ABSTRACT
This article seeks to re-open the consensus concerning the interrelationship between family values and public support for government versus private family provision. We offer new results from analyses of 2001 data from the International Social Survey Programme for a wide range of countries and world regions. Refining conventional scholarly wisdom, Southern European publics' have high levels of traditional family values, but mainly in contrast to other European countries; familism is also notably strong in Eastern Europe and several of the English-speaking democracies. Even more surprising, family values support is strongly and positively associated with support for public child care provision. We discuss implications of results for understanding the nature of public attitudes and familism in cross-national perspective, and the limits of theorizing identifying family values as the primary cause of welfare state development in Southern Europe.
1. Introduction1
Italy, Spain and other Southern European countries have been seen by scholars as exemplifying the most traditional model of family in the developed world. An influential tradition of scholarship points to Mediterranean ‘familism’ as key to explain the demographic and social policy distinctiveness of Southern Europe (Jurado and Naldini 1997; Reher 1998; Moreno 2002; Bettio et al.2006; Callegaro and Pasini 2008; Gal 2010). There, low fertility and traditional caregiver arrangements coexist with cash-heavy welfare programs and limited service provision for youth and the aged.
Despite the widespread acceptance of Mediterranean familism among scholars, the coming of ‘post-modern’ family forms and the growing importance of non-family cleavages chafes with the assumed centrality of traditional family values to policy and society in Southern Europe. Moving beyond the Mediterranean rim, the widely debated restructuring and/or policy drift of modern welfare states suggests a second set of developmental processes that may exert pressures complicating the interrelationship between family and welfare institutions.
The presumption of much comparative scholarship is that the national publics of Southern Europe endorse family-based social provision, and that these patterns of attitudes are connected to, and reinforced by, traditional family structures (Reher 1998; Callegaro and Pasini 2008). But is this indeed the case?
In this study, we seek to advance scholarly understanding of familism and the Mediterranean regime with three pointed questions: First, to what extent are patterns of family structure and values connected to ideal-typical regimes? Second, what is the connection between family structures and family values? Third, using support for public child care provisions as an indicator of the legitimacy of modern, service-oriented welfare states, how are attitudes on this issue related to the pattern of family structure/values across countries?
2. Care preferences, familism and welfare regimes
The 1990s witnessed the emergence of a pair of highly influential classifications of welfare regimes, one devised by Gosta Esping-Andersen (1990), the other by Walter Korpi and Joakim Palme (1998) (see also Korpi 2000). Other classifications soon appeared, adding a pair of new regimes to those already defined: the ‘Radical’ model (Castles and Mitchell 1993), and the ‘Southern’ model (Leibfried 1992; Ferrera 1996; Bonoli 1997). Of these two, it is the operation of a Southern welfare regime that has elicited the greatest debate and attention among scholars (Rhodes 1996; Moreno 2002; Gal 2010).
The Southern/Mediterranean welfare state is said to be characterized by a fragmented system of income maintenance that combines generous programs (pensions, unemployment) with severely underdeveloped areas (basic security, family and care policies); differentiation from the corporatist traditions in the field of health care (with universal NHS's); and a mix of public and private providers of welfare in which the family constitutes the main insurance for risks not covered by public welfare.
Attesting to the theoretical significance of family institutions, Kohli et al. (2008: 170) propose the existence of a North-South gradient, ‘with the Scandinavian countries generally having the least traditional family structure, the Mediterranean countries (Spain and Italy more so than Greece) the most traditional one, and the continental countries lying somewhere in-between’. Mediterranean countries have the highest levels of intergenerational cohabitation and spatial proximity of family members; care activities carried out by family members are more time-intensive in the South than in other countries (Attias-Donfut et al.2005, 2008a), and adult children provide personal care to their elders (dressing, bathing, eating) much more frequently than in the Nordic or Continental countries, where professional service providers take on these tasks (Brandt et al.2009).
The intensity of care for family members has, in turn, been cited as a factor explaining why helping people outside of the household is less common in Southern Europe (Croda and González-Chapela 2005; Attias et al. 2008). Financial transfers between parents and adult children involve larger sums in Mediterranean countries. The frequency of financial transfers to distant family members is likewise notable (Albertini et al.2007; Attias et al. 2008b).
This distinctive coupling of low family service provisions by government with a reliance on private, household-level assistance leads inevitably to questions about the role played by citizens’ values in the development of Mediterranean welfare systems, and the additional scenario argue for a causal relationship. Citing differences in time spent caring between Continental and Mediterranean countries, for instance, Reher (1998) posits attitudes to family as being critical. A central explanatory factor behind the Mediterranean welfare state is a belief that family members have the duty to help each other, and that certain tasks pertaining to the family do not belong to national government. As Callegaro and Passini (2008: 200) set forth this line of thinking: ‘family ties are stronger in the Mediterranean countries, and they induce adult children to think to formal care as something to avoid as long as family members are able to help their elderly relatives’.
This type of ‘natural’ family model, although largely discarded by historical research (Gillis 1997), may explain how idealizations of Southern families often parallel fears that public care policies could ultimately erode family solidarity. But the later narrative is not well supported by evidence, which puts into question the explanatory power of family values to understand Mediterranean distinctiveness. Feelings of closeness and affection appear equally strong in Northern and Southern countries, as indicated by the proportion of elderly that feel ‘very or extremely close’ to their children and are getting along ‘very or extremely’ well with them (Daatland and Lowenstein 2005). Moreover, perceived social isolation among the elderly is highest in Greece and Portugal, even if cohabitation is much higher there than in the Nordic countries (Ogg 2005).
The willingness to perform caring duties of Mediterranean families also appears open to debate. Saraceno (2004) and Rizzi et al. (2008), contend that low fertility rates in Italy and Spain indicate women's growing preference to escape from caregiving roles. Bettio et al. (2006: 272) argue that immigration and the availability of cheap labour have transformed the Southern model of family into a ‘migrant in the family’.
The assumed interrelationship between citizen attitudes and family-based provision could, in principle, be the product of independently existing preferences or instead an adaptation to the absence of adequate public provision. But whether it operates as cause or effect, the underlying assumption among many scholars is that family solidarity lies at the core of the Mediterranean welfare regime. A strong preference for private, family-based provision over government assistance, and a linkage between such preferences and traditional family arrangements, is the key expectation. It is this expectation that focuses our research in this study.
3. Research questions
We situate family solidarity, family values and care provision preferences within a comparative perspective. The thrust of much past scholarship is that family forms should be closely related to family values. If so, we expect to find a correlation between traditional family forms and family values at the macro level of countries and at the micro-level of individuals.
But the connection between family values, welfare preferences and country-specific contexts may be rife with contingencies. Existing institutions may not necessarily represent the wishes of the majority but instead the group(s) with the power to shape rules (North 2006; see also Korpi 1985). Women, the main care givers in Mediterranean countries, cannot be considered a group with power to question and negotiate institutional designs. If this is the case, the underdevelopment of public care policies may not be related to individuals’ preferences for public provision.
Also, the comparative underdevelopment of family policies in the Mediterranean may result from a past combination of traditional family values and the availability of family solidarity (extended families that remain close and are willing to help each other). But if the traditional model of extended family is fading, especially in countries like Spain, with the lowest fertility rate in the EU (Kohli et al.2005), the relationship between family values and preferences for public policies may be mediated by the availability of family-based care. Familism should thus depress support for public care policies, but mainly when the individual is situated within an extended family network. Otherwise, family values may be unrelated or even positively related to support for state intervention in this area.
Together, the preceding considerations and lines of theorizing lead to the following questions:
Are family values stronger within Mediterranean countries than in the rest of Europe, and do levels of familism follow the borders of welfare regimes?
Are family values associated with the traditional family model, in which there is frequent contact and a high level of mutual help?
What effect does familism have on individual preferences for public care policies?
Does the effect of values on preferences for public child care depend on the degree of available family support?
4. Data and methods
4.1. Data
We analyze data from the International Social Survey Program's ‘Social Relations and Support Systems’ survey, fielded in 2001. As far as we know, this is the most recent comparative survey that contains both family solidarity indicators and attitudes to public care programs.
We include data for Australia, Austria, Brazil, Canada, Chile, Cyprus, Czech Republic, Denmark, Finland, France, Germany (samples from East and West collapsed), Hungary, Italy, Japan, Latvia, New Zealand, Norway, Poland, Russia, Slovenia, Spain, Switzerland, United Kingdom (samples from Great Britain and Northern Ireland collapsed) and the USA. This leaves us with data for 32,218 individuals clustered in 24 countries.
Our comparisons go beyond European or even Western borders, helping us to place the Mediterranean regime in broader context. The inclusion of non-European ‘liberal’ welfare states provides useful points of cross-national reference. Data for East European and South American countries (and also Japan) lend further scope to the analysis.
4.2. Dependent variable
As a measure of attitudes to public care policies, we selected an ISSP item that concerns support for state intervention in child-care. The wording of the question is as follows:
Q. ‘On the whole, do you think it should or should not be the government's responsibility to provide childcare for everyone who wants it?’
Definitely should not be; Probably should not be; Probably should be; Definitely should be. 2
4.3. Independent variables: Family solidarity and family values
Guided by the theory of intergenerational solidarity developed by Bengtson and Roberts (1991), we build three measures of family solidarity and one of family values. In this scholarship, six dimensions of solidarity are identified, but the data at hand compel us to focus on four of these: (1) associational solidarity – frequency of intergenerational interaction; (2) functional solidarity – patterns of support or resource sharing; (3) structural solidarity – the availability of family members for interaction; and (4) normative solidarity, i.e., familism – norms or expectations of individual obligations to the family.
4.3.1. Structural solidarity
We build a proxy for the number of adult family members each individual has. The number of brothers and adult children are included as ‘1’ for each person, as well as ‘1’ for mother being alive, and ‘1’ for father being alive. Other family relatives likewise add only ‘1’ for each category. For uncles, cousins, parents-in-law, brothers-in-law and nieces/nephews, one point is added when the interviewee has at least one living relative of each type.
4.3.2. Functional solidarity
The 2001 ISSP contains six items asking respondents about the first and second persons to which they could turn for help in case of illness, financial problems or the experience of depression. The questions are worded as follows:
Q. ‘First, suppose you had the flu and had to stay in bed for a few days and needed help around the house, with shopping and so on. Who would you turn to first for help?’
Husband, wife, partner; mother; father; daughter; daughter-in-law; son; son-in-law; sister; brother; other blood relative; other in-law relative; close friend; neighbor; someone you work with; someone at a social services agency; someone you pay to help; someone else; no one.
Q. ‘And who would you turn to second if you had the flu and needed help around the house?’
We transform the six items into dummy variables that score ‘1’ when the individual turns to a family member for help, and ‘0’ otherwise. Our indicator of functional solidarity is a weighted sum of the six dummies. It adds one point when the individual turn to a family member for help in the first place, and ‘0.5’ for a second place rating of a family member.
4.3.3. Associational solidarity
Our measure of associational solidarity is based in six 2001 ISSP survey items for the frequency of visits between family members. After recoding all responses categories into ‘visits per month’3 we combine items into an additive scale. This scale indicates the number of times per month that each individual visits or is visited by one family member. We list below question wordings for items.
*Q.2 ‘Of your adult brothers and sisters, with whom do you have the most contact?’
Q.3 ‘How often do you see or visit this brother or sister?’
*Q.6 ‘Of your children aged 18 and older, with whom do you have the most contact?’
Q.7 ‘How often do you see this son or daughter?’
Q.9 ‘And now some questions about your father. How often do you see or visit your father?’
Q.11 ‘And what about your mother? How often do you see or visit her?’
-Daily; At least several times a week; At least once a week; At least once a month; Several times a year; Less often.
Q.14 ‘Please indicate how often you have been in contact with any of the following types of relatives in the last four weeks’. (a) Uncles or aunts; (b) Cousins; (c) Parents in law; (d) Brothers or sisters in law; (e) Nieces and nephews.
–More than twice in last four weeks; once or twice in last four weeks; Not at all in last 4 weeks; I have no living relative of this type.
4.3.4. Family values/familism
Our measure of familism, or family values, is constructed by summing responses to a pair of items with which interviewees can agree or disagree: ‘Adult children have a duty to look after their elderly parents’; ‘You should take care of yourself and your family first, before helping other people’. Response options: Agree strongly; Agree; Neither agree nor disagree; Disagree; Disagree strongly. (Response options are coded so that higher values indicate greater support for familism.)
4.4. Control variables
Familism and its expected linkages with family structure/solidarity and public child care preferences may be confounded by additional variables. Principal among these are gender, age, religious participation and household factors. To obtain defensible estimates, we include measures of the latter in our models.
Gender is a binary variable, coded 1 for male and 2 female. Age is a continuous variable, coded in years (it ranges from 15 to 101). Religiosity measures the frequency of attendance at religious services, where response categories are as follows: 6 ‘Once a week or more’; 5 ‘2–3 times a month’; 4 ‘At least once a month’; 3 ‘Several times a year’; 2 ‘Less frequently a year’; 1 ‘Never’. Household composition is measured as follows: 1 ‘single’; 2 ‘only adults’; 3 ‘1 adult+children’; 4 ‘2 adults+children’; 5 ‘3 or more adults+children’. Household employment status is measured as follows: 1 ‘two members working full-time’; 2 ‘one member working full time, or one member working full time and another one working part-time’; 3 ‘one or two members working part-time’; 4 ‘at least one member retired’; 5 ‘None working nor retired’. This last variable was built to be a proxy of the time available in each family for caring duties.
The former controls have been selected on the grounds of previous literature, which has pointed to welfare attitudes as having two general determinants: ‘self-interest’ and ‘ideology’ (Svallfors 1997; Lipsmeyer and Nordstrom 2003). Self-interest refers to the interest an individual has in the maintenance of welfare programs, whilst ideology has to do with moral values, particularly justice beliefs and egalitarianism versus individualism. Gender, age and religiosity reference ‘ideology’ sources of citizens’ attitudes, pointing to socialization processes that may lead to particular understandings of the role of the family. Household employment status and composition approximate ‘interest’ determinants, referring to existing care needs and resources.
4.5. Methods
In the first step, cluster analysis is used to assess the correspondence between the distribution of family solidarity and welfare regimes, as well as to evaluate the distinctiveness of Mediterranean family patterns. Next, we apply linear multilevel regression models to disentangle the determinants of family values and, especially, to verify the association between these values and objective patterns of family solidarity. Results of multilevel models are confirmed with linear regression models fitted to country-specific samples. Finally, we use linear multilevel models with random slopes to analyze the effect of familism on attitudes to childcare. Ordinal logistic regression models fitted to country-specific data provide us with a confirmation of multilevel models results.
5. Results
5.1. Familism and family solidarity in the Mediterranean and other regions
Our results start by considering the patterning of family relationships and family values across countries and regions. Table 1 presents region-specific means using our three measures of family solidarity, and our single measure of family values. In line with past scholarship, the Mediterranean region has higher levels of both family solidarity and familism in comparison to other established European regional contexts, and also the Liberal (EU countries only) regime type.
| Structural solidarity | Associational solidarity | Functional solidarity | Familism | |
|---|---|---|---|---|
| Nordic | 7,53 | 21,48 | 2,65 | 7,02 |
| Continental | 7,51 | 26,45 | 2,86 | 7,24 |
| Liberal EU | 7,82 | 27,12 | 2,84 | 7,36 |
| Liberal (all) | 8,47 | 25,51 | 2,84 | 7,41 |
| Mediterranean | 7,88 | 51,76 | 3,14 | 7,92 |
| Eastern | 7,06 | 35,99 | 2,94 | 8,20 |
| South America | 10,26 | 43,81 | 2,93 | 8,78 |
| Japan | 8,45 | 35,20 | 3,32 | 7,36 |
| Av. Western Europe | 7,67 | 31,30 | 2,87 | 7,37 |
| Structural solidarity | Associational solidarity | Functional solidarity | Familism | |
|---|---|---|---|---|
| Nordic | 7,53 | 21,48 | 2,65 | 7,02 |
| Continental | 7,51 | 26,45 | 2,86 | 7,24 |
| Liberal EU | 7,82 | 27,12 | 2,84 | 7,36 |
| Liberal (all) | 8,47 | 25,51 | 2,84 | 7,41 |
| Mediterranean | 7,88 | 51,76 | 3,14 | 7,92 |
| Eastern | 7,06 | 35,99 | 2,94 | 8,20 |
| South America | 10,26 | 43,81 | 2,93 | 8,78 |
| Japan | 8,45 | 35,20 | 3,32 | 7,36 |
| Av. Western Europe | 7,67 | 31,30 | 2,87 | 7,37 |
Nordic: Finland, Norway, Denmark; Continental: France, Germany, Austria; Liberal: UK, Switzerland, Canada, New Zealand, the US; Mediterranean: Spain, Italy, Cyprus; Eastern: Hungary, Czech Republic, Slovenia, Poland, Russia and Latvia; South America: Chile, Brazil; Average Western Europe: Includes only countries included in the Nordic, Continental and Mediterranean models, plus the UK and Switzerland.
But our empirical portrait of the Mediterranean region begins to change when we broaden the comparisons to include Eastern Europe and Western, non-European countries, areas not normally included in previous comparisons. Considering these new comparisons in Table 1, the Mediterranean countries fall from the top position of family solidarity on structural solidarity (the liberal welfare regime has the highest level) and family values (the Eastern regime is characterized by higher levels of familism). As regards the associational and functional solidarity, these new comparisons do not dislodge the Mediterranean region as still scoring the highest on these two measures of family solidarity.
Additional comparisons with the South American region and with Japan extend our cross-national understanding. The South American region scores the highest on both structural solidarity and familism across all world regions that we consider. For its part, Japan has higher scores on structural and functional solidarity than the Mediterranean region. When combined with the results of our additional comparisons across European and Western, non-European regions, the initial distinctiveness of the Mediterranean region thus recedes considerably.
There is a second finding emerging from these analyses. Results from our measures of family solidarity and family values appear to fit poorly with established classifications of countries based on welfare regimes. Liberal democracies as a whole, for instance, score higher on structural solidarity than Mediterranean countries, and the otherwise disparate regions of Eastern Europe, South America and Japan all tend to have high levels of familism and family solidarity.
Given the preceding findings, it is instructive to proceed further and conduct a ‘theory-blind’ classification of family solidarity and family values. We use our four measures as aggregation variables (standardized), with countries as units to be clustered. Graph 1 present the resultant cluster structure.
The cluster analysis separates our sample of countries into two groups. The first group includes mainly Continental and Northern Europe, with Australia, New Zealand and Canada as a distinct sub-group, and Finland and Latvia as moderate outliers. The second group is composed of a heterogeneous mix of Eastern and Southern European countries, plus Japan, the USA and Chile. Brazil is the outlier within this dual structure.4
Looking at country averages in each of the aggregation variables (results not presented) helps us to understand additional components behind the clustering patterns. Countries with relatively low patterns of intergenerational solidarity: Continental European nations and Northern Europe, alongside the Anglo-Saxon countries (but not the USA), all have generally low levels of family solidarity. This contrasts with Southern and Eastern Europe, Japan and the United States as regions/countries characterized by much higher levels of family solidarity.
In summarizing these results, the overall fault line separating countries with high versus low levels of familism and family solidarity is not one that corresponds to the existing and most strongly carved out distinctions among welfare regime types. While Mediterranean countries do have higher levels of family solidarity and family values than other Western European countries, our results also show that there are at times larger differences when we bring in non-European nations into the comparison.
5.2. Are family values associated with the traditional family model?
We now turn to the interrelationship between family values and family solidarity, analyzing multilevel models with our measure of family values as dependent variable. In Table 2's estimates, both functional and associational solidarity are positively related with familism, while structural solidarity has negative effects. We estimated OLS models by country with the same variables and these confirm results. Functional solidarity has a significant and positive relation with familism in 16 out of 24 countries. Associational solidarity has a significant and positive relation with familism in 11 countries. In no country do we find negative relationships. OLS models diminish the importance of the negative association between family values and family size for structural solidarity. The relationship is significant only in seven countries, although negative in all cases.
| Dependent var.: Familism | M0 | M1 | M2 | M3 | M4 | M5 |
|---|---|---|---|---|---|---|
| Fixed part | ||||||
| Sex | −0,225** | −0,231** | −0,231** | −0,231** | −0,230** | |
| Age | −0,002** | 0,001 | 0,001 | 0,001 | 0,001 | |
| Education | −0,138** | −0,136** | −0,136** | −0,136** | −0,136** | |
| Religiosity | 0,054** | 0,048** | 0,048** | 0,048** | 0,048** | |
| Structural solid | −0,012** | −0,012** | −0,012** | −0,012** | ||
| Functional solid | 0,095** | 0,095** | 0,095** | 0,095** | ||
| Associational solid | 0,003** | 0,003** | 0,003** | 0,003** | ||
| Av. Structural sol. | 0,100 | |||||
| Av. Functional sol. | 0,232 | |||||
| Av. Associational sol. | 0,030* | |||||
| Random part | ||||||
| cons | 7,729** | 8,370** | 7,972** | 7,995** | 7,975** | 8,026** |
| Var(cons) | 0,397 | 0,364 | 0,350 | 0,340 | 0,346 | 0,256 |
| Var(Resid) | 2,122 | 2,035 | 2,017 | 2,017 | 2,017 | 2,017 |
| ICC | 0,158 | 0,152 | 0,148 | 0,144 | 0,147 | 0,113 |
| Dependent var.: Familism | M0 | M1 | M2 | M3 | M4 | M5 |
|---|---|---|---|---|---|---|
| Fixed part | ||||||
| Sex | −0,225** | −0,231** | −0,231** | −0,231** | −0,230** | |
| Age | −0,002** | 0,001 | 0,001 | 0,001 | 0,001 | |
| Education | −0,138** | −0,136** | −0,136** | −0,136** | −0,136** | |
| Religiosity | 0,054** | 0,048** | 0,048** | 0,048** | 0,048** | |
| Structural solid | −0,012** | −0,012** | −0,012** | −0,012** | ||
| Functional solid | 0,095** | 0,095** | 0,095** | 0,095** | ||
| Associational solid | 0,003** | 0,003** | 0,003** | 0,003** | ||
| Av. Structural sol. | 0,100 | |||||
| Av. Functional sol. | 0,232 | |||||
| Av. Associational sol. | 0,030* | |||||
| Random part | ||||||
| cons | 7,729** | 8,370** | 7,972** | 7,995** | 7,975** | 8,026** |
| Var(cons) | 0,397 | 0,364 | 0,350 | 0,340 | 0,346 | 0,256 |
| Var(Resid) | 2,122 | 2,035 | 2,017 | 2,017 | 2,017 | 2,017 |
| ICC | 0,158 | 0,152 | 0,148 | 0,144 | 0,147 | 0,113 |
N° of groups M0–M5=24; N° obs. M0=30747; N° obs. M2-5: 27033;
p<0.000; *p<0.05.
Level II variables have been centered to the grand mean (mean of all country averages).
Among individuals, there is clear evidence for an association between familism and two aspects attributed to the traditional family model (high frequency of contacts and use of the family network to deal with risks). However, the size of the family, and possibly the size of the corresponding care burden, constrains in some countries the adoption of family values.
The extension of the traditional family in the country where an individual lives also has an influence on his/her adoption of familistic values, as we can see by the statistical significance of the average of associational solidarity in each country. This indicates that an individual's position within a polity where a large number of families have traditional characteristics tends by itself to increase the level of support for family values.
5.3. What is the effect of familism on preferences for public care policies?
To assess whether familism depresses support for state intervention in child-care policies, we apply multilevel models with support for public child-care as the dependent variable. In Table 3's models, the key finding concerns the positively signed coefficient for familism. This indicates that support for traditional family values does not decrease support for public child care. Far from being zero-sum or inversely related, familism and government provision of child care services appear to have significant affinity.
| Dep. Vble.: Support for public Childcare | M0 | M1 | M2 | M3 | M4 | M5 | M6 | M7 |
|---|---|---|---|---|---|---|---|---|
| Fixed part | ||||||||
| Sex | 0,074** | 0,011** | 0,085** | 0,085** | 0,085** | 0,085** | 0,083** | |
| Age | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | |
| Religiosity | −0,013** | −0,013** | −0,016** | −0,016** | −0,016** | −0,016** | −0,015** | |
| Household composition. Reference category: household with a single person only. | ||||||||
| Only adults | −0,001 | −0,010 | −0,009 | −0,008** | −0,009 | ‘0,008 | −0,011 | |
| 1 adult+children | 0,163** | 0,159** | 0,162** | 0,163* | 0,164** | 0,163** | 0,161** | |
| 2 adults+children | 0,067* | 0,058* | 0,057* | 0,058 | 0,059* | 0,059* | 0,056* | |
| 3 adults+children | 0,062* | 0,050* | 0,044 | 0,044 | 0,044 | 0,045 | 0,041 | |
| Household Work Status. Reference category: two full time workers in the household. | ||||||||
| 1 full time+1 full time and 1 part time | 0,001 | 0,000 | −0,001 | −0,001 | −0,001 | −0,002 | −0,002 | |
| 1 or 2 part times | 0,092** | 0,093* | 0,087* | 0,086* | 0,087* | 0,087* | 0,085* | |
| 1 or 2 retired | 0,080** | 0,082** | 0,076 | 0,076** | 0,076** | 0,076** | 0,076** | |
| No adults working/pensioners | 0,096** | 0,096** | 0,085** | 0,085** | 0,085** | 0,085** | 0,084** | |
| Structural solid | 0,002 | 0,002 | −0,015 | 0,002 | 0,002 | 0,002 | ||
| Functional solid | 0,005 | −0,002 | −0,002 | −0,001 | −0,027 | −0,002 | ||
| Associational solid | 0,000 | 0,000 | 0,000 | −0,003* | 0,000 | 0,000 | ||
| Familism | 0,048** | 0,031* | 0,035** | 0,039** | 0,052** | |||
| Structur*Familism | 0,002* | |||||||
| Functional*Familism | 0,003 | |||||||
| Associational*Familism | 0,0004* | |||||||
| Cons | 3,101** | 3,099** | 3,068 | 2,710** | 2,845** | 2,806** | 2,779** | 2,670** |
| Random part | ||||||||
| Var(cons) | 0,220 | 0,223 | 0,223 | 0,209 | 0,209 | 0,209 | 0,209 | 0,259 |
| Var(Resid) | 0,722 | 0,723 | 0,723 | 0,716 | 0,716 | 0,716 | 0,716 | 0,713 |
| ICC | 0,234 | 0,236 | 0,236 | 0,226 | 0,226 | 0,226 | 0,226 | |
| Dep. Vble.: Support for public Childcare | M0 | M1 | M2 | M3 | M4 | M5 | M6 | M7 |
|---|---|---|---|---|---|---|---|---|
| Fixed part | ||||||||
| Sex | 0,074** | 0,011** | 0,085** | 0,085** | 0,085** | 0,085** | 0,083** | |
| Age | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | −0,004** | |
| Religiosity | −0,013** | −0,013** | −0,016** | −0,016** | −0,016** | −0,016** | −0,015** | |
| Household composition. Reference category: household with a single person only. | ||||||||
| Only adults | −0,001 | −0,010 | −0,009 | −0,008** | −0,009 | ‘0,008 | −0,011 | |
| 1 adult+children | 0,163** | 0,159** | 0,162** | 0,163* | 0,164** | 0,163** | 0,161** | |
| 2 adults+children | 0,067* | 0,058* | 0,057* | 0,058 | 0,059* | 0,059* | 0,056* | |
| 3 adults+children | 0,062* | 0,050* | 0,044 | 0,044 | 0,044 | 0,045 | 0,041 | |
| Household Work Status. Reference category: two full time workers in the household. | ||||||||
| 1 full time+1 full time and 1 part time | 0,001 | 0,000 | −0,001 | −0,001 | −0,001 | −0,002 | −0,002 | |
| 1 or 2 part times | 0,092** | 0,093* | 0,087* | 0,086* | 0,087* | 0,087* | 0,085* | |
| 1 or 2 retired | 0,080** | 0,082** | 0,076 | 0,076** | 0,076** | 0,076** | 0,076** | |
| No adults working/pensioners | 0,096** | 0,096** | 0,085** | 0,085** | 0,085** | 0,085** | 0,084** | |
| Structural solid | 0,002 | 0,002 | −0,015 | 0,002 | 0,002 | 0,002 | ||
| Functional solid | 0,005 | −0,002 | −0,002 | −0,001 | −0,027 | −0,002 | ||
| Associational solid | 0,000 | 0,000 | 0,000 | −0,003* | 0,000 | 0,000 | ||
| Familism | 0,048** | 0,031* | 0,035** | 0,039** | 0,052** | |||
| Structur*Familism | 0,002* | |||||||
| Functional*Familism | 0,003 | |||||||
| Associational*Familism | 0,0004* | |||||||
| Cons | 3,101** | 3,099** | 3,068 | 2,710** | 2,845** | 2,806** | 2,779** | 2,670** |
| Random part | ||||||||
| Var(cons) | 0,220 | 0,223 | 0,223 | 0,209 | 0,209 | 0,209 | 0,209 | 0,259 |
| Var(Resid) | 0,722 | 0,723 | 0,723 | 0,716 | 0,716 | 0,716 | 0,716 | 0,713 |
| ICC | 0,234 | 0,236 | 0,236 | 0,226 | 0,226 | 0,226 | 0,226 | |
p <0.000; *p<0.05.
| Estimate | [95% Conf. Interval] | |
|---|---|---|
| Var(Familism) | 0,002 | 0,0009 // 0,0038 |
| Cov(Familism, cons) | −0,011 | −0,0230 // 0,0014 |
| Estimate | [95% Conf. Interval] | |
|---|---|---|
| Var(Familism) | 0,002 | 0,0009 // 0,0038 |
| Cov(Familism, cons) | −0,011 | −0,0230 // 0,0014 |
We extend these results by estimating ordinal logistic regressions by country. These results (not shown) reveal that family values have a significant effect on attitudes to public child care in 14 out of 24 countries, and when the effect is significant, it is always positive in direction. Consistent with the multilevel results, the ordinal logistic regressions unveil a tendency for familism to possibly increase (and never erode) citizens’ demands for public care policies. The results hold if we run country-specific models including alternative measures for education and the number of dependent children (in place of household work status and composition), and also when we use only the item about the caring duties of adult children as indicator of family values.
To assess the scope of our results, we estimated ordinal regression models by country with the same independent and control variables as in Model 3 (Table 3) but now using as a dependent variable an item related to the elderly: ‘On the whole, do you think it should or should not be the government's responsibility to provide a decent standard of living for the old?’ The effect of family values on support for state intervention on elderly care is positive and statistically significant in 18 out of 24 countries.
5.4. Is the effect of familism on attitudes to public child care mediated by the availability of family support?
To this point, results for the analyses suggest that support for family values do not depress support for state intervention in child-care even after controlling for the characteristics of the family. However, family values may still shape individual's preferences for child-care policies in ways that depend on the kind of available family support arrangements. Family values may, for instance, increase support for child care policies more strongly when families are small and there are limited possibilities of family-based care provision for dependents.
To evaluate this possibility, we analyze three multilevel models with the same individual variables as in Model 3, but now including an interaction between: (M4) structural solidarity and familism; (M5) functional solidarity and familism; (M6) associational solidarity and familism. Three sets of ordinal logistic regression models by country (one interaction in each set) were also run to firm up results. In these ordinal models we include only variables related to family solidarity and values, excluding controls to see if interactions achieve statistical significance.
Models 4, 5 and 6 (Table 3) show that the effect of family values on support for public child care is not mediated by the extension of the family or by their functionality. The interaction between functional solidarity and familism is not statistically significant in the multilevel models with controls (or in ordinal logistic regression models estimated for specific countries). An interaction between structural solidarity and familism that does reach statistical significance in the multilevel model is due to the existence of a small positive interaction in only two countries: Cyprus and Denmark.
The interaction between associational solidarity and familism is perhaps more interesting as the effect is significant in four countries (Australia, Norway, Czech Republic and Japan). There, we can find a positive interaction between associational solidarity and the effect of family values on support for child care, where family values increase support for public child care as the level of contact with family member's increases. A possible explanation is that contact with family members raises awareness of individuals’ needs, and this is translated into increased support for state intervention in care policies. This refines our understanding of the association between familism and government child care support, highlighting again the unexpectedly positive nature of this relationship.
6. Conclusion
A presumption shared by the large majority of comparative scholarship (and also by regionally oriented theory and research) is that the Mediterranean region is distinctive or ‘special’ when it comes to the organization and political impacts of family institutions. As we discussed in the introduction, there are several established expectations. Not only do scholars expect to find cross-nationally high or even exceptional levels of support for the traditional family, these support patterns are often thought to be grounded in historically robust patterns of family interaction and dependence. In turn, some authors have seen Mediterranean familism as constraining citizens’ willingness to support public service provision, potentially helping to explain comparatively lower levels of welfare state development in Southern Europe.
What, then, do we find when we put to the test expectations of the Mediterranean familism model? Corroborating one initial expectation, family values support is indeed high in the Mediterranean region when we focus on traditional comparisons with Nordic and continental European nations.
But when our cross-national analysis includes more novel comparisons with Eastern Europe, South America and liberal democracies outside of the European region, the picture becomes more complicated. These comparisons quickly lead to substantial recalibration of the phenomenon of Mediterranean familism. We find that high levels of family support and family values in Southern Europe have some important parallels in Eastern Europe and in a number of non-European countries. Moreover, while there are clearly linkages between key dimensions of family solidarity and support for family values, such linkages can be found outside the Mediterranean region as well.
Our analysis of public child care preferences offer a second set of results that put in further perspective theory and research on Mediterranean familism and welfare regimes. Here, we find evidence that the relationship between family values support and attitudes toward government child care provision is decidedly positive in nature. Far from eroding support for public child care services, family values instead appears to increase (or at least be positively associated with) preferences for public provision.
Not only is the Mediterranean regime potentially far less distinctive for its levels of family solidarity and family values than previously thought, the political thrust of familism is toward greater pressure for welfare state development. This offers to scholars an empirically grounded caution against tendencies toward romanticizing low levels of welfare state development in Southern Europe. Far from going hand in hand with one another, there appears to be a disconnection between the policy implications and demands imposed by familism and the low levels of public services provision for families in Mediterranean countries. The current economic crisis and the adoption of cost-containment measures within many countries are likely to widen the gap between family needs and public resources.
In the Mediterranean countries, where cuts in public expenditures endanger the very existence of public care programs, families may be forced to the limit in their role as providers of care and risk protection. Being perhaps accustomed to a lack of public support, Southern European citizens may conceivably accept cuts in care programs with less turmoil or simply more resignation than other nationals. However, the overburdening of families is also likely to be an undesirable situation in the eyes of Mediterranean citizens and, as such, strategies to make the care load manageable are to be expected. Looking at the past, typical strategies include to postpone and limit the number of children; hiring cheap and illegal care workers – generally but not always of migrant origin; and a move to part-time jobs – especially among women. Yet any of these strategies, when generalized, may tend to have negative consequences for future social security budgets and, in all likelihood, with respect to the development of a more modern political economy.
Footnotes
The authors gratefully acknowledge the help of Rosa Gutiérrez and of three anonymous reviewers. This article was written under the framework of the research project ‘Social policies for the elderly and children: preference formation and welfare reform’, funded by Fundación CSIC-Caixa.
Response categories for the item have been recoded so that higher values indicate higher support.
Recodes for questions 3,7,9 and 11: Lives in the same household=30 visits per month; Daily=25 visits per month; At least several times a week=16 v.p.m.; At least once a week=6 v.p.m.; At least once a month=2 v.p.m; Several times a year=0,5 (6 times a year); Less frequently=0,2 (2,4 times a year). For question 14: More than twice in last 4 weeks=4 visits per month; Once or twice in last 4 weeks=2 visits per month; Not at all in last 4 weeks=0,3 visits per month (3,6 times per year); No living relative of this type=0.
This structure remains stable using different linkage methods (within-group, centroids, medians, Ward's). Interval: Euclidean Squared Distance.
References
Inés Calzada, Researcher at the Spanish National Research Council and Lecturer of Sociology at Linköping University. She works in topics related to comparative social policy, public opinion and migration. E-mail: [email protected]
Clem Brooks, Rudy Professor of Sociology at Indiana University. His interests are in political sociology, comparative politics, political psychology, and quantitative methods. With Jeff Manza, he is the author of Social Cleavages and Political Change, Why Welfare States Persist, and Whose Rights? E-mail: [email protected]