Jump to ContentJump to Main Navigation
Dividing the DomesticMen, Women, and Household Work in Cross-National Perspective$

Judith Treas and Sonja Drobnič

Print publication date: 2010

Print ISBN-13: 9780804763578

Published to Stanford Scholarship Online: June 2013

DOI: 10.11126/stanford/9780804763578.001.0001

Show Summary Details
Page of

PRINTED FROM STANFORD SCHOLARSHIP ONLINE (www.stanford.universitypressscholarship.com). (c) Copyright Stanford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in SSO for personal use (for details see www.stanford.universitypressscholarship.com/page/privacy-policy). Subscriber: null; date: 18 December 2018

Beliefs about Maternal Employment

Beliefs about Maternal Employment

Chapter:
(p.147) Chapter Eight Beliefs about Maternal Employment
Source:
Dividing the Domestic
Author(s):
Maria Charles, Erin Cech
Publisher:
Stanford University Press
DOI:10.11126/stanford/9780804763578.003.0008

Abstract and Keywords

This chapter describes the variability in women's attitudes toward maternal employment across 32 industrial, transitional, and developing countries and territories. Drawing on surveys for these countries, this chapter demonstrates the cross-national variation in public opinion regarding maternal employment. Exploring this variability can help contextualize the ideologies of motherhood and can provide insights on the ideals of full-time maternal care for children in the home.

Keywords:   maternal employment, cross-national variation, public opinion, motherhood, maternal care

Beliefs about the needs of children and about appropriate modes of mothering exert a powerful moral force in American society, influencing parents' perceived options, their feelings of self worth, and ultimately their decisions about how to organize work and family lives. The dominant American ideal, recently dubbed “intensive motherhood” (Hays 1996), represents children as vulnerable and precious beings whose proper care demands an extraordinary commitment of time, emotion, and nurturance by one primary caregiver, preferably their own mother (Phoenix, Woollett, and Lloyd 1991; Zelizer 1985). Like other cultural ideologies, intensive motherhood derives much of its power from its naturalized, taken-for-granted quality.

Although social expectations and personal feelings of extreme maternal devotion are pervasive, the appropriate enactment of this devotion remains the subject of much emotionally charged controversy in the contemporary United States. One particularly intensive variety of motherhood requires that women forgo paid employment and dedicate themselves exclusively to their maternal roles. From this perspective, women's interest in paid employment conflicts with children's need for around-the-clock mothering. This model of full-time maternal care is rejected on principle by many women, especially the young, the highly educated, and the nonreligious, and it is unfeasible for the many others who lack the requisite financial or personal resources. However, even those American women who do not actively embrace ideals of full-time mothering are cognizant of this cultural schema and know that others may hold them morally accountable to it (Blair-Loy 2004; Hochschild 2003; Lareau 2003; Taylor 1996). It is in fact not uncommon for social movement activists, policy makers, religious leaders, and public intellectuals to represent this as the only ethical or socially viable form of mothering.

But although specific practices may be naturalized within local contexts, previous research shows that norms about maternal care, familial gender roles, and child-rearing styles vary a great deal across time, space, and demographic groups (Ariès 1965; Evans and Kelley 2001; Kremer 2006; Fuller 2007; (p.148) Gottlieb 1993; Lareau 2003). Comparativist scholars have suggested that ideals of motherhood are influenced by national institutional structures and social policy arrangements that legitimize certain family forms and make certain child care arrangements more or less viable.1 Exploring such variability can help contextualize dominant cultural ideals and practices, and provide insights regarding the individual and structural factors that are associated with alternative conceptualizations of motherhood. Normative understandings of motherhood are important because they help shape women's public sphere roles and children's early life experiences, which in turn influence other divisions of domestic labor.2

The primary goal of the current analysis is to provide a detailed description of variability in women's attitudes toward maternal employment across 32 industrial, transitional, and developing countries and territories. We focus on women's views because we believe that they provide an especially good measure of the normative penetration of ideals of full-time maternal care. Application of multivariate logistic regression models allows us to examine international differences after controlling for demographic factors that are known to be correlated with gender role attitudes (i.e., women's educational attainment, employment status, religiosity, age), and to determine the extent to which observed patterns of cross-national variation map on to standard classifications of welfare state regimes.

Although we do not have the comparative data necessary to assess arguments about attitudinal effects of specific policy measures or social expenditures formally, we do draw some preliminary conclusions about these relationships based on information drawn from secondary sources. Results are consistent with arguments suggesting that institutional and social policy structures help shape cultural understandings of motherhood and childhood, and that these understandings in turn help sustain path-dependent trajectories of policy development (Esping-Andersen 1990, 1999; Mahon 2002a; Morgan 2006; Orloff 1993; Sainsbury 1999b). We consider implications for mainstream and feminist theories of the welfare state.

Data, Methods, and Descriptive Statistics

Data are from the 2002 Family and Gender Roles III module of the International Social Survey Programme (ISSP). National surveys were administered between 2001 and 2004. Three countries (Bulgaria, Russia, and Taiwan) were omitted because of missing information on religiosity, an important demographic variable. Countries considered are predominantly advanced industrial market economies, but our sample also includes seven formerly socialist countries of eastern and central Europe, and four middle-income developing countries. Results presented here are based on unweighted samples.3

(p.149) We consider cross-national variability in support for full-time mothering based on two indicators: (1) respondents' beliefs that mothers should not be employed “when there is a child under school age” and (2) respondents' beliefs that mothers should not be employed “after the youngest child starts school.”4 These are the most direct and unambiguous available indicators of belief in the desirability of full-time motherhood, an economically costly form of child rearing that depends upon a full-time breadwinner.5

Table 8.1 shows country scores on indicators of support for full-time maternal care. Values give the proportion of a country's respondents who agree with the given statement. Much cross-national variability is evident. With respect to mothers of preschool children, Israel and New Zealand occupy extreme ends of the distribution: 10% of women in Israel agree that mothers with very young children should stay home, compared with a full 63% of women in New Zealand. In all countries there was more support for full-time mothering of younger than older children, and the view that women with school-age children should stay home was held by only a minority of women (15% on average compared with an international average of 38% with respect to preschoolers). Nonetheless, we also find large cross- national differences regarding care of older children, with values ranging from 2% in Denmark, the Netherlands, and Sweden to 32% in the Philippines. In the United States, support for full-time maternal care is slightly above the international mean with respect to preschool children (42%), but below the mean with respect to school-age children (5%).

The low values shown in Table 8.1 for Sweden and Denmark are unsurprising in light of the reputably gender-egalitarian cultures characterizing Scandinavian welfare states. Other values, particularly the high percentages of women advocating stay-at-home mothering of preschoolers in many English-speaking countries, were unexpected. Similar patterns of cross-national variability were found for a sample that included male respondents.6

Before further consideration of these international differences, it is useful to determine whether they can be attributed to differences in the sociodemographic composition of the respective national populations. Past research suggests that support for gender-differentiated family roles is greater among older persons, highly religious persons, and those without college degrees (Knudsen and Wærness 2001; Morgan 2006; Sundstöm 1999). The stronger support for maternal employment in Nordic countries, for example, may be partly attributable to low levels of religiosity or high levels of educational attainment in those societies.

To understand better the nature of observed cross-national differences, we compute a set of multivariate logistic regression models. This allows us to calculate, for each country, the predicted probability of espousing full-time maternal care while holding constant the relevant individual-level attributes.7 (p.150)

TABLE 8.1 Means on attitudinal variables: Women, 2002

Country [Regime]

Mothers of Preschoolers Should Stay Home

Mothers of School Children Should Stay Home

Australia [L]

0.55

0.05

Austria [C]

0.45

0.16

Belgium (Flanders) [C]

0.26

0.05

Brazil [D]

0.34

0.19

Chile [D]

0.44

0.18

Cyprus [C]

0.18

0.10

Czech Republic [S]

0.35

0.16

Denmark [SD]

0.21

0.02

Finland [SD]

0.38

0.07

France [C]

0.37

0.05

Germany, East [S]

0.12

0.04

Germany, West [C]

0.44

0.13

Great Britain [L]

0.55

0.06

Hungary [S]

0.43

0.14

Ireland [L]

0.35

0.08

Israel [L]

0.10

0.07

Japan [C]

0.52

0.14

Latvia [S]

0.39

0.13

Mexico [D]

0.39

0.31

Netherlands [C]

0.27

0.02

New Zealand [L]

0.63

0.05

Northern Ireland [L]

0.51

0.22

Norway [SD]

0.28

0.05

Philippines [D]

0.48

0.32

Poland [S]

0.54

0.22

Portugal [C]

0.38

0.12

Slovak Republic [S]

0.49

0.19

Slovenia [S]

0.25

0.10

Spain [C]

0.35

0.14

Sweden [SD]

0.18

0.02

Switzerland [C]

0.40

0.30

United States [L]

0.42

0.05

Liberal

0.45

0.08

Conservative

0.37

0.11

Social-democratic

0.27

0.04

Formerly socialist

0.39

0.15

Developing

0.39

0.27

All

0.38

0.13

Data are taken from the 2002 wave of the International Social Survey Program (ISSP).

Means can be interpreted as the proportion of women agreeing with the corresponding statement.

C, conservative regime; D, developing regime; L, liberal regime; S, formerly socialist regime; SD, social-democratic regime.

The explanatory variables included in our regression models were selected based on results of previous attitudinal research on motherhood and gender roles. Age is measured in years, and all other covariates are “dummy coded,” with the value 1 indicating presence of the corresponding attribute (p.151) and the value 0 indicating its absence. Highly religious persons are defined as those attending religious services at least once a week. Those coded 1 on the “university degree” and “employed” variables reported having completed a university degree and having a paid job, respectively.8 Respondents coded 1 on the “married” variable reported being married or “living as married.” The presence of one or more children at home is coded from information on household composition.9 To determine whether employment differentially affected attitudes for women with and without children at home, we include an interaction term, “child × employed,” in our models. Information on the employment histories of respondents' own mothers was used to construct an additional dummy indicator: Did your mother ever work for pay for as long as 1 year, after you were born and before you were 14?10

Means for all covariates, pooled and by country and regime, are shown in Table 8.2. For all variables but age (measured in years), means can be interpreted as the proportion of respondents with the respective attribute (i.e., the proportion coded 1). On the religiosity variable, for example, the pooled mean of 0.24 indicates that 24% of all respondents attend services once a week or more. In Denmark, the corresponding value is only 2%; in Mexico and Ireland, it is 65%.

What Predicts Support for Full-Time Maternal Care?

Table 8.3 shows results from a series of multivariate logistic regression models where the dependent variables are the two attitudinal measures described earlier. Values give covariate effects on the logged odds of holding the belief in question; the exponent of these values give the multiplicative change in the odds corresponding to a 1-unit increase in the covariate.11 For example, according to the “age” coefficient in the first column, the log-odds of advocating full-time maternal care of preschoolers increases by 0.015 point as respondents' ages increase by 1 year. The exponent of this value (exp(0.015) = 1.015) tells us that the odds of holding this belief grows by 1.5% with each year of age.

Three models are presented for each attitudinal indicator. The first (model a) includes individual-level sociodemographic covariates only, the second (model b) adds variables identifying country of residence (with the United States the omitted reference category), and the third (model c) includes individual effects and variables representing five “welfare regime” types (liberal, conservative, social–democratic, formerly socialist, and developing, with liberal the omitted reference category). Our classification of welfare regimes, shown in Table 8.3, is discussed later. (p.152)

TABLE 8.2 Means on independent variables: Women, 2002

Country [Regime]

University Degree

Highly Religious

Employed

Child in Household

Child × Employed

Married

Age (y)

Mother Had Job

Australia [L]

0.24

0.20

0.52

0.41

0.23

0.66

46.77

0.42

Austria [C]

0.08

0.24

0.46

0.38

0.23

0.48

46.27

0.51

Belgium (Flanders) [C]

0.06

0.13

0.48

0.33

0.24

0.62

48.86

0.35

Brazil [D]

0.06

0.49

0.37

0.69

0.25

0.43

39.27

0.52

Chile [D]

0.10

0.11

0.59

0.33

0.20

0.45

49.14

0.45

Cyprus [C]

0.17

0.14

0.62

0.48

0.37

0.68

40.55

0.52

Czech Republic [S]

0.07

0.07

0.58

0.40

0.24

0.59

43.15

0.89

Denmark [SD]

0.09

0.02

0.61

0.36

0.30

0.55

45.66

0.58

Finland [SD]

0.14

0.05

0.58

0.38

0.23

0.67

44.18

0.58

France [C]

0.27

0.07

0.64

0.51

0.36

0.55

41.86

0.57

Germany, East [S]

0.11

0.05

0.42

0.30

0.19

0.64

49.67

0.85

Germany, West [C]

0.10

0.12

0.42

0.32

0.15

0.60

46.74

0.50

Great Britain [L]

0.14

0.14

0.54

0.33

0.23

0.53

48.69

0.58

Hungary [S]

0.04

0.14

0.35

0.29

0.16

0.49

51.42

0.61

Ireland [L]

0.13

0.65

0.46

0.45

0.25

0.57

44.26

0.25

Israel [L]

0.26

0.17

0.56

0.55

0.33

0.67

42.08

0.56

Japan [C]

0.07

0.06

0.41

0.42

0.17

0.70

49.33

0.58

Latvia [S]

0.19

0.04

0.57

0.48

0.31

0.47

43.80

0.89

Mexico [D]

0.12

0.65

0.44

0.62

0.29

0.58

40.83

0.29

Netherlands [C]

0.27

0.09

0.57

0.37

0.26

0.51

43.90

0.34

New Zealand [L]

0.16

0.16

0.62

0.39

0.27

0.61

48.64

0.44

Northern Ireland [L]

0.14

0.47

0.42

0.38

0.20

0.44

49.43

0.34

Norway [SD]

0.26

0.03

0.58

0.42

0.27

0.56

44.57

0.54

Philippines [D]

0.01

0.58

0.37

0.81

0.32

0.76

38.50

0.42

Poland [S]

0.11

0.59

0.39

0.48

0.23

0.56

48.65

0.57

Portugal [C]

0.11

0.38

0.46

0.35

0.22

0.56

48.97

0.45

Slovak Republic [S]

0.07

0.45

0.55

0.42

0.30

0.61

43.63

0.71

Slovenia [S]

0.08

0.21

0.42

0.37

0.22

0.57

47.78

0.51

Spain [C]

0.07

0.26

0.38

0.34

0.16

0.56

46.92

0.31

Sweden [SD]

0.27

0.03

0.64

0.36

0.28

0.51

46.78

0.55

Switzerland [C]

0.06

0.26

0.36

0.66

0.26

0.49

44.23

0.43

United States [L]

0.22

0.37

0.59

0.36

0.22

0.48

45.51

0.63

Liberal

0.18

0.30

0.53

0.40

0.24

0.56

46.59

0.47

Conservative

0.13

0.17

0.50

0.39

0.24

0.56

46.02

0.46

Social-democratic

0.19

0.03

0.60

0.38

0.27

0.58

45.20

0.56

Formerly socialist

0.09

0.24

0.48

0.40

0.24

0.56

46.49

0.71

Developing

0.07

0.49

0.38

0.69

0.28

0.55

40.79

0.42

All

0.13

0.24

0.50

0.44

0.25

0.56

45.37

0.51

Data are taken from the 2002 wave of the International Social Survey Program (ISSP). For all variables but age, means can be interpreted as the proportion of women with the corresponding attribute.

C, conservative regime; D, developing regime; L, liberal regime; S, formerly socialist regime; SD, social-democratic regime.

Effects of Individual-Level Attributes

We turn first to the specifications that include sociodemographic variables only (models Ia and IIa). Higher education greatly reduces women's support for full-time maternal care, whereas personal religiosity increases support. Having a university degree, for instance, more than doubles the odds that a woman will find it acceptable for mothers with school-age children to work for pay.12 High religiosity—specifically, attending religious services at least (p.153)

TABLE 8.3 Espousal of full-time maternal care among women in 32 countries

Variable

Mothers of Preschoolers Should Stay homec

Mothers of School Children Should Stay homed

Sociodemographic

Model Ia

Model Ib

Model Ic

Model IIa

Model IIb

Model IIc

    University degree

−0.601***

−0.608***

−0.616***

−0.878***

−0.660***

−0.722***

    Highly religious

0.327***

0.258***

0.263***

0.541***

0.160**

0.284***

    Employed

−0.291***

−0.392***

−0.295***

−0.664***

−0.612***

−0.584***

    Child in household

0.297***

0.252***

0.299***

0.412***

0.081

0.130

    Child × employed

−0.365***

−0.290***

−0.364***

−0.286**

−0.154

−0.179

    Married

0.039

0.061

0.048

0.058

0.145**

0.122*

    Age (y)

0.015***

0.013***

0.015***

0.015***

0.019***

0.018***

    Mother had job

−0.133***

−0.210***

−0.140***

−0.232***

−0.272***

−0.249***

Countrya

    Australia

0.478**

0.173

    Austria

−0.035

1.212***

    Belgium (Flanders)

−0.949***

−0.264

    Brazil

−0.615***

1.415***

    Chile

−0.380**

2.027***

    Cyprus

−1.173***

0.857**

    Czech Republic

−0.292*

1.444***

    Denmark

−1.070***

−1.194**

    Finland

−0.311*

0.236

    France

−0.062

0.338

    Germany, East

−1.834***

−0.667

    Germany, West

−0.096

0.896***

    Great Britain

0.500***

0.104

    Hungary

−0.248*

0.842***

    Ireland

−0.627***

0.319

    Israel

−1.853***

0.507

    Japan

0.285*

0.945***

    Latvia

0.019

1.289***

    Mexico

−0.385**

2.136***

    Netherlands

−0.691***

−1.372**

    New Zealand

0.827***

−0.066

    Northern Ireland

0.039

1.529***

    Norway

−0.595***

−0.114

    Philippines

−0.056

2.169***

    Poland

0.225

1.536***

    Portugal

−0.336**

0.897***

    Slovak Republic

0.215

1.540***

    Slovenia

−1.057***

0.556*

    Spain

−0.557***

0.924***

    Sweden

−1.219***

−1.070*

    Switzerland

−0.095

1.375***

Regimeb

    Conservative

−0.299***

0.343***

    Social-democratic

−0.731***

−0.778***

    Formerly socialist

−0.214***

0.716***

    Developing

−0.348***

1.412***

Constant

−1.072***

−0.585***

−0.770***

−2.535***

−3.518***

−3.093***

G2 (df)

1078.64 (8)

2139.41 (39)

1232.07 (12)

977.51(8)

1924.29(39)

1529.47 (12)

BIC

−1000.03

−1756.20

−1114.16

−898.89

1541.03

−1411.54

(a) United States is the reference category for models Ib and IIb.

(b) Liberal regime is the reference category for models Ic and IIc.

(c) N = 18,506

(d) N = 18,530

Values are additive coefficients from logistic regression models. G2 gives the difference in −2LL between the respective model and the constant-only model. Except for age, all covariates are dummy coded, with 1 indicating presence of the respective attribute and 0 indicating its absence.

(*) p 〈 0.05,

(**) p 〈 0.01,

(***) p 〈 0.001.

BIC = −G2 + df(ln(N)).

(p.154) once a week—works in the opposite direction, significantly increasing belief in full-time maternal care for both preschoolers and school-age children. This may mean that different understandings of childhood and/or motherhood are inculcated within university and religious institutions, that educational and religious institutions attract persons holding different ideological beliefs concerning the role of women in society, or both. These findings are consistent with those from previous American and international research on gender role ideologies (Evans and Kelley 2002; Knudsen and Wærness 2001).

As expected, effects of motherhood on attitudes depend upon current employment status. For example, the odds of advocating stay-at-home mothering of preschoolers are nearly twice as high among nonemployed mothers as among employed mothers.13 Although these cross-sectional data do not allow us to assess the causal mechanisms underlying this relationship, we presume that it is bidirectional (i.e., that women's attitudes about appropriate maternal practices influence their propensity to combine motherhood with paid employment, and that women's experiences as either employed or stay-at-home mothers influence their attitudes about maternal employment). Among childless women, employment reduces the odds of espousing full-time maternal care. Being married seems to make little difference to women's beliefs about appropriate child-rearing practices.14

Also consistent with past research, we find that younger women and women whose own mothers were employed are generally more supportive of combining market work with motherhood (Brayfield, Jones, and Alder 2001; Knudsen and Wærness 2001).

To assess possible differences by country in the predictors of beliefs about appropriate mothering, we also ran a set of 32 single-country models for each dependent variable. These models (not shown) yield slope coefficients that are either consistent in direction to those displayed in Table 8.3 or statistically nonsignificant. We thus find little evidence of important cross-national differences in these individual-level relationships.

We now return to our earlier question concerning the compositional dependence of cross-national differences in normative understandings of motherhood. In particular, we wish to know whether the differences revealed in Table 8.1 persist after we have taken into account international differences in the sociodemographic variables discussed earlier. For example, are women with similar values on age, education, religiosity, employment status, and other relevant attributes more likely to support full-time maternal care if they live in New Zealand than if they live in Israel?

How Do Norms of Motherhood Differ by Country?

Models Ib and IIb in Table 8.3 allow us to examine cross-national differences while holding sociodemographic attributes constant. This information (p.155) can be garnered from the strength and direction of the 31 dummy “country” indicators. Because the omitted reference category is the United States, a positive value on a particular country term indicates a stronger propensity for women to hold this belief in the respective country than in the United States; a negative value indicates a weaker propensity.

Country effects shown in Table 8.3 reveal rank orderings that are roughly consistent with those for the unadjusted percentages displayed in Table 8.1. We find, for instance, the same set of highest and lowest scoring countries on both dependent variables. This result tells us that the international differences described earlier are not simple artifacts of differences in respondents' demographic attributes, but hold even net of such compositional differences. Cross-national variability is substantial even if we consider only advanced industrial countries. The odds of supporting full-time mothering of preschoolers are, for instance, more than three times higher among U.S. women than among their demographically comparable counterparts in Sweden.15

How can we explain these considerable attitudinal differences even among countries with similar traditions of feminist mobilization and comparable levels of postindustrial development and affluence? Scholars of welfare state and gender regimes have suggested that ideals of good motherhood and beliefs about children's needs are influenced by prevailing institutional and policy arrangements that set the context for family decisions and make certain child care arrangements more or less viable (Esping-Andersen 1999; Kremer 2006; Mahon 2002a, b; Pfau-Effinger 2000, 2004; Sainsbury 1999b). When consolidated, public policy regimes gain constituencies and may become inscribed in national culture. Existing family arrangements and systems of public provision thereby come to be taken for granted as natural and desirable. In the following section we consider the extent to which ideals of maternal care cohere within categories of standard welfare state classification.

Exploring Regime Effects

Welfare state regimes refer to the different ways in which welfare production is allocated between the state, the market, and the family. The most influential typology, which distinguishes three basic regime types (liberal, social–democratic, and conservative), was put forward by Danish sociologist Gøsta Esping-Andersen in his 1990 classic, The Three Worlds of Welfare Capitalism. Policy logic in liberal regimes gives primacy to the market, with individuals and families expected to care for themselves or purchase services. Benefits are modest, means tested, and aimed at a small group of welfare recipients.16 Nonparental child care is provided mostly by low-wage workers employed in private-sector markets. The liberal cluster is primarily comprised of Anglo-Saxon countries, with the United States representing (p.156) the ideal/typical case. In social–democratic regimes, found in Sweden and other Scandinavian countries, taxes are high, benefits are universal and are paid to individuals, and state policies aim to promote equality at a relatively high material level (as opposed to covering minimal needs). Policies are based on “universal earner” principles, with the aim of promoting full employment of women and men in part through heavy state investment in child care services and parental leave allowances. Conservative regimes are mostly continental European countries, with (West) Germany as the ideal/ typical case. Labor force participation is an important source of entitlement in conservative regimes because benefits are generally conferred on heads of households and are often linked to occupational status and earnings. Policies aim to temper negative effects of unfettered market competition through transfers to families. High male wages, employment-based pension funds, and progressive joint taxation promote family divisions of labor between a full-time domestic caregiver and a full-time breadwinner.

Although Espmg-Anderson's (1990) analysis focused on advanced capitalist societies, the data collected for the ISSP allow us to examine norms of motherhood in formerly socialist and developing economies as well. We therefore add two categories to Esping-Andersen's original three. Our fourth “regime” is comprised of seven eastern and central European countries that share a communist past and a recent postsocialist market transition. Marxist doctrine, with its avowed allegiance to principles of gender equality and even the “withering away” of the family, provided a strong ideological basis for state efforts to facilitate female employment through provision of child care and other services. However, resource shortfalls and precommunist legacies of care often meant considerable gaps between ideological commitment and actual services (Michel 2006).

The fifth regime type considered here is comprised of four developing countries (Brazil, Chile, Mexico, and the Philippines). Because this group is restricted to middle-income, largely Roman Catholic societies—three Latin American and one East Asian—it can by no means be regarded as representative of all developing countries. However, these four do share with other developing countries a relatively small tax base and a limited state capacity to buffer risk and provide social welfare services to individuals and families. Moreover, all of these countries have been exposed, since at least the 1980s, to pressures from international financial institutions (e.g., The World Bank, International Monetary Fund) for neoliberal restructuring of their economies and a reduced state role in welfare provision. Child care policies in developing countries are also influenced by concepts of early care and education propagated by such multilateral organizations as the United Nations' Educational, Scientific, and Cultural Organization (or UNESCO).17

(p.157) Since publication of Esping-Andersen's (1990) Three Worlds, numerous alternative classification systems have been proposed by welfare state scholars and feminist theorists. Although compelling arguments have been advanced that France, Belgium, the Netherlands, the Mediterranean region, the Antipodes, and/or Japan constitute distinct national types (Blossfeld and Drobnic 2001; Castles 1996; Huber and Stephens 2001; Jones 1993; Leibfried 1992; Miyamoto 2003; O'Connor 1999), we provisionally follow Esping-Andersen (1997, 1999) in assigning all of these to the conservative category.18 Israel has been described as a hybrid of liberal and social–democratic regime types (Sabbagh, Powell, and Vanhuysse 2007). We classify it as “liberal” based on observations by Ben-Arieh, Boyer, and Gajst (2004). We will assess arguments about the cultural distinctiveness of these and other national societies by examining within-regime variability in beliefs about maternal employment. Our operational classification of countries by regime type can be seen in Tables 8.1 and 8.2.

Models Ic and IIc in Table 8.3 allow us to ascertain the extent to which patterns of cross-national variability in support for full-time maternal care map on to our classification of regime types. Effects are measured relative to the liberal category. For example, the value 0.343 in the last column tells us that the odds that women will advocate full-time maternal care of schoolage children are about 41 percent greater in conservative than in liberal regimes (exp(0.343) = 1.409), holding constant other variables. All regime effects are statistically significant at the 0.001 level.

Consistent with the predictions of standard welfare state regime theories, results show that women in social–democratic countries are least likely to advocate full-time maternal care of children in either age group. We suspect that a strong state role in harmonizing family and work obligations, subsidizing costs of quality child care, and disseminating alternative normative conceptualizations of childhood and motherhood contributes to an erosion of support for full-time mothering of preschoolers in social–democratic regimes. In other words, state policies have helped propagate a normative “universal worker” ideal and a less mother-centered understanding of childhood.

In liberal welfare states, normative models of motherhood appear to depend strongly on children's ages. Employment of mothers with school-age children is deemed relatively unproblematic in liberal regimes, with only social–democratic regimes showing weaker support for full-time mothering in that age group. However, we were surprised to find that it is in the liberal, not conservative, regimes where women are most likely to believe that mothers with preschool children should “stay home.” Cultural tendencies for individualistic attribution of preferences and outcomes in liberal countries, (p.158) combined with a weak state role in facilitating (and legitimizing) female employment, may mean that maternal labor force participation is widely understood as a “choice”—and perhaps one that places a mother's personal interests ahead of her children's. Market-and school-based care provisions for older children are reasonably compatible with at least part-time maternal employment in liberal regimes. In contrast, irregular (and in some cases even half-day) school schedules greatly increase structural and normative pressures for stay-at-home mothering of school-age children in such conservative-group countries as West Germany, Austria, and Switzerland (Bird and Gottschall 2004; Buchmann and Charles 1995; Hagemann 2006).

Regression coefficients for models Ic and IIc provide no evidence that the Marxist critique of bourgeois family norms has resulted in an across-the- board weakening of women's support for full-time mothering in the formerly communist countries of eastern and central Europe. Women's attitudes toward child rearing in fact differ little in the aggregate from those in conservative state regimes. This suggests that state-imposed norms of maternal employment are not automatically internalized. We may also be witnessing an ideological backlash. Given frequent shortages of goods and services, and the absence of serious state efforts to increase men's participation in domestic work, mothers' “double shift” was often onerous during the communist era (Bicskei 2006; Heinen 2002; Michel 2006). As a result, some eastern and central European women may now view the male breadwinner family as relatively emancipatory, especially where public investment in child care was inadequate to support the state-prescribed dual-earner families. The relative weakness of the feminist movement and the symbolic association of “gender equality” with the communist state may also increase support for western-style male breadwinner family models in these contexts.19

In developing regimes, women's attitudes are quite distinctive with regard to school-age (but not preschool-age) children. The odds of supporting full-time maternal care of children in the older age group are in fact nearly nine times higher in developing than in social–democratic societies,20 although still only a minority of women in developing countries advocate stay-at-home mothering of these children. The difference between more and less affluent societies in this regard may be attributable to the poor conditions facing many workers in developing countries (Heymann 2006). Women are strongly overrepresented in the unregulated and poorly paid informal sector of less developed economies, especially in Latin America. When their parents work long and unpredictable hours, children are often left unsupervised after school.

To determine whether effects of individual-level covariates vary across regime types, we have run a series of regime-specific models (not shown). In only one case did a statistically significant coefficient deviate in sign from (p.159) those shown for models Ia and IIa. The exception concerns the interaction of motherhood and employment. Being an employed mother in a developing country makes women more, not less, likely to advocate full-time maternal care of preschoolers. This may again reflect differences in the nature and meaning of female labor force participation in more and less economically developed countries. Mothers who believe that full-time maternal care is advantageous can more often afford to stay at home with their preschool children in affluent than in developing societies.21

The relative predictive power of models displayed in Table 8.3 can be assessed by comparing the Bayesian information criterion (BIC) statistics across specifications.22 The preferred model is that with the smallest (i.e., the most negative) BIC score. Not surprisingly, the best fit is obtained when attitudes are allowed to vary freely by country (i.e., in models Ib and IIb). Nonetheless, regime distinctions do substantially improve prediction relative to the individual-effects models, especially with regard to beliefs about mothering of older (i.e., school-age) children. Comparing BIC scores between the regime- and the country-effects models suggests greater within- regime cohesion of attitudes concerning school-age than preschool-age children. As discussed later, state and private provisions for the care of very young children (e.g., early education and child care services, maternal and paternal leave policies) vary a great deal even within regime types.

Within-regime variability in attitudes toward maternal employment is evident in Figure 8.1, which displays adjusted country effects garnered from models Ib and IIb, grouped by regime. Bars represent, for each country, the predicted probabilities of espousing full-time maternal care, holding constant individual-level factors (see equation in note 5). Specifically, values give probabilities for a hypothetical 30-year-old employed woman who is single, has no children at home, has no university degree, is not highly religious, and whose own mother was employed.23 The corresponding representation of regime effects (based on coefficients from models Ic and IIc) can be found in Figure 8.2, where bars represent predicted probabilities for each regime.

Figure 8.1 reveals considerable within-regime heterogeneity in support of full-time motherhood. In the following section, we consider these and other country-level differences in connection with recent literature on welfare states and family policy.

Implications for Welfare State Theory

Comparativist scholars see both material and cultural significance in distinctions among welfare state regimes. The material effects of variability in state-funded social services and income transfers are obvious. The cultural effects are the taken-for-granted understandings of work, gender roles, (p.160)

Beliefs about Maternal Employment

Figure 8.1. Espousal of full-time maternal care: Probability of agreement, by country.

NOTE: Values are predicted probabilities of agreement with the corresponding statement, calculated from models lb and lib of Table 8.3. Probabilities are for a hypothetical 30-year-old employed woman who is single, has no children, has no university degree, is not highly religious, and whose mother was employed. The top and bottom bar graphs are presented with different scales to facilitate inter-country comparisons within each panel.

(p.161)
Beliefs about Maternal Employment

Figure 8.2. Espousal of full-time maternal care: Probability of agreement, by regime.

NOTE: Values are predicted probabilities of agreement with the corresponding statement, calculated from models Ic and lie of Table 8.3. Probabilities are for a hypothetical 30-year-old employed woman who is single, has no children, has no university degree, is not highly religious, and whose mother was employed.

family relationships (including the nature of childhood), and the role of the state that congeal over time. These normative understandings in turn strengthen public support for specific brands of welfare statism, resulting in path-dependent trajectories of social policy development. Regime-specific cultures of the market, the state, and the family are thereby crystallized, as suggested in Esping-Andersen's (1999) observation that: “Genetics clearly do not create preferences and beliefs. What might account for this is society itself with all its institutions, incentive systems, and inscribed norms of proper conduct” (p. 172).24

Family policy experts and feminist scholars have, however, challenged the material and cultural coherence of the Esping-Andersen typology on the grounds that it pays insufficient attention to within-regime variability in state-sanctioned gender relations and family structures. The mainstream welfare state literature suffers, they argue, from a preoccupation with income maintenance and job security of core workers (“decommodification”) and glosses over important differences in the degree to which national social policies facilitate female employment and help ease burdens associated with family care (Daly 2000; O'Connor 1999; Orloff 1993, 1996; Sainsbury 1996).25 For example, in countries where the state seeks to increase female labor force participation (or stem fertility declines), generous child care services (p.162) are provided, and dominant normative understandings of childhood and parenthood do not include long periods of full-time maternal care. When, on the other hand, welfare policy is built around a male breadwinner model, the state supports lengthy maternity leaves and/or subsidizes the (male) family wage to allow mothers to withdraw completely from the labor force. The state-sanctioned gendered division of earning and caring roles is thereby legitimized and naturalized (Mahon 2002b).

In the following paragraphs, we briefly discuss existing evidence on within-regime variability in state provision of child care services, a key indicator of “defamilialization” (i.e., loosening of households' welfare and caring responsibilities).26 We then consider the extent to which these policy differences correspond to the patterns of cross-national variability in attitudes that are revealed in Figure 8.1.

Among countries classified as “conservative” in the Esping-Anderson typology, much heterogeneity in family policy provisions has been documented. The French and Belgian states are commonly identified as exceptional within this cluster because they provide preschool education for all children age 3 to 6 years and offer relatively good publicly funded services for younger children.27 The Dutch state has recently been moving toward a relatively generous provision of child care support as well, although it long resisted loosening of its strong familialist traditions (Bussemaker and van Kersbergen 1999; Morgan 2006). Substantial differences are also found among the highly familialist societies in the conservative group. For example, the southern European countries of Italy and Spain have relatively well-developed systems of public preschool that accommodate most 3- to 5-year-olds (Mahon 2002a), whereas Japan (sometimes described as a “Confucian” welfare state) is characterized by high rates of dependency on multigenerational families and very little in the way of publicly funded preschool or child care (Jones 1993).

Significant internal variability in state support for maternal employment has also been observed within the social–democratic group, where Norway provides the least generous support for public child care (Leira 1992; Sainsbury 1999a); and within the liberal group, where Australia and Canada have been identified as better providers of maternity leave and child care coverage (Mahon 2002a; O'Connor 1999).

The comparative literature likewise suggests much heterogeneity among formerly socialist societies. Both during and after the communist era, child care coverage has been good in the former German Democratic Republic (GDR),28 but relatively poor in Poland and the region that is now the Czech Republic, perhaps because of the continuous cultural influence of Catholic religious doctrine in the latter two regions. Hungary appears to have fallen somewhere in between Poland and the GDR with regard to child care policy (p.163) (Hagemann 2006; Heinen 2002; Mahon 2002a; Michel 2006). Slovenia, a relatively affluent country in the group, developed an extensive system of public care and education for young children that has been kept more or less intact throughout the transition period (Stropnik 2001).

Evidence is also emerging that “varieties of welfare capitalism” exist within the developing world (Pribble 2006; Rudra 2007). This heterogeneity is attributable in part to cross-national differences in exposure to the global economy and pressures from international financial institutions for neoliberal-style structural adjustments (Rudra 2002, 2007).29 Although available information on child care and preschool coverage does not allow systematic comparison across our developing countries, some features of Brazilian society would seem to support more generous provisions. These include a more inward-directed development strategy, more universalistic welfare state traditions (which, however, fall considerably short in practice), and the expressed commitment of the Brazilian state and the national feminist movement to expanding child care provisions.30

To what extent do these national differences in family policy map on to patterns of within-regime variability in women's attitudes? Within both the social–democratic and conservative regimes, we find that support for full-time maternal care tends to be weaker in countries that have been identified as having the most extensive public-sector provisions for child care and education. Danish and Swedish women are, accordingly, less likely than their Finnish and Norwegian counterparts to support full-time maternal care of children regardless of age. With respect to school-age children, France, Belgium, and the Netherlands stand out within their conservative group as particularly supportive of maternal employment. French and Dutch women are, however, no less likely than their counterparts in other “conservative” countries to support full-time maternal care of very young children. This age dependence may be attributable to state subsidies for lengthy maternity leaves in France and the Netherlands. Women in the highly familialist countries of the Mediterranean region (i.e., Spain, Portugal, and Cyprus) and Japan show no greater tendency to support stay-at-home mothering than do their counterparts in other conservative states.

Within the liberal group, we find much heterogeneity but no consistent pattern of cross-national variability across indicators. Attitudinal patterns are strongly dependent on the age of the child. With regard to employment of mothers with preschool children, the odds that a woman will support stay-at-home mothering is, for example, several times stronger in New Zealand than in Israel. Both countries fall close to the liberal group average with respect to care of school-age children, however.31 Israeli women's views on the care of preschool children may be attributable to that country's long tradition of community-based education (often religiously oriented) for (p.164) preschool-age children.32 We have no ready explanation for the very high value in New Zealand on the same measure. The difference between Australia and New Zealand with respect to employment of mothers with very young children may be partially attributable to the relatively well-developed Australian system of child care supports that grew out of a feminist–labor alliance during the 1970s and ′80s.33

In the group of formerly socialist societies, we find correspondence of attitudes with national cultural traditions as well with the types of institutional support for female labor force participation that were developed under communism. East German and Slovenian women, for example, view maternal employment considerably more favorably than their counterparts in Poland and the Czech Republic, where the historical influence of Catholic religious doctrine on family practices and state policies was evident even during the communist era.34 This contrast highlights the resilience of traditional customs and cultures in the face of massive institutional transformation. But the large differences observed between East and West German women with respect to maternal employment suggest that family policy regimes can have enduring normative effects. Attitudes in the East and West may eventually converge in reunited Germany, but our data point to strong cultural legacies from socialist institutions in the Eastern Länder (see also Rosenfeld, Trappe, and Gornick 2004; Rudd 2000). The odds of advocating full-time maternal care of school-age children is, for example, nearly five times higher for West German women than for their counterparts in the East, who showed roughly similar patterns of employment prior to World War II.35

Turning to the group of developing nations, Brazilian women are less likely to object to maternal employment than their counterparts elsewhere in the developing world. As suggested earlier, this difference may be attributable to the stronger government and popular support for publicly funded early-childhood education and the weaker influence of neoliberal antistatism in Brazil. A broader sample of developing countries is necessary before any general conclusions can be drawn about this group.

Women's views about maternal employment indeed vary across standard welfare state regime types (Figure 8.2), presumably because their experiences with state policies related to family welfare supports, taxation, and employment influence their understandings of what constitutes “normal” motherhood and childhood.36 These understandings in turn help sustain feedback loops and path-dependent trajectories of policy development within regime types. But our results also provide support for feminist critiques, which hold that standard typologies of welfare state regimes obscure crossnational variability in defamilialization of caring responsibilities. Norms of maternal care documented here do in fact correspond to important withinregime differences in child care and family policies. A strong state role in (p.165) harmonizing market and family roles may help disseminate “family values” that legitimize a greater public-sector role in the care and education of young children.

Future research should use over-time data and multilevel modeling to explore further the spatial variability in norms of motherhood. Application of over-time data (for either qualitative or quantitative analysis) would allow researchers to identify more clearly the direction of causal relationships and to gain a better sense of the relative importance of attitudinal divergence and convergence in global economies and societies. More rigorous examination of the intermediary mechanisms driving observed patterns of cross-national variability may be facilitated by a formal multilevel modeling approach. The theoretical literature points to a number of macrolevel covariates that might be relevant in this regard. These include rates of female labor force participation, levels of economic prosperity, urban concentration of the population, timing and speed of industrialization, preschool enrollment rates, service-sector size, strength of the feminist movement and leftist parties, and female representation in parliaments. Given recent evidence of significant micro/macro interactions in the determination of household divisions of labor, researchers should also attend carefully to how national and regional factors may mediate effects of individual-level variables on norms of motherhood.

Conclusion

Our analyses support the following empirical conclusions: First, beliefs about the appropriateness of maternal employment vary a great deal cross-nationally, even among women with similar configurations of sociodemo-graphic characteristics. Second, some of this cross-national variability in beliefs about mothering corresponds to differences in welfare state regime types. And third, within-regime variability corresponds to differences in national family policy provisions that have not yet received much attention from mainstream welfare state scholars. These findings are consistent with arguments positing a coconstitutive relationship between welfare state regime types on the one hand and societal understandings about work, gender roles, and childhood on the other. They also suggest that feminist scholars are right to call for greater attention to within-regime variability in policy efforts aimed specifically at reconciling market and caring work.

There is much evidence that individual behavior and household divisions of labor are strongly conditioned by welfare-and family-policy regimes and the associated norms of motherhood (Cooke 2007; Esping-Andersen 1999; Gornick and Meyers 2003; Lewis 2006; Mandel and Semyonov 2006; Mayer 2001).37 The very low rate of maternal labor force participation in (p.166) contemporary West Germany, for instance, reflects the interaction of powerful ideological and structural forces:

[M]others abide by the institutional steering of their life courses prescribed for them by legislation and backed up by normative expectations of colleagues, friends, and families. These dual measures—the external policy instrument and the expectations of their normative environment—combine to produce a nearly insurmountable prescribed pattern for the labor-market participation of mothers with children under three years old: that is, they stay at home. (Bird and Gottschall 2004, p. 296)

The overdetermination of German stay-at-home motherhood stands in stark contrast to the normative and institutional environment in Sweden, where:

Swedish policy makers effectively legislated the demise of the male breadwinner family in the late 1960s and early 1970s, making it financially onerous for one parent to be home full-time…. This created strong demand for the continued expansion of the Swedish day care system while softening divisions between advocates of working mothers and defenders of the male-breadwinner model. (Morgan 2006, p. 114)

Both of these accounts recognize the mutually constitutive relationship between established social policy regimes and normative expectations regarding maternal employment and the role of the state. Countries are by no means culturally homogeneous,38 but they do clearly differ in what Kremer (2006) calls dominant “ideals of care.”

Although international differences in state-sanctioned ideals of motherhood and the family are striking, it is possible that the worldwide diffusion of neoliberal ideology and trends toward regional integration and economic globalization may alter the logic of path dependence in national welfare state provisions. Rianne Mahon (2002b), for instance, suggests that the expansion and institutional formalization of the European Union has contributed to a growing “hybridization” in family policies and injection of new ideas into national societies. Such tendencies could imply a gradual convergence in normative ideals of care across countries and regime types.

In interpreting results of our analyses, we have suggested that notions regarding where and by whom child care should be provided are attributable in part to shared experiences with national welfare- and family-policy regimes. These experiences congeal into common normative understandings of motherhood, childhood, and the role of the state, and they influence the extent to which mothering is seen as compatible with paid employment. The resulting cultural understandings of motherhood and childhood are reflected in the actions and agendas of policy makers, employers, and social movement activists, and they are manifested in subsequent waves of policy development. An active state role in harmonizing market and family obligations (p.167) legitimizes maternal employment and public child care and further increases demand for policies that defamilialize care. A powerful feedback loop is thus completed.

Notes

We thank Lynn Prince Cooke, Claudia Geist, and the volume's editors for helpful comments.

(1.) On the relationship between family policy arrangements and child-rearing ideals, see Mahon (2002a,b) and Kremer (2006). Esping-Andersen (1999) and Svallfors (2007) offer more general discussions of normative feedback effects emanating from welfare state institutions. The alternate causal pathway—from public preferences to policy provisions—is considered by Brooks and Manza (2006).

(2.) American and international studies indicate, for instance, that nonemployed, partnered women do a larger share of routine housework than their employed counterparts (Bianchi et al. 2000; Fuwa 2004; Geist 2005).

(3.) We have also computed weighted models with very similar results. Any significant differences are noted.

(4.) R esponses were taken from the following survey question: Do you think that women should work outside the home, full-time, part-time, or not at all under the following circumstances: after marrying and before there are children, when there is a child under school age, after the youngest child starts school, after the children leave home.” We do not distinguish between full- and part-time employment because of tremendous cross-national variability in definition and prevalence of part-time work.

(5.) Although we considered other survey items, many of them tapped into multiple attitudes and therefore present ambiguities of interpretation and crossnational comparison.

(6.) As expected, men are generally more likely to espouse full-time maternal care than are women.

(7.) The logistic regression equation takes the following form: log[Pi/(1 − Pi)] = a + bXi, where i denotes the ith sample respondent, Pi represents the probability that respondent i holds the belief in question, a is the model intercept, and bXi represents a vector of covariates (Xi) and their slopes (b). Predicted probabilities are calculated as Beliefs about Maternal Employment.

(8.) Women reporting “full-time,” “part-time,” or “less than part-time” activity are counted as employed. Overall, less than 2% of women reported working “less than part-time.” Most of these women are from countries where part-time employment is defined as 15 to 35 hours per week (10–29 hours in Great Britain and Denmark, 10–39 hours in the Philippines). The “less-than-part-time” category also includes a small number of women who were temporarily not working.

(9.) The “household composition” variable is not available for Slovenia, Israel, Cyprus, and Brazil. For Australia, the total given on this variable appears to include missing values. For these four countries, we determined presence of (p.168) children based on variables giving the numbers of preschool- and school-age children in the household.

(10.) Although we would have liked to control for differences in economic class position, such comparisons were unfeasible as a result of complications involved in determining class positions of nonemployed and marginally employed women.

(11.) Multiplying a coefficient by –1 gives the effect on the log-odds that a woman disagrees with the respective statement (i.e., finds maternal employment acceptable).

(12.) exp(0.878) = 2.406.

(13.) exp(0.297)/(exp(–0.365 + 0.297 + –0.291)) = 1.927. Because we have included main and interaction terms for employment and the presence of children, the main “employment” effects pertain to women with no children at home, and the main “child” effect pertains to nonemployed women.

(14.) When sample weights are applied, the positive effect of marriage becomes statistically significant in models Ia and IIa, and the effects of having children (for both employed and nonemployed women) become statistically significant in model Ic.

(15.) exp(1.219) = 3.384.

(16.) Qualification for means-tested benefits is based on financial need. Universal welfare benefits, in contrast, are targeted toward all those who fall within particular social categories (e.g., families with children, citizens older than age 65, the unemployed).

(17.) To allow for greater investment in primary education, these organizations tend to emphasize low-cost, nonformal provisions for young children (Rosemberg 2003).

(18.) Esping-Andersen (1997) in fact describes the current Japanese welfare state as a hybrid system, which fuses the corporatism and familialism of the conservative regime with the market emphasis of the liberal regime. Because our focus is on family-related policy and values, we classify Japan as a conservative regime. We follow Esping-Andersen's 1999 typology in classifying the Dutch welfare state as conservative. See Kouloumou (2006) on family policy in Cyprus.

(19.) Although no clear link is evident between the strength of women's movements and the extensiveness of family policy provisions (Morgan 2006), feminist organizations undoubtedly have an ideological and agenda-setting effect within their respective national contexts. See also Sainsbury (1999b) on different strands of feminism and their policy implications.

(20.) exp(1.412)/(exp(–0.778) = 8.935.

(21.) We also note the following interregime differences in strength of covariate effects: Coefficients for religiosity tend to be smaller in developing and liberal countries, and the presence of children (among nonemployed women) is a weaker predictor of attitudes in developing and socialist regimes.

(22.) Application of the Bayesian information criterion (BIC) allows us to consider both parsimony and explanatory power in selecting the best-fitting model. By this standard, inclusion of country or regime effects must be justified on the (p.169) basis of improved model fit. The computational formula for BIC is shown in the footnote to Table 8.3. For more information, see Raftery (1995).

(23.) We have also computed regime-specific models for which parameter estimates for all individual-level covariates are allowed to vary across regime types. Country effects for these models correspond very closely to those shown in Figure 8.1 for the pooled models.

(24.) In Esping-Anderson's (1999) terms, therefore, Homo liberalismus, Homo familius, and Homo socialdemocraticus become culturally dominant in the respective welfare state regimes. These ideal typical homines exhibit, he argues, preferences and cultural logic that are both reflected in and promoted by the corresponding policy regimes.

(25.) Decommodification refers to the “degree to which the individual's typical life situation is freed from dependence on the labor market” (Esping-Andersen and Korpi 1987, p. 40). Feminist scholars have pointed out that women who specialize in extramarket care work are dependent on families, rather than markets, for their welfare. Their independence therefore requires not decommodification, but a lessening of their reliance on families (Orloff 1993). Esping-Andersen (1999, pp. 60–67) acknowledges important intra-regime difference in relations between the family and the state, but argues that his original classification of regimes remains useful for capturing some differences in state efforts to defamilialize caring responsibilities.

(26.) A familialist system is one in which public policy is organized around the presumption that families will carry the principle responsibility for members' welfare.

(27.) R egarding provisions in France and Belgium, see Bussemaker and van Kersbergen (1999); Meyers, Gornick, and Ross (1999); Mahon (2002a), and Morgan (2006). Important ideological dissimilarities with the Nordic countries have been identified. In particular, programs in France and Belgium are intended to increase fertility and improve early childhood education more than to facilitate maternal employment and promote gender equality.

(28.) All-day supervision for nursery-, preschool- and school-age children was provided on a universal basis in the former GDR. This supported—some would say enforced—female labor force participation rates that were high even compared with other socialist countries. Although child care provisions have been scaled back in the East German Läder, they continue to be considerably more generous than in the West (Rosenfeld, Trappe, and Gornick 2004).

(29.) Insulation of the economy from international competition allows greater state discretion to intervene in the economy and expand its role in provision of social services. Interestingly, Rudra (2002) finds that exposure to international markets decreases state welfare spending in developing, but not developed, countries.

(30.) The Brazilian feminist movement has placed priority on child care since the 1970s. The national constitution of 1988 mandates federal government provision of free daycare and preschool to all children younger than 7 years of age. This policy has not yet been implemented, however. Similar gaps between de jure and de facto coverage have been identified with respect to other sorts of universal (p.170) entitlements in Brazil. On Brazilian social policy, see Connelly, DeGraff, and Levison (1996); Huber (1996); and Barrientos (2004).

(31.) Northern Ireland is a notable outlier on this indicator. Even there, only a small minority of women object to employment of mothers with school-age children.

(32.) The Israeli welfare regime has both liberal and social–democratic elements (Sabbagh, Powell, and Vanhuysse 2007). With regard to attitudes toward employment of women with very young children, Israeli women look more like their Nordic than their liberal counterparts.

(33.) In a comparative analysis of family life in Australia, New Zealand, and Canada, Baker (2001) notes that child care provisions for infants are in especially short supply in New Zealand. She also describes strong social pressure for athome mothering in both Australia and New Zealand.

(34.) Because our models control for individual religiosity, we here refer to societal-level effects. The policy provisions and norms of family life that grow out of the dominant religious doctrine are presumed to influence attitudes of even those persons who are not highly religious.

(35.) exp(0.896)/(exp(−0.667) = 4.773. See also Cooke (2007) on divisions of domestic labor in East and West German Länder.

(36.) See also Geist (2005) on interregime differences in the gendered division of domestic labor.

(38.) Variability within regions of Switzerland, Hungary, and the United Kingdom have, for example, been described by Bühler (1998), Bicskei (2006), and Wincott (2006), respectively. See also Duncan (2000).

(37.) Recent comparative analyses suggest that living in a country characterized by conservative family policies or low rates of female employment may imply more gendered divisions of housework regardless of an individual woman's employment status or personal attitudes (Breen and Cooke 2005; Fuwa 2004; Hook 2006).

References

Bibliography references:

Ariès, Philippe. 1965. Centuries of childhood: A social history of family life. New York: Vintage.

Baker, Maureen. 2001. Families, labour and love. Vancouver, BC: UBC Press.

Barrientos, Armando. 2004. Latin America: Towards a liberal–informal welfare regime. In Insecurity and welfare regimes in Asia, Africa and Latin America: Social policy in development contexts, ed. Ian Gough and Geof Wood, 121–168. New York: Cambridge University Press.

Ben-Arieh, Ahser, Uifat Boyer, and Idit Gajst. 2004. Children's welfare in Israel: Growing up in a multi-cultural society. In Children's welfare in ageing Europe, volume II, ed. An-Magritt Jensen, Asher Ben-Arieh, Cinzia Conti, Dagmar Kutsar, Máire Nic Ghiolla Phádraig, and Hanne Warming Nielsen, 775–810. Trondheim: Norwegian Center for Child Research.

Bianchi, Suzanne M., Melissa A. Milkie, Liana C. Sayer, and John P. Robinson. 2000. Is anyone doing the housework? Trends in the gender division of household labor. Social Forces 79: 191–228.

(p.171) Bicskei, Éva. 2006. “Our greatest treasure, the child”: The politics of child care in Hungary, 1945–56. Social Politics 13: 151–188.

Bird, Katherine, and Karin Gottschall. 2004. Erosion of the male-breadwinner model? Female labor-market participation and family-leave policies in Germany. In Equity in the workplace: Gendering workplace policy analysis, ed. Heidi Gottfried and Laura Reese, 281–303. Lanham, MD Lexington.

Blair-Loy, Mary. 2004. Competing devotions: Career and family among women executives. Cambridge, MA: Harvard University Press.

Blossfeld, Hans-Peter, and Sonja Drobnič. 2001. Theoretical perspectives on couples' careers. In Careers of couples in contemporary societies, ed. Hans-Peter Blossfeld and Sonja Drobnič, 16–50. New York: Oxford University Press.

Brayfield, April, Rachel K. Jones, and Marina A. Adler. 2001. Harmonizing work and family in the European Union: Public perceptions of children as an obstacle to women's employment. In Women's employment in comparative perspective, ed. Tanja van der Lippe and Liset Van Dijk, 179–202. New York: Aldine de Gruyter.

Breen, Richard, and Lynn Prince Cooke. 2005. The persistence of the gendered division of domestic labour. European Sociological Review 21: 43–57.

Brooks, Clem, and Jeff Manza. 2006. Social policy responsiveness in developed democracies. American Sociological Review 71: 474–494.

Buchmann, Marlis, and Maria Charles. 1995. Organizational and institutional factors in the process of gender stratification: Comparing social arrangements in six European countries. International Journal of Sociology 25: 66–95.

Bühler, Elisabeth. 1998. Economy, state or culture? Explanations for the regional variations in gender inequality in Swiss employment. European Urban and Regional Studies 5: 27–39.

Bussemaker, Jet, and Kees van Kersbergen. 1999. Contemporary social–capitalist welfare states and gender inequality. In Gender and welfare state regimes, ed. Diane Sainsbury, 15–46. New York: Oxford University Press.

Castles, Francis G. 1996. Needs-based strategies of social protection in Australia and New Zealand. In Welfare states in transition: National adaptations in global economies, ed. Gøsta Esping-Andersen, 88–115. Newbury Park, CA: Sage.

Connelly, Rachel, Deborah S. DeGraff, and Deborah Levison. 1996. Women's employment and child care in Brazil. Economic Development and Cultural Change 44: 619–656.

Cooke, Lynn Prince. 2007. Persistent policy effects on the division of domestic tasks in reunified Germany. Journal of Marriage and Family 69: 930–950.

Daly, Mary. 2000. The gender division of welfare: The impact of the British and German welfare states. New York: Cambridge University Press.

Duncan, Simon. 2000. Introduction: Theorizing comparative gender inequality. In Gender economy and culture in the European Union, ed. Simon Duncan and Birgit Pfau-Effinger, 1–23. New York: Routledge.

Esping-Andersen, Gøsta. 1990. The three worlds of welfare capitalism. Cambridge: Polity Press.

Esping-Andersen, Gøsta. 1997. Hybrid or unique? The Japanese welfare state between Europe and America. Journal of European Social Policy 7: 179–189.

Esping-Andersen, Gøsta. 1999. Social foundations of postindustrial economies. New York: Oxford University Press.

(p.172) Esping-Andersen, Gøsta, and Walter Korpi. 1987. From poor relief to institutional welfare states: The development of Scandinavian social policy. In The Scandinavian model: Welfare states and welfare research, ed. Robert Erikson, Erik Jørgen Hansen, Stein Ringen, and Hannu Uusitalo, 39–74. New York: M.E. Sharpe.

Evans, Mariah D. R., and Jonathan Kelley. 2001. Employment for mothers of preschool children: Evidence from Australia and 23 other nations. People and Place 9: 28–40.

Evans, Mariah D. R., and Jonathan Kelley. 2002. Attitudes towards childcare in Australia. The Australian Economic Review 35: 188–196.

Fuller, Bruce. 2007. Standardized childhood: The political and cultural struggle over early education. Stanford, CA: Stanford University Press.

Fuwa, Makiko. 2004. Macro-level gender inequality and the division of household labor in 22 countries. American Sociological Review 69: 751–767.

Geist, Claudia. 2005. The welfare state and the home: Regime differences in the domestic division of labour. European Sociological Review 21: 23–41.

Gornick, Janet C., and Marcia K. Meyers. 2003. Families that work: Policies for reconciling parenthood and employment. New York: Russell Sage Foundation.

Gottlieb, Beatrice. 1993. The family in the western world from the black death to the industrial age. New York: Oxford University Press.

Hagemann, Karen. 2006. Between ideology and economy: The “time politics”of child care and public education in the two Germanys. Social Politics 13:217–260.

Hays, Sharon. 1996. The cultural contradictions of motherhood. New Haven, CT: Yale University Press.

Heinen, Jacqueline. 2002. Ideology, economics, and the politics of child care in Poland before and after the transition. In Child care policy at the crossroads: Gender and welfare state restructuring, ed. Sonya Michel and Rianne Mahon, 71–92. New York: Routledge.

Heymann, Jody. 2006. Forgotten families: Ending the growing crisis confronting children and working parents in the global economy. New York: Oxford University Press.

Hochschild, Arlie Russell. 2003. The commercialization of intimate life: Notes from home and work. Berkeley, CA: University of California Press.

Hook, Jennifer L. 2006. Care in context: Men's unpaid work in 20 countries, 1965–2003. American Sociological Review 71: 639–660.

Huber, Evelyne. 1996. Options for social policy in Latin America: Neoliberal versus social democratic models. In Welfare states in transition: National adaptations in global economies, ed. Gøsta Esping-Andersen, 141–191. Newbury Park, CA: Sage.

Huber, Evelyne, and John D. Stephens. 2001. Development and crisis of the welfare state: Parties and polices in global markets. Chicago: University of Chicago Press.

Jones, Catherine. 1993. The Pacific challenge: Confucian welfare states. In New perspectives on the welfare state in Europe, ed. Catherine Jones, 198–217. New York: Routledge.

Knudsen, Knud, and Kari Wærness. 2001. National context, individual characteristics and attitudes on mothers' employment: A comparative analysis of Great Britain, Sweden and Norway. Acta Sociologica 44: 67–79.

(p.173) Kouloumou, Toula. 2006. Children's welfare and everyday life in Cyprus: A family affair with intergenerational implications. In Children's welfare in ageing Europe, volume II, ed. An-Magritt Jensen, Asher Ben-Arieh, Cinzia Conti, Dagmar Kutsar, Máire Nic Ghiolla Phádraig, and Hanne Warming Nielsen, 591–647. Trondheim: Norwegian Center for Child Research.

Kremer, Monique. 2006. The politics of ideals of care: Danish and Flemish child care policy compared. Social Politics 13: 261–285.

Lareau, Annette. 2003. Unequal childhoods: Class, race, and family life. Berkeley: University of California Press.

Leibfried, Stephan. 1992. Towards a European welfare state: On integrating poverty regimes in the European community. In Social policy in a changing Europe, ed. Zsuzsa Ferge and Jon Eivind Kolberg, 245–280. Frankfurt: Campus Verlag.

Leira, Arnlaug. 1992. Welfare states and working mothers: The Scandinavian experience. New York: Cambridge University Press.

Lewis, Jane. 2006. Employment and care: The policy problem, gender equality and the issue of choice. Journal of Comparative Policy Analysis 8: 103–114.

Mahon, Rianne. 2002a. Gender and welfare state restructuring: Through the lens of child care. In Child care policy at the crossroads: Gender and welfare state restructuring, ed. Sonya Michel and Rianne Mahon, 1–27. New York: Routledge.

Mahon, Rianne. 2002b. Social care: Toward what kind of “social Europe?” Social Politics 9: 343–379.

Mandel, Hadas, and Moshe Semyonov. 2006. A welfare state paradox: State interventions and women's employment opportunities in 22 countries. American Journal of Sociology 111: 1910–1949.

Mayer, Karl Ulrich. 2001. The paradox of global social change and national path dependencies: Life course patterns in advanced societies. In Inclusions and exclusions in European societies, ed. Alison Woodward and Martin Kohli, 89–110. New York: Routledge.

Meyers, Marcia K., Janet Gornick, and Katherin E. Ross. 1999. Public childcare, parental leave, and employment. In Gender and welfare state regimes, ed. Diane Sainsbury, 117–146. New York: Oxford University Press.

Michel, Sonya. 2006. Introduction: Perspectives on child care, east and west. Social Politics 13: 145–150.

Miyamoto, Taro. 2003. Dynamics of the Japanese welfare state in comparative perspective: Between “three worlds” and the developmental state. The Japanese Journal of Social Security Policy 2: 12–24.

Morgan, Kimberly J. 2006. Working mothers and the welfare state: Religion and the politics of work–family policies in western Europe and the United States.Stanford, CA: Stanford University Press.

O'Connor, Julia S. 1999. Employment equality strategies in liberal welfare states. In Gender and welfare state regimes, ed. Diane Sainsbury, 47–74. New York: Oxford University Press.

Orloff, Ann Shola. 1993. Gender and the social rights of citizenship: The comparative analysis of gender relations and welfare states. American Sociological Review 58: 303–328.

Orloff, Ann Shola. 1996. Gender in the welfare state Annual Review of Sociology 22: 51–78.

(p.174) Pfau-Effinger, Birgit. 2000. Conclusion: Gender cultures, gender arrangements and social change in the European context. In Gender economy and culture in the European Union, ed. Simon Duncan and Birgit Pfau-Effinger, 262–276. New York: Routledge.

Pfau-Effinger, Birgit. 2004. Socio-historical paths of the male breadwinner model: An explanation of cross-national differences. British Journal of Sociology 55: 377–399.

Phoenix, Ann, Anne Woollett, and Eva Lloyd, eds. 1991. Motherhood: Meanings, practices and ideologies. Newbury Park, CA: Sage.

Pribble, Jennifer. 2006. Women and welfare: The politics of coping with new social risks in Chile and Uruguay. Latin American Research Review 41: 84–111.

Raftery, Adrian E. 1995. Bayesian model selection in social research. Sociological Methodology 26: 111–163.

Rosemberg, Fúlvia. 2003. Multilateral organization and early child care and education policies for developing countries. Gender and Society 17: 250–266.

Rosenfeld, Rachel A., Heike Trappe, and Janet C. Gornick. 2004. Gender and work in Germany: Before and after reunification. Annual Review of Sociology 30: 13–34.

Rudd, Elizabeth C. 2000. Reconceptualizing gender in postsocialist transformation. Gender and Society 14: 517–539.

Rudra, Nita. 2002. Globalization and the decline of the welfare state in less-developed countries. International Organizations 56: 411–445.

Rudra, Nita. 2007. Welfare states in developing countries: Unique or universal? The Journal of Politics 69: 378–396.

Sabbagh, Clara, Lawrence Alfred Powell, and Pieter Vanhuysse. 2007. Betwixt and between the market and the state: Israeli students' welfare attitudes in comparative perspective. International Journal of Social Welfare 16: 220–230.

Sainsbury, Diane. 1996. Gender equality and welfare states. New York: Cambridge University Press.

Sainsbury, Diane. 1999a. Gender and social democratic welfare states. In Gender and welfare state regimes, ed. Diane Sainsbury, 75–114. New York: Oxford University Press.

Sainsbury, Diane. 1999b. Gender, policy regimes, and politics. In Gender and welfare state regimes, ed. Diane Sainsbury, 245–275. New York: Oxford University Press.

Stropnik, Nada. 2001. Child care in Slovenia: An example of a successful transition. Child Care, Health and Development 27: 263–278.

Sundstöm, Elizabeth. 1999. Should mothers work? Age and attitudes in Germany, Italy and Sweden. International Journal of Social Welfare 8: 193–205.

Svallfors, Stefan. 2007. Conclusion: The past and future of political sociology. In The political sociology of the welfare state: Institutions, social cleavages, and orientations, ed. Stefan Svallfors, 258–280. Stanford, CA: Stanford University Press.

Taylor, Verta. 1996. Rock-a-by baby: Feminism, self-help, and postpartum depression. New York: Routledge.

Wincott, Daniel. 2006. Paradoxes of new labour social policy: Toward universal child care in Europe's “most liberal” welfare regime? Social Politics 13: 286–312.

Zelizer, Viviana A. 1994. Pricing the priceless child: The changing social value of children. Princeton, NJ: Princeton University Press.

Notes:

(1.) On the relationship between family policy arrangements and child-rearing ideals, see Mahon (2002a,b) and Kremer (2006). Esping-Andersen (1999) and Svallfors (2007) offer more general discussions of normative feedback effects emanating from welfare state institutions. The alternate causal pathway—from public preferences to policy provisions—is considered by Brooks and Manza (2006).

(2.) American and international studies indicate, for instance, that nonemployed, partnered women do a larger share of routine housework than their employed counterparts (Bianchi et al. 2000; Fuwa 2004; Geist 2005).

(3.) We have also computed weighted models with very similar results. Any significant differences are noted.

(4.) R esponses were taken from the following survey question: Do you think that women should work outside the home, full-time, part-time, or not at all under the following circumstances: after marrying and before there are children, when there is a child under school age, after the youngest child starts school, after the children leave home.” We do not distinguish between full- and part-time employment because of tremendous cross-national variability in definition and prevalence of part-time work.

(5.) Although we considered other survey items, many of them tapped into multiple attitudes and therefore present ambiguities of interpretation and crossnational comparison.

(6.) As expected, men are generally more likely to espouse full-time maternal care than are women.

(7.) The logistic regression equation takes the following form: log[Pi/(1 − Pi)] = a + bXi, where i denotes the ith sample respondent, Pi represents the probability that respondent i holds the belief in question, a is the model intercept, and bXi represents a vector of covariates (Xi) and their slopes (b). Predicted probabilities are calculated as Beliefs about Maternal Employment.

(8.) Women reporting “full-time,” “part-time,” or “less than part-time” activity are counted as employed. Overall, less than 2% of women reported working “less than part-time.” Most of these women are from countries where part-time employment is defined as 15 to 35 hours per week (10–29 hours in Great Britain and Denmark, 10–39 hours in the Philippines). The “less-than-part-time” category also includes a small number of women who were temporarily not working.

(9.) The “household composition” variable is not available for Slovenia, Israel, Cyprus, and Brazil. For Australia, the total given on this variable appears to include missing values. For these four countries, we determined presence of (p.168) children based on variables giving the numbers of preschool- and school-age children in the household.

(10.) Although we would have liked to control for differences in economic class position, such comparisons were unfeasible as a result of complications involved in determining class positions of nonemployed and marginally employed women.

(11.) Multiplying a coefficient by –1 gives the effect on the log-odds that a woman disagrees with the respective statement (i.e., finds maternal employment acceptable).

(12.) exp(0.878) = 2.406.

(13.) exp(0.297)/(exp(–0.365 + 0.297 + –0.291)) = 1.927. Because we have included main and interaction terms for employment and the presence of children, the main “employment” effects pertain to women with no children at home, and the main “child” effect pertains to nonemployed women.

(14.) When sample weights are applied, the positive effect of marriage becomes statistically significant in models Ia and IIa, and the effects of having children (for both employed and nonemployed women) become statistically significant in model Ic.

(15.) exp(1.219) = 3.384.

(16.) Qualification for means-tested benefits is based on financial need. Universal welfare benefits, in contrast, are targeted toward all those who fall within particular social categories (e.g., families with children, citizens older than age 65, the unemployed).

(17.) To allow for greater investment in primary education, these organizations tend to emphasize low-cost, nonformal provisions for young children (Rosemberg 2003).

(18.) Esping-Andersen (1997) in fact describes the current Japanese welfare state as a hybrid system, which fuses the corporatism and familialism of the conservative regime with the market emphasis of the liberal regime. Because our focus is on family-related policy and values, we classify Japan as a conservative regime. We follow Esping-Andersen's 1999 typology in classifying the Dutch welfare state as conservative. See Kouloumou (2006) on family policy in Cyprus.

(19.) Although no clear link is evident between the strength of women's movements and the extensiveness of family policy provisions (Morgan 2006), feminist organizations undoubtedly have an ideological and agenda-setting effect within their respective national contexts. See also Sainsbury (1999b) on different strands of feminism and their policy implications.

(20.) exp(1.412)/(exp(–0.778) = 8.935.

(21.) We also note the following interregime differences in strength of covariate effects: Coefficients for religiosity tend to be smaller in developing and liberal countries, and the presence of children (among nonemployed women) is a weaker predictor of attitudes in developing and socialist regimes.

(22.) Application of the Bayesian information criterion (BIC) allows us to consider both parsimony and explanatory power in selecting the best-fitting model. By this standard, inclusion of country or regime effects must be justified on the (p.169) basis of improved model fit. The computational formula for BIC is shown in the footnote to Table 8.3. For more information, see Raftery (1995).

(23.) We have also computed regime-specific models for which parameter estimates for all individual-level covariates are allowed to vary across regime types. Country effects for these models correspond very closely to those shown in Figure 8.1 for the pooled models.

(24.) In Esping-Anderson's (1999) terms, therefore, Homo liberalismus, Homo familius, and Homo socialdemocraticus become culturally dominant in the respective welfare state regimes. These ideal typical homines exhibit, he argues, preferences and cultural logic that are both reflected in and promoted by the corresponding policy regimes.

(25.) Decommodification refers to the “degree to which the individual's typical life situation is freed from dependence on the labor market” (Esping-Andersen and Korpi 1987, p. 40). Feminist scholars have pointed out that women who specialize in extramarket care work are dependent on families, rather than markets, for their welfare. Their independence therefore requires not decommodification, but a lessening of their reliance on families (Orloff 1993). Esping-Andersen (1999, pp. 60–67) acknowledges important intra-regime difference in relations between the family and the state, but argues that his original classification of regimes remains useful for capturing some differences in state efforts to defamilialize caring responsibilities.

(26.) A familialist system is one in which public policy is organized around the presumption that families will carry the principle responsibility for members' welfare.

(27.) R egarding provisions in France and Belgium, see Bussemaker and van Kersbergen (1999); Meyers, Gornick, and Ross (1999); Mahon (2002a), and Morgan (2006). Important ideological dissimilarities with the Nordic countries have been identified. In particular, programs in France and Belgium are intended to increase fertility and improve early childhood education more than to facilitate maternal employment and promote gender equality.

(28.) All-day supervision for nursery-, preschool- and school-age children was provided on a universal basis in the former GDR. This supported—some would say enforced—female labor force participation rates that were high even compared with other socialist countries. Although child care provisions have been scaled back in the East German Läder, they continue to be considerably more generous than in the West (Rosenfeld, Trappe, and Gornick 2004).

(29.) Insulation of the economy from international competition allows greater state discretion to intervene in the economy and expand its role in provision of social services. Interestingly, Rudra (2002) finds that exposure to international markets decreases state welfare spending in developing, but not developed, countries.

(30.) The Brazilian feminist movement has placed priority on child care since the 1970s. The national constitution of 1988 mandates federal government provision of free daycare and preschool to all children younger than 7 years of age. This policy has not yet been implemented, however. Similar gaps between de jure and de facto coverage have been identified with respect to other sorts of universal (p.170) entitlements in Brazil. On Brazilian social policy, see Connelly, DeGraff, and Levison (1996); Huber (1996); and Barrientos (2004).

(31.) Northern Ireland is a notable outlier on this indicator. Even there, only a small minority of women object to employment of mothers with school-age children.

(32.) The Israeli welfare regime has both liberal and social–democratic elements (Sabbagh, Powell, and Vanhuysse 2007). With regard to attitudes toward employment of women with very young children, Israeli women look more like their Nordic than their liberal counterparts.

(33.) In a comparative analysis of family life in Australia, New Zealand, and Canada, Baker (2001) notes that child care provisions for infants are in especially short supply in New Zealand. She also describes strong social pressure for athome mothering in both Australia and New Zealand.

(34.) Because our models control for individual religiosity, we here refer to societal-level effects. The policy provisions and norms of family life that grow out of the dominant religious doctrine are presumed to influence attitudes of even those persons who are not highly religious.

(35.) exp(0.896)/(exp(−0.667) = 4.773. See also Cooke (2007) on divisions of domestic labor in East and West German Länder.

(36.) See also Geist (2005) on interregime differences in the gendered division of domestic labor.

(37.) Recent comparative analyses suggest that living in a country characterized by conservative family policies or low rates of female employment may imply more gendered divisions of housework regardless of an individual woman's employment status or personal attitudes (Breen and Cooke 2005; Fuwa 2004; Hook 2006).

(38.) Variability within regions of Switzerland, Hungary, and the United Kingdom have, for example, been described by Bühler (1998), Bicskei (2006), and Wincott (2006), respectively. See also Duncan (2000).