independent variable (Model C) makes the effect of the family phases vanish. This suggests that family phases may be strongly correlated with women’s employment rate and that this correlation is to a large extent responsible for their effect on organization of domestic tasks. Models D and E address this issue.

Table 9 Phases of family life as predictors of women’s share of domestic work and women’s employment rate (standardized coefficients (beta)

As shown in Models D and E, the pattern of influence found in Models A and B has its equivalent on the side of employment: the women’s employment rates also depend strongly on the family phases, even more so than their family-work share, as the higher R 2 indicate. Even when age and other control variables are included (Model E), family phases still retain a strong influence on women’s employment rates. Especially phases with a dependent child (preschool- and school-child families in particular) are associated with much lower female employment rates than other phases, as graph 3 already showed. Post-child families and no-child couples present employment rates much closer to those of prechild families. These results suggest that for women, employment rates and share of domestic tasks are systemically interrelated, both following the logic of family phases that typically represent different amounts of work. The burden represented by children strongly influences both the women’s share of family work and, thus, their engagement in paid work. When this burden decreases (as the family enters a phase where children get more independent), women can get more involved in occupational activities and therefore increase their involvement in paid employment. The difference of the R2 between the two dependent variables – higher for employment rate than for family-work share - suggests that the effect of the family phases on the distribution of family work might be mediated by the women’s employment rate. This is corroborated by Model C that adds the female employment rate to Model B: in Model C, the women’s employment rate “takes over” the significant parts of the family phase effects on family-work share while it is itself quite well explained by the family phases (Models D and E).

What about men? Are their employment rates also sensitive to family phases and do they remain so when control variables are included? Table 10 answers these questions.

Table 10 Phases of family life as predictors of men’s employment rate (standardized coefficients (Beta)

Model A shows that family phases characterized by the presence of children do not have any effect on men’s employment rates: contrary to their effect on women, they do not promote a lower implication of men in occupational activity. Only when children have left the parental home do we find a significant effect: the employment rates of men decrease, in sharp contrast with what we found for women. This effect is likely to be associated with age, as men usually reach their age of retirement in this phase of family life.

The inclusion of control variables (Model B)36 has a very different effect for men than for women: the effects of all family phases become highly significant and, unlike for women, positive. This means that when a control of age is set (which has, as the table shows, a much bigger influence on men’s employment rates than on women’s), the family phases after the prechild phase increase men’s employment rates (it is interesting to note that the effect of the post-child phase decreases with the introduction of age, confirming that this effect is more age-related than phase-related). Thus, the logic of men’s and women’s professional trajectories appears to be opposite. Family phases with dependent children decrease the women’s employment rates and increase the men’s employment rates. Only couples without children do not follow this logic, as shown by the insignificance of this phase for women and its marginal significance for men.

Putting all our results together, we can conclude that there is a strong relationship – which we can interpret as an influence on the basis of the arguments developed at the beginning of this section and our quantitative findings - between the family phases and women’s employment, and next to no relationships with or influence on men’s employment.37 We find a strong exclusionary relationship between women’s employment rate and their family-work load – the time spent for work at home cannot be spent for work in an occupation, and vice-versa; but none for men. And we find practically no relationship between men’s employment rates and their investment in family-work.38

This can be reformulated in the sense that if a strict sexual segregation of activities along the internal/external (or private/public) divide, typical of the traditional model of family organization, has become rare, nevertheless a complementary relationship between male and female participation profiles emerges on a rather large range: the family tasks prevail for women and limit their occupational participation, the occupational tasks prevail for men and limit their family participation. As the load of women’s family tasks varies with the phases of family life, women’s occupational participation varies as well, but more or less to the extent that their family obligations allow. As men’s employment rates vary only slightly, their family investment is also relatively constant, but it seems even more constant than it should be for that reason only. This is highlighted by the fact that retired men do hardly more family work than fully employed men – sheer availability seems of no relevance here. To use an economic concept: women’s employment rates have a rather high family-task elasticity whereas men’s family-work rates are employment-inelastic. Plain availability does not seem to account for these findings. Rather, the reason for this fact is likely to be found in the low social valuation of family work, especially for men – or put differently, in the lasting sex-typing of these activities under changed circumstances which this paper documents.

In order to develop a fuller interpretation of our results, we can take up a term inspired by an article of Hughes (1945), master status, sometimes used by feminist scholars but in a modified sense. In accordance with Hughes’ argumentation, sex itself is usually considered to be a master status, in the sense that it outweighs other status criteria and creates tensions when it is not in accordance with the cultural definition of inferiority or superiority of other status criteria (Laws 1979). The concept is mainly used with reference to identity; this may be the reason why it seems to be used more often in the socialpsychological than in the sociological literature (e.g., Ferree & Smith 1979). In the context of our analysis, it appears to be more promising to use this concept in a more complex way, i.e. to define master status as meaning the (socially determined) dominance of one status area over the others in an individual’s participation profile, and to apply this definition to the complementarity of sex-specific master-statuses of the partners in a couple. We leave open, for the time being, the question of the reasons responsible for this dominance, but then at the same time underline its importance. Our findings point exactly in this direction:

· the female master status area is the family, employment is subsidiary to it;
· the male master status area is employment, the family is subsidiary to it;
· together, they form a complementary system that can exist in several variants, the traditional model being just one extreme case where each partner’s sex-specifically dominant status area excludes the other one; in most empirical cases today the dominance of one field is less exclusive and moreover, its expression varies across the family phases. This does not mean however that is has entirely disappeared.

So we come to the conclusion that the sexist gender order in the family has become flexible, but that it remains largely alive in a modernized form.

6. Conclusion: Family Traditionalism is not Dead, it is Rather Being Modernized

Our exploration of recent Swiss data on couples and their internal organization has produced convergent evidence in accordance with many results of other studies from different countries (see once more the detailed literature overviews by Shelton & John 1996, Shelton 1999, Bielby 1999, or Coltrane 2000); for our purpose, it can be summarized in the following way:

1. In a synchronic perspective, different types of (internal) family regulation vary considerably, especially the dimension of task inequality between women and men, whereas their general change over the last thirty years seems relatively moderate. No single type predominates, neither the fully traditional one which still exists, nor any opposite or inverse type. The main departure from the traditional model is the more widespread employment of non-single women, much more than certain trends towards less inequality in the family.
2. We find a strong asymmetry concerning the partners’ patterns of handling the parallelism of family and employment obligations. Male participation profiles are mainly organized around employment that remains, to a very large extent, fixed on full-time enrolment, leaving only little room for family duties. Female participation profiles pivot around the family. They entail work loads that vary strongly between typical phases of family life and affect women’s employment while men’s remains unaffected.
3. Thus, according to our evidence, the male breadwinner model has much less disappeared than folk and media wisdom would have it and than statistical data on male and female employment seem to reveal (if not longitudinal and properly differentiated). It has, however, changed its appearance: men’s employment careers are steady and tendencially ascending, women’s careers are broken and static or descending.39
4. The overall result is that only a minority of women living in couples stay permanently out of the field of employment (but still about one third), others attune their occupational life to lesser and not necessarily constant extents to the family life cycle, which itself varies according to a calendar or agenda that is largely set by extra-familial institutions.

Once a child has arrived, the calendar of family life is in fact largely determined by the social institutionalization of children’s upbringing, to the point that their school career has traditionally been used to define the phases of the family life cycle. This implies that individual couples have a limited leeway to influence the conditions under which the tasks and work loads typical of each phase can be honored. This gives to the institutional environment a considerable influence on a couple’s choices of internal organization. This influence is likely to be affected again by socialpolitical measures. If we found many signs of wide-spread normative egalitarianism among our respondents, it appears that this belongs more to the area of generalized beliefs and opinions than to personal convictions that would be strong enough to motivate decisions going against choices that are generally considered to be rational (like renouncing to parts of employment that induce greater losses of income – usually the man’s - for the sake of gender equality in the couple). It would be interesting to enlarge our questioning to the conditions and consequences of this situation. Let us just mention the beginning comparative research literature on the influence of the institutional frame-work in various societies on the rules relevant for the couples’ choices concerning the partners’ employment. These comparisons are working with adapted typologies of regimes of social policy (Pfau Effinger 1999, Künzler & Walter 1999, Korpi 2000, Boje & Almqvist 2000, Fux & Baumgartner 2001). Their interesting results show that the Swiss situation, that has been our focus, belongs to an especially « liberal », i.e., non-interventionist and de facto inequality-fostering regime otherwise typical of the Anglo-Saxon countries.

This makes us think that the modernization of family traditionalism which we have found instead of a simple disappearance of the traditional model, expresses a resilience of traditional elements that may be due to a large extent to the extrafamilial context of living and its ways of functioning. In this perspective, a considerable part of the family structure appears to be institutionalized indirectly by exogenous social rhythms and structures as much as or even more than by the personal beliefs and interpersonal negotiations of individual partners. This gives some credibility to the institutional rather than to the autopoïetic, constructivist, or resource-theoretical interpretations, without contradicting complementary elements of utility-optimization or of interindividual « doing gender ».40

In this context, let us come back to the concept of master status. As already pointed out, the actual use of the concept in the literature is strongly limited to cultural mechanisms like sex-specific prestige differentials, sex-role expectations and possibly gender performance (« doing gender ») and their inter-individual enactment (although the doing-gender literature hardly refers to that term). With respect to the interconnections between the different (gendered) ways of combining or not combining family work and occupational work and the (equally gendered) life course, Krüger & Levy (2001) have discussed the necessity not to focus the problematic of doing gender preferentially or even exclusively on the inter-individual, micro-social level, as the initial ethnomethodological orientation of this concept did, but to apply it to the sometimes macro- but more frequently meso-social, institutional level as well. Institutions in the sense of organizations, and not only « moral » or cultural institutions or ideologies, are doing gender as well as or even more effectively than individual actors, and a gendered functioning of the institutional environment where a couple lives may create strong incentives to accept sex-specific master-statuses even if the partners have other, e.g., egalitarian, normative convictions.41

At the end, a methodological consequence of our findings should also be pointed out. The introduction of a typology of family-life configurations or phases has not only added strong evidence to our substantial conclusion. It makes it plain that unlike for men, strictly synchronic data about women’s participation profiles have to be considered as nothing more than mean values that are sociologically meaningless because of the strong variation induced by these phases. This implies, more specifically, that women’s employment rates are at a time over - and then underestimated by such figures, since there are typical phases with much higher rates than the mean and other no less typical phases with much lower rates. For the study of systematically changing participation profiles in the individuals’ lifetime and the consequences of these changes (which this contribution has no room to treat), the inclusion of the life-course dimension and of longitudinal data proves to be indispensable.


Beck, Ulrich (1986) Risikogesellschaft. Auf dem Weg in eine andere Moderne. Frankfurt/Main: Suhrkamp.

Beck-Gernsheim, Elisabeth (1998, 2000 2 ) Was kommt nach der Familie? Einblicke in neue Lebensformen. München: Beck.

Berger, Peter & Hansfried Kellner (1964) « Marriage and the Construction of Reality: An Exercise in the Microsociology of Knowledge. » Diogenes 46: 1-32.

Bertram, Hans (1991) Die Familie in Westdeutschland. Stabilität und Wandel familialer Lebensformen. Opladen: Leske + Budrich.

Bielby, Denise D. (1999) « Gender and Family Relations ». in: Janet Saltzman Chafetz (ed.) Handbook of the Sociology of Gender, Dordrecht / Boston / London: Kluwer Academic Publishers: 391-406.

Bielby, William T. & Denise D. Bielby (1989) « Family Ties: Balancing Commitments to Work and Family in Dual Earner Households ». American Sociological Review 54 (5): 776-789.

Blood, Robert O. and Donald H. Wolfe (1960) Husbands and Wives. The Dynamics of Married Living. New York: Free Press.

Boje, Thomas P. & Anna-Lena Almqvist (2000) « Citizenship, Family Policy, and Women’s Patterns of Employment ». in: Thomas P. Boje & Arnlaug Leira (eds.) Gender, Welfare State and the Market. London: Routledge.

Buchmann, Marlis and Manuel Eisner (1997) « The transition from the utilitarian to the expressive self: 1900-1992. » Poetics 25: 157-175.

Calot, Gérard et al. (1998) Deux siècles d’histoire démographique suisse. Album graphique de la période 1860-2050., Berne et Paris: Office fédéral de la statistique et Observatoire démographique européen.

Cherlin, Andrew J. (1981, 1992 2 ) Marriage, Divorce, Remarriage. Cambridge: Harvard University Press.

Chodorow, Nancy (1978) The Reproduction of Mothering. Berkeley: California University Press.

Coltrane, Scott (2000) « Research on Houdehold Labor: Modeling and Measuring the Social Embeddedness of Routine Family Work ». Journal of Marriage and the Family 62(4): 1208-1233.

Coltrane, Scott (1996) Family Man: Fatherhood, Housework, and Gender Equity. New York: Oxford University Press.

Coverman, S. (1985) « Explaining Husbands’ Participations in Domestic Labor ». The Sociological Quarterly 26: 81-97.

Crompton, Rosemary (2001) « Gender, Comparative Research and Biographical Matching . « European Societies 3: 167-190.

Crompton, Rosemary (ed., 1999) Restructuring Gender Relations and Employment: The Decline of the Male Breadwinner. Oxford: Oxford University Press.

de Singly, François, (ed., 1991) La famille. L’état des savoirs. Paris: La Découverte.

Eichler, Margrit (1973) « Power, Dependency, Love and the Sexual Division of Labour. A Critique of the Decision-Making Approach to Family Power and an Alternative Approach » in: M. Stephenson (ed.) Women in Canada. Toronto: General Publishing: 38-55.

England, Paula & George Farkas (1986) Households, Employment and Gender: A Social, Economic and Demographic View. New York: Aldine DeGruyter.

Ferree, Myra Marx & Eliot R. Smith (1979), « A Cognitive Approach to Social and Individual Stigma ». Journal of Social Psychology 109, 87-97.

Friedan, Betty (1963) The Feminine Mystique. New York: Norton.

Fux, Beat & A. Doris Baumgartner (2001) « Impact of Population Related Policies on Selected Living Arrangements: Comparative Analyses at the Regional Level in Belgium, the Netherlands, and Switzerland ». in: Eric Klijzing & Martine Corijn (eds.) Flagship Conference: Partnership and Fertility a Revolution?, Vol. 2, A Selection of Contributed Papers, Geneva: UNECE.

Giele, Janet Zollinger & Glen H. Elder (eds., 1998) Methods of Life Course Research: Qualitative and Quantitative Approaches. Thousand Oaks/London: Sage.

Goode, William J. (1963) World Revolution and Family Patterns. Glencoe: Free Press.

Held, Thomas (1978) Soziologie der ehelichen Machtverhältnisse. Neuwied: Luchterhand.

Held, Thomas & René Levy (1974) Die Stellung der Frau in Familie und Gesellschaft. Eine soziologische Analyse am Beispiel der Schweiz. Frauenfeld: Huber.

Held, Thomas & René Levy (1983) « Economic Recession and Swiss Women’s Attitudes Towards Marital Role Segregation ». in: Eugen Lupri (ed.) The Changing Position of Women in Family and Society. Leiden: Brill: 373-395.

Hettlage, Robert (1992) Familienreport. Eine Lebensform im Umbruch. München: Beck.

Hoffmann-Nowotny, Hans-Joachim et al. (1984) Planspiel Familie. Diessenhofen: Rüegger.

Höpflinger, François (1986) « Die Wahrnehmung familialen Verhaltens im Paarvergleich. Ergebnisse einer Schweizer Ehepaarbefragung ». Zeitschrift für Soziologie 15(1): 68-74.

Höpflinger, François & Maria Charlers (1990) « Innerfamiliale Arbeitsteilung: mikrosoziologische Erklärungsansätze und empirische Beobachtungen ». Zeitschrift für Familienforschung 2: 87-113.

Hughes, Everett Cherrington (1945) « Dilemmas and Contradictions of Status. » American Journal of Sociology 50: 353-359.

Inglehart, Ronald (1971) « The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies. » American Political Science Review 65(4): 991-1017.

Inglehart, Ronald (1977) The Silent Revolution. Changing Values and Political Styles Among Western Publics. Princeton: Princeton University Press.

Inglehart, Ronald (1990) Culture Shift in Advanced Industrial Society. Princeton: Princeton University Press.

Kellerhals, Jean, René Levy, Eric Widmer, Raphael Hammer & Michèle Ernst (2003, forthcoming) Les couples contemporains. Cohésion, régulation et conflits. Zurich: Seismo.

Kellerhals, Jean, J.F. Perrin, Geneviève Steinauer-Cresson, Laura Vonèche & G. Wirth (1982) Mariages au quotidien. Inégalités sociales, tensions culturelles et organisation familiale. Lausanne: P.M. Favre.

Korpi, Walter (2000) « Faces of Inequality: Gender, Class, and Patterns of Inequalities in Different Types of Welfare States ». Social Politics, 127-191.

Krüger, Helga & René Levy (2001) « Linking Life Courses, Work, and the Family: Theorising a not so Visible Nexus between Women and Men ». Canadian Journal of Sociology 26(2): 145-166.

Künzler, Jan & Wolfgang Walter (1999) « Division of Household Labour. The Role of Welfare States, Gender Roles and Economic Resources ». Paper, ESA Conference, Amsterdam August 18-21.

Laws, Judith Long (1979) The Second X: Sex Role and Social Role. New York: Elsevier.

Levy, René & Michèle Ernst (2002) « Lebenslauf und Regulation in Paarbeziehungen: Bestimmungsgründe der Ungleichheit familialer Arbeitsteilung ». Zeitschrift für Familienforschung 14(2), 103-131.

Levy, René, Dominique Joye, Olivier Guye & Vincent Kaufmann (1997) Tous égaux? De la stratification aux représentations. Zürich: Seismo.

Levy, René (1996) « Toward a Theory of Life Course Institutionalization ». in: Ansgar Weymann & Walter Heinz, Society and Biography. Interrelationships between Social Structure, Institutions and the Life Course. Weinheim: Deutscher Studienverlag: 83-108.

Luhmann, Niklas (1995) Social Systems. Stanford: Stanford University Press.

Maruani, Margaret (1993) L’emploi dans l’Europe des Douze. Bruxelles: Commission des Communautés Européennes.

Mitterauer, Michael (1984) « Die Entwicklung zum modernen Familienzyklus ». In: Michael Mitterauer & Reinhard Sieder (eds.) Vom Patriarchat zur Partnerschaft. Zum Strukturwandel der Familie. München, Beck: 64-91.

Mitterauer, Michael (1979) « Entwicklungstrends der Familie in der europäischen Neuzeit ». In: Rosemarie NaveHerz & Heinz Markefka (eds.) Handbuch der Familien und Jugendforschung, Bd. 1, Familienforschung. Neuwied: Luchterhand: 179-194.

Murphy, M. (1987) « Measuring the Family Life Cycle: Concepts, Data, and Methods ». in: Alan Bryman et al. (eds.) Rethinking the Life Cycle. London: Macmillan: 30-50.

Parsons, Talcott & Robert F. Bales (1955) Family, Socialization and Interaction Pro-cess. Glencoe, Free Press.

Peterson, Gary W. & Suzanne K. Steinmetz (1999) « Introduction: Perspectives on Families as We Approach the Twenty-First Century Challenges for Future Handbook Authors ». in: Marvin B. Sussman, Suzanne K. Steinmetz & Gary W. Peterson, (eds.) Handbook of Marriage and the Family. New York/London: Plenum Press: 1-10.

Pfau Effinger, Birgit (1999) Kultur, Wohlfahrtsstaat und Frauenarbeit im europäischen Vergleich. Leverkusen: Leske+Budrich.

Popenoe, David (1993) « American family decline, 1960-1990: A Review and Appraisal ». Journal of Marriage and the Family 55(3): 527-542.

Roussel, Pierre (1989) La famille incertaine. Paris: Odile Jacob.

Safilios Rothschild, Constantina (1975) « The Dimensions of Power Distribution in the Family ». in: J. Christ & H. Grunebaum (eds.) Marriage Problems and Their Treatment. Little Brown, Boston.

Safilios-Rothschild, Constantina (1970) « The Study of Family Power Structure: A Review 1960-1969. » Journal of Marriage and the Family 32(4): 539-552.

Sardon, Jean-Paul (2000) « La conjoncture démographique: l’Europe et les pays développés d’OutreMer ». Population 4-5: 730-757.

Shelton, Beth Anne (1999) « Gender and Unpaid Work ». in: Janet Saltzman Chafetz (ed.) Handbook of the Sociology of Gender, Dordrecht / Boston / London: Kluwer Academic Publishers: 375-390.

Shelton, Beth Anne & Daphne John (1996) « The Division of Household Labor ». Annual Review of Sociology 22, 299-322.

Strub, Silvia & Tobias Bauer (2002) Wie ist die Arbeit zwischen den Geschlechtern verteilt ? Bern: BASS.

Szinovacz, Maximiliane E. (1987) « Family Power ». in: Marvin B. Sussmann & Suzanne K. Steinmetz (eds.) Handbook of Marriage and the Family. London, New York: Plenum Press: 651-693.

Teachman, Jay D., Lucky M. Tedrow & Kyle D. Crowder (2000) « The Changing Demography of America’s Families. » Journal of Marriage and the Family 62: 1234-1246.

Teachman, Jay D., Karen A. Polonko & John Scanzoni, « Demography and Families ». in: Marvin B. Sussman, Suzanne K. Steinmetz & Gary W. Peterson, (eds.) Handbook of Marriage and the Family. Plenum, New York/London 1999, 39-76.

Vaskovics, Laszlo A. (1994) « Wiederentdeckung familialer Lebenswelten ein Trend? » in: Laszlo A. Vaskovics (ed.) Familie Soziologie familialer Lebenswelten. Soziologische Revue (Sonderheft 3): 4-17.

Widmer, Eric et al. (2003, under review) « Entre standardisation, individualisation et sexuation: une analyse des trajectoires personnelles en Suisse ». Revue suisse de sociologie.

Willemsen, Tineke M. & Menno J. G. Jacobs (2001) « Looking for New Patterns of Work and Care: An International Overview ». paper, Tilburg University.


1. This paper is heavily indebted to the common work with our collaborators Michèle Ernst and Raphaël Hammer; it uses mainly the data produced by our common research project « Social stratification, cohesion and conflict in contemporary families » (Kellerhals et al. 2003), financed by the Swiss National Science Foundation (grant # 5004-47772). It goes without saying that the responsibility for this text and its shortcomings is entirely the authors’. All comments are welcome; please address them to R. Levy, Lemanic institute for life-course studies, University of Lausanne, Bâtiment Provence, CH 1015 Lausanne, Switzerland, or [email protected]

2. A demographic peculiarity with respect to current ideas about the « demographic transition » that Switzerland shares with other Western countries is a marked baby boom during the fourty years starting with World War II (1938-1978), culminating in the middle of the 60ies (Calot et al. 1998).

3. This caustic remark may need an explication. Under the hypothesis that important parts of actual family structures may be more influenced by the institutional environment in which they are embedded than by the partners’ values – we will come to this thesis later – it is easy to imagine that partners who have, say, egalitarian convictions, but are more or less forced to adapt to their context in a non-egalitarian way, may develop, together or – more plausibly– individually, strong tensions that may eventually produce the couple’s dissolution.

4. Certainly one of the oldest hypotheses in this area, starting no later than with sociological functionalism; any quotation would be arbitrary.

5. The feminist literature generally identifies this border as separating the private from the public sphere, securing women’s traditional confinement to the former.

6. Our concept of cohesion designates the way a couple handles its boundaries: internal boundaries between the individual member and the group (mainly in terms of autonomy vs. fusion), and external boundaries between the group and its environment (mainly in terms of closure vs. openness towards the extrafamilial world), and the primary orientation of the group (internal vs. external locus of its main projects or objectives).

7. Although this reference from political sociology is rarely present in family power studies, it may be helpful even if there are important differences between a local community and a family or a couple.

8. A series of cluster analyses of 25 items belonging mostly to four batteries of questions concerning relational roles, instrumental roles, decisional power and routinization produced a five-fold typology combining the components of these dimensions in varying profiles, showing their intricate if not mechanical interrelationships. The presentation and use of this typology is beyond the purpose of the present contribution (see chapter 3 in Kellerhals et al. 2003). According to this analysis, 23% of our couples can be classified as « traditionally gendered », another 23% is relatively close to it. About 36% of the couples have a clearly more egalitarian structure. The female share in family work is clearly highest in the two most inegalitarian types.

9. This couple response rate would correspond to individual response rates of 56% if combined without interdepence (which may or may not be a realistic assumption in our case).

10. Since we explicitly include care, but also in order to stress the existential similarity between occupational and family work, we shall consistently use the latter term instead of housework or the like. This should not be understood as a neglect of the sociologically important differences existing between these two types of work as a function of the very different contexts that organize them; it is on the contrary meant to underline them.

11. We do not systematically compare our results with other Western countries, which would be another study. Tendencies towards more egalitarian forms of organization have already been identified in the 80ies (Höpflinger & Charles 1990), but remain generally modest. Readers familiar with repeated European or larger surveys like the ISSP or Eurobarometer (like one anonymous referee) might wonder why no reference is made to these. Unfortunately, Switzerland started to participate in these two and other internationally comparable surveys only recently, beginning in 1998 (ISSP). Before, there were only two limited participations in the ISSP (1987, 1993). The recent international Family and Fertility Survey has also been conducted in Switzerland, but its questions about family organization are framed in a way that does not allow for comparisons with earlier Swiss studies. The same holds true for the Swiss labour force survey (SAKE) that included modules on unpaid work only in 1997 and 2000, a period too short for our present interests. The best source of information about existing social science data on Switzerland is SIDOS, the Swiss Information and Date Archive Service for the Social Sciences (http://www.sidos.ch).

12. Unfortunately, even semantically very similar items have not always been worded identically. While the first and third source (preceding footnote) used identical wordings of the questions and answers, Hoffmann-Nowotny et al. (1984: 192) had a not quite comparable formulation of answer items. Thus, their figures must be interpreted as very approximative indications with respect to the other surveys. Their question concerned the frequency of men’s helping with five household chores and have been recoded in a rather generous way: « regularly » as man, « frequently » as joint, and « sometimes », « rarely » and « never » as « woman ». Kellerhals et al. (2003) asked for the proportion of all the work being done by the respondent as compared with her/his partner; « practically all » and « about three quarters » are considered to mean that the respective activity is accomplished mainly by the respondent, « half » to mean « joint », « quarter » or less have been equated to the partner. Moreover, the reader should bear in mind that most of these surveys have been conducted on the national level, which implies, in the case of Switzerland, the use of questionnaires in two or three languages (German, French, Italian). Comparisons have therefore to allow for some variation due to purely linguistic reasons; making them depend on the condition of strictly identically wording would simply make no sense. Obviously, the same holds true for the English captions in the tables of this paper which have in fact never been used as such in these surveys.

13. The sources are, by year of survey: 1971: Held & Levy 1974; 1980: Hoffmann-Nowotny et al. 1984; 1991: Levy et al. 1997; 1998: Kellerhals et al. (2003). The sample of the 1980 survey (Hoffmann-Nowotny et al.) was restricted to relatively young, married Swiss couples, i.e., national couples married between 1970 and 1980, and living in the two major language regions (German and French, without the Ticino, the Italian-speaking part of the country which tends to be rather traditionalistic in family matters and accounts for 8% of the total population). The three others included the adult age range between 20 and 70 years and were conducted in all linguistic regions of the country; the 1971 sample was representative of married couples (which left most probably only very few unmarried couples out, given the period), the 1991 one included individuals without consideration of their marital status, but the relevant questions were only asked to the subsample of people living together with a partner; the 1998 sample, also, included heterosexual couples living together since at least one year without consideration of their being married or not. Given all these differences and the N, calculating significance tests does not seem to be adequate. We cannot statistically exclude that the sampling differences account for some part of the percentage differences we wish to interpret, but we feel that there can only be speculations about their influence (especially, the 1980 sample may be biased towards less traditionalism since it excludes the Italian-speaking minority and older couples). As a safeguard, we only take into consideration reasonably large differences over the whole period.

14. Nevertheless, Strub & Bauer (2002) indicate that retrospective self-report questions, compared to time-budget accounts, underestimate the number of weekly hours for family work by about three for women and overestimate it by about three for men, resulting in a serious underestimation of women’s share. However, since our interest here is not in precise statistics about time volumes, but in interrelations between partners’ activities, this problem needs not be addressed specifically in our further analyses.

15. Strub & Bauer (2002) find some rather modest decreases in inequality for a recent and short period, 1997 through 2000, and some indications going in the same direction for a longer period (1980-2000), for which, however, the available data lack the appropriate precision.

16. The items can be found in table 1. Let us repeat that according to Strub & Bauer (2002) our figures probably underestimate the extent of inequality.

17. In the same vein, for the USA, Cherlin (1981, 1992, p. 58) reports a strong decrease between 1970 and 1989 of the acceptance of statements like « It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family » (ca. 80% -> 25%), or « A preschool child is likely to suffer if his or her mother works », and a corresponding increase of « A working mother can establish just as warm a relationship with her children as a mother who does not work ».

18. Such a test had not been planned at the outset. As some answer distributions (concerning several aspects of cohesion and regulation) provoked a suspicion of normative bias, we had to improvise in order to control for such effects at least approximately. The logic used consisted in counting among all the items in each of five answer batteries the supposedly most normative answer and giving a respondent one « point of conformism » if he/she gave consistently the most normative answer to all the items of the same battery. We used various forms of this logic according to the variables in the analysis in order to avoid tautologies. We did find confirmations of a desirability bias, but it is restricted to simple, one-dimensional answer displacements. The answer distributions of several of our dimensions of family functioning turned out to be influenced by normative conformism. However, the relations we found when analyzing these dimensions, e.g., concerning their association with life course phases or status va-riables (mainly, regression coeffi-cients), where not affected.

19. Same sources as table 1.

20. Asking directly could be another way of avoiding this problem. Our questionnaire included the outright question « All well considered and frankly speaking: who is the boss in your couple? ». This way of ques-tioning can be assimilated to the reputational approach to power identification. The question allowed for two non-typed answers, « both equally » and « none ». The answers showed again the tendency toward equality, combined with a clear leaning toward male dominance.

21. In table 3 and the following analyses, the two evaluations of each descriptive statement (OK, resent) have been collapsed as this dimension is not of direct interest here. It goes without saying that the answers indicating dependence have been inverted for one partner (equating his declaring of his dependence with her declaring of her partner’s dependence and the reverse) so that the ? can really be interpreted.

22. In the first category, 86% of the men and 70% of the women are retired. About half of the remaining men (5.4%) are unemployed, the other half are ill or handicapped (6.0%). The second reason of nonemployment of the female partners is housework or children at home. In the atypical category, 37.9% of the men are not employed, 54.7% of them because of retirement, 16.0% because of unemployment, and 20% because of illness or handicap. The other two thirds of the men in these couples work part-time, i.e., their employment rate is less than 90%.

23. Retrospective biographical data, otherwise not used in this paper (cfr. Widmer et al. 2003), confirm this statement: about 85% of our male respondents have employment trajectories that clearly correspond to the classical 3-phase model: education, full-time employment, retirement. Only about 15% have trajectories with interruptions or spells of part-time work. The female trajectories fall into four patterns (and a tiny residual category od 4%) the most frequent of which counts 34% of all the cases. This pattern comes closest to the traditional model. It consists of a sequence of initial education followed by a rather short period of full-time employment, followed in turn by a long period of full-time family work, and no subsequent return to the labour market. We can add that age differences appear to intervene only very specifically in these patterns, their main impact being among women who stay in the labour market, in that the older do so more often through part-time work, while the younger more often maintain a full-time employment. Unfortunately, these truly longitudinal data cannot be used for the analyses of this article since it would make no sense to ask people in later phases of their lives about their earlier work load for the family and employment, expressed in hours, for periods that may lie many years back.

24. This does of course not exclude that many of these women maintain their employment for other than strictly monetary reasons, such as autonomy, intrinsic work motivation, or the intention to increase their occupational activity later on.

25. The income indications are graded into 11 classes.

26. It is also interesting to note that while the proportions of the correlations are roughly the same for the two estimates of the household income, the correlation of the individual income with the person’s own indication of the household income is higher than with the partner’s indication, possibly indicating a higher coherence « inside » each partner’s perceptional universe.

27. The men’s answers show the same pattern, but supplanted by an additional tendency to name slightly more often the man and less often the woman as the one who would adjust. Nevertheless, overall the woman is clearly preferred for this function by both sexes.

28. It may not be safe to interpret the difference between the two symetric cases of one-sided dependency, but the difference between the 34% in independent couples and 59% in « symetrically » interdependent couples is striking. As we know nothing about the semantic or conceptual difference between the two intermediate situations, an ordinal correlation coefficient is problematic (let us nevertheless note that gamma would be ?=.24); Cramer’s V = .12 (highly significant).

29. In interpreting these results, we should not forget that about 17% of the men in our sample are retired whereas 75% (or 90% of the non-retired) are full-time employed.

30. This interpretation is confirmed by results of Strub & Bauer (2002) and comparable ones from seven other European countries (Willemsen & Jacobs 2001), showing, among others, relatively constant and similar overall working hours for men as well as for women living in couples when family work and occupational work are added up.

31. For a discussion of American studies of the relationship between spouses’ occupational and family work hours, see, e.g., Shelton 1999. Our main result is consistent with her summary statement that « paid work time has a greater effect on women’s than on men’s household labor » (p. 383). Let us add that our results give only limited credibility to the time availability approach. Quite substantial and problematic additional suppositions about the differential « urgency » of employment and family activities would be needed to derive such results from this explanatory model. See also Bielby & Bielby (1989).

32. The age distribution of the question whether childless respondents planned to have some later showed no single positive response above 36 years of age (women), a limit that seems to correspond to a wide-spread perception of a biological deadline (« biological clock »). We use this age limit to distinguish childless couples (woman over 36) from pre-child couples.

33. A comparison of the parallel relationship of male and female employment rates with age and with family phases shows quite clearly that male rates are more strongly related to age, with the official retirement age (65) as a point of inflexion, whereas female rates are more strongly related to the family phases. According to various checks that cannot be shown here for reasons of space, there are no clear-cut indications of cohort differences within family phases.

34. The relatively high rate of atypical employment constellations in the two initial phases and the last one is largely due to the man having not yet completed his initial education for the first two, and being retired for the last. Couples without children have been added to this table (while not present in the graphs) to give the possibility to compare; this group is defined as having no children on both sides and the woman’s age being over 36. As already stated, these couples have an extremely low probability to have a child.

35. The median hours in childless couples are 36.4 for women and 6.2 for men, with a ratio of 5.9.

36. Men’s Model B has been chosen to be a mirror image of women’s Model B. Since these regression analyses are not our main interest, we do not systematically compare all logically possible configurations of model variables.

37. A multiple regression analysis, published in another paper (Levy & Ernst 2002), confirms this intepretation.

38. Let us note in passing that resource theory would suggest a negative correlation, since it would predict that on the basis of his financial contribution to the satisfaction of the couple’s needs, a full-time employed man (and more so if his income is high, which we did not include) would be able to get a complete discharge from the less prestigious family work. Put cynically, we could say that this discharge exists anyway quite generally and seems to need no backing by specific resource contributions. Nevertheless, some other results tend to confirm resource theory, especially the association between women’s education and their employment rate (and this rate’s resilience « against » the family phases).

39. This is a reference to findings of a previous study on stratification and mobility (Levy et al. 1997) not presented here. Our findings correspond to German results and are also confirmed by Crompton (2001) for other European countries (Britain, Norway). We hypothesize that the specific form, variability and extent of sex-specific career models depend on societal gender regimes that in turn vary according to each society’s institutional profile and its respective basic social-political orientations (such as those identified by Korpi 2000).

40. More generally, we feel that in the complex system of interdependencies of gender, family organization, life course and social inequalities, it makes little sense to expect any single theory to be clearly more relevant than its contenders. It seems much more fruitful to seek to establish which of the various mechanisms identified by different theoretical approaches are relevant for which element or under which circumstances, since they appear to be complementary rather than mutually exclusive.

41. This interpretation is partly in tune with the theoretical tendency Coltrane (2000) calls « institutional influences », except for the fact that we establish a direct connection between the dimension of the (fairly institutionalized) life-course of individuals and the family which he treats as a separate aspect.