A Socio-Structural Perspective on Family Model Preferences, Gender Roles and Work–Family Attitudes in Spain

Abstract

Since the early 1990s, the diversity of work–family arrangement models has increased in Spain. It is difficult to understand this phenomenon without attending to the Spanish population’s preferences for such models. This article analyses the attitudes towards gender roles, and family model preferences within a normative and socio-structural framework. Using data from the International Social Survey Programme 2012, we developed descriptive and explanatory analyses. The findings reveal contradictions between attitudes towards the mother’s and father’s work intensity and gender roles that seem to be resolved through preferences for a “hybrid” or “adaptive” family model. We also identified the determinants of family model preferences for both men and women. The results show that gender plays a significant role in explaining preferences (women are less likely than men to prefer the male-breadwinner family model) and that socio-structural factors such as age, education level, immigrant condition, religious status and social class influence the preferences of men and women differently. Ultimately, these results contrast with Hakim’s Preference Theory, which emphasises individuals’ choices over socio-structural factors as determinants of family models, and align with Crompton’s and Pfau-Effinger’s theories.

1. Introduction

Since the 1990s, diversity of family models has increased in Spanish society, suggesting that the traditional model of the male breadwinner is becoming a thing of the past. The diversity and often ambiguity of this social reality cannot be understood without attending to the factors that determine attitudes toward gender roles and preferences for different family models (structure and values are interrelated—that is, they influence each other). Similar to Pfau-Effinger (2004a, 2004b, 2004c), we argue that relationships of mutual influence exist between the cultural and structural levels (attitudes and socio-structural factors, respectively) and the social actors’ actions. This argument is based on the theoretical contributions of Weber ([1922] 1964), Lockwood (1964) and Archer (1995), who consider the conceptualisation of these interpellations as the foundation for analysis of social change. We understand the theoretical conceptualisation of preferences for family models to contain the prevailing ideas in a society (in this case, Spanish) of the most desirable way as the best way to attend to family and work outside the home, a concept that also implies specific expectations about gender relations. In Pfau-Effinger’s words, the concept of “model” indicates typical societal ideal representations, norms and values regarding the family and the societal integration of women and men. One or several such models in combination may dominate culturally in the gender arrangement of a society (Pfau-Effinger 2004a, p. 382). This study analysed the attitudes toward gender roles and work–family preferences for family models from a socio-structural and gender perspective. First, the study analysed Spaniards’ attitudes toward gender roles through their degree of agreement/disagreement with statements such as “Both should contribute to household income” and “When mom works, the preschool child is likely to suffer”; and Spaniards’ opinions about the woman’s ideal work situation when there are children under six years old in the household. Second, the study determined the social factors that influence both men’s and women’s preferences for the different family models: the mother stays at home and the father works full-time (Mother at home/F FT), the mother works part-time and the father works full-time (M PT/F FT), both the mother and the father work full-time (M/F FT), and both the mother and the father work part-time (M/F PT).

Prior studies have analysed family models in Spain and the contrasts between reality and the ideal (Hakim 2003), participation of men and the changes their participation undergoes in different countries (González et al. 2009), co-responsibilities of fathers and mothers (González and Jurado 2015) and lifestyle preferences and strategies of Spanish working mothers (Campillo and Armijo 2017). The field lacks studies of the ideal plane, however, particularly studies of attitudes toward work–family models and gender roles in Spain. More specifically, no research has determined what factors influence individuals’ preferences for the different family models (especially the two-breadwinner model). Our paper fills this gap through in-depth analysis of attitudes toward gender roles and preferences for family models in a context of constant significant socio-economic and family change. This study is confirmed to be relevant by international research proving the effectiveness of studying attitudes and values to understand trends in family model changes (Amato et al. 1995; Greenstein 1996; Kaufman 2000; Paxton and Kunovich 2003; Cunningham 2005, 2008).

We examined the attitudes of the Spanish population (both men and women) towards gender roles and its preferences for family models based on theoretical antecedents and data from the 2012 “Family and Changing Gender Roles IV” Module of the International Social Survey Programme (ISSP). Spain is a representative southern European country where the family/kinship solidarity model predominates (also termed the familism model by Calzada and Brooks 2013; Reher 1998; Micheli 2012; Ferrera 1996; Naldini 2003). Spain also has a reductionist welfare state with very low public support for the family (Liebfried 1993; Ferrera 1998; Saraceno 2003) and is a society in which the traditional family model and egalitarian new normative family models coexist (Moreno Mínguez et al. 2017; Villa 2007; León and Migliavacca 2013; Naldini and Jurado 2013). The prior studies by Ortega-Gaspar (2013) demonstrate, as argued by Gauthier (1996), that Spain belongs to the group of countries termed pro-traditional countries, which focus on preserving the family even if the government provides little aid to support this goal. This argument enables us to understand why the male-breadwinner and the father FT/mother PT models coexist as preferred family models in contemporary Spain, even though it is a reductionist social state.

Drawing on Esping-Andersen and Billari’s (2015) Multiple Equilibrium Framework and the trend observed toward support for a dual-earner model due to a process of attitudinal change in recent decades, this study attempted to confirm these changes by analysing attitudes to gender roles (whether they are becoming more egalitarian) and preferences for family models. The paper’s originality lies in its identification of the factors that condition men’s and women’s preferences for the different models. Specifically, it breaks new ground by studying the various models indicated, which include the dual-earner models (i.e., both parents work full-time and both work part-time).

Our general hypothesis was that attitudes towards the different family models reflect the intricate crossing of values inherited historically (dominance of patriarchal culture) and values of a society advancing toward new ways of distributing work and family responsibilities. In other words, attitudes may be due partly to the influence of the culture in the country studied, Spain. If the general hypothesis were confirmed, our study would support Pfau-Effinger’s thesis (2004a, 2004b, 2012). Pfau-Effinger demonstrates the influence of the bourgeois culture (with its specific norms, values, attitudes, beliefs, etc.) on the imposition of a specific family model, the housewife model of the male breadwinner family as the predominant family structure. Our Hypothesis H0 led us to consider attitudes toward gender roles and preferences for family models as determined to a significant extent by socio-structural factors and as determined differently for women and men. If this were the case, the cultural influence hypothesis would be confirmed. We would stress that Pfau-Effinger confirms her theory by observing how social practice reinforces the imposition of the male-breadwinner family as the dominant family model, whereas our analysis focused on the ideal level, that of attitudes and not of social practices.

On the one hand, prior studies such as those by Crompton and Harris (1998) and Ortega-Gaspar (2011), among other studies on attitudes toward gender roles in diverse countries, enable us to confirm the influence of specific socio-structural factors such as sex, country of origin, education and profession on these attitudes. We extended this analysis by attempting to determine the influence of the socio-structural factors on the three factors indicated—the ideal work situation for the mother and father when the children are under six, attitudes toward gender roles and family model preferences. This goal was based on the premise that the socio-structural factors that influence attitudes to gender roles must also influence the other two factors. On the other hand, this hypothesis was supported by studies such as those by Pfau-Effinger (2004a), who argues that role acquisition occurs in the internalisation of the structure of family roles during the socialisation process. Pfau-Effinger recalls that, for Parsons and Bates (1955), gender role differentiation has a normative function.

In this sense, Craig and Mullan (2010) and Hochschild and Machung (1989) argue that traditional gender norms and ideologies internalised through the socialisation process lead, for example, to men and women dividing work inside and outside the home unequally. This gendered difference in internalising expectations about behaviour leads us to expect that the different socio-structural factors affect men and women differently. When examining the factors that explain the high level of work–family conflict in Spain and Great Britain for both men and women, Ortega-Gaspar (2013) observes that various socio-structural factors influence the two social contexts studied differently. This finding leads us to reflect on the effect of culture and its supposed influence on the different weight of structural factors in men’s and women’s attitudes. Finally, Pfau-Effinger (2004a) argues that family models in modern societies are images that specify the criteria for achieving the best integration and employment of women, men and children in society. This concept refers to two central social institutions, the family and the labour market, which must be arranged on the foundation of the gendered division of labour.

The results reveal contradictions in family–work model preferences, fundamentally among women. We observe preference for a model that enables mothers to work part-time when children in the household are under school age, a preference that could be interpreted as a strategy to resolve contradictions (women are more likely than men to express this preference). These attitudes confirm the tendency towards the “one and a half” model proposed by Trifiletti (2012). The contradictions also suggest a possible rupture at the level of preference with the more traditional model and openness to attitudes that invite adoption of an “adaptive” model (the one-and-a-half family model, in our case father full-time/mother part-time). These results may be due to the limitations of the public family protection system (childcare leave legislation developed very slowly; Iglesias and Meil 2001), rapid change in the Spanish labour market (from 1992 to 2016, the percentage of Spanish women ages 25–54 with paid employment rose from 38.8% to 65.6%; Eurostat 2016; Meil et al. 2018) and the enduring influence of patriarchal culture. Families are shifting from a single-breadwinner model to models with two earners (Valiente 2010; Moreno Mínguez et al. 2017). In Spain in 1990, 62% of families fit the male-breadwinner model, as opposed to 28% in 2012 (United Nations Economic Commission for Europe, UNECE 2012). This change was possible primarily thanks to the family network (Tobío 2012) and has been understood as a strategy to reduce the risk of poverty during the economic crisis (González and Jurado 2015). The results also show that preferences for the most desirable family–work models vary according to socio-structural factors such as age, nationality, education, partner’s working status, number of children under six, religion, social class, whether one’s mother worked for pay, number of paid-work hours, hours of housework and place of residence. These factors influence men and women differently and depend on their age.

The rest of the paper is organised as follows. Section 2 reviews the literature on family model preferences to establish the theoretical framework for this study. Section 3 presents the study goal, working hypotheses, data and empirical methodology. Section 4 presents the empirical results of both the descriptive and the regression analysis. Section 5 draws conclusions and implications.

2. Literature Review

The ongoing debate on changing family models due to decline of the male-breadwinner family model has generally focused on a possible shift towards a dual full-time adult worker model with more equal gender balance (Daly 2011; Lewis et al. 2008). In a context of growing diversity of family models, it is important to attend to normative change in preferences, values and attitudes on work–family arrangements in Spain, where we observe considerable and growing support for the dual-earner model, especially among the youngest generations—among other reasons, as a result of the economic crisis Spain has suffered (González Ferrer et al. 2014; González and Jurado 2015). Studies on preferences suggest that they are context-sensitive and are constructed from cultural elements such as values. Warren et al. (2010) hold that preferences may be calculated during a process of trial and choice and are influenced by objectives, knowledge limitations and experience. We thus assumed that better understanding of preferences toward family models improves understanding of the context of the family’s reality in Spain. More precisely, Thomas and Znaniecki (1918) define attitudes as “individual mental processes which determine real or possible activity of the individual counterpart of the social value”. Following this psychological tradition, our study considers preferences and attitudes as on the same level. A series of theories argues the differences between attitudes and preferences, among them, the Theory of Planned Behaviour. Due to length restrictions and to the fact that our study was based on theories that assimilate preferences to the attitudes of the individuals (and that the Theory of Planned Behaviour is thus not the focus of our analysis), we do not develop these further. Values, on the other hand, refer to the norms or rules that tell members of the group how they should behave (Rokeach 1968; Leong and Ward 2006). Values are the cultural substratum from which attitudes emerge as a normative belief system that predisposes the individual to practice behaviour (Schwartz 2011). According to Schwartz (1999), “cultural values represent the implicitly and explicitly shared ideas about what is good, right and desirable in a society”.

Changes in family models (decline of male-breadwinner family model or increase in dual-earner couples) are related to the transformation of attitudes and values brought about by the increasing incorporation of women into higher education and the labour market, and by men’s increasing participation in family tasks (Ortega-Gaspar 2011; Schofer and Meyer 2005; Evertsson et al. 2018). The changes discussed above affect both the level of work–family conflict (Ortega-Gaspar 2013; Moreno Mínguez 2015) and values concerning family–work arrangements (Cunningham 2008; Cooke 2010; Daly 2011; Daly and Scheiwe 2010; O’Brien 2013). According to Sjöberg (2010), structural and institutional conditions can explain ambivalence toward attitudes that emphasise motherhood vs. individuality and incorporation into the labour market.

Although the theoretical models used by economists and sociologists to explain differences between men’s and women’s attitudes towards paid and unpaid work do not include some of the factors that influence attitudes towards family models, they could be useful for focusing our study. Becker’s (1985) economic theory would attribute asymmetric gender preferences toward domestic and paid work to a utility function. Women are expected to spend more hours on domestic tasks and to opt for part-time work because, among other reasons, the responsibilities for care of children and domestic tasks fall to married women more than to men. Since these tasks require a lot of attention, women tend to dedicate less time to work outside the home. In fact, Becker finds significant and different implications for the remuneration and labour status of men and women. That is, the family may provide greater incentives to choose this alternative. It is assumed that spouses’ work–family arrangements are the result of a neutral decision. Sociological theories indicate that economistic theories do not take into account the normative elements related to previously internalised gender roles (Brines 1994), the influence of social and economic contexts on men’s and women’s choices (Crompton et al. 2007), the importance of egalitarian division of labour in both the private and the public sphere or the significant role of social policies (Orloff 2002; Lewis 2009; Pfau-Effinger 2004b; Daly 2005). On the other hand, Hakim’s (2000) theory of preferences argues that men and women have different preferences based on their personal choices. For Hakim, the role of the welfare state has been over-estimated, and individuals’ preferences rather than social policies and historical legacy (traditions) determine family–work lifestyles. This interpretation emphasises the relevance of preferences and individual choice over social determinants. Based on the results of various empirical studies, Crompton and Lyonette (2005) question Hakim’s emphasis on individual choice over the constraints imposed by social context (McRae 2003; Procter and Padfield 1998; Crompton and Harris 1998).

Economists have recognised the existence of “gender norms” and have tried to find a rational explanation. According to Baker and Jacobsen (2007), spouses’ coordination on the “marital hold-up problem” is related to both the gendered division of labour and gendered preferences. This explanation justifies the existence of preferences but does not explain how individuals internalise them and adapt their behaviour according to them. In formulating an economic model that assumes gender identity, Akerlof and Kranton (2000) argue that couples tend to choose family and work options that maximise utility involving gender roles, thereby reinforcing gender identity. According to Auspurg and Gundert (2015), the influence of gender identity on individual utility may explain why women do more housework than their male partners and why women prefer a traditional family model, even when their earnings and hours of paid work are higher than or equal to their partners’. In this approach, gender identity, in the form of internalised gender norms, influences men and women’s attitudes, in turn influencing behaviour.

These theoretical and empirical analyses offer new interpretative models to explain the extent to which men’s and women’s different attitudes respond to internalised and rationalised normative principles. They do not provide much empirical evidence about how social factors—such as education, occupational status, place of residence, number of children or number of working hours—influence men’s and women’s familial and labour attitudes in different institutional, cultural and national contexts. Several studies conclude that education is a strong predictor of family–work arrangement attitudes (Crompton and Lyonette 2005; DeBacker et al. 2008; Van Wel and Knijn 2007). Studies by Gershuny et al. (2005) show that men tend to adapt to these new family expectations more slowly than women and that level of education matters. Some authors conclude that the higher is the level of education, the faster is the adaptation to more egalitarian practices, regardless of gender. This tendency is observed even in countries with a “familistic” tradition characterised by minimal social support to dual-earner families (Raley et al. 2006). Others studies show a U-shaped relationship, as is the case in Spain (Flaquer et al. 2016). In fact, families with the highest and the lowest level of education are more likely to opt for the most egalitarian family models (both parents work full-time), for different reasons (Flaquer et al. 2016).

More generally, Crompton (2006) provides evidence that gender roles and attitudes are still quite traditional in Southern Europe and that these attitudes seem to be incompatible with labour market demands. In the same vein, Mósesdottir (2001) highlights that the traditional social norms that determine the traditional gender-based division of households may be incompatible with family economic conditions and the labour market. More specifically, Tavora’s research (2012) on Portuguese society concludes that the couple’s economic need, particularly in couples with lower education levels, is a relevant factor that determines greater probability of choosing a family model in which both spouses work (especially in the case of women). Tavora insists that these imposed practices do not correlate with more egalitarian household tasks and division of care. In the same study, the author observes that women with lower levels of education in Portugal are not oriented to paid work, although there is a high female employment rate due to economic need (Tavora 2012; Torres 2008). These conclusions reject some of Hakim’s findings.

Moreno Mínguez et al. (2017) argues the need to reflect on the importance of understanding the normative level of family attitudes to comprehend the complex cultural interrelations that support contemporary family models in Spain.

Sociologists provide interesting contributions to the economic analyses of work–family preferences and division of gender roles. According to the “Multiple Equilibrium Framework” proposed by Esping-Andersen and Billari (2015), family preferences have remained stable because the dynamic of family changes results from the transformation of the gender roles that comes after a normative change. This framework explains the erosion of the male breadwinner model as a general but slow and progressive tendency. The industrial era’s Stable Family Equilibrium thus gave way to a long period of uncertainty and normative confusion affecting new family identities. Spain is in this stage. In fact, the attempt of different governments’ family policies and family relations to adapt to new family expectations is a sign of this social change.

3. Method

Based on the theoretical approaches presented above, this study aimed to analyse the attitudes towards gender roles and work–family preferences for family models from a socio-structural and gender perspective. Taking Esping-Andersen and Billari’s Multiple Equilibrium Framework as a theoretical reference, we determined the factors that underlie attitudinal support for different types of family models in Spain (one-earner and dual-earner family model). The social factors that determine these preferences were analysed for both men and women.

The indicators of attitudes toward gender roles in the empirical part of this paper are the following:

• Level of agreement with the statement “Both should contribute to household income”. Respondents were given four response options ranging from “Strongly disagree” to “Strongly agree” (

  Table 1. Percentage of respondents who agree or strongly agree with some statements on female working role working role ¹.

  Total sample	age ≤ 49	age > 49	Total	N
  Both should contribute to household income	94.7	91.5	93.2	2520
  When mom works, the preschool child is likely to suffer	47.9	58.6	52.7	2488
  Males	age ≤ 49	age > 49	Total	N
  Both should contribute to household income	92.8	90.2	91.7	1175
  When mom works, the preschool child is likely to suffer	51.1	63.9	56.8	1170
  Females	age ≤ 49	age > 49	Total	N
  Both should contribute to household income	96.3	92.5	94.6	1345
  When mom works, the preschool child is likely to suffer	45.0	54.0	49.1	1318

  ¹ The remaining answer levels are disagree and strongly disagree. Disagree. Source: Author’s calculations based on the ISSP (2012) survey: Family and Changing Gender Roles IV.

  ).
• Level of agreement with the statement “When mom works, the preschool child is likely to suffer”. Respondents were given the same response options as for the above statement (Table 1).
• Opinions on the work intensity of women when there are children under six years old at home were assessed (

  Table 2. Responses to the question “Should women work when there is a child under school age at home?” (%).

  Total Sample	age ≤ 49	age > 49	Total	N
  Work full-time	17.3	9.8	13.9	334
  Work part-time	64.1	55.1	60.0	1445
  Stay at home	18.6	35.0	26.2	630
  Total	100.0	100.0	100.0	2409
  Males
  Work full-time	20.4	9.5	15.4	173
  Work part-time	57.7	48.3	53.3	601
  Stay at home	21.9	42.2	31.3	353
  Total	100.0	100.0	100.0	1127
  Females
  Work full-time	14.6	10.1	12.6	161
  Work part-time	69.7	61.2	65.8	844
  Stay at home	15.7	28.6	21.6	277
  Total	100.0	100.0	100.0	1282

  Source: Author’s calculations based on the ISSP (2012) survey Family and Changing Gender Roles IV.

  ).

Crossing the variables “When mom works, the preschool child is likely to suffer” with opinions about work intensity when there are children under six years old at home (

Table 3. Answers to the question “Should women work when there is a child under school age at home?” according to degree of agreement with “Family life suffers when woman has a full-time job”.

	Strongly Agree or Agree			Disagree or Strongly Disagree
	Women	Men	Total Sample	Women	Men	Total Sample
Work full-time	6.9	6.3	6.6	25.5	22.1	29.0
Work part-time	65.1	51.1	58.7	61.6	66.7	56.4
Stay at home	28.1	42.6	34.7	12.9	11.2	14.6
Total	100.0	100.0	100.0	100.0	100.0	100.0
Number of observations	787	655	1442	894	456	438

Source: Author’s calculations based on the ISSP (2012) survey Family and Changing Gender Roles IV.

) was understood to indicate ambivalence towards gender roles and family model preferences.

Responses to the most desirable family model when children are under six years old were interpreted as indicators of family–work preferences in this situation (

Table 4. Most desirable parental work situation when a child is under school age.

	Age ≤ 49	Age > 49	Total	N
Total Sample
Mother at home, father FT	16.1	37.5	25.8	625
Mother PT, father FT	45.2	38.7	42.2	1022
Mother FT, father FT	14.6	9.3	12.1	294
Mother PT, father PT	23.1	13.8	18.9	457
Mother FT, father PT	1.1	0.8	1.0	24
Total	100.0	100.0	100.0	2422
Males
Mother at home, father FT	19.3	44.9	30.9	351
Mother PT, father FT	44.5	34.0	39.8	452
Mother FT, father FT	14.8	9.7	12.5	142
Mother PT, father PT	20.4	10.5	15.9	181
Mother FT, father PT	1.0	1.0	1.0	11
Total	100.0	100.0	100.0	1137
Females
Mother at home, father FT	13.2	31.0	21.3	274
Mother PT, father FT	45.7	42.8	44.4	570
Mother FT, father FT	14.3	8.9	11.8	152
Mother PT, father PT	25.5	16.7	21.5	276
Mother FT, father PT	1.3	0.7	1.0	13
Total	100.0	100.0	100.0	1285

Source: Author’s calculations based on the ISSP (2012) survey Family and Changing Gender Roles IV.

).

The family models were defined by the criteria of the work–family arrangements (Pfau-Effinger 2004a, 2004b, 2012). The classifications used in this study were based on Pfau-Effinger’s typology of family cultural models in which both members of the couple must balance work and family (Pfau-Effinger 2004a, 2004b, 2012). We adapted this classification to Spain. In Pfau-Effinger’s typology, the main family models in post-industrial societies are: (1) the male-breadwinner family model, in which females work part time and care for the children; (2) the dual-earner/external care provider family model (in which care is mainly provided by the state or the market); (3) the dual-earner family model, in which children are cared for by extended family members; and (4) the male-breadwinner family model, in which mothers stay at home and are the main people responsible for caring for children. Cultural models of ideal families can vary among countries based on gender politics and family cultural tradition. Pfau-Effinger (2004b) affirms that these variations are due to differences deeply rooted in the cultural and historical development patterns of European countries. We adapted the previous classification to the labour conditions and family models characteristic of Spanish society.

The different family models studied are: Type 1, families in which the mother stays at home and the father works full-time (M at home/F FT, here understood as the most traditional model and that closest to patriarchal culture.); Type 2, the mother works part-time and the father works full-time (M PT/F FT); Type 3, both the mother and the father work full-time (M/F FT); and Type 4, both the mother and the father work part-time (M/F PT). Type 5, the father works part-time and the mother works full-time (F PT/M FT) has not been studied due to its small sample size.

3.1. Objectives and Hypotheses

The general hypothesis was that attitudes towards the different family models reflect the intricate crossing of values inherited historically (dominance of patriarchal culture) and values of a society advancing toward new ways of distributing work and family responsibilities. In other words, attitudes may be due partly to the influence of the culture (Pfau-Effinger 2004b) in the country studied, Spain. H0: This study considered attitudes toward gender roles and preferences for family models as determined to a significant extent by socio-structural factors and as determined differently for women and men. If this were the case, the cultural influence hypothesis would be confirmed. This general hypothesis was tested more concretely through the following working hypotheses:

Hypothesis 1 (H1).

Spanish women are expected to show greater preference than Spanish men towards the family model in which the mother works part-time and the father full-time and less preference for the breadwinner family model when the family has children under six years old at home. In the case of Spanish society, this preference could be understood as a result of the influence of traditional culture (Pfau-Effinger 2004b) historically internalised differently by Spanish women and men (women have internalised the role of caretakers and men that of breadwinners).

Hypothesis 2 (H2).

Due to the ambivalent process of social and family change, contradictions are expected among attitudes toward ideal work situation for mother and father when children are under six, gender roles and the interviewees’ preferences for different family models. Such attitudinal ambivalences are considered as indicators of social changes related to family preferences. This assumption aims to enrich Multiple Equilibrium Theory (Esping-Andersen and Billari 2015).

Hypothesis 3 (H3).

Socio-structural factors such as age, education level, immigrant condition, religion status or social class influence family model preferences of women and men differently. In contrast to the Individual Preference Theory (Hakim 2000), socio-structural factors (Crompton et al. 2007) exert a significant influence on attitudes towards family model preference.

3.2. Data and Empirical Methodology

Based on the theoretical framework presented above, this study analysed the attitudes of the Spanish population concerning family models. The empirical analyses were performed using individual-level survey data for Spain from the 2012 “Family Changing Gender Roles” module of the ISSP. The working sample of 2595 individuals is nationally representative of the Spanish adult population (over 18 years of age).

We performed two types of analysis. Firstly, we implemented an exploratory analysis by gender and age to describe the Spanish population’s attitudes towards the role the mother should play with respect to paid work. This analysis provided information about attitudes related to gender role division and the ideal work situation for the mother and father when the children are under six. It also enabled detecting ambivalent attitudes among the Spanish population. Secondly, a regression analysis was conducted to determine the different factors that correlate with predisposition to consider a particular family model as the most desirable. The ISSP (2012) Spanish survey measures cultural influence through individual preferences for family models captured using Question Number 9 of the questionnaire:

Consider a family with a child under school age (“Under school age” means under the age of regular/compulsory school). What, in your opinion, is the best way for them to organise their family and work life? (Please tick one box only).

1. 

The mother stays at home and the father works full-time. (Mother at home/F FT);

2. 

The mother works part-time and the father works full-time. (M PT/F FT);

3. 

Both the mother and the father work full-time. (M/F FT);

4. 

Both the mother and the father work part-time. (M/F PT);

5. 

The father works part-time and the mother works full-time. (F PT/ M FT);

6. 

Can’t choose.

This set of response options is in line with the family model typologies of Pfau-Effinger (2004b) and Naldini (2003) in attending to work–family arrangements.2013

The proposed econometric specification for data processing was a multinomial logit model (see Greene (1998) for details on estimation process and interpretation of marginal effects for this non-linear model). Our dependent variable, Y, included the possible responses, i.e., “Mother at home/father working full-time” (Y = 0), “Mother part-time/father full-time” (Y = 1), “Mother full-time/father full-time” (Y = 2) and “Mother part-time/father part-time” (Y = 3). Respondents who indicated the “Father part-time/mother full-time” option were eliminated from the sample because they were so few (1% of the total sample). We assumed that individual i is rational in the formation of his/her preferences and would choose the alternative j if and only if U_ij > U_ik for all j ≠ k, where U_ij and U_ik represent the utilities associated with the options j and k, respectively. This interpretation adapted the analysis to the economic model of generation of preferences.

If N represents the number of respondents included in the sample, the probability in the multinomial model of preferring alternative j is given by

$$P(Y_i=j)=\frac{\mathit{exp}\left({\beta }_j^{\prime }{X}_i\right)}{1+{\displaystyle \sum _{j=1}^{j=3}\mathit{exp}\left({\beta }_j^{\prime }{X}_i\right)}}i=1,2,\dots ,N\mathrm{and}j=1,2,3.$$

(1)

where β_j is the vector of parameters associated with each alternative j and X_i is the vector of explanatory variables observed for each individual—variables assumed to be related to the generation of preferences. The structural variables included in vector X_i are the following: sex, age, non-Spanish origin, educational level and subjective health status (on a scale of 1–4). This last variable was added as a control for the individual’s mood—possible optimism or pessimism—which could affect the responses. Other structural variables included are the couple’s work intensity, number of preschool children in the household, number of children aged 6–17 living at home, religion, subjective social class (scale from 1 to 10) and whether the mother had or has any paid work. Respondent’s weekly paid work hours, type of environment in which the individual lives and region of residence completed the set of independent variables. This set of explanatory variables was drawn from earlier studies on the influence of social factors on gender roles and family model preferences, as explained in detail in Section 2. We believe it is important to know how these structural variables behave as determinants of Spanish women’s and men’s preferences towards family models. It was understood that the analysis of the preferences towards family models offered information about the cultural values that contribute to construct the idea of the “ideal family form”, since family models can capture the influence of a normative gender culture (Pfau-Effinger 2012). For example, younger and more educated individuals may have been exposed to a more egalitarian culture, whereas the opposite may be true of more religious individuals or those from more traditional countries.

Table A1. Summary statistics of variables included in multinomial models for most desirable parental childcare option when child is under school age. Total sample ^a.

Variable	Mean	Variable	Mean
Female	0.528	Place of residence:
Age	48.4	A big city	0.184
	(17.4)	Suburbs or outskirts of a big city	0.093
Non-Spanish origin	0.089	A town or a small city	0.256
Education level:		A country village	0.438
No formal education	0.080	A farm or home in the country	0.029
Primary school	0.195	Mother’s work harms children	0.532
Lower secondary	0.235	Region:
Upper secondary	0.124	Andalucía	0.182
Post-secondary	0.156	Aragón	0.026
Lower tertiary	0.110	Asturias	0.025
Upper tertiary	0.099	Baleares	0.021
Health status (good/very good)	0.368	Canarias	0.032
Working Status of partner:		Cantabria	0.014
No partner	0.269	Castilla-La Mancha	0.046
Currently in paid work	0.394	Castilla-León	0.062
Paid work before but not now	0.264	Cataluña	0.171
Never had paid work	0.074	Comunidad Valenciana	0.114
Number of toddlers (0–5 years) in household	0.2	Extremadura	0.029
	(0.5)	Galicia	0.074
Number of children (6–17 years) in household	0.4	Madrid	0.093
	(0.8)	Región de Murcia	0.036
Religion:		Navarra	0.014
No religion	0.208	Pais Vasco	0.051
Catholic	0.752	La Rioja	0.010
Other	0.040
Social class (appreciation reported in survey):
Levels [1–4]	0.255
Level 5	0.434
Levels [6–7]	0.288
Levels [8–10]	0.023
Mother worked for pay	0.375
Working hours reported on survey:
Not working	0.498
[1–15]	0.021
[16–35]	0.102
[36–45]	0.266
>46	0.113
Hours of household work reported on survey	15.8
	(15.2)
Number of observations	2090	Number of observations	2090

^a Standard Deviation in parentheses for continuous variables. Source: Author’s calculations based on ISSP (2012) survey Family and Changing Gender Roles IV.

(see Appendix A) contains the descriptive statistics of the above full set of factors.

4. Results

The empirical results are presented in two parts. The first describes the attitudes related to gender roles division among the Spanish population. The second presents the social factors that influence Spanish interviewees’ preferences towards a specific family model.

4.1. Descriptive Analysis of Attitudes Related to Gender Roles

Table 1 shows the percentages of respondents who agreed or strongly agreed (hereinafter referred to as “agreed”) with certain statements about women’s work role. The figures give an idea of the Spanish population’s attitudes toward gender role division. We observed very broad consensus among respondents that both partners should contribute to household incomes (above 90%), especially among women and the youngest group (under 49). These results contrast with the 52.7% of respondents who indicated that the preschool child is likely to suffer if the mother works. We thus observe contradictions in attitudes toward gender roles, partially supporting Hypothesis H2. A higher percentage of young people than older people think this way (47.9% vs. 58.6%). These differences between the two age groups indicate both a rupture with the past status quo, in which predominant preference for the patriarchal family model was the norm, and a step toward a new situation of vagueness in family model preferences, as we found throughout this study.

Table 2 shows the interviewees’ answers to the question “What should the work intensity of women be when there are children under the age of 6 at home?” Of the total sample, slightly more than one quarter of the interviewees (26.2%) agreed with the statement that mothers with this responsibility should stay at home. These responses show a tendency towards preference for a two-earner couple family model, as 74% of interviewees chose the dual-earner family model. More specifically, 60% prefer a model in which the mother works part-time and 13.9% a model in which she works full-time. These results support the explanation of the dual-earner couples or the “one and a half” model (Trifiletti 2012; Esping-Andersen and Billari 2015).

We also observed behavioural differences by age, as 35% of interviewees over 49 affirmed that the mother should stay at home. This percentage is almost twice the result for interviewees under 49. By sex, men prefer that the mother stay at home more strongly than women (31% of men vs. 21% of women). The women surveyed opted more than men for part-time jobs (H1 is supported). Interestingly, men over 49 years of age are more traditional (they prefer the mother to stay at home) than men under 49, a result that may mark the social transition Spain is undergoing (Hypothesis H2 is supported).

The data also support the presence of ambivalent attitudes toward gender roles. Contradictions exist between attitudes toward gender roles when there is a child under six at home and the way that mother should participate in the labour market (Table 3). Although Spaniards expressed a preference for a dual-earner family model, they still adhere to the cultural heritage of the male-breadwinner model, a system in which women are expected to be the primary caregivers. Women show a slightly higher level of ambivalence than men. Almost 7% of women think that family suffers when the woman has a full-time job but also affirmed that women should work full-time: 65% of women and 51% of men answer that family suffers when the mother works full-time but also affirm that women should work part-time. This evidence supports Multiple Equilibrium Theory (Esping-Andersen and Billari 2015) and thus Hypothesis H2. The implicit attitude reveals a form of adaptation to the labour and family expectations that fall to women understood as welcoming the new changes, but under the pressure of tradition. The traditional normative model is still present and emerges in the interviewees’ attitudes.

4.2. Multinomial Estimates for Family Model Preferences

Table 4 shows the interviewees’ opinions about the most desirable (ideal) parental work situation when having children under school age (0–5 years old). We chose this question to analyse the influence of the Spanish social context, characterised by minimal social protection for the care of children under six years old. The analysis presented in this section aimed to determine the factors that explain these preferences. It is important to stress that we analysed a hypothetical option.

The results agree with those analysed previously (Table 2). Note that 42% of interviewees opted for a family model in which the mother works part-time and the father full-time. Only 26% of interviewees chose a traditional model (mother at home and father working full-time). It is significant that nearly 20% of Spaniards indicated Mother and Father PT as the most desirable model when children are under five. It is also not insignificant that 12% chose the model of both parents FT. These results reveal a diversity of family models in Spain on the ideal level as well, that is, on the level of preferences. It is interesting to pause over the very small number of Spaniards who expressed a desire for the model Mother FT and Father PT. Given these results, we should ask what reasons underlie this attitude. They might be related to the nature of women’s participation in the Spanish labour market. They might be the result of availability of work on the market, its flexibility or lack of flexibility. Finally, they might stem from the different position of women in the labour market or even reflect the presence of different expectations for men’s and women’s roles.

By age, we see that the most desirable models for interviewees under 49 are primarily the two in which both members of the couple work (45.2% Mother PT/Father FT; 23% Mother PT/Father PT). It is interesting to see that the model Mother at home/Father FT loses weight in this younger group (16%) while the model Mother and Father FT (14.6%) gains in preference. These results support the thesis of Esping-Andersen and Billari (2015) while also providing more details about change occurring at the level of gender attitudes and their influence on preferences for family models in Spain.

Disaggregating by sex, we note a slightly higher percentage of women than men opting for part-time jobs for women and full-time jobs for men (44% and 40%, respectively). The flip side of this preference is that the percentage of men who chose the most traditional family model is higher than the percentage of women (31% and 21%, respectively). It is interesting that more women than men (21.5% and 15.9%) chose as most desirable the model in which both members of the couple work part-time, indicating that more women tend to prefer a model that shares roles in an egalitarian way (at least in the area of work). Interesting differences by sex emerge when we control for age. Among men over 49, 45% chose the traditional family model as the most desirable family model, whereas only 19% of men under 49 think this way. At the same time, men are more traditional than women of the same age (only 13.2% of women share this view). The models in which both father and mother work PT is the most desirable for 25.5% of women under 49 and 20.4% of men under 49. The most striking difference involves the model Mother and Father FT, registered between the two age groups studied, with hardly any differences by sex when we control for age. Specifically, 14.8% of men and 14.3% of women under 49 chose this option as the most desirable, vs. 9.7% and 8.9% of men and women over 49, respectively. This evidence partially fulfils Hypothesis H1, as sex exerts an influence, but that influence is attenuated when we control for age.

The multinomial model to explain the probability of opting for a particular family model was estimated both for men and women separately (

Table 5. Multinomial estimation by gender for the most desirable parental work situation option when child is under school age (marginal effects ^a).

	Women b				Men b
Variables	M home/F FT	M PT/F FT	M FT/F FT	M PT/F PT	M home/F FT	M PT/F FT	M FT/F FT	M PT/F PT
Age	0.003 ***	0	−0.001 ^	−0.002 ^	0.005 ***	−0.002 ^	0	−0.003 ***
Non-Spanish origin	0.111 **	−0.133 **	0.048 +	−0.026	0.073	0.219 ***	−0.122 **	−0.171 **
Educational degree [ref.: No formal education]
Primary school	−0.015	−0.006	−0.037	0.058	−0.051	0.046	−0.052	0.056
Lower secondary	−0.049	−0.008	−0.024	0.082 ^	−0.005	0.095	−0.113 +	0.023
Upper secondary	−0.131 **	−0.034	0.018	0.148 **	−0.094 +	0.085	−0.049	0.058
Post-secondary	−0.122 **	−0.006	−0.001	0.129 **	−0.139 *	0.119 +	−0.046	0.066
Lower-level tertiary	−0.149 ***	−0.089	0.042	0.196 ***	−0.238 ***	0.169 *	−0.004	0.073
Upper-level tertiary	−0.110 ^	−0.209 **	0.041	0.278 ***	−0.251 ***	0.160 *	−0.098	0.189 **
Health status (good/very good)	0.014	0.007	0	−0.021	0.064 **	−0.033	0.008	−0.038 ^
Occupational status of partner [ref.: Without Partner]
Currently in paid work	0.013	0.017	−0.013	−0.016	−0.132 ***	0.099 **	0.034	−0.001
Paid work before but not now	0.047 ^	−0.025	−0.001	−0.02	−0.024	0.035	−0.051 *	0.041
Never had paid work	0.259 *	−0.271 *	−0.139 ***	0.151	0.031	−0.018	−0.067 **	0.055
Number of toddlers [0–5] in household	−0.031	0.091 **	−0.001	−0.059 *	0.053 ^	0	−0.028 +	−0.025
Number of children [6–17] in household	0.011	−0.024	0.011	0.002	0.002	−0.004	−0.001	0.003
Religion [ref.: No religion]
Catholic	0.082 **	−0.028	0.015	−0.069 *	0.105 ***	0.027	−0.052 *	−0.079 **
Other	0.266 ***	−0.121	−0.015	−0.131 ^	0.290 ***	−0.087	−0.049	−0.154 **
Social class [Ref.: levels [1–4]]
Level 5	−0.016	−0.037	0.066 **	−0.012	0.007	−0.029	0.063 **	−0.042
Level [6–7]	0	−0.042	0.003	0.039	0.01	−0.049	0.058 *	−0.019
Level [8–10]	−0.150 **	0.112	0.151 *	−0.113 *	−0.106	0.055	0.196 *	−0.145 ***
Mother worked for pay	−0.012	−0.015	−0.003	0.03	−0.022	0.026	0.007	−0.012
Working hours reported on survey [ref.: Not working]
[1–15]	−0.066	0.053	−0.027	0.04	−0.123 ^	0.093	0.015	0.015
[16–35]	−0.067 *	0.102 **	−0.025	−0.011	−0.080 +	−0.180 ***	0.101 **	0.160 ***
[36–45]	−0.079 **	0.118 ***	−0.001	−0.038	0.01	−0.049	0.015	0.023
46+	0.032	−0.002	−0.022	−0.009	0.026	−0.091 *	0.048 ^	0.016
Hours of household work reported on survey	0.001	0	0	−0.001	−0.003 **	0.001	0	0.001 +
Place of residence [ref.: Big city]
Suburbs or outskirts of a big city	−0.060+	0.049	0.027	−0.017	−0.061	0.027	0.073 *	−0.039
A town or a small city	0.021	−0.081 *	−0.001	0.062 ^	−0.081 *	−0.015	0.104 ***	−0.008
A country village	0.03	−0.013	−0.022	0.005	−0.029	0.039	0.015	−0.025
A farm or home in the country	−0.025	0.052	0.021	−0.048	0.065	−0.059	−0.051	0.045
Mother’s work harms children	0.054 **	0.057 *	−0.084 ***	−0.028	0.208 ***	−0.068 **	−0.119 ***	−0.021
Numbers of observations	1063				960
Pseudo-R2	0.1337				0.2042
Log pseudolikelihood	−1194.388				−970.58267
Wald test (chi2)	1778.37 ***				848.32 ***

^a Marginal effects calculated following Greene (1998). Estimations on weighted data. The abbreviation ref. indicates reference category for categorical variables. ^b Mother at home/Father full-time (M home/F FT); Mother part-time/Father full-time (M PT/F FT); Mother full-time/Father full-time (M FT/F FT); Mother part-time/Father part-time (M home/F FT). The symbol ***/**/*/^/+ means statistical significance at 1%/5%/10%/15%/20% level. Source: Author’s calculations based on the ISSP (2012) survey Family and Changing Gender Roles IV.

) and for the total sample (

Table A2. Multinomial estimation for most desirable parental childcare option when child is under school age (marginal effects). Total sample ^a,b.

Variables	M Home/F FT	M PT/F FT	M FT/F FT	M PT /F PT
Women	−0.062 ***	0.009	−0.024 ^	0.077 ***
Age	0.004 ***	−0.001	−0.001	−0.002 ***
Non-Spanish origin	0.070 **	0.024	−0.011	−0.083 *
Education level [ref.: No formal education]
Primary school	−0.008	−0.003	−0.044	0.056 ^
Lower secondary	−0.005	0.013	−0.065 ^	0.056 ^
Upper secondary	−0.102 **	−0.008	−0.006	0.116 ***
Post-secondary	−0.115 **	0.028	−0.019	0.106 ***
Lower tertiary	−0.176 ***	0.003	0.022	0.150 ***
Upper tertiary	−0.171 ***	−0.058	−0.019	0.248 ***
Health status (good/very good)	0.035 *	−0.008	−0.001	−0.026 +
Working status of partner [ref.: Without Partner]
Currently in paid work	−0.044 *	0.066 **	−0.002	−0.02
Paid work before but not now	0.036 ^	−0.002	−0.033 ^	−0.001
Never had paid work	0.102 **	−0.073 ^	−0.083 ***	0.054
Number of toddlers [0–5] in household	0.007	0.040 ^	−0.007	−0.039 *
Number of children [6–17] in household	0.004	−0.015	0.009	0.002
Religion [ref.: No religion]
Catholic	0.093 ***	0.008	−0.018	−0.083 ***
Other	0.259 ***	−0.061	−0.047	−0.151 ***
Social class [Ref.: levels [1–4]]
Level 5	−0.005	−0.041 +	0.068 ***	−0.023
Level [6–7]	0.001	−0.048 ^	0.034 *	0.013
Level [8–10]	−0.102 ^	0.072	0.147 **	−0.117 ***
Mother worked for pay	−0.019	−0.006	0.004	0.021
Working hours reported on survey [ref.: Not working]
[1–15]	−0.073 +	0.042	−0.008	0.038
[16–35]	−0.045	−0.004	0.01	0.038
[36–45]	−0.018	0.017	0.008	−0.006
>46	0.025	−0.044	0.016	0.002
Hours on household work (survey)	0	0.001	0	0
Place of residence [ref.: Big city]
Suburbs or outskirts of a big city	−0.058 ^	0.044	0.038+	−0.025
A town or a small city	−0.031	−0.042	0.043 *	0.03
A country village	0.005	0.011	−0.008	−0.009
A farm or home in the country	0.025	−0.012	−0.019	0.006
Mother’s work harms children	0.124 ***	−0.008	−0.097 ***	−0.019
Numbers of observations	2023	2023	2023	2023
Pseudo-R2	0.1381
Log pseudolikelihood	−2254.0701
Wald test (chi2)	2431.47 ***

^a Marginal effects calculated following Greene (1998). Estimations on weighted data. Abbreviation ref. indicates the category of reference for categorical variables. ^b Mother at home/Father full-time (M home/F FT); Mother part-time/Father full-time (M PT/F FT]); Mother full time/Father full time (M FT/F FT); Mother part-time/Father part-time (M home/F FT). The symbol ***/**/*/^/+ means statistical significance at 1%/5%/10%/15%/20% level. Source: Author’s calculations based on ISSP (2012) survey Family and Changing Gender Roles IV.

, Appendix A). This last model was estimated solely to detect gender differences in family model preferences. For the total sample, the results show that women were 6.2% less likely than men to choose the mother at home/father working full-time model. The probability of preferring the model in which the mother and father work part-time is higher in the case of women (7.7%). In contrast, women were 2.4% less likely than men to choose the model of mother and father working full-time. These results are in line with those shown in Table 3 and support part of Hypothesis H1, that women show lower preference than men for the male-breadwinner family model but that gender is not a statistically significant variable that determines preference for the mother part-time father full-time model, although it is in the other models (Table A2, Appendix A).

Table 5 shows the estimated marginal effects when the multinomial model is performed separately for women and men. Three general conclusions emerge. Firstly, most of the regressors show explanatory power for probability to opt for one family model or another. Secondly, the sign, magnitude and statistical significance of the marginal effects differ for women and men. Thirdly, in the case of women, most of the significant effects observed correspond to the traditional model and the model in which both women and men work part-time. In the case of men, the influence of the covariates considered extends to the other family models.

For both women and men, age is positively related to the probability of choosing the most traditional family model, although the marginal effect is greater for men. Age has a negative effect, however, on probability of choosing the other models. Foreigners are more likely than native Spaniards to prefer the more traditional model. Women of non-Spanish origin were 11.1% more likely than Spanish women to choose this model (the variable is statistically significant for women but not for men) and 4.8% more likely than Spanish women to choose the model M FT/F FT. Women of non-Spanish origin were less likely (13.3%) than Spanish women to indicate the M PT/F FT model as more desirable. Non-Spanish men showed a higher probability than Spanish men (21.9%) to choose the model M PT/F FT as the most desirable. Although the respondents’ origin was quite diverse, their countries of origin shared an attachment to patriarchal cultures.

Education level is a strong predictor of attitudes to family–work arrangements for both women and men. Interviewees who attended school beyond primary level were less likely to choose the traditional model as the most desirable. This evidence aligns with that observed by Crompton and Lyonette (2005), Van Wel and Knijn (2007) and DeBacker et al. (2008) and reinforces the argument by Cloïn (2010) that higher education has a positive impact on women’s employment and the egalitarian family arrangement. The peculiarities of the Spanish labour market and institutional policies of family conciliation could influence different preferences of men and women by educational level. Women with education levels from upper-secondary to upper-level tertiary showed lower probability of choosing the “mother at home father working full-time” model than women who do not have formal education. This group of women also shows greater preference for a model in which both members of the couple work part-time. This result could be explained by the fact that these women are younger (the results show that the lower is their age, the more likely the respondent is to choose this model) and have had the opportunity to have a formal education (from upper secondary to upper-tertiary level), possibly reflecting new attitudes toward work and family. Men with post-secondary to upper-level tertiary education, in turn, show a higher probability than those without university education to prefer the model in which the mother works part-time and the father full-time. These men chose a two-breadwinner model that maintains the traditional division of work and family roles between the sexes.

The partner’s occupational situation also determines family model preferences. Men with working partners have a lower probability of choosing the male breadwinner model as the best option and a higher probability for choosing the mother part-time/father full-time model (always when compared those who have no partner). The presence in the household of children under six years seems to influence women positively to opt for the mother part-time/father full-time model but to have a negative influence on desirability of the model in which both spouses work part time. Economic or structural reasons could explain the lack of harmony between these two results if we attribute the latter to the Spanish labour market. Men in households with children under six were 5.3% more likely than men without younger children to choose the male-breadwinner family model and are less inclined to the mother and father full-time model.

Taken as a whole, these results align with human capital theory, which emphasises that factors such as education, number of children, occupation and economic status can explain the preferences for different family models. Human Capital Theory partially supports the hypothesis since it gives a sense (although not a complete explanation) of a process in which parents (mothers and fathers) are open to education even when they express ambivalence about it. The analysis adds more details about how those factors influence women’s and men’s preferences in different ways.

Both men and women who profess religious belief showed a greater preference for the more traditional model than those who profess no religion, and the marginal effect is greater for those with a religion other than Catholicism. This population was less likely to choose part-time work for both partners. Subjective social class also determines the option chosen. Both men and women who consider themselves as upper-class (levels 8–10 in the subjective scale) and lower-middle class (level 5) were more likely to opt for one of the more egalitarian models (father and mother working full-time) than those who consider themselves lower class (levels 1–4). There may be different reasons for this result (development and/or demand at the professional level for the former, economic need for the latter). Although the effect is not very significant, we observed that women whose mothers have work experience prefer reduced-hour jobs for both spouses.

Women who work 16–45 h a week showed a lower probability than others of choosing the most traditional model and a greater predisposition for the “mother part-time and father full-time” model. Women who work more than 45 h a week may be experiencing a situation of exhaustion due to long working hours that influences their preference towards a more traditional family model. For males, the probability of choosing the male-breadwinner model decreased when their level of involvement with care and domestic tasks increased. This result could be interpreted as a change in the Spanish male view of gender roles. Place of residence is relevant in explaining preference for a specific family model. Women living in suburbs or on the periphery of large cities were less likely to opt for the more traditional model than those living in a big city, probably for economic reasons.

As expected, the respondents’ attitudes toward the ideal work situation for mother and father when children are under six have significant explanatory power for family model preferences. Women who agree or strongly agree with the statement “When mom works, the preschool child is likely to suffer” were more likely to opt for traditional family models over the model in which both mother and father work. Interestingly, men with these attitudes were less likely to opt for the model in which the mother works part-time and the father full-time than are those who disagreed or strongly disagreed with the statement.

In general, the multinomial analysis confirms Hypothesis H3. Socio-structural factors influence women’s and men’s preferences towards family models differently, as described in this section.

5. Conclusions

This study enabled us to confirm that attitudes to gender roles and family model preferences are determined to a significant extent by socio-structural factors supporting the thesis of Crompton et al. (2007) and rejecting that of Hakim (2000). It also provides new insight by confirming that this influence differs for men and women. This insight in turn enabled us to determine the influence of culture on attitudes toward gender roles and preferences for family models.

One of the main conclusions to emerge from this study reinforces the theory of gender culture (Pfau-Effinger 2012, p. 533): “The gender culture of a country comprises one or more dominant cultural family models at the macro-level of society. These are fundamentally cultural values that together construct the idea of the ‘ideal’ family form. Cultural family models connect cultural values in relation to the family-employment relationship of the adult family members, the gender division of labour within the family, and the most suitable form of care for children”.

We thus propose a new way of classifying family models to complement Pfau-Effinger’s classification of cultural family models in order to incorporate both the predominant family models observed (real behaviour) and the predominant family models preferred (ideal level). This classification should fit the “gender culture” of each country better. Such a classification would include Spain in the “dual-earner breadwinner model with childcare of children under age 6 in the extended family” (reality) and in the “male breadwinner/female part-time work and care model” (preferences, ideal level).

These results point to an unresolved dilemma in attitudes toward work–family arrangements. They support Sjöberg’s (2010) thesis that the greater is the “lag” between rates of change in female education and in institutions that could reconcile paid work with motherhood, the greater is the attitudinal ambivalence concerning gender roles. These ambivalent attitudes reflect the uncertainty resulting from changes in the normative family in Spanish society and the lack of family policies to improve work–family balance and gender equality. Women’s preferences showed a higher level of ambivalence. In some cases, they maintain traditional attitudes towards family and gender roles. At the same time, they expressed a lower level of agreement with the most traditional family model. These results generally agree with the “Multiple Equilibrium Framework” developed by Esping-Andersen and Billari (2015), while also providing an interesting advance over this theory. Multiple Equilibrium Framework theory affirms that family preferences have remained stable because the dynamic of family changes is a result of the transformation of gender roles that occurs following a normative change. More precisely, the results analysed here enable us to see the dynamic of the trend in attitudes to gender roles that runs parallel to the changes occurring in performance of men’s and women’s roles. These results are also redrawing the map of preferences for family models in Spain, which we understand as tending to be more dynamic than is usually thought and that could break the state of multiple equilibrium proposed by Esping-Andersen and Billari. It remains to be seen whether the change occurring at the level of attitudes develops into greater diversity of family models in reality.

The ambivalent attitude toward gender roles observed in this study sheds light on the situation of uncertainty created by these changes in preferences toward the family models that Spaniards are experiencing. The results enable us to see some changes in attitudes toward gender roles which, following Esping-Andersen and Billari, shift toward changes in preferences for family models.

The results confirm that age and education influence women’s ambivalent expectations concerning desirable family models more strongly than men’s. Preference for the family model of the part-time mother and father is growing in younger generations (most notably, this factor is statistically significant in the case of men), a finding that could be understood as an indicator of attitudinal changes that might simultaneously influence future family model preferences. The preference for a family model of mother and father working full-time is also growing among young people, but here the factor is statistically significant only for women. The results show that the younger are the women, the higher is the probability of choosing a model that offers the same work situation for men and women when there are children under six years old at home. The higher is the level of education and the lower is the age, the higher is probability that women opt for a hybrid work–family model compared to men. Men’s choices of more egalitarian vs. traditional family models tend to be based on education level and age. In sum, structural factors such as education level and age have helped partially to correct gender inequality concerning the desirable family model, although the results reveal contradictions. However, we see some indications of traditional gender stereotypes associated with low educational level, age, religious beliefs and condition as immigrants.

The study shows that attitudes towards the different family models reflect the intricate crossing between traditional and modern culture (understood as closer to vs. farther from patriarchal influence). The results contribute to supporting Pfau-Effinger’s Theory of the influence of historical, cultural, institutional and structural factors on family–work arrangements, more specifically in this case on preferences for family models. This study makes an important contribution to sociology of the family because it offers an important explanation that has not been developed before and strengthens the idea shared by many researchers that gender differences cannot be explained solely by economic factors.

The idea that analysis of attitudes toward gender roles can help to explain how mothers and fathers construct the work–family dilemma from a microperspective is frequently present in the literature about southern European countries (Moreno Mínguez 2010; Crompton and Lyonette 2005; Fortin 2005; Steiber and Haas 2009; Hobson 2011; Stam et al. 2013). Studies that focus on such contradictions in these contexts are less common, however, even though the traditional family model has had a significant presence in the singular construction of men’s and women’s familial and labour expectations in southern European countries.

The study shows contradictory attitudes toward gender roles that seem to be resolved through preferences for a “hybrid” or “adaptive” family model as the ideal work situation for mother and father when children are under six (Mother Part-time, Father Full-time). These results are in line with the thesis of Akerlof and Kranton (2000) that couples tend to opt for family and labour options that maximise utility related to the gender roles, a thesis that reinforces gender identity. This thesis is, in turn, strongly related to the explanation by Tavora (2012), Mingione (2001) and Moreno Mínguez (2013) of southern European family models. Their conclusions enabled us to understand how these countries’ historical and cultural legacies have helped to produce the coexistence of a traditional family cultural model based on traditional gender roles with growing female participation in the labour market and a traditional value system associated with care and family responsibilities. Family–work preference models are related to the cultural roots of the prevailing family models in each cultural context. According to this idea, gender identity, in the form of internalised gender norms, influences men’s and women’s preferences.

This article offers a more comprehensive understanding of the relationship between attitudes toward gender roles and family model preferences. It contributes to explaining the relative ambivalence reflected in the finding that women’s preferences on family and work do not yield the same returns in family model expectations as do men’s in Spain, a country characterised by familism and an undeveloped Welfare State.

We acknowledge that this empirical study was not without limitations. The most obvious of these is that the data analysed are cross-sectional; that is, we have a single time observation for each survey respondent. This means that some risk is involved in translating the associations estimated into relationships of causality. In fact, in estimating equations in reduced form with this type of data, the correlations estimated would be consistent with various sociological—or psychological—mechanisms. Having panel data available would have helped to mitigate this problem (Hsiao 2007). It would also have enabled us to control for the impact of omitted or unobserved variables. However, we believe that the results presented and our discussion of them constitute a good foundation for deepening the research on attitudes towards gender roles and family model preferences within a normative and socio-structural framework.

In addition, other structural factors could be included when explaining family-model-related preferences and ambivalences. Due to the data used, however, it was not possible to include additional factors in the empirical analysis performed here. We would indicate especially the specific characteristics of the Spanish labour market (very long working days, inadequacy of school schedules to match working hours, etc.) and the limitations of family policies to encourage work–family balance. All of these conditions are very likely to impinge on preferences regarding work–family models, especially to encourage preference for the “one and a half” worker model (he works full-time, she works part-time). The characteristics of the Spanish labour market and the difficulties of combining work and care may also influence men’s and women’s respective preferences.