Are Mothers and Daughters Most Important? How Gender, Childhood Family Dissolution and Parents’ Current Living Arrangements Affect the Personal Care of Parents

: This study examines adult children’s propensity to provide personal care to older mothers and fathers. The theory of intergenerational solidarity facilitates the understanding of commitment and support between adult children and parents. Solidarity may depend on childhood events as well as the current situation, and we therefore focus on whether there was a parental breakup in childhood and the parent’s current living arrangements. We also focus on the gendered aspects of the relations as earlier research has found stronger matrilinear relationships. The propensity for personal care was analyzed with regression analysis using the 2012 Swedish Generations and Gender Survey. The results show that daughters are more likely than sons to provide personal care to both parents. Parental breakup in childhood does not change the propensity of personal care to any parent. The probability of receiving care is higher for lone mothers than for mothers living with the father, but not for repartnered mothers. Adult children’s care provision does not differ for lone fathers and fathers living with the mother, but children are more likely to provide care to lone fathers than to repartnered fathers. We interpret this to indicate that repartnering weakens ties to fathers but not mothers. The results indicate that the child’s gender and the parent’s living arrangements operate differently with regard to care for mothers and fathers. The most common pattern is care provided from daughters to mothers. For example, daughters of lone mothers are more likely to provide care than sons in the same situation. We conclude that intergenerational solidarity is not affected by parental breakup in childhood but that present living arrangements affect such solidarity in gendered ways.


Introduction
Parent-child relationships are more or less enduring over the life course, although strengths and dependencies differ by context and social group. These relationships become increasingly crucial in aging populations where the welfare system does not match spending to the increasing needs of an aging and diverse population. This situation has become a gendered issue because daughters more often provide help and mothers receive more help (Bonsang 2007;Grigoryeva 2017). The mother-daughter bond is also the strongest among the parent-child dyads (Swartz 2009). The concept of intergenerational solidarity facilitates an understanding of the relationships, commitment, support and care between adult children and their parents. This solidarity may depend on childhood events and the current situation, which may entail an older parent needing help.
In this study, we investigate factors that may influence the provision of personal care to older parents by adult children. We ask whether parental breakup during childhood and parents' current living arrangements shape the provision of personal care to older parents by their children-for example, when parents are living alone or repartnered. Personal care involves helping mothers and fathers to eat, get up, dress, bathe, and/or use the toilet. These are intimate tasks that are much less commonly provided in general than other forms of care (Szebehely et al. 2014; National Board of Health and Welfare 2012), and are provided only when individuals are ill and require regular care. Although personal care is often related to other forms of care (Jegermalm 2006), it may not be directly generalizable to all types of parental care. Studying personal care separately from other forms of care is important because this type of care is often time-consuming, demanding and crucial to recipients, who are dependent on their care providers (Fine and Glendinning 2005).
This study poses the questions of who performs and who receives personal care in Sweden, a generous welfare state where eldercare is, by law, a municipality matter, and family responsibility supposedly occurs voluntarily (Jegermalm et al. 2019). Sweden is also known for its strong norms of gender equality, including gender-egalitarian ideas of care responsibilities. Research has indicated that there has not been substantial changes in policies about eldercare (for overview, see Peterson 2017;Meagher and Szebehely 2019) or norms of gender equality during the last decade (Nyman et al. 2018;Gustafsson Sendén et al. 2019).
Sweden has one of the world's most aged populations, but public spending on eldercare has not increased in proportion to increases in the size of the aging population. The availability of eldercare has thus decreased since the 1980s (Szebehely and Meagher 2018). Although a statutory right to eldercare exists, informal caregiving has increased among both women and men (Jegermalm and Grassman 2012;Szebehely et al. 2014), and the role of intergenerational caregiving is predicted to become increasingly important (Jegermalm et al. 2019). Most Swedish studies have combined research on care to anyone outside and/or inside the household, and in a few cases, older parents have been distinguished (e.g., Stanfors et al. 2019; National Board of Health and Welfare 2012; Szebehely and Ulmanen 2012). The care for mothers and fathers by adult daughters and sons has not earlier been separated by the four parent-child dyads. Jegermalm (2006) showed descriptively the care provided to mothers and fathers but not whether it was the daughter or son who provided the care. We extend earlier research on Sweden by identifying both the provider and the receiver in adult child-parent relationships. Distinguishing parent-child dyads may be vital to allocate resources in the welfare state (Ulmanen 2013). The circumstances of eldercare have become even more critical following the COVID-19 pandemic, further revealing inadequacies in Swedish eldercare.
In Sweden, divorce and separation rates have been high since the mid-1970s and remain among the highest worldwide (González-Val and Marcén 2012). The increase in divorce rates started already in the 1920s and was largely related to women's labor force participation (Sandström 2011). For example, the relative divorce risk for women in the early 1970s was less than half (0.4) the risk in 1980 and further increased to 40% by the early 1990s (Andersson 1995). The separation rate for cohabiting couples is even higher, also among older cohorts (Henz and Jonsson 2003;Hoem and Hoem 1992). If divorce and separations are combined, the disruption rate in Sweden is higher than that in most other countries (Andersson 2003). From the children's perspective, 20% experienced parental dissolution before age 16 among individuals born 1933-1979, of which 65% had cohabiting parents (Härkönen et al. 2021). The high separation rates indicate that a large proportion of children come from homes in which parents live apart. Sweden also has one of the world's highest proportions of older individuals living alone; 48% of individuals older than 75 lived alone in 2012 (Padyab et al. 2019;Statistics Sweden 2021;Tomassini et al. 2004). Single living often has a range of negative consequences for older adults, and this group is often considered particularly vulnerable (Shaw et al. 2018), especially from a care perspective. Partnered parents often receive support from their partner, but single living parents may need to rely on their children's support to a larger extent (Silverstein and Giarrusso 2010). This group may also be more likely to rely on public and private services.
Historically, the death of a parent was often the cause of a challenging situation for surviving parents and children. In some cases, the surviving parent repartnered.
Today, in addition, the emergence of new family constellations due to separation/divorce makes families increasingly complex and presents more challenging situations that may impact family relationships and cause greater boundary ambiguity (van der Pas and van Tilburg 2010). For example, a potential source of ambivalence is that biological ties are often stronger than marital ties (Martin-Uzzi and Duval-Tsioles 2013). Stepchildren or stepparents tend to be excluded from individuals' perceptions of who belongs to the family, although this is rare in biological two-parent families (Castrén and Widmer 2015).
Stepmothers are often perceived as significant kinkeepers but fathers seem to have a more distancing role (Schmeeckle 2007). Differences in the bond strengths may be moderated for step relationships of longer duration (Becker et al. 2013), demonstrating the complexity of step ties. Some potential issues which can be challenging in complex families are specific to older parents, based on weakening intergenerational solidarity (e.g., conflict about inheritance; Jensen et al. 2019). Boundary ambiguity may generate a mismatch between role expectations and responsibilities.
Demographic changes will become even more relevant as younger generations with more complex family constellations enter old age (Fingerman et al. 2012). With an aging population, a large proportion of older individuals, of whom the majority are parents, need help. The questions of whether and how children engage in personal care for aging parents with breakup histories and different living arrangements are crucial to understanding the potential support deficits that older individuals will likely face. These deficits may create gendered inequalities both in old age and between caregiving daughters and sons. The results from this study are relevant both for countries in which public services are strained, as in Sweden, and in countries in which eldercare is a family matter but the family may have increasing difficulties to cover the needs.

Research on Care for Older Parents According to Parental Breakup and Living Arrangements
Research has shown that adult children's potential caregiving may be influenced by earlier life events, such as childhood parental breakup, which may indicate a more distant relationship with parents during both adolescence and adulthood (Coleman et al. 1997;Silverstein and Bengtson 1997;Webster and Herzog 1995). There is extensive literature on adult children's support for older parents; nevertheless, there is a research gap on the gendered aspect of relationships between adult children and parents, parental breakup and parents' current living arrangements (for an overview, see Herlofson 2013). A few studies have investigated how adult children's experience of parental breakup in childhood is associated with care for aging parents. A Dutch study found that adult children with a parental breakup in childhood provided less support to parents, particularly to fathers (measured by a support scale including several support items, including practical support) (Kalmijn 2007). An earlier study based on U.S. data from 1988 found that parental divorce during childhood was associated with an increase in time transfers to mothers and a decrease in that to fathers (Furstenberg et al. 1995). Ermisch (2014) found that divorced/separated mothers and fathers receive more help from children than married parents.
Other studies have indicated that children's provision of care to parents may be a gender issue. Mureşan (2017) studied the association between parental breakup in childhood, as well as widowhood, and the different types of support adult children provide to parents in a cross-national setting using the Generations and Gender Survey (GGS, excluding Sweden). She found that adult children provide less personal care to divorced older fathers than to partnered fathers, which is similar to the findings of Lin (2008). Mureşan (2017) also found that help to mothers was the same regardless of the mother's relationship status with the father. These findings contradict the finding by Glaser et al. ( 2008) that divorced mothers receive more support than their married counterparts. The timing of parental breakup also seems to matter for parental care. For example, earlier studies have indicated that the younger the child was at the parental breakup, the less contact the child had with older parents and the less support the child provides to older parents later in life (Furstenberg et al. 1995;Kalmijn 2013).
Parents' living arrangements (often in the form of partner status) and the care provided by children have been investigated in different ways. For example, remarried parents, particularly fathers, have been shown to be less likely to receive support from adult children (Kalmijn 2007). A UK study also showed that remarried fathers received less support from their children than married fathers, with no difference in support to remarried and married mothers. Divorced mothers received more support than mothers who were still married (Glaser et al. 2008). Parents who live without partners have been shown to receive more care from children (Gannon and Davin 2010;Leopold et al. 2014;Luppi and Nazio 2019;Silverstein et al. 2006). In summary, the findings from earlier studies indicate large variations in who receives care according to parents' living arrangements and gender. The evidence is somewhat inconsistent, potentially due to the different time periods and contexts of previous studies.

Theoretical Framework and Hypotheses
Intergenerational ties are found to be strongest between biological parents and their children. From the child's perspective, these bonds may create the strongest sense of obligation for reciprocity when parents need care (Manderson and Warren 2013). Mothers are generic kinkeepers in most societies, who maintain, organize and conceptualize relationships within a family system (Kalmijn et al. 2019). One explanation is gender-specific employment patterns that lead to women's higher familial responsibilities (Chesley and Poppie 2009).
To understand what leads to stronger ties and more readiness to help older parents, we draw on the intergenerational solidarity theory, which suggests that the child-parent relationship is embedded in past and present family structures (Bengtson et al. 2002;Bengtson and Oyama 2010). The theory treats solidarity as multidimensional with both positive and negative aspects. Family members, such as adult children, represent a latent resource for older parents that can be activated in times of need as functional solidarity. Whether an older parent needs and receives support from a child is linked to structural solidarity-e.g., proximity to family members, family size and the health of family members. Another central component of our study is normative solidarity, which is the strength of the commitment to meet familial obligations. Positive aspects may include feelings of affection and closeness between generations, while negative aspects include conflicts, such as divorce and a new marriage, which lead to weakened bonds and, in the worst case, a terminated relationship. Events early in life may have significant effects on family solidarity later in life. The relatively static role of structures and behavioral expectations is underscored because it is challenging to break free of role expectations and responsibilities. The theory also includes affectual solidarity, which reflects the sentiments held toward family members; consensual solidarity, which is the degree of agreement in, for example, opinions and attitudes; and associational solidarity, which is the frequency and type of contact between generations. Structural, functional, and normative solidarity are direct parts of our framework, but all dimensions are considered to be related.
Gender differences in the bonds and solidarity between parents and children have been found repeatedly (see, for example, Daatland 2007;Kalmijn et al. 2019;Silverstein et al. 2006). The type of care varies, but a common finding across contexts is that daughters provide care for family members more often than sons (Bonsang 2007;Kalmijn 2007). For example, Swedish (mainly descriptive) studies indicate that women provide more informal care in general and are more engaged in personal care than men (Jegermalm 2006;Jegermalm et al. 2014;Szebehely et al. 2014;Ulmanen 2015).
Based on these findings, the study's first hypothesis is that adult daughters are more likely to provide personal care for older parents than adult sons (Hypothesis 1). We expect gender differences among children to be relatively small in Sweden because both women and men engage in paid labor throughout the life course, which has been shown to reduce the gender gap in the provision of help to parents (Sarkisian and Gerstel 2004). Consistent with this reasoning, gender differences have been shown to be greater in family-oriented welfare states than in countries in which the state has the major responsibility for eldercare, as in Sweden (Schmid et al. 2012). As there may be a higher sense of "reciprocity care" to a mother, mothers are expected to receive more care than fathers. This expectation is difficult to test even when controlling for the health status of parents because unobserved needs likely exist that differ between older women and men, including mothers' higher propensity to both ask for and receive help from their adult children. We cannot test this expectation using the available data.
The solidarity theory predicts that divorce and remarriage weaken support to parents. In particular, normative solidarity in the form of familial obligations may be weaker in families with divorce (Bengtson et al. 2002). Empirical studies often find that divorce and remarriage have negative long-term effects on the child-parent relationship and may produce weaker feelings of solidarity, even after friendly separations (Albertini and Garriga 2011;Daatland 2007;Ganong and Coleman 1999;Kalmijn 2013;Kaufman and Uhlenberg 1998;Webster and Herzog 1995). Consequently, growing up with divorced parents may negatively influence intergenerational functional solidarity.
According to the theoretical and empirical findings, we expect that adult children who did not experience parental breakup in childhood are more likely to provide personal care to older parents than adult children who experienced parental breakup in childhood (Hypothesis 2).
Daughters' potentially higher filial responsibility and more supportive behavior compared with those of sons (Silverstein et al. 2006) may make them more likely to conform to the role of caregiver even after experiencing parental breakup in childhood. Hence, we expect that adult daughters who experienced parental breakup in childhood are more likely to provide personal care to older parents than adult sons with the same experience (Hypothesis 3).
Structural solidarity predicts that parents' living arrangements indicate different care needs and lead to different obligations and opportunities for adult children to provide care (Bengtson et al. 2002). As a coresidential union is often seen as protective, studies have consistently found that older individuals living without a partner are more vulnerable than partnered individuals and have fewer socioeconomic resources and higher morbidity and mortality (Burstrom et al. 2010;Weitoft et al. 2004). An older parent living without a partner is likely to expect and require more support from an adult child than a partnered parent (Gannon and Davin 2010;Ikkink et al. 1999;Kalmijn 2007).
Thus, we expect that adult children are more likely to provide personal care to older parents who live without partners than to older partnered parents (Hypothesis 4). Following the assumptions of normative solidarity that parental repartnering often changes ties to parents, we also expect the association to differ depending on whether the parent is with the "original" parent or has repartnered (Daatland 2007;Stuifbergen et al. 2008). Hence, we compare parents in two steps. We first compare parents who live without a partner with partnered parents, and then compare parents who live without a partner with intact and repartnered parents, separately.
In the first hypothesis, we suggest that daughters provide more care to parents in general. However, when parents live alone, both daughters and sons may be similarly likely to provide care. The intergenerational solidarity theory assumes that the norms of reciprocity are strong between generations and for women and men. There is a strong belief in a cyclical process of support over the life course. By helping older parents who live without a partner, adult children may expect to be supported in turn by succeeding generations, particularly when they are in a vulnerable situation of living alone as structural solidarity predicts. It is also likely that it is easier for both daughters and sons to care for parents who live alone because there is no other individual in the household who may have opinions about care or related issues.
Thus, we expect that no gender difference exists among adult children in providing personal care to older mothers and fathers who live without partners (Hypothesis 5).
As we touched upon in the fourth hypothesis, normative and structural solidarity predict that different types of partnerships are associated with different needs and levels of closeness and commitment to the parent. We suggest that children are more likely to provide personal care to older parents who live without partners than to older partnered parents. It is also important to separate intact and repartnered as children's ties and feelings of reciprocity to stepparents are often weaker than to biological parents Coleman 2006a, 2006b). Adult children with a repartnered parent also tend to provide less support to the parent than adult children with two "original" parents living together (Ganong and Coleman 2006a;Kalmijn 2007Kalmijn , 2013. Following this line of reasoning, it is important to compare repartnered and intact parents. We expect that adult children are less likely to provide personal care to older repartnered parents than to parents who live with the "original" parent (Hypothesis 6). As maternal bonds are often stronger than paternal bonds and may become stronger after a parental divorce (Kalmijn 2007;Kaufman and Uhlenberg 1998;Silverstein and Bengtson 1997), it is vital to examine the consequences of parental breakup and parents' living arrangements on caregiving for mothers and fathers separately.

Other Predictors for the Care of Older Parents
In addition to the association between parental breakup and parents' living arrangements, other predictors are important in the analysis of children's care for parents. For example, adult children of all ages engage in care for parents, but increasing age leads to a greater likelihood of providing care (Brandt et al. 2009). Whether and how much an adult child provides care may also depend on the child's partner status, for which the data are mixed across countries. A cross-national European study found that partnered individuals provide less care to parents (Brandt et al. 2009), while a Canadian study found that men's involvement in care to parents does not differ by marital status (Campbell and Martin-Matthews 2003). Swedish studies have indicated that partnered individuals engage most in caregiving in general (Jegermalm 2006;Jegermalm and Grassman 2012).
Regarding the associations between the adult child's education level and parent-care involvement, the results are mixed, often by the type of care. For example, Mureşan (2017) found that middle-and highly educated children are more likely to provide financial and emotional support for parents but not personal care. Winqvist (1999) also found that social class matters for parental care. She found differences in how much time children spent with their elderly parents and in the perception of parental care between social classes, which shaped whether and how children provided personal care. For example, careeroriented children did not prioritize parental care, while wage earners perceived parental care as a family matter. In another cross-national study, higher education was not found to be associated with more time spent on care for parents (Bonsang 2007), and a Swedish study found no differences in the education levels of those who provided the most care (Jegermalm and Grassman 2012).
Some studies have found that gainful employment and longer work hours are associated with less help to parents (Doty et al. 1998;Jegermalm and Grassman 2012), while other studies have found no effect (Brandt et al. 2009). In Sweden, Jegermalm and Grassman (2012) found a decreasing gap in overall caregiving over time between those who were gainfully employed and those who were not. Other obligations, such as taking care of dependent children, may negatively influence caregiving for older parents (Brandt et al. 2009;Sarkisian and Gerstel 2004). Previous research has indicated that having siblings eases the burden of care to parents (Gerstel and Gallagher 2001), and a greater number of siblings decreases the likelihood of caregiving (Bonsang 2007;Brandt et al. 2009;Grundy and Read 2012;Jegermalm and Sundström 2015). The number and gender of siblings seem to particularly affect the care intensities of sons (Roquebert et al. 2018), while some of the higher care intensities of daughters occur in situations where they have older brothers (Arnault and Fontaine 2018).
Other factors that positively influence adult children's care for parents are geographical closeness (Mureşan 2017), good health of the caregiver (Bonsang 2007), and good subjective relationship quality (Ganong and Coleman 2006a). In summary, certain factors can hinder or facilitate adult children's provision of care to aging parents. We will consider some of these factors in our analyses.

Data and Methods
We use the Swedish GGS for 2012/2013 (Thomson et al. 2015), which is representative of the Swedish population aged 18-79 years. In total, 9688 individuals participated in the survey, corresponding to a response rate of 54% (Statistics Sweden 2014). The questionnaire includes items on personal care to parents, relationship histories, labor market attachment, socioeconomic status, health and well-being, and childhood events. As we focus on children's provision of personal care for older mothers and fathers, we select respondents with a living mother and/or father and those at the ages most likely to provide care to parents-i.e., 35-75 years old at the time of the interview. Our effective subsamples comprise 3571 respondents with a living mother and 2466 individuals with a living father and include children who provide care to parents and those who do not. Our design also means that a respondent can be in both subsamples if both parents are alive.

Dependent Variable
We employ logistic regression models in which the dependent dichotomous variables are personal care for the mother and father separately, which is our measure of functional solidarity. The respondents were asked whether they regularly provided personal care (e.g., eating, getting up, dressing, bathing, and using the toilet). In total, 5% of daughters and 3% of sons reported providing personal care to the mother, and 4% of daughters and 2% of sons reported providing personal care to the father (Table 1). We stratified the models by parent gender so that gender differences could be easily distinguished. Other Swedish studies on care to anyone have shown higher proportions of providing care (see summarizing tables in Jegermalm et al. 2014). These studies may not be comparable to ours because we focus on regular personal care, and the differences in prevalence between our study and others are likely due to different operationalizations of care and the use of different samples. Although personal care is related to other types of care (Jegermalm 2006), it is a separable and intense type of care in which fewer people regularly engage. Jegermalm (2006) found that 6% of the total sample provided personal care to someone outside or inside the household, of which two-thirds were women. Similarly, the National Board of Health and Welfare (2012) found that 5% of the total sample did so. Szebehely and colleagues (2014) found that about 6% of women and 3% of men provided personal care (of respondents in the sample) and that personal care was the least common care type.
To further validate our data and findings, we estimated the prevalence of regularly providing personal care to parents in Sweden using another high-quality and frequently used data source, the Survey of Health, Ageing and Retirement in Europe (SHARE). The fourth wave of SHARE for Sweden from 2011/2012 is highly comparable to the Swedish GGS because the questions on personal care are very similar. We used a subsample of individuals of the same ages (1492 individuals) as our sample. In total, 6% reported regularly providing care to a mother and 2% reported providing care to a father in SHARE.
These numbers agree with those found in our sample. We concluded that the GGS was preferable for our analysis because it contains information on younger individuals who may have experienced parental breakup and parents' current living arrangements, unlike SHARE. Mureşan (2017) also used the GGS to conduct a cross-national study (not including Sweden) and found that similar proportions of respondents reported providing personal care to their parents in Norway (4%), France (3%), Germany (2%) and Poland (3%), while the percentage was slightly higher in six other European countries. The incidences of personal care are low across countries. These numbers are similar to those presented by Rostgaard and Szebehely (2012), who found that 3% of the older Swedish respondents living in their own home reported that they needed help with bathing (compared to cleaning and grocery shopping at 16 and 14%, respectively). Due to the potential substantial impacts of children's personal care on both parents and adult children, it is important to study rare circumstances. Rostgaard and Szebehely (2012) aimed to provide an example of the percentages of older individuals reporting needing help with personal care, but comparing reports from potential providers and potential receivers is not optimal, as they do not overlap.

Independent Variables
Our central independent variables are the gender of the adult child, parental breakup in childhood and parents' current living arrangements. For parental breakup, we follow Mureşan (2017) definition of parents' divorce or separation before the child is 15 years of age. We distinguish four types of parental living arrangements: (1) parent living with the "original" parent; (2) parent living with a new partner (either married or cohabiting); (3) parent living without a partner; and (4) other living arrangements (e.g., relatives, nursing home or the adult child). Unfortunately, we cannot distinguish between parents who are divorced, never-married or widowed in the "parent lives without partner" category because of data limitations. We sometimes refer to this category as "lone parent" and "parent who lives alone" for better text flow. The fourth category, other living arrangements, is part of the analyses but not the focus of our research; thus, we will not report on these results. Table 2 displays the descriptive statistics of parental breakup and parents' current living arrangements. Descriptive tables of adult children's personal care for older parents separated by parental breakup and parents' current living arrangements are presented in Appendix A (Tables A1 and A2).
We adjust the models for the characteristics of the adult child that are likely to influence personal care-namely, age, activity status, education level, partner status, child living in the household, sibling composition, distance to parents (in time) and health status. Table 2 displays the descriptive statistics for these variables. Age is categorized in the descriptive statistics but is continuous in the analytical models because it is expected to be linearly correlated with personal care. We categorize the adult child's activity status as (1) employed, (2) unemployed, (3) retired, or (4) other. The last category includes parental leave, sick leave, and studying. The variable representing children living in the household is dichotomous and measures potentially conflicting multigenerational responsibilities. By contrast, having siblings may ease the burden of care for aging parents. The variable representing sibling composition is categorized as (1) no siblings alive or ever; (2) one or more brothers; (3) one or more sisters; and (4) a mix of siblings. Distance to parents is assessed in terms of time in hours to the mother and father (separately). We include it as a continuous measure because distance is expected to be linearly correlated with personal care (displayed as a categorical variable to show the distribution in Table 2). Regarding the respondent's health status, we employ the question, How is your health in general? Would you say it is very good, good, fair, poor or very poor? The variable is categorized as (1) very good or good and (2) fair, poor or very poor (with 3% reporting poor or very poor health).  Table 3 displays the results from the multivariate logistic regression models in which the outcomes depict whether (yes/no) the adult child regularly provides personal care to an older mother and an older father. Notably, our analytical strategy and data enable us to show whether children provide personal care to either the mother or father but not the prevalence of providing care to the mother and/or father separately; that is, we cannot argue that children provide care to mothers more often than fathers. We present the models stepwise by first including parental breakup and parents' living arrangements in separate models (Models 1a-1b and 2a-2b). The final models include both variables (Models 3a-3b). All models are adjusted for the characteristics of the adult child. Bivariate results for the explanatory variables shown are similar to those of the multivariate models (not presented here). First, we test whether adult daughters and sons differ in their likelihood of helping mothers and fathers eat, get up, dress, bathe, or use the toilet. In Models 1-3 in Table 3, the results confirm the hypothesis-namely, that adult daughters are more likely than sons to provide this type of personal care to both mothers and fathers (Hypothesis 1). Additionally, in Models 1a-b in Table 3, we find that adult children who experienced parental breakup in childhood do not provide less or more personal care, and the nonsignificant results are the same for care given to mothers and fathers. Thus, we cannot confirm the second hypothesis that the provision of personal care to parents in late life is influenced by parental breakup in childhood.

Results
We further hypothesize that adult daughters who experienced parental breakup in childhood are more likely than adult sons to provide personal care to parents (Hypothesis 3). To test this hypothesis, we generate two interaction terms between gender and parental breakup in childhood and include models that are otherwise identical to Models 3a-3b in Table 3. The selected results are displayed in Table 4. We find that the probability of providing personal care is higher among daughters than sons who experienced a breakup. This result is significant at the 5%. We note that the size of the interaction term for fathers is similar to that for mothers but is nonsignificant, likely due to the limited sample size. In the next step, we examine whether adult children are more likely to provide personal care to older parents who live without a partner than to parents who live with a partner (Hypothesis 4). First, we compare parents without a partner with partnered parents. The multivariate analyses (models not presented because of space limitations) show that adult children have a higher probability of providing personal care to lone mothers and fathers than to partnered mothers and fathers (OR mothers: 2.44, p-value: 0.000, OR fathers: 2.30, p-value: 0.019). Second, we further separate parents who still live together and repartnered mothers and fathers because these family bonds often differ and may produce different probabilities of personal care. The results are displayed in Table 5. Model 1a shows that children are more likely to provide personal care to a lone mother than to a mother who lives with the father. In Model 1b, we do not find that children are more likely to provide personal care to lone mothers than to repartnered mothers. With regard to care for fathers, Model 2a does not show differences in the personal care provided to fathers when comparing lone fathers with fathers who live with the mother. Adult children are more likely to provide personal care to a lone father than to a repartnered father, as shown in Model 2b. In summary, we find that children are more likely to provide personal care to lone mothers than to mothers living with the father, and are more likely to provide care to lone fathers than to repartnered fathers. Functional solidarity (in the form of health conditions) may be different among these parents. We conclude that parents living alone may trigger latent solidarity and generate incentives for the greater provision of personal care for lone parents relative to that of other groups of parents. This finding is gendered. In vulnerable situations, care is more likely to follow the mother; repartnered mothers receive the same care as lone mothers, while repartnered fathers may be more likely to receive help from the new partner's family. This indicates that the matrilineal family system may still be important in Sweden, consistent with the idea that mothers are kinkeepers (Kalmijn et al. 2019).
We do not expect to find gender differences between adult children providing personal care to older parents living alone (Hypothesis 5). To assess this hypothesis, we include two separate interaction terms between the child's gender and the mother's/father's living arrangements in models otherwise identical to Models 3a-3b in Table 3 (one model for each sample). Selected results from the interactions for parents who live alone are displayed in Table 6, where the reference group is adult sons with a mother/father who lives without a partner (in both models). The results do not indicate gender differences in providing personal care to lone fathers. Adult daughters with lone mothers are statistically more likely to provide personal care than adult sons with lone mothers. We conclude that the daughter-mother relationship may be closer than the son-mother relationship in situations of vulnerability. Table 6. Selected results from multivariate logistic regression models of the child's personal care to parents, interactions between the sex of the child and parents who live without a partner in old age, stratified by the sex of the parent 1 . Finally, we expect that adult children are less likely to provide personal care to older repartnered parents than to original parents living together (Hypothesis 6). Models 3a-b in Table 3 offer no support that personal care provision differs for mothers living with the father and repartnered mothers. The results show that children are less likely to provide personal care to repartnered fathers, a finding that is statistically significant at the 10% level (5% in Model 2b, Table 3 when parental breakup is excluded).

Child's Personal Care to
In Table 3, we find that the older the children are, the more likely they are to provide personal care to the mother and father. We also find that children are less likely to care for mothers when a longer time is required to travel to the mother, but this does not change the care provided to fathers. Unemployed children are more likely to provide care to mothers, but retired children are less likely to provide care to mothers. Care for fathers does not vary by children's activity statuses. Children are more likely to help a father with personal care when they themselves have a nonresident partner or are divorced and currently single. The relationship statuses of children do not change their willingness to provide personal care to mothers. An exception is that children with a nonresident partner are more likely to care for mothers when living arrangements are not adjusted (statistically significant at the 10% level). Children's education level, health, and whether they have children living in the household do not matter for their likelihood of providing care to their mother or father. We also note statistically nonsignificant, gendered patterns for sibling composition. An only child is the most likely to provide personal care to a mother and a child with only sisters is the least likely. Children with only brothers are most likely to provide care for fathers, and those with only sisters are the least likely.

Sensitivity Analyses
We perform several sensitivity analyses to test whether the results for personal care to parents are robust. We address multiple indicators of the quality of the child-parent relationship. First, we include measures of relationship quality with parents until 15 years of age and present relationship quality with parents. None of these predictors produced statistically significant results or changed the main associations. Second, as a proxy for the bonds between parents and children during childhood, we include whether the child lived with both parents up to 15 years of age instead of parental breakup in childhood. Similar to parental breakup, the results are nonsignificant for personal care provided to fathers. For mothers, the result is statistically significant, indicating that daughters who lived with both parents are more likely to provide help to their older mothers. When we compare daughters and sons, we find that daughters who did not live with both parents are more likely to provide care to mothers than sons. Third, we also consider intergenerational family ties by operationalizing siblings without considering their gender. The result is nonsignificant, indicating that the number of siblings is not associated with the prevalence of providing personal care to aging parents in our study. Thus, the coherence between our main results and the sensitivity analyses with different measures of the strength of the relationship between the adult child and parent and intergenerational family ties supports our findings.

Discussion and Conclusions
The present study investigated adult children's propensity to regularly help older parents with personal tasks such as eating, getting up, dressing, bathing, and using the toilet based on earlier life events and current situations. The genders of the children and parents were of particular interest because earlier studies have indicated the importance of women as kinkeepers and caregivers within various family systems. We applied two central concepts of the intergenerational solidarity theory, structural and normative solidarity, and argued that parental breakup in childhood may weaken later-life functional solidarity and, hence, the prevalence of providing personal care to parents. We further argued that normative and structural solidarity operate differently when the parent lives with the other parent, with a new partner or without a partner. In particular, we anticipated matrilineal kinkeeping to be central with regard to who receives personal care from adult children.
Consistent with previous research, we found that, compared with sons, daughters were more likely to provide personal care to mothers and fathers (Grigoryeva 2017). This result is in accordance with women's greater importance in maintaining intergenerational ties. The gender difference was not large, perhaps because Sweden aims to be a genderegalitarian society in which the most basic needs of older individuals are, or at least should be, provided by the municipality (Schmid et al. 2012). In contrast to expectations, parental breakup in childhood was not negatively associated with personal care provision. Three reasons may explain this finding. First, conflict and solidarity can coexist; as conflicts resolve, the relationship may improve (Bengtson et al. 2002). Second, the negative effects of a breakup on family ties may weaken over time as divorce becomes more common and less stigmatized (Glaser et al. 2008;Sigle-Rushton et al. 2005), as it has in Sweden. Third, as others have discussed, parental divorce may have long-lasting effects but may be overcome when older parents are in need of care (Lin 2008).
In contrast to Mureşan (2017), we found that adult daughters who experienced a parental breakup, compared with their male counterparts, tend to provide personal care more often to mothers in late life, with no gender difference in the personal care provided to fathers. After a breakup, daughters likely engage in their mother's life more than sons. Daughters and sons may have similarly strong (or weak) normative solidarity with fathers after a breakup. Previous international studies have found that care is predominantly provided by women for women across generations. We find a similar pattern for Sweden.
Adult children were shown to be more likely to provide personal care to lone mothers than to mothers living with the father. A potential explanation is that structural solidarity is stronger among lone mothers-i.e., mothers living without a partner are more vulnerable, and partnered mothers may have better social and economic resources. We also found that adult children with fathers who live without a partner were similarly likely to provide personal care as children with fathers who lived with the mother. Fathers living without partners were not perceived as more vulnerable than their married counterparts. These fathers may also have a more distanced relationship with children compared to married fathers. Thus, they may experience weaker normative solidarity, a finding also reported in other studies (e.g., Lin 2008). The motivation to provide or, perhaps more accurately, not to provide personal care to fathers likely differs here. One plausible explanation for this finding is that fathers who live without a partner are less likely to receive help because of a distant relationship with children. A spouse is often the primary caregiver; thus, married partners may receive help from the adult child's mother. Although we do not contrast care to mothers and fathers, we can speculate that matrilineal solidarity can explain why lone mothers are potentially more likely to receive personal care while lone fathers receive less. Lone fathers tend to have weaker ties to children. Compared with lone fathers, lone mothers may also experience stronger structural solidarity in that they are more often older and frailer; thus, they need more help. In general, women may also have better skills to communicate how they are doing and what help they need (Ek 2015;Weisman and Teitelbaum 1989).
Repartnering may reduce parents' needs for support from children, but a new partner does not seem to benefit the mother; we found that repartnered mothers were similarly likely to receive help as mothers living without a partner. We also found that re-partnered fathers were less likely to receive personal care than fathers who lived with the child's mother. A different explanation could underlie why children help fathers. For example, children may help disabled fathers to ease the mother's burden. Consistent with findings by Kalmijn (2007), we found that repartnered fathers receive less care than lone fathers. This finding may imply that the normative solidarity between children and fathers is weaker and that care to repartnered fathers may instead be provided by the new partner's family. Although our explanation is speculative, it is plausible because repartnered older men often have younger partners who may have a greater ability and better resources to care for them. The findings partly agree with conclusions drawn by Silverstein and Bengtson (1997), who found that children feel more obligated to care for divorced/separated/widowed mothers and that children have a more detached relationship with divorced/separated/widowed fathers.
Consistent with many other studies on parent-child bonds (Furstenberg et al. 1995;Grigoryeva 2017;Larsson and Silverstein 2004), our study notes the importance of matrilineal care structures: the most common pattern was personal care provided through the mother's line and from daughters to mothers. For example, in a recent study on intergenerational ties from the Netherlands comparing bonds with stepparents and biological parents, Kalmijn et al. (2019) concluded that biology was more important for mothers and that a partnership premium was more important for fathers. Thus, intergenerational solidarity theory would benefit from the integration of gender and lineage (i.e., whether the relationship is matrilineal or patrilineal) as analytical dimensions to facilitate an understanding of solidarity within the growing complexity of family ties.

Policy Implications
Eldercare in Sweden is the responsibility of the municipalities, and solutions should be formulated at this level. Nonetheless, there are groups of adult children that provide care to aging parents. Providing care to aging parents may have negative consequences for adult children. Frequent and time-consuming caregiving, such as personal care, may hinder full labor market participation and participation in other activities, such as leisure and recuperation (Berecki-Gisolf et al. 2008;Lilly et al. 2007;Szebehely et al. 2014). Informal caregiving has also been shown to adversely influence the well-being of adult children (Borg and Hallberg 2006;Pinquart and Sörensen 2006;Ulmanen 2015). Relying on children's care to parents rather than on available care provided by the municipality may also lead to inequalities between older individuals, at least when stratified by living arrangements and gender, as seen in our study. Before granting eldercare services (such as homecare), local authorities assess the older individual's needs, for example, by asking about their health impairments and living situation, including partner and family members. An individual with a partner or adult child may receive less support as they may be assessed as being in less need. Policies on the eligibility of eldercare should acknowledge that some groups of older women and men may be more exposed because of their living arrangements. Even if it may look like their needs are covered by, for example, a partner, they may in fact not be covered. The social network and kinship ties are very different across different family constellations of older women and men. According to our findings, parents who live without a partner, and to some extent also repartnered parents, seem to be in an elevated vulnerable situation. Municipalities should be more observant of the help (or no help) that older individuals may receive from their children and partners, and not assume that such help is accurately provided (or not provided).
In countries with strong welfare support, such as Sweden, older individuals predominantly prefer to receive formal care (Eurobarometer 2007). When the child is expected to provide help with getting up, eating, bathing, dressing and using the toilet, the relationship quality to the parent may be endangered. If personal care and many other care types are sufficiently provided by the municipality, then the child-parent relationship can focus on providing emotional support, joy, and love. The negative aspects of a care deficit may increase as life expectancy increases, thereby extending the period of filial responsibility for adult children. This situation is likely to affect daughters more than sons. When daughters provide more eldercare, they may invest less in their paid work and career, and have an increased risk of poor health and sick-leave. This situation may deprive the economy of women who provide personal care for their parents also in the long term, and it can have consequences for the work force.

Study Limitations
The study has at least five limitations. First, the relatively low response rate and small case numbers for some groups limit our analyses. We tested the validity of our findings compared with other data sources and found our results to be comparable and robust. A small sample size may explain why we obtained nonsignificant results-for example, for parental breakup and siblings, for which other studies have found support. We recommend testing using a larger-scale survey. The Swedish GGS has a low response rate (54%) compared with the 67% average response rate for countries participating in the first wave of the GGS in the Generations and Gender Programme. Analyses of nonresponses found that the representation of the population was good in many aspects, such as age and gender, which is more critical than the response rate. Selectivity was not shown to be a problem among the nonresponses, except among individuals born outside Europe who were much less likely to participate (Statistics Sweden 2014). Second, the measure of functional solidarity was dichotomous, which may, arguably, be too crude. The question captures children who help a parent with typical daily activities regularly. We therefore eliminate the gray zones between little help and no help. This may be why we have a relatively small group who report regularly providing help to parents. Personal care in the form of helping the parent eat, get up, dress, bathe, and/or use the toilet is also relatively less common than other types of care, such as housework and grocery shopping (Szebehely et al. 2014). Previous research has found that men are less likely to perform instrumental care. We did not observe this finding, likely because of the phrasing of the question. Third, we do not know the health statuses of the parents; thus, we do not know whether the parents need help from children (or anyone else). Fourth, as a control variable, it would be beneficial to have information on whether the parents use other types of care, such as formal care provided by municipalities. By focusing on the caregivers, we also do not know whether the needs of the parents are covered by formal and/or informal care or how the different types of care complement each other. Notably, this study was conducted in Sweden, where the welfare state rather than the family is the primary provider of eldercare. Finally, we acknowledge that our study does not provide information on preferences for care from either children or parents, which may also be gendered. Institutional Review Board Statement: The study was conducted according to the ethical review and approval of the Swedish Ethical Review Authority (diary number 2011/736-31/5).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data used for analysis in this study is openly accessible on the webpage of the Generation and Gender Programme (https://www.ggp-i.org/data/ accessed on 29 April 2021).

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.