Typology of Parent-Child Ties Within Families: Associations with Psychological Well-Being

Abstract

Middle-aged adults often have relationships with multiple family members (e.g., children and parents). The constellation of parent-child relationships within families may have implications for individuals’ psychological well-being. This study created typologies of parent-child ties by combining multiple dimensions of relationships, and examined the extent to which middle-aged adults showed variability across typologies of parent-child ties within multigenerational families. Using 2,252 parent-child ties across three generations from 633 middle-aged adults, this study identified typologies of parent-child ties based on five indicators (i.e., contact, downward and upward support, and positive and negative relationship qualities), and examined the associations of specific typologies of parent-child ties as well as within-family variability in typologies with middle-aged adults’ psychological well-being. This study found seven types of parent-child ties as distinct combinations of contact, support exchanges, and relationship quality. Within-family variability in these types was associated with more depressive symptoms, and having types characterized by conflicted ties was associated with more depressive symptoms and lower life satisfaction. Middle-aged adults seem to be happiest when they are able to maintain homogeneous, harmonious patterns of relationships with their parents and grown children. Findings were discussed with regard to factors that also may predict greater variability in family relationship patterns.

Intergenerational ties are of considerable importance to adults of all ages; young adults rely on their parents for tangible and non-tangible support, older parents often consider their grown children their most important social ties, and middle-aged adults report strong feelings, frequent contact, and support exchanges with grown children and with older parents (Rossi & Rossi, 1990; Swartz, 2009). Research has consistently demonstrated the significance of parent-child ties for individuals’ health and well-being (Umberson, Pudrovska, & Reczek, 2010). In examining associations of parent-child ties with individuals’ well-being, however, the complexity of these relationships has not been well addressed. First, studies have examined associations between a single dimension of parent-child ties (e.g., contact, support, relationship quality) and well-being. An emerging perspective suggests that dynamics of parent-child ties – as naturally occurring can be better captured by combining multiple dimensions of relationships (Mandara, 2003). We need to know which combinations of dimensions of parent-child ties may contribute to better well-being. Second, most adults have multiple parent-child ties with upper and lower generations, but studies have typically examined the impact of the relationship with one member (e.g., focal offspring or parent) on individual well-being. This dyadic approach in effect ignores other relationships that individuals have within families (Suitor et al., 2018). The assumption that information about a single relationship within a family is representative across all relationships is likely to bring biased results (i.e., overestimated or underestimated) regarding the impact of parent-child ties on individual well-being.

To fill these gaps in the literature, we considered (a) multiple dimensions that characterize a parent-child tie and (b) multiple parent-child ties that individuals have within three-generational families. Specifically, we employed a typology approach to incorporate multiple dimensions of intergenerational ties. Taking into account all ties that middle-aged adults have with each of parents and offspring within families, we classified parent-child ties into distinct typologies combining five indicators of intergenerational relationships (i.e., contact, support given and received, and positive and negative relationship qualities) and examined which members are more likely to show specific types of ties with middle-aged adults. We then examined implications of typologies of parent-child ties for individual well-being. Some adults may maintain similar relationships across their parents and adult offspring, but others may show considerable variability in relationship patterns among family members. Thus, we examined associations of (a) specific typologies of relationships and (b) within-family variability in typologies with middle-aged adults’ psychological well-being (i.e., depressive symptoms and life satisfaction).

Multiple Dimensions of Parent-Child Ties

Intergenerational ties vary along a variety of dimensions; how often parents and grown children visit, how often they help each other, how much affection the parties feel as well as the level of conflict or distress. These dimensions of family functioning may not occur in isolation. Solidarity theory suggests that positive features of relationships co-occur (e.g., people who report positive relationships also report frequent support; Bengtson & Roberts, 1991). By contrast, the ambivalence model suggests that positive and negative relationship qualities also co-occur in parent-child ties (Lüscher & Pillemer, 1998). Scholars have used a typology approach as an effective tool to integrate multiple dimensions of relationships (e.g., contact, support, and relationship qualities) in different populations (e.g., Guo, Chi, & Silverstein, 2012; Silverstein & Bengton, 1997; van Gaalen & Dykstra, 2006). These typology studies differ depending on the specific dimensions of parent-child ties included for the typologies; some studies focused only on types of support exchanged between parents and adult offspring (e.g., Hogan, Eggebeen, & Glogg, 1993; Kim, Zarit, Fingerman, & Han, 2015), and other studies included relationship qualities in parent-child ties (e.g., Ferring, Michels, Boll, & Filipp, 2009; Silverstein, Gans, Lowenstein, Giarrusso, & Bengtson, 2010).

Drawing on solidarity and ambivalence theories as well as prior typology studies, we considered three dimensions of parent-adult child relationships (five indicators): (a) structural, (b) functional, and (c) affectional dimensions. For structural aspects, we included in-person contact, which may provide structural opportunities or constraints on the content and frequency of interactions between parents and offspring (Deane, Spitze, Ward, & Zhuo, 2016). For functional aspects, we considered both support given and received. Intergenerational support exchanges occur frequently across the life course and make important contributions to parents’ and children’s well-being in later years (Silverstein, Conroy, Wang, Giarrusso, & Bengtson, 2002). Given that middle-aged adults as a pivotal generation often support upper and lower generations simultaneously (Grundy & Henretta, 2006), support exchanges may be a more important indicator of parent-child ties than the frequency of contact. Finally, we included positive and negative relationship qualities as affectional aspects of these ties. As ambivalence theory has indicated, affection and conflict often coexist in family ties; recent typology studies have incorporated both affection and conflict to reveal emotional complexity of parent-child relationships (e.g., Silverstein et al., 2010; van Gaalen & Dykstra, 2006).

Multiple Parent-Child Ties Within Families

Most families include multiple parent-child relationships (Ward, 2008); that is, middle-aged adults have more than one child, and often have two living parents. A family systems theory has emphasized dynamics among multiple parent-child relationships in families and how these multiple relationships function as a whole (Cox & Paley, 1997). Recently, researchers have begun to pay more attention to the existence of substantial variation in parents’ relationships with adult children in the same family (Suitor et al., 2018). In a given family system, certain relationships may be viewed as more important, more demanding, or more problematic than other relationships (i.e., within-family variability). Thus, including multiple parent-child relationships enables us to examine within-family characteristics differentiating distinct relationships within the family.

Prior typology studies have also suggested variability within families, showing distinct relationship patterns with mother and father (e.g., Ferring et al., 2009; Silverstein & Bengtson, 1997). However, regarding relationships with adult children, most typology studies have considered a single relationship with one focal child (e.g., van Gaalen & Dykstra, 2006) or aggregate children (e.g., Guo, Stensland, Li, & Dong, 2019) per family. Examining one focal child is likely to be biased by selection effects (e.g., closest or well-off child), and using aggregate children does not allow to examine children’s individual characteristics. Thus, we do not know whether parents are more likely to have distinct types of relationships with specific children.

Further, in the context of multigenerational families, middle-aged adults simultaneously interact with older parents and grown children, and they may have different relationships depending on the generation of members (i.e., parents or offspring; Grundy & Henretta, 2006). Prior typology studies have examined relationships with parents or adult offspring, focusing on two-generation families. The generational stake hypothesis suggests that parents are more invested in their children and experience better quality parent-child ties than do their children (Bengtson & Kuypers, 1971). However, given the generational stake hypothesis mainly focuses on relationship quality (emotional), less is known about how relationship patterns including other relationship dimensions (structural and functional) reveal generational differences between older parents and adult children.

To create typologies of parent-child ties, this study used all parent-child ties that middle-aged adults have within three-generation families (i.e., ties with each parent and each offspring). We also addressed within-family differences in typologies of parent-child relationships by examining which members (e.g., generation, gender, biological relation, sociodemographic status, health status, and geographic distance) are more likely to have specific types of relationships with middle-aged adults.

Parent-Child Ties and Psychological Well-Being

Prior studies have consistently shown that ties to parents and grown children affect individual well-being (Umberson et al., 2010). Contact has been regarded as an important factor for maintaining parent-child relationships (Deane et al., 2016); frequent in-person contact is associated with better daily mood (Fingerman, Kim, Birditt, & Zarit, 2016). Distinct literatures have also confirmed that positive relationships with parents or adult children are beneficial and negative or ambivalent relationships are detrimental for individuals’ well-being (Fingerman, Pitzer, Lefkowitz, Birditt, & Mroczek, 2008; Lee & Szinovacz, 2016; Umberson, 1992). However, frequent support exchanges have shown mixed results for associations with well-being (Bangerter, Kim, Zarit, Birditt, & Fingerman, 2015; Byers, Levy, Allore, Bruce, & Kasl, 2008). Regarding these mixed findings, researchers have suggested that implications of support exchanges for well-being need to be understood in the emotional context of families (Lowenstein, Katz, & Gur-Yaish, 2007). For example, excessive support from children can have negative consequences for older parents’ well-being, because it hurts parents’ autonomy and increases feelings of dependency (Silverstein, Chen, & Heller, 1996), while the negative consequences can be avoided if the relationships with children are close and affectionate (Merz, Consedine, Schulze, & Schuengel, 2009). Thus, in the current study, we addressed which combinations of relationship dimensions (e.g., contact, support, and relationship quality) contribute to better well-being by examining associations between relationship typology and well-being.

Another unique feature of this study involves consideration of multiple parent-child ties for the middle-aged adult’s well-being. First, we considered variability across typologies of parent-child ties to characterize families as a whole. As family systems theory posits (Cox & Paley, 1997), families differ in similarity or differentiation of intergenerational ties. In some families, middle-aged adults may have distinct relationship patterns with each parent and grown child, whereas others may show more similar relationships (i.e., between-family differences in within-family variability). Variability across typologies of parent-child ties within families may reflect long-standing problems in some dyads within a family. Conversely, middle-aged adults with highly differentiated relationships may be responding more effectively to the individualized needs of parents and grown children as contingency theory suggests (Eggebeen & Davey, 1998). Given that in most western countries, there are cultural norms calling for parents to treat their children equally as well as beliefs about obligation toward family members (Kalmijn, 2013 a; Rossi & Rossi, 1990), however, having differentiated relationships across family members may have harmful impact on individual well-being. Studies examining differential treatment among siblings found that when parents favored some siblings over other siblings, the less favored sibling reported poorer mental health, even in adulthood (Davey, Tucker, Fingerman, & Savla, 2009; Pillemer, Suitor, Pardo, & Henderson, 2010). Yet, studies have not investigated the implications of having different relationship patterns across family members from the perspective of the middle-aged adult. Second, we also considered the existence of specific types of ties within families for well-being. Studies found that having problematic relations with at least one child could lower parental well-being (Fingerman, Cheng, Birditt, & Zarit, 2012; Ward, 2008). Given that negative emotions are more intense and detrimental for well-being than positive ones (Gilligan, Suitor, Feld, & Pillemer, 2015a; Rook, 2001), having parent-child ties characterized by high conflicts would be associated with poorer well-being for the middle-aged adult.

Method

Data and Sample

Data were from the Family Exchanges Study, Wave 1 (Fingerman, 2008). The original sample included 633 middle-aged adults (aged 40 to 60) who had at least one child over age 18 and one living parent in the Philadelphia Metropolitan Area. The study identified potential participants via listed samples from Genesys Corporation supplemented with random digit dialing within geographic area codes. Individuals residing in high density minority areas and lower socioeconomic households were oversampled, including 31% of African Americans. The sample was comparable to the Philadelphia Primary Metropolitan Statistical Area (PMSA) in terms of income, but participants were slightly better educated (Pennsylvania State Data Center, 2001; U.S. Census Bureau, 2008). Participants completed Computer Assisted Telephone Interviews lasting approximately one hour in 2008. Middle-aged participants (G2) responded to questions about their own background (e.g., age and education) and each of their living parents (G1; n = 868) and grown offspring (G3; n = 1,384). This study analyzed 2,252 parent-child ties (G1-G2 n = 868; G2-G3 n = 1,384) that 633 participants reported to identify typologies of parent-child ties in adult families (See Table 1 for sample characteristics).

Table 1.

Characteristics of Middle-Aged Participants and Their Parents and Grown Offspring

	Middle-aged participants (G2; N = 633)		Family members (N = 2,252)
			Aging parents (G1; n = 868)		Grown offspring (G3; n = 1,384)
Variable	M	(SD)	M	(SD)	M	(SD)
Sociodemographic characteristics
Age	50.70	(4.99)	77.05	(6.81)	25.19	(5.80)
Malea, %	48		37		52
Years of education	14.18	(2.02)	12.17	(2.81)	13.73	(1.92)
Household incomeb	4.40	(1.45)	—	—	—	—
Physical healthc	3.48	(1.07)	2.78	(1.09)	4.26	(0.93)
Married or remarriedd, %	70		49		18
Workinge, %	76		11		69
Racial/ethnic minorityf, %	37		—		—
Number of parents	1.37	(0.48)	—	—	—	—
Number of siblings	3.24	(2.45)	—	—	—	—
Number of adult children	2.19	(1.24)	—	—	—	—
Distance from G2 (miles)	—	—	235.77	(660.70)	188.66	(613.36)
Coresiding with G2g, %	—		9		23
Biological tie with G2h, %	—		97		87
Relationship indicators
In-person contacti	—	—	5.07	(2.01)	5.57	(2.15)
Downward supportj	—	—	3.55	(1.40)	4.68	(1.68)
Upward supportj	—	—	4.16	(1.59)	3.63	(1.46)
Positive relationship qualityk	—	—	3.98	(0.90)	4.02	(0.90)
Negative relationship qualityk	—	—	1.97	(0.95)	2.05	(0.87)

Notes. 2,252 parent-child ties nested within 633 participants.

^a1 = male and 0 = female.

^bRated from 1 (less than $10,000) to 6 ($100,000 or more).

^cRated from 1 (poor) to 5 (excellent).

^d1 = married or remarried and 0 = not married.

^e1 = working full-or part-time and 0 = not working.

^f1 = racial/ethnic minority and 0 = non-Hispanic White.

^g1 = coresiding with the middle-aged participant and 0 = not coresiding with the middle-aged participant.

^h1 = biological parent or child and 0 = not biological (step, adopted, or other).

ⁱRated from 1 (less than once a year or never) to 8 (daily).

^jMean of 6 types of support rated from 1 (less than once a year or never) to 8 (daily).

^kMean of 2 items rated from 1 (not at all) to 5 (a great deal).

Measures

Indicators of parent-child ties.

We used five indicators to characterize parent-child ties in adult families: (a) contact, (b) upward support (i.e., provided to parents), (c) downward support (i.e., provided to children), (d) positive relationship quality, and (e) negative relationship quality (See Table 1). For contact, middle-aged participants rated frequency of in-person contact with each parent and offspring in the past 12 months (1 = less than once a year or not at all to 8 = daily; Rossi & Rossi, 1990). To assess the frequency of support provided to and received from each parent and grown child, participants completed the Intergenerational Support Index (Fingerman, Miller, Birditt, & Zarit, 2009), including six types of support (i.e., financial, practical, emotional support, advice, companionship, and listening to talk; 1 = less than once a year or not at all to 8 = daily). Mean scores across the six items were calculated (α = .88 for downward support; α = .87 for upward support). Because we included ties across three generations, we distinguished support exchanges – downward vs. upward support, depending on the direction of support flows between generations (e.g., Dykstra & Fokkema, 2011).Participants also rated positive and negative relationship qualities towards each of parents and grown children (Umberson, 1992). Two items for positive relationship quality included “how much does your (father/mother/child) love and care for you?” and “how much does your (father/mother/child) understand you?” Two items for negative relationship quality included “how much does your (father/mother/child) criticize you?” and “how much does your (father/mother/child) make demands on you?” Responses were rated on a 5-point scale (1 = not at all to 5 = a great deal). Mean scores across two items were calculated (ρ = .74 for positive relationship quality and ρ = .63 for negative relationship quality).

Psychological well-being.

We assessed two dimensions of psychological well-being: (a) depressive symptoms and (b) life satisfaction. For depressive symptoms, we used five items from the depression 6-item subscale of the Brief Symptom Inventory-18 (Derogatis, 2001). Participants rated how often they experienced these feelings in the past 7 days: lonely, blue, worthless, hopeless about the future, and no interest in things (1 = not at all to 5 = extremely), and a mean score was calculated (α = .83). Life satisfaction was measured by one item (1 = not at all satisfied to 10 = completely satisfied).

Covariates.

Participants reported their own demographic and health characteristics: age, gender (1 = male, 0 = female), years of education, household income (1 = less than $10,000, 2 = $10,001-$25,000, 3 = $25,001-$40,000, 4 = $40,001-$75,000, 5 = $75,001-$100,000, and 6 = $100,001 or more), marital status (1 = married or remarried, 0 = not married), work status (1 = working, 0 = not working), physical health (1 = poor to 5 = excellent), and race/ethnicity (1 = racial/ethnic minority, 0 = non-Hispanic White). Participants reported family size, including numbers of living parents, siblings, and adult children. Participants also provided information about each parent and grown child (generation; coded 1 = parent, 0 = offspring): age, gender (1 = male, 0 = female), years of education, marital status (1 = married or remarried, 0 = not married), work status (1 = working, 0 = not working), physical health (1 = poor to 5 = excellent), geographic distance from each member (miles), living arrangement (1 = coresiding, 0 = not coresiding), and biological relation (1 = biological and 0 = not biological).

Analytic Strategy

We included relationships with all living parents and adult offspring – reported by the participants. We treated each family member (i.e., parents and offspring) as a “separate” but “nested” relationship (Level 1; N = 2,252) within middle-aged participants (Level 2; N = 633) by employing multilevel models to handle the family-nested data. Thus, each participant can have separate multiple data lines for each of parents and offspring in the data structure. To cluster each parent-child tie into distinct typologies, we applied a latent profile analysis (LPA; based on 2-level mixture procedures) to five indicators of parent-child ties. We used standardized scores of the five indicators (T-scores; M = 50, SD = 10) for ease of interpretation. We selected the best model for latent classes of parent-child ties based on model fit statistics, including the likelihood ratio chi-square test statistic, the Akaike information criterion (AIC), the Bayesian information criterion (BIC), and the Lo-Mendell-Rubin adjusted likelihood ratio test (LMR LRT).

To characterize the derived typologies of parent-child ties, we estimated a 2-level multinomial logistic regression model. For the dependent variable (i.e., types of each parent-child tie nested within participants), we used the “most-likely” class variable which was constructed via the latent class posterior distribution. For predictors (i.e., family members’ characteristics) and covariates (i.e., participants’ characteristics), we first checked correlations among variables listed in Table 1. Among family members’ characteristics, we eliminated age and coresidence, due to high correlations with other variables (e.g., generation and distance, respectively). Among participants’ characteristics, we excluded household income, which was highly correlated with education, marital status, and minority status. Therefore, the final model for typology membership included family member’s characteristics (level-1 variables; i.e., generation, gender, years of education, physical health, marital status, work status, geographic distance, and biological relation), controlling for middle-aged participants’ characteristics (level-2 variables; i.e., gender, age, years of education, physical health, marital status, work status, race/ethnicity, number of adult offspring, and number of siblings).

To examine the association between typology of parent-child ties and psychological well-being, we estimated separate regression models for depressive symptoms and life satisfaction. Since the psychological well-being measures were participant-level variables, we created two sets of predictors by summarizing parent-child typologies at the participant level: (a) variability over typologies (i.e., entropy scores) and (b) existence of each typology of parent-child ties. First, entropy scores were calculated to represent dispersion across typologies of parent-child ties within participants (Ram, Conroy, Pincus, Hyde, & Molloy, 2012):$-\frac{1}{\text{ln}m}{\sum }_{j=1}^{j=m}{p}_{\mathit{ij}}\text{ln}{p}_{\mathit{ij}}$, where the entropy score for person i is a function of the proportions (p_ij) of the parent-child ties that were in typology j (typology 1 to m). The entropy scores were scaled to range from 0 (i.e., all parent-child ties in a single typology; no variability within families) to 1 (i.e., equal number of parent-child ties in each of the typologies; complete variability within families; See Supplementary Figure 1 for examples of different levels of entropy scores). Second, for the specific effect of each typology of parent-child tie, we created a set of dummy variables representing whether participants had each typology of parent-child ties within the family (1 = having that typology, 0 = not having that typology). We also controlled for participants’ characteristics that were used in the previous model, except for two family size variables (i.e., number of siblings and number of adult children), which did not show any significant correlations with well-being outcomes.

Results

Identification of Typologies for Parent-Child Ties

We classified 2,252 parent-child ties (nested within 633 middle-aged participants) into distinct typologies based on five indicators of relationships (i.e., contact, downward and upward support, positive and negative relationship quality). The LPA identified seven latent classes of parent-child ties as optimal (See Supplementary Table 1 for the model fit statistics). Figure 1 describes the mean level profiles of the indicators (See Supplementary Table 2 for the exact mean values of T-scores and raw scores) for these typologies of parent-child ties, including membership probabilities.

Figure 1.

Seven latent typologies of parent-child ties. The original scales of indicators were transformed into standardized scores, T-scores (M = 50, SD = 10) for ease of interpretation.

The first three types (Types 1, 2, and 3; 30%) were characterized by lower levels of contact and support exchanges. These types were distinct from each other in terms of relationship qualities. The first type was labeled “detached” (6%); positive relationship quality was reported as very low and negative relationship quality was moderate. The second type, “not engaged-harmonious” (17%) showed harmonious relationships (i.e., high positive relationship quality and low negative relationship quality) despite lower contact and support exchanges. The third type, “not engaged-conflicted” (7%) showed low positive relationship quality and high negative relationship quality. The fourth type, “low engaged helping-ambivalent” (8%) also showed lower contact but medium levels of support exchanges. Since both positive and negative relationship qualities were high for parents and offspring in this type, we labeled “ambivalent” ties. The last three types involved more frequent contact. The fifth type, “actively engaged-conflicted” (6%) was characterized by frequent in-person contact but medium levels of support exchanges. In this type, the positive relationship quality was low and negative relationship quality was very high. The sixth type, “actively helping-ambivalent” (29%) showed high levels of all indicators. Thus, parent-child ties in this type showed frequent contact and support exchanges, and both positive and negative relationship qualities were high as well. The last type, “medium engaged-harmonious” (28%) was characterized by medium levels of contact and support exchanges. Parent-child ties in this type were harmonious (i.e., high positive relationship quality and low negative relationship quality).

Prediction of Typologies for Parent-Child Ties

Next, we examined how the derived typologies of parent-child ties were distinguished in terms of family members’ characteristics (Table 2). The 7th type (medium engaged-harmonious) was served as a reference type in the 2-level multinomial logistic regression. With regard to the family members’ generation and gender, middle-aged adults were more likely to have “not engaged-harmonious” ties with older parents and “actively helping-ambivalent” ties with adult children. Also, middle-aged participants were more likely to have “detached” and “not engaged-harmonious” ties with male members (i.e., son or father) and “low engaged helping-ambivalent” and “actively helping-ambivalent” ties with female members (i.e., daughter or mother). Geographic distance from members was the most consistent factor differentiating derived typologies; dyads who lived at a greater distance from one another were more likely to belong to less engaged typologies (Types 1 to 4). Not surprisingly, dyads living in closer proximity were more likely to show more engaged relationships (Types 5 and 6). Regarding sociodemographic status, middle-aged participants showed diminished likelihood of having more engaged or helping relationships (Types 4 to 6) with married parents and offspring. Also, participants were less likely to have “detached” and “conflicted” relationships (Types 1, 3, and 5) with parents and offspring with better health. Middle-aged adults were more likely to have “not engaged-harmonious” with family members who are working. Finally, middle-aged participants were less likely to have “detached” and “not engaged-conflicted” ties with their biological parents and offspring.

Table 2.

Relative Risk Ratios of Multilevel Multinomial Regression Model for Typology of Parent-Child Ties

	Type 1	Type 2	Type 3	Type 4	Type 5	Type 6	Type 7
Variable	Detached	Not engaged- Harmonious	Not engaged- Conflicted	Low engaged helping- Ambivalent	Actively engaged- Conflicted	Actively helping- Ambivalent	Medium engaged- Harmonious
G1/G3’s characteristics
Upper generationa	1.26	3.47***	0.69	1.40	0.53	0.50**	(1.00)
Maleb	2.49**	1.46*	0.98	0.64*	1.06	0.71**	(1.00)
Years of education	0.97	1.03	0.91	1.01	0.92	1.02	(1.00)
Physical healthc	0.64**	0.90	0.74**	1.30	0.78*	1.02	(1.00)
Married or remarriedd	0.63	0.93	0.72	0.31***	0.32**	0.34***	(1.00)
Workinge	1.32	2.29***	1.14	0.90	1.16	0.77	(1.00)
Distance from G2f	1.71***	2.14***	1.18*	1.88***	0.52***	0.56***	(1.00)
Biological tie with G2g	0.22***	0.76	0.30***	1.38	0.95	1.38	(1.00)
G2’s characteristics
Maleb	0.99	1.85**	1.30	0.55*	0.86	0.67**	(1.00)
Age	0.99	1.01	0.95*	0.99	0.99	0 94***	(1.00)
Years of education	0.88	0.94	1.11	1.07	1.02	1.02	(1.00)
Physical healthc	1.05	0.90	0.92	1.02	0.90	0.99	(1.00)
Married or remarriedd	1.43	1.07	0.69	0.98	0.91	1.19	(1.00)
Workinge	0.44**	0.80	0.89	0.45*	0.69	0.82	(1.00)
Racial/ethnic minorityh	1.99*	0.90	0.79	1.16	0.58	1.05	(1.00)
Number of adult children	1.12	1.12	1.31**	0.92	0.87	1.04	(1.00)
Number of siblings	1.02	1.08*	0.96	1.06	1.02	0.95	(1.00)

Notes. Parent-child tie N = 2,252 (nested within 633 participants). G1 = upper generation (aging parents); G2 = middle generation (middle-aged participants); G3 = lower generation (grown offspring). Reference category = Type 7 (Medium engaged-Harmonious).

^a1 = upper generation (G1) and 0 = lower generation (G3).

^b1 = male and 0 = female.

^cRated from 1 (poor) to 5 (excellent).

^d1 = married or remarried and 0 = not married.

^e1 = working full- or part-time and 0 = not working.

^fLogged miles.

^g1 = biological parent or child and 0 = not biological (step, adopted, or other).

^h1 = racial/ethnic minority and 0 = non-Hispanic White.

^*p < .05.

^**p < .01.

^***p < .001.

Typologies of Parent-Child Ties and Psychological Well-Being

To examine how typologies of parent-child ties within families are associated with middle-aged adults’ psychological well-being, we conducted multiple regressions for depressive symptoms and life satisfaction (Table 3). First, we examined the associations between within-family variability over typologies of parent-child ties and psychological well-being. Entropy scores as a measure of within-family variability of parent-child ties, ranged from 0.00 to 0.76 (M = 0.39, SD = 0.16) and were normally distributed (Skewness = −0.38; Kurtosis = −0.17). Based on the level of entropy score, we can see whether participants may have mixed patterns (i.e., higher entropy scores) or homogeneous patterns (i.e., lower entropy scores) across different parent-child ties. For example, some participants with higher entropy scores may have a “detached” tie with father, an “actively helping-ambivalent” tie with one adult offspring, and “not engaged-harmonious” ties with the other two offspring. In contrast, other participants with lower entropy scores may have only a single typology of parent-child ties; all ties show a “medium engaged-harmonious” type (See Supplementary Figure 1). Greater variability in ties was significantly associated with more depressive symptoms (p = .027), but marginally associated with lower life satisfaction (p = .081; Model 1 in Table 3).

Table 3.

Regression Models for Associations Between Parent-Child Ties Within Families and Psychological Well-Being

	Depressive symptoms						Life satisfaction
	Model 1			Model 2			Model 1			Model 2
Variable	B	(SE)	β	B	(SE)	β	B	(SE)	β	B	(SE)	β
Parent-child typology variabilitya	0.32*	(0.14)	0.08	—	—	—	−0.66†	(0.38)	−0.06	—	—	—
Having each type of parent-child tiesb
Type 1 (Detached)	—	—	—	0.09	(0.06)	0.05	—	—	—	−0.30 †	(0.17)	−0.07
Type 2 (Not engaged–Harmonious)	—	—	—	0.03	(0.05)	0.02	—	—	—	0.14	(0.13)	0.04
Type 3 (Not engaged–Conflicted)	—	—	—	0.12†	(0.06)	0.07	—	—	—	−0.30 †	(0.17)	−0.07
Type 4 (Low engaged helping–Ambivalent)	—	—	—	0.10	(0.07)	0.06	—	—	—	0.01	(0.17)	0.00
Type 5 (Actively engaged–Conflicted)	—	—	—	0.14*	(0.07)	0.08	—	—	—	−0.58**	(0.18)	−0.12
Type 6 (Actively helping–Ambivalent)	—	—	—	0.04	(0.05)	0.03	—	—	—	0.01	(0.14)	0.00
Type 7 (Medium engaged–Harmonious)	—	—	—	−0.02	(0.05)	−0.02	—	—	—	−0.03	(0.14)	−0.01
Controls
Male c	−0.05	(0.05)	−0.04	−0.04	(0.05)	−0.03	−0.20	(0.13)	−0.06	−0.20	(0.13)	−0.06
Age	−0.00	(0.01)	−0.03	−0.00	(0.01)	−0.02	0.02	(0.01)	0.05	0.02	(0.01)	0.05
Years of education	−0.02	(0.01)	−0.05	−0.02	(0.01)	−0.05	0.02	(0.03)	0.02	0.00	(0.03)	0.01
Physical healthd	−0.14***	(0.03)	−0.23	−0.13***	(0.03)	−0.22	0.50***	(0.07)	0.31	0.48***	(0.07)	0.30
Married or remarried e	−0.27**	(0.06)	−0.19	−0.27***	(0.06)	−0.19	0.74***	(0.15)	0.20	0.72***	(0.15)	0.19
Working f	−0.23**	(0.06)	−0.15	−0.22***	(0.06)	−0.14	0.18	(0.16)	0.05	0.15	(0.16)	0.04
Racial/ethnic minority g	−0.16**	(0.06)	−0.12	−0.16**	(0.06)	−0.12	0.37*	(0.15)	0.10	0.36*	(0.15)	0.10
Adjusted R2	.16			.17			.17			.19

Notes. Participant N = 633.

^aEntropy scores over seven types of parent-child ties (ranged 0 = no variability to 1 = complete variability).

^bl = having that type of parent-child ties within family and 0 = not having that type ofparent-child ties within family.

^c1 = male and 0 = female.

^dRated from 1 (poor) to 5 (excellent).

^el = married or remarried and 0 = not married.

^fl = working full- or part-time and 0 = not working.

^gl = racial/ethnic minority and 0 = non-Hispanic White.

^†p < .10.

^*p < .05.

^**p < .01.

^***p < .001.

Next, we examined the associations between each specific typology of parent-child ties and psychological well-being (See Supplementary Table 2 for proportions of participants who had each type of parent-child ties within the family). Having “actively engaged-conflicted” ties was significantly associated with more depressive symptoms (p = .035) and lower life satisfaction (p = .001; Model 2 in Table 3). Another type characterized by conflicted ties, “not engaged-conflicted” showed marginally significant associations with more depressive symptoms (p = .051) and lower life satisfaction (p = .072); having “distant” ties was also marginally associated with lower life satisfaction (p = .068).

Discussion

Intergenerational ties are critical for emotional and physical well-being as well as other positive outcomes in adulthood (e.g., work and parenthood) across the life span. Although prior studies have described typologies of families in adulthood (e.g., Silverstein et al., 2010; van Gaalen & Dykstra, 2006), this study addressed several gaps in the literature. We constructed typologies from multiple dimensions of parent-child relationships, including frequency of contact, receipt and provision of support, and positive and negative relationship qualities. Furthermore, we examined within-family variability of relationships with multiple family members across three generations. By combining multiple relationship dimensions from multiple family members, we found a greater number of patterns (i.e., 7-class model) than prior studies which have found only four or five patterns (e.g., Silverstein & Bengtson, 1997; van Gaalen & Dykstra, 2006). Most importantly, we found links between these relationship typologies and the middle-aged adults’ well-being, both in the types of relationships and in the variability of those relationship patterns across family members.

Typologies of Parent-Child Ties

This study contributes to the burgeoning literature regarding family typologies by integrating multiple dimensions to yield a more nuanced understanding of the variability in parent-child ties. In prior studies with typologies, there is typically one category that is deemed “detached” or “weak” (Dykstra & Fokkema, 2011; Silverstein & Bengtson, 1997). Those findings were consistent with solidarity theory; the relationships seemed to lack cohesion. Yet, the studies were limited by a focus on a single dimension of the relationship – relationships qualities (Silverstein et al., 2010) or support exchanges (Dykstra & Fokkema, 2011).

By examining multiple dimensions of relationships, we found three distinct types of relationships characterized by low involvement: a typology in which all indicators were low (detached), a typology in which contact and exchanges were low, but affection was high (not engaged-harmonious) and a typology in which contact and exchanges were low, but conflict was relatively high (not engaged-conflicted).

Only one typology fit the prototype for high solidarity, the “engaged-harmonious” type. This typology accounted for over a quarter of the relationships. But, other typologies shed new light on solidarity theory, suggesting that positive feelings can occur without frequent contact or support. Other studies have reported that parents can derive benefits from their relationships with grown children even when they do not see them often (Buber & Engelhardt, 2008), but these ideas have not permeated typologies of relationships. Moreover, the typology characterized by low frequency of contact and high conflict may reflect relationships that are estranged (Gilligan, Suitor, & Pillemer, 2015b). However, it should be noted that families may also use different means (e.g., phone, text, email, and video chat) than in-person contact to interact each other. Future studies should consider different modes of contact to capture the variety of interactions among family members.

In addition to the disengaged and conflicted typology, three other typologies were either conflicted or ambivalent. These typologies involved greater engagement and helping (van Gaalen & Dykstra, 2006). The findings suggest that great attention is warranted in understanding the occurrence of ambivalence and conflict in intergenerational ties. Prior studies have suggested a more limited range of ambivalent situations reflecting conflicts in norms and values or in expectations and values (Pillemer & Suitor, 2002). In this study, over half of relationships were classified as ambivalent or conflicted, suggesting that negative feelings occur under a variety of circumstances (Birditt, Miller, Fingerman, & Lefkowitz, 2009).

Within-Family Characteristics Differentiating Typologies of Parent-Child Ties

By including ties with multiple family members, we were able to focus on within-family factors that predict membership in distinct typologies of parent-child ties (i.e., which members are more likely to have specific types of relationships with middle-aged adults), controlling for between-family differences attributed to middle-aged participants’ characteristics. Furthermore, because we included family members from three generations, we were able to identify generational disparities in typologies. Ties to older parents were over three times more likely to fall in the “not engaged-harmonious” type than were ties to adult children. This finding was interesting because middle-aged adults typically report more positive relationship qualities with their grown children than with their parents (Birditt, Hartnett, Fingerman, Zarit, & Antonucci, 2015) and also report exchanging more support with grown children than with parents (Fingerman et al., 2011). When considering multiple relationship dimensions simultaneously, however, ties to parents may be simpler and less emotionally intense than ties to grown children, at least when the parents are still healthy.

A number of characteristics of family members also differentiated the typologies. Consistent with other typology studies, geographic distance was associated with each of the less engaged typologies (Dykstra & Fokkema, 2011; Silverstein & Bengtson, 1997). These findings suggest that family members must be in proximity to interact each other and provide practical help. Nonetheless, we found that approximately 8% of ties fell into a “low engaged helping” typology, and characteristics of this typology warrant attention in future research (e.g., helping from a distance; Horowitz & Boerner, 2017). In addition, we found differences in the relationship patterns by gender of family members, consistent with women’s role in families (e.g., providing support to a family member in need; Rossi & Rossi, 1990). While ties with fathers or sons tend to involve less interactions, ties with mothers or daughters were likely to involve more interactions and supporting behaviors as well as be emotionally mixed (van Gaalen & Dykstra, 2006). Gender of the middle-aged adults also appeared to be consistently significant; thus, males were more likely to show the less engaged ties across family members.

Regarding sociodemographic status of family members, midlife adults tended to have less engaged (or helping) ties with parents/offspring who are married and working. Marriage has been often referred as a “greedy institution” when it comes to time available for kin members (Sarkisian & Gerstel, 2008); work status may also represent their limited time for other family members (Sarkisian & Gerstel, 2004). Also, family members’ physical health differentiated the patterns of parent-child ties; middle-aged adults were less likely to have distant or conflicted ties with parents or offspring with better health. Given that health problems may be one of main reasons for eliciting family support, it would be worthwhile to examine how health declines among family members engender changes in relationship patterns in future research (e.g., Schenk & Dykstra, 2012). Finally, out findings revealed significant differences in relationship patterns by biological relation; middle-aged adults tended to have problematic ties (“detached” and “not engaged-conflicted”) with non-biological parents and offspring (Kalmijn, 2013b).

Parent-Child Typologies and Well-Being

Although researchers have examined typologies of parent-child ties elsewhere, studies have not examined how these typologies as combinations of different relationship dimensions are associated with individual well-being. Participants who had actively engaged but conflicted relationships reported poorer well-being. This finding is consistent with other research linking negative quality parent-child relationships with poorer well-being (Umberson, 1992). It is also indicated that the negative component in the parent-child relationships is more salient than positive aspects in explaining well-being (Gilligan et al., 2015a; Rook, 2001). Research also has suggested that ambivalent relationships with parents and grown children are associated with poorer well-being (Fingerman et al., 2008). However, that was not the case in this study, and future research might examine how factors such as support exchanges and contact may mitigate effects of ambivalent ties. Rather, we found that having detached relationships with parents or offspring was marginally associated with lower life satisfaction, suggesting that family estrangement could violate the norms of family ties, which lowers individuals’ well-being (Gilligan et al., 2015b). Given that our data is based on middle-aged participants’ cross-sectional reports on all relationships, however, it is also possible that participants’ poorer psychological well-being contributes to conflicted or detached relationships with parents or offspring.

This study also provided unique insights regarding relationships with multiple family members. When participants experienced greater variability in the qualities of their ties, they also reported more depressive symptoms. From the perspective of offspring, prior studies have suggested the degree of parental differentiation was associated with grown children’s emotional well-being, with siblings who are favored reporting better well-being than siblings who are not favored (Davey et al., 2009; Pillemer et al., 2010). However, our study examined implications of within-family differentiation from a middle-aged adults’ perspective. The detrimental effect of within-family variability may reflect different reasons for variability. It is possible that families with greater variability have some family members who suffer long-standing issues, which may not be captured by recent health or other life problems. Further, while responding to different needs and situations of family members, middle-aged adults may view their differentiated relationships as violating the norms of equality (Kalmijn, 2013a), which can lower their psychological well-being. Given the cross-sectional nature of our data, however, variability in parent-child ties may reflect middle-aged adults’ limited resources, including material and psychological resources. Thus, middle-aged adults who have fewer material and psychosocial resources may differentiate family members more rather than maintaining equitable relationships across members (Crouter, McHale, & Tucker, 1999).

Some limitations of this study should be noted. First, because we relied on middle-aged adults’ reports, the parent-child ties might be biased by the reporter’s perception and specific characteristics (e.g., personality traits; Fingerman et al., 2008). Reports from other members in the families (e.g., aging parents and grown offspring) would paint a more comprehensive picture of within-family dynamics. Second, regarding family members’ characteristics, we included mainly sociodemographic variables, so we could not explain detailed situations which may contribute to greater variability across family members. Third, this study attempted to capture multiple relationships within families, but there are still missing important intergenerational relationships, including in-laws and grandparents/grandchildren (Kim et al., 2015; Suitor et al., 2018). Fourth, because our analysis is based on a cross-sectional data, we could not confirm the causal association between the typologies and psychological well-being. A longitudinal study might have allowed to better determine how the typologies of parent-child ties affect their well-beings rather than being reflections of their characteristics. Finally, given that the data came from the specific regional areas, the findings may not be generalized to other groups that were underrepresented in our sample (e.g., Asian or Hispanic populations).

Despite such limitations, this study is one of few to consider multiple dimensions of parent-child ties and multiple ties in families. Our finding provided valuable information about theoretically and empirically relevant parent-child types as different combinations of relationship dimensions across generations. Researchers should look more closely at individual and family contexts which may lead to each of different relationship typologies within the same family. Our findings also have practical implications for middle-aged adults’ mental health; understanding of whole family constellations, not just specific dyads might be critical to promote better health and well-being. The effort to understand multigenerational family contexts systematically seems to be particularly important when designing interventions for middle-aged adults. It would be important for future research to develop theories to explain dynamics between multiple members within families as a whole.