The Relationship between Familizing and Individualizing Policies and Mental Health in Parents in Europe

Abstract

Previous studies suggest the relative importance of the impact of childcare policies on mental health in parents. There have also been studies showing that welfare states have differing policy packages, consisting of a mixture of familizing and individualizing policy measures. This study builds on and extends this knowledge by carrying out a European comparison of the association between mental well health and family policies. We use Lohmann and Zagel’s familizing and individualizing policy indices to describe family policies. Our main interest is differences in mental health depending on the country, household, and individual-level characteristics. Therefore, we apply a multilevel model to 26 countries included in the 2013 wave of the European Union Statistics on Income and Living Conditions survey (N = 141,648). The analysis found that, in general, parents of children under 13 have better mental health than other adults. We found individualizing policy measures to be positively related to mental health in parents, while familizing policies had a negative relationship. No evidence was found for the combined presence of individualizing and familizing policies making a difference to mental health in parents. These results suggest that welfare states could help parents by promoting individualizing policies to make parenthood a less stressful experience.

1. Introduction

Parenthood is a two-sided experience. On the one hand, it brings joy to parents, while on the other, it creates additional stress. In this paper, we intend to examine how different family policy configurations in Europe relate to mental health in parents with young children. We build on evidence that there is a variation in mental health among parents across Europe (Glass et al. 2016). Previous research attributes a substantial part of this variation to disparities in work–family reconciliation policies and their diverse effects on parental mental health, e.g., (Boye 2011; Aassve et al. 2012; Glass et al. 2016; Pollmann-Schult 2018).

One of the aims of family policies is to influence the distribution of caretaking tasks. Policy measures vary in terms of to whom they attribute the responsibility for care provision and to what extent. The way parents divide care responsibilities depends on a wide array of individual and couple characteristics. However, policy measures (e.g., care services, leave facilities, and cash transfers) also play a role (Ciccia and Verloo 2012). The literature on welfare states often distinguishes between two types of policies: familizing and individualizing (also referred to as defamilizing). Familizing policies place the responsibility for childcare within the sphere of the private household, such as through the provision of relatively long periods of parental leave and inadequate availability of formal childcare. Parental leave is thus necessary to fulfill the fundamental human need for giving and receiving care (Kurowska 2018). However, extended and poorly paid leave can also result in women shouldering the major share of childcare and experiencing the “child penalty” on their return to the labor market (Saraceno 2011).

In contrast, individualizing policies outsource care from the family to formal institutions (state and/or market), intending to make individuals more independent of their families: both those who would need to provide care and also those in need of care (Daly 2011). In some countries, family policies are homogeneous—they consistently delegate caretaking to the same provider (within or outside the family). However, in the majority of countries, both familizing and individualizing measures have been introduced; therefore, they are heterogeneous from the family policy point of view (Bambra 2007). Individualizing and familizing policies can affect families in different, and sometimes contradictory, ways.

Previous studies on the association between these policies and mental health in parents conceal the complexity of family policies by either focusing on a single policy measure (e.g., childcare facilities) while ignoring other policies (e.g., parental leave), e.g., (Pollmann-Schult 2018), or by capturing several policy measures in a single dimension (e.g., generosity), e.g., (Glass et al. 2016). Therefore, they overlook the fact that, in most cases, both familizing and individualizing policies are in place at the same time. For example, a study by Glass and colleagues summed scores on multiple indicators into one index, which led to the same index score representing very different situations (e.g., a situation where few children are in formal childcare, but long-paid maternity leave is available versus a situation where extensive formal care use and short-paid maternity leave are available). Thus, using this method, these studies tend to disregard the reality of the dual presence of both policy trends.

In addition, to the best of our knowledge, there has been no previous research conducted on mental health in parents and family policies utilizing this holistic approach. Parents’ personal preferences in childcare—the realization of which would promote better mental health—are limited by what options policy interventions make available for them, and not having access to a different care arrangement could negatively affect their mental health. Therefore, we take into consideration both kinds of policies to test the above hypothesis.

Furthermore, while several studies found that men benefit more from parenthood because the burdens involved are lesser for them (Nelson et al. 2013; Nelson-Coffey 2018), there is a gap in the literature on whether gender differences in parental mental health might be explained by the level of individualization and familization of family policies.

Our work addresses these gaps in the literature by using two comprehensive indices measuring the levels of familization and individualization, respectively, and testing their association with mental health in parents with young children compared to individuals who do not have young children in their households. We examine whether there is a correlation between the level of individualization and familization of family policies and mental health in parents. In addition, we investigate whether there is a significant gender difference in this association. Covering 26 European countries, we apply a multilevel framework using data from the 2013 wave of the European Union Statistics on Income and Living Conditions (EU-SILC) survey. Family policies are measured using an updated version of the Lohmann and Zagel (2016) indices of familizing and individualizing policies, while mental health in parents is measured using the five-item Mental Health Inventory. In the following, “parents” refers to the group of parents with children below the age of 13 years. We compare this group to both childless individuals and individuals who have children 13 years or older.

2. Theoretical Framework and Literature Review

2.1. Parenting and Mental Health

There is no scientific consensus regarding the relationship between parenthood and mental health. Some studies have found a positive association (Nelson et al. 2013; Balbo and Arpino 2016), others have found a negative one (Evenson and Simon 2005), and some others have not found any connection (Rothrauff and Cooney 2008). Previous studies give numerous theoretical explanations for either the positive or the adverse mental health effects of parenthood. Role strain, time crunch, work–family conflict, pressure deriving from the intensive mothering, and involved fathering ideologies, as well as financial strain and marital strain, are the most common negative factors identified. In contrast, emotional reward, self-fulfillment, a sense of purpose, and social embeddedness are, in general, identified as the positive factors (Nomaguchi and Milkie 2003). Although there is evidence of both a positive and negative association between parenting and mental health, there seems to be scientific consensus that raising young children can be particularly stressful for many reasons. Taking care of young children can be very time intensive; it requires a high level of mental effort and involves many logistical tasks. Arranging quality childcare is one of the greatest challenges for parents due to the difficulties in gaining access to affordable and good-quality childcare, space and time flexibility/rigidity, and the financial burden that childcare entails (McLean et al. 2017). Therefore, we expect that parents of young children have poorer levels of mental health than parents with older children or people who have no children (H1).

Research that focuses on gender differences in the association between parenthood and mental health often traces the origin of these gender differences back to social role theory. Today, fathers spend significantly more time undertaking childcare than their fathers did; however, their involvement still differs from that of mothers. Mothers undertake more routine-like and stressful tasks, such as feeding, bathing, and management, while fathers mostly participate in leisure activities (Musick et al. 2016). Since gender beliefs are stronger in relation to mothers of young children than in the case of older children, these mothers experience a higher level of work-related guilt than fathers of the same age group (Borelli et al. 2017). Furthermore, mothers are more subject to work–family conflict than fathers, and are thereby also more at risk of reduced mental health (Piccinelli and Wilkinson 2000). Therefore, we also expect that having young children will be more negatively associated with mothers’ mental health than with that of fathers (H2).

2.2. Family Policies

A state’s family policy measures may embody different, sometimes even contradictory, objectives, including economic support for families, enhancing children’s well-being and development, boosting fertility, activating women on the labor market, granting families the right to care, or enhancing gender equality (Ciccia 2016). Welfare states differ in which of these objectives they prioritize and in what way they achieve them. In addition, a policy measure in itself can serve different purposes depending on the way it is designed. For instance, parental leave can enhance gender equality if it is well paid, is of moderate duration, and includes a father quota. However, when leave is long and poorly reimbursed through low and flat payments, it may strengthen traditional gender roles and preserve informal caregiving as the responsibility of the family, which most often means the mother (Akgunduz and Plantenga 2013).

To operationalize this complexity of family policies and to create a measurement of welfare state family policy intentions, the current study uses an updated version of the familization and defamilization indices created by Lohmann and Zagel (2016), who consider dependencies to be at the core of this classification. For them, “defamilization” is a set of measures that aim to reduce intrafamilial dependencies—both gender-specific and intergenerational—and to provide freedom from family members and/or the market. Since human beings are unable to take care of themselves at the beginning and—very often—at the end of their lives, someone else must provide care for them. For this reason, Lohmann and Zagel included both childcare-related and elderly care-related policies in their instruments. Building on Leitner’s work (Leitner 2003), Lohmann and Zagel do not consider familization and defamilization as two opposite poles of a single continuum but instead as two separate scales that exist simultaneously alongside each other. In agreement with Daly (2011), who criticized the use of the term “defamilization” because of the “de,” which gives formal care policies a negative overtone, as if familial care is something bad, the authors of this paper prefer to use the term “individualization”.

To the best of our knowledge, there is no previous research that has measured the levels of familization and individualization separately but simultaneously and compared their association with mental health in parents across several countries. Studies that have examined the relationship of mental health in parents and individual policy measures have delivered negative, e.g., (Brodeur and Connolly 2013; Herbst and Tekin 2014), positive, e.g., (Schober and Stahl 2016; Schmitz 2020), and neutral results, e.g., (Healy and Dunifon 2014); however, they did not look at entire policy packages. Glass and colleagues (2016) created a comprehensive policy index that consisted of three indicators (paid leave available to mothers, paid vacation and sick leave, and work flexibility). They found this to have a strong and positive effect on both the general population’s and parents’ mental health. This index, however, did not intend to classify welfare states based on their policy intentions, nor did it compare the effects of different packages. Other studies have looked at the relationship between welfare state typologies and mental health. However, they used a regime typology based on Esping-Andersen’s work, which considered decommodification measures from the market but did not take dependency on the family into consideration.

While family policies may reflect existing social norms in society, they can also have the intention to change the prevailing norms, hence the interactive relationship between norms and policies (e.g., the use of a father quota in parental leave schemes often intends to change traditional gender patterns in caring duties) (Ma et al. 2020). In societies where the dominant parenting norms and family policy intentions are not congruent, parents’ beliefs—if they are based on the norms—may be in conflict with the policies; therefore, gendered norms concerning parenthood need to be controlled for during the analysis. Parents’ care decisions are not independent of the care regime they live in, because policies provide them with opportunities (e.g., childcare places, long or short parental leave, etc.). However, since family policy regimes can support different norms, in agreement with Leitner (2003) and Kurowska (2018), the ideal policy regime is that which enables parents to make care arrangements in accordance with what they believe to be the best option, rather than only offering one kind of care option. In other words, we expect parents to have a better level of mental health in states that are high on both familizing and individualizing indices, therefore enabling parents to choose what they consider to be the best care option (H3).

At an aggregate level, the negative consequences of intrafamilial dependencies, reflected in high levels of familization, are greater for women than for men, since in most European societies, women carry out the majority of care tasks (Eurostat 2020). As a result, women’s careers may suffer long-term setbacks both financially and in terms of opportunities (Akgunduz and Plantenga 2013). Furthermore, being bound to stay at home due to not having state-supported formal care options—and, therefore, not realizing desired employment plans—also has its toll on women’s mental health (Schmitz 2020). When looking at women and men separately, researchers have found that women who live in less generous welfare states have poorer levels of mental health than men (Chung et al. 2013). This was explained by a lack of family policy measures in those countries, which affect women more adversely than men. Therefore, we expect women’s mental health to have a more positive association with the level of individualization than that of men (H4).

3. Methods

3.1. Data

We used the 2013 wave of the EU-SILC (Eurostat 2017)1, which included an ad hoc module on well-being. The data collection was carried out on the basis of multiple, nationally representative sampling designs and is therefore representative of all individuals living in private households. In addition to personal interviews, the data collection also included information from register data. We made a number of selections in order to establish our study sample. First, the sample was restricted to countries which are covered by the modified policy indices (this excluded Bulgaria, Switzerland, Cyprus, Croatia, Iceland, and Serbia). As a result, 26 countries were included in the analysis. Second, we concentrated only on those members of the active-age population, who were in the age range for potentially having a child under 13 (18–50 years old). Third, on control variables that had less than 10% missing information, we applied listwise deletion. In the original sample, employment status had a higher percentage of missing cases (12%). Therefore, we included a separate category to determine whether the missing cases differed from the rest of the sample.

3.2. Variables

Dependent variable—The main outcome variable, mental health in parents, was measured using the Mental Health Inventory (MHI-5). This composite index includes five items (being nervous; feeling happy; down in the dumps; calm and peaceful; and downhearted and depressed), each of which is evaluated by respondents on a one to five-point scale. The MHI-5 scale has been validated in the general population (Rumpf et al. 2001; Yamazaki et al. 2005). We confirmed this validity through multigroup confirmatory factor analyses at the level of full metric invariance (CFI: 0.920, RMSEA: 0.023). We also confirmed its reliability (Cronbach’s alpha: 0.79). We summed the scores of the items and then transformed the results into a 0 to 100 scale, where higher and lower scores represent better and worse mental health, respectively.

Independent variables—The core explanatory variables were developed based on Lohmann and Zagel’s indices (2016). Nevertheless, as mentioned above, in agreement with Daly (2011), who criticized the use of the term “defamilization,” we call the instruments individualization and familization indices. We updated the original index scores with the most recently available data (2009) from the Multilinks Database (Keck and Saraceno 2012). The two indices utilize multiple indicators rather than a single indicator for each, which makes the results more trustworthy in differentiating between welfare states (Kunißen 2019). For the most part, policies rather than policy outcomes were included, since the latter are the result of many other factors combined and cannot be considered as direct indicators of the intention of the policymaker (Kurowska 2018). For indicators that were not available in the Multilinks dataset for 2009, we either substituted similar ones from other sources or omitted them (see

Table 1. Indicators used for defamilization and familization indices (Lohmann and Zagel 2016).

Defamilizing Policies	Familizing Policies
Early years care	Allowances and taxes
Individual entitlement to childcare for children under 3 years (yes/no); coverage rate for children under 3 years (share of children aged 0–2 years); full-time childcare usage of children under 3 years (share of children aged 0–2 years); enrolment rates in early childhood education and care services and primary education for children aged 3–5 years (share of children aged 3–5 years)	Eligibility condition for child allowances: universal benefit (yes/no); child allowance for one child (share of net average income); child allowance for three children (share of net average income); tax deduction or tax credit for children (yes/no)
Elderly care	Support of the elderly
Access to care services (yes/no); home-based care service recipients (share of population aged 65 and older); persons living in care institutions (share of population aged 65 years and older)	Children’s legal obligation to support parents (yes/no)
Parental leave	Parental leave
Length of well-paid leave (months); duration of paternity leave (days)	Length of unpaid leave (months; constructed as the sum of maternity leave and parental leave minus the months of paid leave)

and

Table 2. The indicators constructing the individualizing and familizing policy indices used in the present paper.

Individualizing Policies	Familizing Policies
Early years care	Allowances and taxes
Individual entitlement to childcare for children under 3 years (yes/no); full-time childcare usage of children under 3 years (share of children aged 0–2 years); enrolment rates in early childhood education and care services and primary education for children aged 3–5 years (share of children aged 3–5 years)	Eligibility condition for child allowances: universal benefit (yes/no); child allowance for one child (share of net average income); child allowance for three children (share of net average income); tax deduction or tax credit for children (yes/no)
Elderly care	Support of the elderly
Access to care services (yes/no); coverage rates for residential care (share of population aged 65 and over); coverage rates for formal home care (share of population aged 65 and older)	Children’s legal obligation to support parents (yes/no)
Parental leave	Parental leave
Length of well-paid leave (months); duration of paternity leave (days)	Proportion of underpaid leave (months; constructed as the sum of maternity leave and parental leave minus the months of the well-paid leave divided by the total length of leave)

). We collected the long-term care coverage rates of elderly people from Eurostat, with the reference years varying between 2004 and 2009 (Bettio and Verashcagina 2012). The enrolment rates in early childhood education and care services and primary education for children aged 3–5 years were taken from the OECD Child Well-being Dataset (OECD 2013). The 2004 information regarding adult children’s obligation to support parents was updated in light of changes by 2009 based on an article by O’Mahoney (2015). The scores of the different indicators were summed for each index, with higher scores meaning higher levels of individualization or familization, respectively. Since these two indices were composed of multiple policy measures, similar country scores may be built up in different ways.

Gender was a dichotomous variable with the options of man (1: reference category) and woman (2). Parental status was also dichotomous: having a child under 13 (1) and not having one (0: reference category).

Control variables—To account for the nonlinear association between mental health and age, we included both age and age squared in our analyses. We measured household income with a four-category variable, which represents the household’s total disposable income as a proportion of the national median equivalent income. The categories were as follows: relative poverty (<60% of the median equivalent income), low-income group (60–80% of the median equivalent income), national average level income (80–120% of the median equivalent income: reference category) and relatively high-income group (>120% of the median equivalent income). Regarding the respondent’s employment status, four types were distinguished: (1) full-time worker (reference category), (2) part-time worker, (3) non-working, and (4) missing value. The partner’s employment status had the values of (1) non-partnered, (2) full-time worker (reference category), (3) partnered and partner works part-time, and (4) partnered and partner is not working. In models in which the partner’s employment was not included, we used a dichotomous variable indicating partnership status: (1) partnered (reference category) and (2) non-partnered. The influence of family structure was examined by controlling for the presence of other adults (apart from the partner) in the household. The respondent’s level of education could fall into one of three categories: (1) no or primary-level of education (reference category), (2) secondary education or professional, or (3) higher education. At the macrolevel, we controlled for GDP (2012), aggregate total spending on social security (Eurostat 2012) and parenting norms2 (EVS 2016).

Table A1. Descriptive statistics.

	Range	N (%) Mean (Std. Dev.)
Individual
Dependent
MHI-5	0–100	68.97 (18.62)
Independent
Sex
Male	Ref.	64,480 (45.5%)
Female	0/1	77,240 (54.5%)
Age	18–50	36.18 (9.27)
Age2	324–2500	1394.77 (649.13)
Has child(ren) under 13
No	Ref.	86,422 (61.01%)
Yes	0/1	55,226 (38.99%)
Other adult(s) living in the household
No	Ref.	82,975 (58.58%)
Yes	0/1	58,673 (41,42%)
Household income
<60% of median income	0/1	23,344 (16.48%)
60–80% of median income	0/1	21,439 (15.14%)
80–120% of median income	Ref.	45,092 (31.83%)
>120% of median income	0/1	51,773 (36.55%)
Level of education
No or primary education	Ref.	27,720 (19.57%)
Secondary education or professional	0/1	71,100 (50.19%)
Tertiary education	0/1	42,828 (30.24%)
Employment status
Full-time work	Ref.	84,142 (59.40%)
Part-time work	0/1	15,024 (10.61%)
Non-working	0/1	30,497 (21.53%)
Missing value	0/1	11,985 (8.46%)
Partner’s employment status
Full-time work	Ref.	61,411 (43.35%)
Part-time work	0/1	8673 (6.12%)
Non-working	0/1	16,635 (11.74%)
Not partnered	0/1	54,929 (38.78%)
Macrolevel
Familizing Policy Index 2009	0–4.68	3.59 (0.734)
Individualizing Policy Index 2009	0–4.83	3.04 (1.095)
GDP	1–10	34.49 (12.80)
Aggregate level of social expenditure	0–1.4	0.383 (0.31)
Aggregate gender norms	0–1	0.295 (0.14)

Source: European Union Statistics on Income and Living Conditions, 2013. N = 141,648.

in the Appendix A provides descriptive statistics for all variables.

3.3. Statistical Procedure

As the first step, we estimated the country-specific mean scores and standard deviations for the MHI-5 (see

Table A2. Individual-level means of mental health and policy index scores by country.

Country	N	Mean MHI-5 Score	(SD)	Individualization Index 2009	Familization Index 2009
Austria	4989	72.73	16.73	3.11	4.83
Belgium	4789	66.13	18.06	4.68	3.72
Czechia	4633	63.82	17.03	2.80	2.36
Germany	7958	67.23	16.60	4.46	4.04
Denmark	2017	77.56	16.42	5.92	4.42
Estonia	4810	70.88	16.61	4.43	3.18
Greece	6761	60.01	22.35	2.45	2.61
Spain	13,143	68.86	20.15	2.31	2.90
Finland	4817	73.54	14.86	4.32	2.60
France	6645	64.89	19.43	3.81	4.26
Hungary	8365	66.49	18.69	2.62	4.32
Ireland	2975	74.85	17.83	2.68	3.14
Italy	10,031	64.64	19.92	3.00	3.41
Lithuania	2875	66.31	15.86	2.32	4.53
Luxemburg	3053	66.20	19.43	3.09	4.12
Latvia	4164	70.78	18.06	1.91	3.77
Malta	2840	68.49	18.51	3.78	4.60
Netherlands	4703	76.46	15.88	3.69	4.26
Norway	2958	79.96	13.47	4.68	2.32
Poland	9289	72.16	15.82	0.99	3.31
Portugal	4451	64.39	21.79	3.46	3.13
Romania	6739	66.41	17.09	2.01	3.42
Sweden	2680	76.89	17.29	5.28	2.16
Slovenia	3196	73.94	15.23	3.41	4.72
Slovakia	6695	71.87	15.74	2.25	4.63
United Kingdom	6072	71.23	19.55	2.03	3.58

in the Appendix A). Respondents who reported the highest average level of mental health were found to reside in Nordic countries, such as Norway, Denmark, and Sweden, whereas in Greece, the Czech Republic, and Portugal, they reported the lowest average mental health. In the latter group, the within-country differences were higher compared to countries with the best mental health results.

In applying a multilevel modeling framework, we developed a three-level model, where individuals (Ni = 141,648) are nested in households (Nj = 100,636), which are nested in countries (Nk = 26). In the analysis, all continuous variables were centered on the mean. The first six models contain the individual-level control variables and investigate the relationship between having children, gender, and mental health. The results from these models are presented in

Table 3. Multilevel regression of mental health on parenthood—main effects at the individual level; total sample, women, and men subsamples.

	Total Sample		Women		Men
	Model 1 β (SE)	Model 2 β (SE)	Model 3 β (SE)	Model 4 β (SE)	Model 5 β (SE)	Model 6 β (SE)
Intercept	1.559 (0.094)	8.398 *** (0.928)	−2.289 * (0.973)	5.046 *** (0.985)	1.840 (1.033)	8.385 *** (1.039)
Has child under 13	−0.051 (0.125)	0.951 *** (0.124)	0.017 (0.156)	0.904 *** (0.156)	0.403 * (0.186)	0.821 *** (0.184)
Gender
Man (ref.)	-	-
Women	−2.856 *** (0.082)	−2.129 *** (0.088)
Household income
<60% of median		−3.694 *** (0.166)		−3.402 *** (0.210)		−3.159 *** (0.227)
60–80% of median income		−1.682 *** (0.162)		−1.574 *** (0.202)		−1.574 *** (0.222)
80–120% of median income (ref.)		-		-		-
>120% of median income		1.936 *** (0.129)		2.022 *** (0.165)		1.741 *** (0.171)
Employment status
Full-time work (ref.)		-		-		-
Part-time work		−1.338 *** (0.160)		−0.279 (0.196)		−2.331 *** (0.334)
Non-working		−5.891 *** (0.126)		−3.721 *** (0.170)		−9.861 *** (0.208)
Missing value		−0.057 (0.210)		0.833 ** (0.302)		−0.744 ** (0.306)
Partner’s employment
Not partnered		−4.260 *** (0.128)		−5.018 *** (0.162)		−2.298 *** (0.205)
Full-time work (ref.)		-		-		-
Part-time work		−0.144 (0.200)		−0.950 * (0.417)		0.581 * (0.250)
Non-working		−3.118 *** (0.154)		−4.560 *** (0.249)		−0.892 *** (0.216)
Variance
National	4.470	4.337	4.350	4.264	4.655	4.561
Household	11.773	11.424	11.177	11.014	10.773	10.354
Individual	13.165	13.008	13.859	13.646	13.624	13.292
N	141,648	141,648	77,240	77,240	64,408	64,408
−2 Log Likelihood	−602,828.18	−600,212.68	−331,514.24	−330,340.8	−274,868.84	−272,927.15

All results are controlled for age, age square, level of education, number of other adults living in the household. Models 1, 3, and 5 also controlled for having a partner, whereas in Models 2, 4, and 6, the variable partner’s employment status contains that information. FT: full-time, PT: part-time, NW: non-working, MV: missing value, SE: standard error, ref.: reference category. * p < 0.050; ** p < 0.010; *** p < 0.001 (two-sided).

. We then examined the effect of the policy indices—the main contextual variables—as well as their interactions with parenthood for women and men separately (see

Table 4. Multilevel regression of mental health on family policy environment and controls—main effects at the macrolevel and cross-level interaction effects; total sample, women, and men subsamples.

	Model 9 β (SE)	Model 10 β (SE)	Model 11 β (SE)	Model 12 β (SE)	Model 13 β (SE)	Model 14 β (SE)
Total sample
Intercept	7.934 *** (0.885)	7.902 *** (0.874)	8.229 *** (0.782)	8.206 *** (0.781)	8.180 *** (0.781)	8.187 *** (0.781)
Has child under 13	1.047 *** (0.126)	1.047 *** (0.126)	1.047 *** (0.126)	1.047 *** (0.126)	1.077 *** (0.126)	1.077 *** (0.126)
Gender
Man (ref.)	-	-	-	-	-	-
Women	−2.114 *** (0.088)	−2.114 *** (0.088)	−2.114 *** (0.088)	−2.114 *** (0.088)	−2.111 *** (0.088)
GDP	0.065 (0.058)	0.056 (0.058)	0.008 (0.052)	0.005 (0.053)	0.005 (0.053)	0.005 (0.053)
Social expenditure	−0.095 (0.167)	−0.045 (0.173)	−0.061 (0.142)	−0.040 (0.148)	−0.038 (0.148)	−0.039 (0.148)
Parenting norms			−4.838 ** (1.525)	−4.699 ** (1.547)	−4.682 ** (1.548)	−4.685 ** (1.548)
FPI 2009	0.104 (0.978)	0.484 (1.048)	1.156 (0.894)	1.295 (0.939)	1.540 (0.941)	1.533 (0.941)
IPI 2009	1.158 (0.824)	1.057 (0.818)	0.176 (0.764)	0.159 (0.762)	0.039 (0.763)	0.044 (0.763)
FPI 2009*IPI 2009		−0.849 (0.925)		−0.378 (0.810)	−0.359 (0.810)	−0.317 (0.812)
FPI 2009*Does not have child under 13 (ref.)					-	-
FPI 2009*Has child under 13					−0.653 *** (0.141)	−0.624 *** (0.146)
IPI 2009*Does not have child under 13 (ref.)					-	-
IPI 2009*Has child under 13					0.300 *** (0.094)	0.287 ** (0.096)
FPI 2009*IPI 2009* Does not have child under 13 (ref.)						-
FPI 2009*IPI 2009*Has child under 13						−0.103 (0.138)
Variance
National	3.978	3.914	3.374	3.359	3.361	3.361
Household	11.548	11.548	11.548	11.548	11.546	11.546
Individual	12.903	12.903	12.903	12.903	12.902	12.902
N	141,648	141,648	141,648	141,648	141,648	141,648
−2 Log Likelihood	−600,403.95	−600,403.53	−600,399.68	−600,399.58	−600,384.1	−600,383.82
Women
	Model 15 β (SE)	Model 16 β (SE)	Model 17 β (SE)	Model 18 β (SE)	Model 19 β (SE)	Model 20 β (SE)
Intercept	4.671 *** (0.960)	4.646 *** (0.954)	4.980 *** (0.854)	4.971 *** (0.855)	4.940 *** (0.856)	4.942 *** (0.856)
Has child under 13	0.950 *** (0.158)	0.867 *** (0.185)	0.951 *** (0.158)	0.951 *** (0.158)	1.034 *** (0.159)	1.034 *** (0.159)
GDP	0.0563 (0.0583)	0.056 (0.059)	−0.00455 (0.0518)	−0.006 (0.052)	−0.005 (0.052)	−0.005 (0.052)
Social expenditure	−0.173 (0.167)	0.067 (0.176)	−0.138 (0.140)	−0.129 (0.147)	−0.127 (0.147)	−0.127 (0.147)
Parenting norms			−5.101 *** (1.506)	−5.046 *** (1.533)	−5.030 ** (1.535)	−5.031 ** (1.535)
FPI 2009	0.003 (0.983)	0.582 (1.068)	1.110 (0.882)	1.165 (0.929)	1.529 (0.933)	1.528 (0.933)
IPI 2009	1.099 (0.828)	1.055 (0.834)	0.064 (0.754)	0.058 (0.754)	−0.105 (0.757)	−0.104 (0.757)
FPI 2009*IPI 2009		−1.091 (0.944)		−0.149 (0.803)	−0.128 (0.804)	−0.115 (0.808)
FPI 2009*Does not have child under 13 (ref.)					-	-
FPI 2009*Has child under 13					−0.866 *** (0.180)	−0.861 *** (0.182)
IPI 2009*Does not have child under 13 (ref.)					-	-
IPI 2009*Has child under 13					0.363 ** (0.120)	0.360 ** (0.121)
FPI 2009*IPI 2009* Does not have child under 13 (ref.)						-
FPI 2009*IPI 2009*Has child under 13						−0.028 (0.180)
Variance
National	3.993	3.983	3.320	3.317	3.321	3.319
Household	11.196	10.558	11.196	11.196	11.19	11.190
Individual	13.502	13.132	13.502	13.502	13.502	13.502
N	77,240	77,240	77,240	77,240	77,240	77,240
−2 Log Likelihood	−330,439.07	−272,967.07	−330,434.31	−330,434.3	−330,418.97	−330,418.96
Men
	Model 21 β (SE)	Model 22 β (SE)	Model 23 β (SE)	Model 24 β (SE)	Model 25 β (SE)	Model 26 β (SE)
Intercept	8.045 *** (0.980)	8.003 *** (0.963)	8.334 *** (0.892)	8.295 *** −0.887	8.273 *** (0.888)	8.299 *** (0.888)
Has child under 13	0.867 *** (0.185)	0.867 *** (0.185)	0.867 *** (0.185)	0.867 *** (0.185)	0.886 *** (0.185)	0.888 *** (0.185)
GDP	0.068 (0.060)	0.056 (0.059)	0.012 (0.055)	0.007 (0.055)	0.008 (0.055)	0.007 (0.055)
Social expenditure	0.003 (0.171)	0.067 (0.176)	0.036 (0.149)	0.072 (0.154)	0.073 (0.154)	0.072 (0.155)
Parenting norms			−4.734 ** (1.599)	−4.500 ** (1.611)	−4.484 ** (1.612)	−4.491 ** (1.613)
FPI 2009	0.095 (1.006)	0.582 (1.068)	1.125 (0.937)	1.359 (0.977)	1.482 (0.980)	1.468 (0.980)
IPI 2009	1.185 (0.848)	1.055 (0.834)	0.225 (0.801)	0.196 (0.793)	0.084 (0.796)	0.098 (0.796)
FPI 2009*IPI 2009		−1.091 (0.944)		−0.639 (0.844)	−0.630 (0.845)	−0.521 (0.848)
FPI 2009*Does not have child under 13 (ref.)					-	-
FPI 2009*Has child under 13					−0.357 (0.192)	−0.293 (0.196)
IPI 2009*Does not have child under 13 (ref.)					-	-
IPI 2009*Has child under 13					0.299 * (0.129)	0.262 * (0.131)
FPI 2009*IPI 2009*Does not have child under 13 (ref.)						-
FPI 2009*IPI 2009*Has child under 13						−0.287 (0.187)
Variance
National	4.085	3.983	3.528	3.488	3.491	3.258
Household	10.558	10.558	10.558	10.558	10.556	10.557
Individual	13.131	13.132	13.132	13.132	13.132	13.131
N	64,408	64,408	64,408	64,408	64,408	64,408
−2 Log Likelihood	−272,967.72	−272,967.07	−272,963.94	−272,963.65	−272,959.32	−272,956.37

All results are controlled for the individual- and household-level control variables. FPI 2009: Familizing Policy Index 2009; IPI 2009: Individualizing Policy Index 2009; SE: standard error, ref.: reference category. * p < 0.050; ** p < 0.010; *** p < 0.001 (two-sided).

). Based on previous research, a variation in mental health in parents across countries was expected (Margolis and Myrskylä 2011); therefore, random slopes for the parenthood variable were included in the level-3 models.

4. Results

Figure 1 (and Columns 5–6 of Table A2 in the Appendix A) presents the scores for the policy indices. The European states are quite equally distributed on the individualization scale; half of them have more than 50% of the highest score, while half have less than 50%. For the familization scale, the majority achieved more than half of the top score. These results show that European countries offer quite similar levels of familialism, with diverse compositions; however, their levels of individualism are much more scattered. On the individualization scale, Poland, Latvia, and Romania scored the lowest, while Denmark, Sweden, and Norway scored the highest. On the familization scale, Sweden, Norway, and the Czechia scored the lowest, while Austria, Slovenia, and Slovakia scored the highest.

In Table 3, we compare the mental health status of women and men paying special attention to the modifying effect of parental status. The unadjusted model shows that parental status is not related to mental health in a significant manner (Model 1), but when adjusted by income, employment status, and partner’s employment status, people raising children under 13 reported significantly better mental health than others, who either have older children or none at all (Model 2). This finding contradicts Hypothesis 1. When looking at the separate models for women and men, we found the same results for women. In the case of men, there is a significantly positive association between their mental health and parenthood, even before controlling for income and employment. This difference suggests that for women, the stressful elements of parenting young children are connected to conflicts with employment and financial strain, while men are not affected as much by these factors. Hypothesis 2 is also falsified by our results; hence, parenthood has a positive association with both men’s and women’s mental health. In fact, after controlling for certain sources of stress, raising a young child proves to be more beneficial for women than for men.

In Table 4, we added the policy indices as well as the country-level control variables. Across the models, the main effect of parenthood is significant and positive. This means that when we take into account individual, household, and macrolevel differences (e.g., general financial status of the country of residence, spending on social expenses, or parenting norms), parents of young children report better mental health than the rest of the sample. In Models 9–10, 15–16, and 21–22, we included two macrolevel control variables, and no main effect was found on the policy indices in any of the samples. The interaction term between familization and individualization did not have a significant effect either, suggesting that the level of one policy index does not modify the association of the other policy index with mental health.

Models 11–12, 17–18, and 23–24 included an indicator for parenting norms. The coefficient of this measurement indicated worse mental health for respondents living in countries where the average support for beliefs that young children suffer when their mothers work is high compared to others living in countries where the general opinion is more supportive of working mothers. This is especially the case for the women’s subsample. Controlling for this aspect, however, still did not change the previous results of the policy indices.

In the final models (Model 13–14, 19–20, and 25–26), we focused on the association between parenthood and the policy environment. The results show that in countries with higher levels of familization, parents of young children report poorer levels of mental health (total sample: b = −0.653) ceteris paribus. However, based on Models 19 and 25, it is women (b = −0.866) for whom raising young children in familizing countries is negative. There is no significant difference for fathers. Regarding individualizing policies, having access to more policy measures is related to better mental health across the total sample and the subsamples. The positive association is somewhat stronger for mothers of young children (b = 0.363) than for fathers of youngsters (b = 0.299). The inclusion of the interaction between the policy indices and parental status did not lead to any significant results.

In summary, these findings do not confirm Hypothesis 3: the simultaneous levels on both policy indices do not show any association with mental health in general or with the mental health of parents in particular. However, the results confirm Hypothesis 4: women do have a stronger mental health benefit when raising a young child in more individualized countries. In fact, raising young children in familizing countries has a negative effect on mothers of young children.

5. Discussion

This article set out to explore whether simultaneous levels of familialism and individualism—as the policy contexts in which parents arrange care—have any relationship with parental mental health. Previous research has investigated the relationship between specific policy measures or policy regimes but thereby lack a more holistic approach. The first aim of this study was to investigate whether parenthood was associated with different levels of mental health compared to not being a parent. The second aim was to test our hypothesis about parents having better mental health in countries where familization and individualization are both at a high level.

In light of the statistical results, we report three key findings. This study supports evidence from previous research, e.g., (Aassve et al. 2015; Nelson et al. 2013), according to which parents have better mental health than nonparents. After we controlled for multiple key individual factors (e.g., educational level, income, and employment status) as well as macrolevel variables (GDP, social expenses, and parenting norms), parents of young children reported better levels of mental health than other adults. Therefore, our results support previous findings of a positive effect of parenthood on mental health. Such a positive relationship has been explained, for example, as due to feelings of meaning in life, fulfillment of social roles, and psychological need satisfaction (Nelson-Coffey 2018).

The findings of the multilevel analysis did not confirm the hypothesis (H3), according to which parents would have better mental health if both types of policies were available to them. Due to the lack of similar previous research, we have no basis to compare these results. However, they may suggest that in societies with one major policy orientation there is harmony between childcare norms and family policies. Therefore, having policy options that support different care ideals which parents can choose from does not appear to affect parents’ mental health on average.

The last important conclusion is that only individualizing policies proved to be beneficial for parents, while familizing policies showed a negative relationship with their mental health, especially for women. This is consistent with our other findings regarding parenting norms: traditional views have a negative association with mental health. Due to the uniqueness of our research on the simultaneous presence of familizing and individualizing policy packages, we cannot compare these findings to other results. Thus, validating them on the basis of different datasets is a task for future research.

One major limitation of our study is the lack of the most appropriate measures for some of the control variables. We acknowledge that the measurement we used to control for parenting norms might not be the most ideal due to its gender bias, but it was the only indicator that was available for the sample of European countries. In addition, in our research, we compared a group of people who were parenting at least one child under 13 who lived in the same household with people who were not doing so. In reality, this latter group cannot be considered as a completely distinct group. Therefore, using them as a comparison is based on our working definition, which is informed by previous research stating that parents of young children have poorer mental health than parents of older children (Luthar and Ciciolla 2016).

The most important strength of our study is its examination of the relationship between mental health in parents and family policies understood as complex packages of measures that have both familizing and individualizing tendencies, taking them both into consideration. We found that mothers and fathers benefited from a higher level of individualization, while familization negatively affected the mental health of mothers of younger children. This should indicate to policymakers that parents benefit more from longer, well-paid parental leave, longer paternity leave, and higher formal care coverage than standard care allowances. Future research should focus on including regional and even company-level data to obtain a more accurate picture of the relationship between family policies and mental health in parents. It would also be interesting to examine the same relationships using a longitudinal approach, which would make it possible to study changes in the patterns of mental health in comparison to changes in the family policy mix.