Who Initiates the Help Older Parents Give to Midlife Children

Abstract

Objectives

Older parents continue to help children after these children have been adults for decades. We utilize a typology approach to assess who initiates the help. We ask whether profiles of help initiation are associated with how often older parents help and how they evaluate their helping behaviors.

Methods

Older parents (N = 241; M_age = 80.12) indicated the extent to which they volunteered to help children and helped per child’s request. Parents reported their resources and obligation to help, child problems, frequency, and evaluation (rewards/stresses) of helping.

Results

Latent profile analysis reveals four profiles representing parents who are initiators (n = 65), responders (n = 56), initiators/responders (n = 50), and uninvolved (n = 69). Resources, needs, and individual beliefs differentiate profiles. Parents offer the same amount of help regardless of who initiates such help. Parents who are initiators/responders view helping as more rewarding than parents who are initiators and more stressful than uninvolved parents.

Discussion

This study reveals variation in the initiation of older parents’ help and refines our understanding of family help in late life. Findings may suggest a parental expectation for children to be competent in adulthood regardless of their resources and willingness to help.

In Western cultures, intergenerational support typically flows downstream from parents to children (Albertini, Kohli, & Vogel, 2007; Silverstein & Giarrusso, 2010). Even after children enter midlife, parents still assist them in many ways, such as by offering comfort, advice, or financial support (Henretta, Van Voorhis, & Soldo, 2018; Huo, Graham, Kim, Birditt, & Fingerman, 2019; Shapiro & Remle, 2011). Parents may hold a lifelong belief that they should help children (Levitzki, 2009). Yet, most parents expect children to be independent by midlife (Birditt, Fingerman, & Zarit, 2010; Furstenberg, 2010; Suitor, Sechrist, & Pillemer, 2007) and their assistance to grown children drops off with age as offspring needs decline (Hartnett, Furstenberg, Birditt, & Fingerman, 2013). Thus, it is crucial to ask whether older parents instigate help or they help in response to children’s specific requests. A key contribution of this study is to examine who initiates older parents’ help and how such initiation influences parents’ perception of helping. This information may refine our understanding of parent–child ties in late life.

We explicitly assess the extent to which older parents (a) volunteer to help midlife children and (b) believe the children request that help. Consistent with the literature on helping behaviors (Iida, Seidman, Shrout, Fujita, & Bolger, 2008; Nadler, 2015), older parents may vary in how they perceive the initiation of help to midlife children. Some parents tend to volunteer to help (initiators), whereas other parents primarily assist per child’s request (responders). There are parents who engage in both approaches (initiators/responders). We also recognize that some parents have detached ties to their children and rarely offer help (uninvolved; Fiori, Smith, & Antonucci, 2007; Litwin & Shiovitz-Ezra, 2011). We utilize a typology approach to test these profiles and examine individual differences in the initiation of older parents’ help. We identify factors associated with these profiles, such as the parent’s resources to assist and the child’s needs for help. We also examine how these profiles are associated with (a) how often older parents help and (b) how they evaluate rewards and stresses of helping.

Factors Associated with Who Initiates Older Parents’ Help

This study draws on the Multidimensional Intergenerational Support Model (Fingerman, Sechrist, & Birditt, 2013), which conceptualizes intergenerational support in adult families. This model considers resources, needs, and individual beliefs about helping as key factors that elicit family help. Here, we use this model to examine profiles of who initiates older parents’ help. Resources and needs influence whether help is offered or required, but the actual help also varies based on whether individuals believe they should help.

Parent Resources

Parent socioeconomic status (SES) may be associated with who initiates their help. Research finds that upper SES parents offer more help to each child but lower SES parents give more support overall (Fingerman et al., 2015; Henretta et al., 2018; Kornrich & Furstenberg, 2013). SES also shapes the way parents help: Upper SES parents may provide more emotional or financial support, whereas lower SES parents tend to have coresident offspring and help in practical ways (e.g., offer grandchild care; Compton & Pollack, 2015; Fingerman et al., 2015; Smits, van Gaalen, & Mulder, 2010; Swartz, 2009). This study does not focus on the link between SES and how often parents help; rather, we build on prior research and examine whether the initiation of such help also varies by SES.

Several theories guide our hypothesis. Social capital theory posits that upper SES parents have more numerous resources and are better able to share their resources (Furstenberg & Kaplan, 2004). They may invest more resources in children by both initiating help and responding to requests (Conger, Conger, & Martin, 2010). Upper SES parents may view support as a way of smoothing economic inequalities across generations, especially if they have more resources than children. Lower SES parents, however, may lack resources (e.g., fewer assets, earlier disability) and primarily assist when children request (Matthews, Jagger, & Hancock, 2006; Matthews, Smith, Hancock, Jagger, & Spiers, 2005).

Exchange theory suggests that parents may provide support in exchange for the help they receive from children (Kohli & Künemund, 2003). Lower SES parents often hold higher expectations for support from adult children, presumably because neither generation can afford to pay for formal help (Lee, Peek, & Coward, 1998; Peek, Coward, & Peek, 2000). Yet, such higher expectation for child support may not necessarily increase these parents’ initiation of help in return. Indeed, given the intergenerational transmission of SES (Conger et al., 2010; Lee & Seshadri, 2019), lower SES parents may also have lower SES children who incur life crises that elicit parental help (Furstenberg, 2010). These parents may focus their limited resources on helping children who ask for help. Together, we expect upper SES parents are more likely to be initiators/responders or initiators, yet lower SES are more likely to be responders or uninvolved.

Child Problems (Needs)

We also explicitly examine whether children’s problems are associated with offspring requests for parental help. Contingency theory in sociology and altruism theory in economics suggest that older parents mobilize resources to help their children in need (e.g., when such needs interfere with children’s attainment of economic independence; Eggebeen & Davey, 1998; Gilligan, Suitor, Rurka, Con, & Pillemer, 2017; Huo et al., 2019; Kohli, 2004; Kohli & Künemund, 2003). Prior studies, however, did not ask whether parents instigate help or these children directly request parental help.

The social support literature suggests that when individuals incur major life problems (e.g., health concerns, emotional problems and financial crises), they turn to their social partners for help (Gleason & Iida, 2015; Thoits, 2011). Likewise, older parents may prioritize assistance to their children’s requests when children incur problems. Some of these parents also may help in the absence of explicit requests from offspring in response to those problems. Thus, we expect older parents with more children having problems to be responders or initiators/responders rather than initiators or uninvolved.

Parent Obligation

We also examine older parents’ perceived obligation to help, which may underlie the reason why parents initiate help. The literature regarding obligation examines children’s filial obligation to care for older parents rather than the reverse (Gans & Silverstein, 2006; Silverstein & Giarrusso, 2010), with a few exceptions (Kim, Fingerman, Birditt, & Zarit, 2016). Filial obligation induces offspring to help their older parents who require help (Silverstein & Giarrusso, 2010) and we expect a similar pattern for parents with greater obligation toward their offspring. That is, older parents who believe they should step in may assist due to their children’s request for help. Yet, these parents may not wait until help is requested. Qualitative research in Israel suggests that older parents who feel responsible for children endeavor to help these children (Levitzki, 2009). These parents may feel motivated to keep an eye on their children’s welfare and initiate help. We expect parents with greater obligation to help are more likely to be initiators or initiators/responders rather than responders or uninvolved.

Implications of Who Initiates Older Parents’ Help

We then explore whether profiles of help initiation are associated with how often older parents help and how they evaluate the rewards and stresses of helping. We focus on these links adjusting for resources, needs, and individual beliefs, which also influence frequency and evaluation of helping (Conger et al., 2010; Fingerman et al., 2015; Kim et al., 2016).

Frequency of Helping

We assess how often parental help occurs among older parents in different profiles. Parents who are initiators/responders may help most often because they are motivated both by their own desire to help and by child’s request (Iida et al., 2008). Yet, the comparison between parents who are initiators and parents who are responders is unclear. The type of support matters. We consider emotional support, advice, practical help, and financial support, which are all commonly provided by parents (Fingerman et al., 2011). Emotional support may reflect feelings of affection or a desire to comfort others’ distress (Burleson, 2003). Parents who primarily instigate help may offer more emotional support than parents who solely respond to requests. Advice is often unsolicited in close ties (Feng & Magen, 2016); parents who are initiators may give advice more often than parents who are responders. Material support such as money or practical help primarily responds to specific needs (Swartz, 2009). Yet, whether parents initiate material support may also vary depending on their own financial resources.

Rewards and Stresses of Helping

We also examine how older parents evaluate their helping behaviors. Research reveals that helping is not always rewarding; rather, individuals may view helping as rewarding or stressful under varying circumstances (Bangerter, Kim, Zarit, Birditt, & Fingerman, 2015). Older parents may enjoy helping when they volunteer to do so. A recent review (Inagaki & Orehek, 2017) proposes that providers find helping beneficial when they can choose when and what to give. It may be the case that parents who initiate help evaluate giving that help as beneficial. Self-determination theory further posits that intrinsic motivation to help may satisfy older parents’ needs to feel autonomous and competent, which renders assistance rewarding (Deci & Ryan, 2000; Gruenewald, Karlamangla, Greendale, Singer, & Seeman, 2007; Weinstein & Ryan, 2010). Further, older parents may feel even more rewarded if they also help children in response to a need that can be addressed. According to the support matching hypothesis, parents may offer optimal help in response to specific requests from children, thereby gaining emotional rewards (Gleason & Iida, 2015; Inagaki & Orehek, 2017).

Yet, older parents can also feel burdened when they primarily help in response to their offspring’s requests. Help on demand may deprive older parents’ sense of control over their own behaviors and augment their distress (Ross, 2017; Thoits, 2011). Also, most parents expect their children to be independent in adulthood and these parents may incur distress when they primarily respond to requests from children (Fingerman, Cheng, Birditt, & Zarit, 2012; Suitor et al., 2007). We expect parents who are initiators or initiators/responders to view their helping behaviors as more rewarding and less stressful compared to parents who are responders or uninvolved.

Other Factors as Covariates and the Current Study

We consider additional factors: parent gender, age, health, marital status, minority status, affection for children, support from parents, coresidence, and number of young grandchildren. Women are more involved in family ties than men (Rossi & Rossi, 1990; Umberson, Pudrovska, & Reczek, 2010). As people age, they have more health concerns, which may limit their ability to initiate help (Fingerman, VanderDrift, Dotterer, Birditt, & Zarit, 2011; Grundy, 2005). Married parents have more resources to help children, whereas unmarried, divorced, or widowed parents offer less help (Isherwood, Luszcz, & King, 2016; Umberson et al., 2010). Compared to European American parents, African American parents may have fewer financial resources but be more involved in their children’s lives (Ajrouch, Antonucci, & Janevic, 2001; Peek et al., 2000). Solidarity theory posits that individuals offer more help in closer ties (Bengtson & Roberts, 1991). Exchange theory suggests parents help more often and enjoy helping more when they also receive help (Gleason & Iida, 2015; Kohli & Künemund, 2003). Coresidence allows older parents to offer more help, especially practical help that requires proximity (Compton & Pollak, 2015; Smits et al., 2010). Older parents provide more help to their midlife children who are parents of young children (Fingerman, Miller, Birditt, & Zarit, 2009; Suitor et al., 2007).

This study utilizes a typology approach to differentiate older parents in terms of who initiates their help to midlife children. We seek to identify factors associated with different profiles of help initiation and also examine whether these profiles are associated with the frequency of help and older parents’ evaluation of helping behaviors.

Methods

We draw on data from the Family Exchanges Study Wave 2 conducted in 2013. The Family Exchanges Study recruited three-generation families from the Philadelphia Primary Metropolitan Statistical Area (Fingerman et al., 2009). Wave 1 data collection began in 2008 with 633 midlife adults (aged 40–60 years) who had at least one living parent and one child aged 18+. Midlife adults provided contact information for their older parents and young adult children, who were also invited to participate in the study. Participants completed a 1-hr computer-assisted telephone interview and received $50.

Among the 455 older parents (i.e., parents of midlife children) invited in Wave 1, 337 (74%) participated; they were younger, healthier, less likely to be disabled and more likely to be female than parents who did not participate. In Wave 2, we contacted these older parents but 126 parents did not return: 58 were deceased, 5 were too ill, and 63 could not be reached. Compared to these 126 parents, older parents who participated in Wave 2 (n = 211) were younger, healthier, better educated, and more likely to be married. We also reached out to older parents who were eligible but did not participate in Wave 1. Thirty older parents newly joined Wave 2 and they did not differ from the 211 returning parents in any background characteristics. We only use Wave 2 data because we did not ask questions about who initiates family help in Wave 1.

Older parents specified ties with up to three midlife children. For parents who had more than three children (46%), these children were selected to be: (a) who was first recruited into the study and provided contact information for the parent, (b) who gave most to the parent, and (c) who gave least. On average, older parents reported on 2.67 midlife children (total n = 645, with 4% step children). Table 1 describes the parent sample.

Table 1.

Background Characteristics of the Sample

	Older parents (N = 241)
	M	SD	Range
Age	80.12	5.93	60–95
Self-rated healtha	2.96	1.08	1–5
Number of adult childrenb	2.67	0.57	1–3
Number of young grandchildrenc	2.09	3.00	0–15
Education	12.97	2.23	0–20
Number of children with problemsd	1.74	1.67	0–10
Parent obligation to childrene	3.73	0.60	1–5
Rewards of helpingf	3.99	1.05	1–5
Stresses of helpingf	1.71	1.11	1–5
Affection for childreng	4.08	0.61	1–5
Help provided to childh
Emotional support	4.40	2.30	1–8
Advice	3.55	2.21	1–8
Practical support	2.52	2.05	1–8
Financial support	2.37	1.49	1–8
Help received from childh
Emotional support	4.40	2.39	1–8
Advice	3.81	2.37	1–8
Practical support	3.56	2.28	1–8
Financial support	2.19	1.63	1–8
		Proportion
Male		.30
Married/Remarried		.39
Minorityi		.31
Coresident with any childj		.19
Profiles k
Parent mainly initiated parental help		.27
Child mainly initiated parental help		.23
Both initiated parental help		.21
Neither initiated parental help		.29

Note: ^a1 (poor), 2 (fair), 3 (good), 4 (very good), 5 (excellent). ^bUp to three focal children selected to be the midlife child first recruited to the study, the child who gave the most and the child who gave the least. ^cNumber of young grandchildren aged under 18. ^dChildren with any of the 10 life problems (e.g., health concerns, financial difficulties, emotional problems, divorce). ^eBeliefs that parents should help adult children, rated 1 (never), 2 (rarely), 3 (sometimes), 4 (often), and 5 (always). ^fRated 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), and 5 (a great deal). ^gAveraged affection across midlife children. Affection for each child was calculated by averaging ratings of two items: whether each child (a) loves and cares for the parent, and (b) understands the parent, from 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), to 5 (a great deal). ^hRated 1 (less than once a year or never), 2 (once a year), 3 (a few times a year), 4 (monthly), 5 (a few times a month), 6 (weekly), 7 (a few times a week), and 8 (daily). The average frequency with which older parents provided and received each type of support from midlife children. ⁱCoded as 1 (a minority), 0 (not a minority). ^jRated 1 (Coresided with any child) and 0 (Did not coresident with any child). ^kBased on the extent to which parents volunteered to help without being asked, and parents gave help because their midlife children asked for it.

Measures

Help initiation

Older parents indicated the extent to which they (a) volunteered to help midlife children without being asked, and (b) gave children help because children had asked, from 1 (not at all) to 5 (a great deal). These variables are not specific to each child or each type of support; rather, they refer to parents’ general tendency to initiate help and respond to requests.

Factors associated with help initiation

Parents completed an obligation measure parallel to the widely used filial obligation measure, indicating how often parents should provide adult children six types of support (e.g., emotional support, practical assistance, advice) on a scale from 1 (never) to 5 (always). We calculate a mean score across the six items (α = .72). We measured older parents’ socioeconomic status based on their self-reported education in years. Parents also indicated whether each midlife child incurred 10 life problems (e.g., health concerns, emotional problems, substance abuse, financial difficulties, divorce) in the prior 2 years (Fingerman et al., 2009; Greenfield & Marks, 2006). Due to a skewed distribution, we recode a binary variable: 1 (this child had at least one problem) and 0 (this child did not have any problems). We then calculate a sum score to assess the number of children with problems.

Frequency of helping

Parents indicated how often they provided six types of support (e.g., listening to talk about daily life, emotional support, companionship, advice, practical help, and financial support) to each child using the Intergenerational Support Scale (ISS; Fingerman et al., 2011). Responses were coded on a scale from 1 (less than once a year or never) to 8 (daily). We examine the frequency of each type of support averaged across multiple midlife children.

Rewards and stresses of helping

Older parents rated in general how (a) rewarding and (b) stressful they found it to help their midlife children (Bangerter et al., 2015). Responses were coded from 1 (not at all) to 5 (a great deal).

Covariates

Older parents reported their age, and gender coded as 1 (male) and 0 (female). They also rated physical health from 1 (poor) to 5 (excellent; Idler & Kasl, 1995). We recode parents’ marital status as 1 (married/remarried) and 0 (not married). We recode parents’ racial/ethnic groups as 1 (a minority) and 0 (not a minority). Parents rated how much each child (a) loved and cared for them and (b) understood them (Fingerman et al., 2011) from 1 (not at all) to 5 (a great deal). We average the two items to measure affection for each child (ρ = .62) and then calculate a mean score across children. Parents also indicated how often they received six types of support (ISS; Fingerman et al., 2011). We create a variable to measure how often parents received each type of support on average. Parents also reported whether they lived with any adult child as 1 (yes) or 0 (no) and the number of grandchildren aged under 18.

Analytic Strategy

We first explore profiles of help initiation and conduct latent profile analysis (LPA) based on two continuous variables: older parents’ reports on the extent to which they volunteer to help and they assist per request (Mplus 7.4; Asparouhov & Muthén, 2008). We test models with different numbers of profiles and compare the fits of these models, considering log likelihood, Bayesian information criterion (BIC), adjusted Bayesian information criterion (ABIC), Lo-Mendell-Rubin (LMR) test and Vuong-Lo-Mendell-Rubin (VLMR) test, and entropy. LPA reveals the probability that older parents exhibit each profile.

We then examine factors associated with these profiles. Given that 68 older parents were couples and reported on the same grown child, we initially estimated two-level models given nested data—parent (level 1) is nested within couple (level 2). The preliminary null model reveals a zero random effect for the couple level, suggesting there is no variability at the couple level. We dropped the couple level and estimate an ordinary least squares (OLS) regression. Predictors are parent education, the number of midlife children with problems, and parent obligation to help adult children; these predictors are entered into the same model. The outcome indicates the likelihood that older parents would show each profile. We estimate multinomial logistic regressions due to the categorical nature of this outcome; we alternate the reference group to conduct pairwise comparison. Models control for parent age, gender, health, marital status, minority status, affection, support receipt, coresidence, and number of young grandchildren.

We further explore how often older parents with different profiles offer each type of help and evaluate rewards and stresses of helping. Older parents specified the frequency and evaluation of helping for each child and they reported on multiple children. Thus, we estimate two-level models to take into account that child is nested within parent. We treat the profiles of who initiates older parents’ help as predictors. For frequency of helping, the outcome variables are how often older parents offer each type of support to each midlife child. For parents’ evaluation of helping, we treat parents’ ratings of rewards and stresses of helping as separate outcomes in two models. We also conduct pairwise comparisons by alternating the reference group for parents’ profiles of help initiation. In addition to the covariates mentioned above, this set of models also adjusts for parent resources, child problems, and parent obligation.

Results

We estimate Latent Profile Analysis and compare model fit statistics for two, three, four, and five profiles of who initiates older parents’ help to midlife children (Supplementary Table 1). The four-profile model exhibits the best fit (see Figure 1 and Supplementary Table 2). We observe four profiles: 65 parents (27%) are initiators who tend to volunteer to help; 56 parents (23%) are responders who tend to respond to requests; 50 parents (21%) are initiators/responders who both volunteer to help and assist per request; and 69 parents (29%) are uninvolved who neither volunteer nor assist per request. One parent has missing data and is not included in later analyses.

Figure 1.

Profiles of who initiates older parents’ help to midlife children.

Factors Associated with Who Initiates Older Parents’ Help

We estimate multinomial logistic regressions to examine how parent resources, child problems, and parent obligation are associated with the likelihood of parent showing each profile of help initiation (Table 2).

Table 2.

Multinomial Logistic Regression Models for Profiles of Who Initiates Older Parents’ Help

Variables	Initiators/responders vs initiators			Initiators/responders vs responders			Initiators/responders vs uninvolved
	B	SE	OR	B	SE	OR	B	SE	OR
Intercept	−2.21	3.97		−9.57*	4.15		−0.28	4.23
Parent education in years	0.20*	0.10	1.22	0.24*	0.11	1.28	0.26*	0.11	1.30
Children with problemsa	0.36	0.20	1.44	0.29	0.22	1.33	0.54*	0.22	1.72
Parent obligationb	1.10**	0.41	2.99	1.44***	0.43	4.21	1.35***	0.42	3.84
Parent covariates
Age	−0.08*	0.04	0.93	−0.01	0.04	0.99	−0.13**	0.04	0.88
Male	−0.13	0.54	0.88	−0.41	0.54	0.67	0.15	0.53	1.16
Physical healthc	−0.18	0.20	0.84	−0.26	0.21	0.77	−0.03	0.21	0.97
Marital statusd	0.75	0.51	2.12	−0.13	0.52	0.88	0.24	0.51	1.27
Minority statuse	−0.28	0.47	0.75	−0.26	0.49	0.77	0.57	0.54	1.76
Affection for childrenf	0.13	0.41	1.14	0.35	0.41	1.42	−0.28	0.42	0.76
Help receivedg	0.13	0.20	1.14	0.13	0.20	1.13	0.58**	0.21	1.79
Coresidenth	0.27	0.54	1.31	−0.40	0.55	0.67	0.56	0.63	1.74
Number of young grandchildreni	−0.00	0.02	1.00	0.02	0.02	1.02	−0.01	0.02	1.00

Note: Older parents n = 240. OR = Odds ratios. Model fit: −2 (pseudo) log likelihood = 650.93. Profiles were generated based on the extent to which parents volunteered to help without being asked, and parents gave help because their midlife children asked for it.

^aChildren with any of the 10 life problems (e.g., health concerns, financial difficulties, emotional problems, divorce). ^bBeliefs that parents should help adult children, rated 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = always. ^c1 = poor, 2 = fair, 3 = good, 4 = very good, 5 = excellent. ^d1 = married/remarried, 0 = not married. ^eCoded as 1 = racial/ethnic minority, 0 = not a minority. ^fAveraged affection across midlife children. Affection for each child was calculated by averaging ratings of two items: whether each child (a) loves and cares for the parent, and (b) understands the parent, from 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), to 5 (a great deal). ^gRated 1 = less than once a year or never, 2 = once a year, 3 = a few times a year, 4 = monthly, 5 = a few times a month, 6 = weekly, 7 = a few times a week, and 8 = daily. The average frequency with which older parents provided and received any type of support from midlife children. ^hRated 1 (Coresided with any child) and 0 (Did not coresident with any child). ⁱNumber of young grandchildren aged under 18.

*p < .05. **p < .01. ***p < .001.

We expect upper SES older parents are more likely to be initiators or initiators/responders than responders or uninvolved. Specifically, we find that upper SES older parents are more likely to be initiators/responders than initiators (odds ratio [OR] = 1.22, p = .04), responders (OR = 1.28, p = .02), or uninvolved (OR = 1.30, p = .01). Parent SES did not differentiate among parents who are initiators, responders, and uninvolved.

We expect older parents who have more children with problems to be responders or initiators/responders than initiators or uninvolved. These parents are more likely to be initiators/responders than uninvolved (OR = 1.72, p = .01). The number of children with problems does not differentiate between any other profiles.

We expect older parents with greater obligation to be initiators or initiators/responders than responders or uninvolved. Indeed, older parents with greater obligation are more likely to be initiators/responders than initiators (OR = 2.99, p = .008), responders (OR = 4.21, p < .001), or uninvolved (OR = 3.84, p < .001). Yet, parent obligation does not differentiate other patterns.

Frequency, Rewards and Stresses of Helping

We estimate models to test links between profiles of help initiation and how often older parents help (Table 3). Parents who are initiators/responders offer more frequent emotional support than parents who are responders (B = 0.67, p = .03). Parents who are initiators/responders (B = 0.77, p = .009), initiators (B = 0.81, p = .002), or responders (B = 0.58, p = .03) give more advice to midlife children than parents who are uninvolved. Parents who are initiators/responders (B = 0.58, p < .05), initiators (B = 0.59, p = .02), or responders (B = 0.59, p = .02) also offer more practical help than parents who are uninvolved. Profiles of help initiation are not associated with financial support.

Table 3.

Regression Predicting Parents’ Frequency, Rewards, and Stresses of Helping from Profiles of Who Initiates Older Parents’ Help

Variable	Emotional support		Advice		Practical support
	B	SE	B	SE	B	SE
Fixed effects
Intercept	0.29	1.86	4.69**	1.79	3.38	1.77
Profiles of help initiationa
Initiators	0.51	0.28	0.81**	0.26	0.59*	0.25
Responders	(Ref.)	(Ref.)	0.58*	0.27	0.71**	0.27
Initiators/Responders	0.67*	0.30	0.77**	0.29	0.58*	0.29
Uninvolved	0.21	0.29	(Ref.)	(Ref.)	(Ref.)	(Ref.)
Parent covariates
Age	−0.01	0.02	−0.02	0.02	−0.03	0.02
Male	0.01	0.25	0.60**	0.23	0.13	0.23
Physical healthb	0.05	0.10	−0.09	0.09	−0.03	0.09
Marital statusc	−0.11	0.23	0.10	0.22	−0.04	0.21
Minority statusd	−0.19	0.24	0.11	0.22	−0.25	0.22
Affection for childrene	0.31	0.18	−0.03	0.17	0.09	0.17
Help receivedf	0.75***	0.10	0.72 ***	0.09	0.54***	0.09
Coresidentg	0.34	0.26	0.40	0.25	0.29	0.24
Number of young grandchildrenh	−0.02	0.01	0.00	0.01	0.01	0.01
Education in years	0.04	0.05	−0.11*	0.04	−0.03	0.04
Children with problemsi	−0.06	0.19	−0.18	0.17	−0.14	0.17
Obligation to helpj	0.20**	0.06	0.05	0.06	−0.04	0.06
Random effects
Intercept VAR	1.01***	0.23	0.86***	0.20	0.82***	0.19
Residual VAR	3.15***	0.22	2.86***	0.20	2.82***	0.20
−2 log likelihood	2,654.7		2,618.0		2,603.5

Note: Older parents n = 240. VAR = Variance.

^aProfiles were generated based on the extent to which parents volunteered to help without being asked, and parents gave help because their midlife children asked for it. ^b1 = poor, 2 = fair, 3 = good, 4 = very good, 5 = excellent. ^c1 = married/remarried, 0 = not married. ^dCoded as 1 = racial/ethnic minority, 0 = not a minority. ^eAveraged affection across midlife children. Affection for each child was calculated by averaging ratings of two items: whether each child (a) loves and cares for the parent, and (b) understands the parent, from 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), to 5 (a great deal). ^fRated 1 = less than once a year or never, 2 = once a year, 3 = a few times a year, 4 = monthly, 5 = a few times a month, 6 = weekly, 7 = a few times a week, and 8 = daily. The average frequency with which older parents provided and received any type of support from midlife children. ^gRated 1 (Coresided with any child) and 0 (Did not coresident with any child). ^hNumber of young grandchildren aged under 18. ⁱChildren with any of the 10 life problems (e.g., health concerns, financial difficulties, emotional problems, divorce). ^jBeliefs that parents should help adult children, rated 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = always.

*p < .05. **p < .01. ***p < .001.

We estimate two-level models to assess whether profiles of help initiation are associated with these parents’ rewards and stresses of helping. Older parents who are initiators/responders view their helping behaviors as more rewarding than parents who are initiators (B = 0.28, p = .03) and more stressful than parents who are uninvolved (B = 0.44, p = .01; Table 4).

Table 4.

Regression Predicting Parents’ Frequency, Rewards, and Stresses of Helping from Profiles of Who Initiates Older Parents’ Help

Variable	Rewards of helping		Stresses of helping
	B	SE	B	SE
Fixed effects
Intercept	0.86	0.85	1.97	1.06
Profiles of help initiationa
Initiators	(Ref.)	(Ref.)	0.24	0.15
Responders	0.10	0.13	0.15	0.16
Initiators/Responders	0.28*	0.13	0.44*	0.17
Uninvolved	0.02	0.13	(Ref.)	(Ref.)
Parent covariates
Age	−0.00	0.01	0.00	0.01
Male	−0.13	0.11	−0.23	0.14
Physical healthb	−0.00	0.04	−0.00	0.05
Marital statusc	0.18	0.10	0.14	0.13
Minority statusd	−0.24*	0.11	0.12	0.13
Affection for childrene	0.54***	0.08	−0.27*	0.10
Help receivedf	0.08*	0.04	0.06	0.05
Coresidentg	−0.20	0.12	0.05	0.14
Number of young grandchildrenh	0.00	0.01	0.01	0.01
Education in years	0.01	0.02	0.02	0.03
Children with problemsi	0.14	0.09	−0.03	0.11
Obligation to helpj	−0.03	0.03	0.02	0.04
Random effects
Intercept VAR	0.14**	0.05	***	0.07
Residual VAR	0.75***	0.06	***	0.06
−2 log likelihood	1,657.6		1,819.0

Note: Older parents n = 240. VAR = Variance.

^aProfiles were generated based on the extent to which parents volunteered to help without being asked, and parents gave help because their midlife children asked for it. ^b1 = poor, 2 = fair, 3 = good, 4 = very good, 5 = excellent. ^c1 = married/remarried, 0 = not married. ^dCoded as 1 = racial/ethnic minority, 0 = not a minority. ^eAveraged affection across midlife children. Affection for each child was calculated by averaging ratings of two items: whether each child (a) loves and cares for the parent, and (b) understands the parent, from 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), to 5 (a great deal). ^fRated 1 = less than once a year or never, 2 = once a year, 3 = a few times a year, 4 = monthly, 5 = a few times a month, 6 = weekly, 7 = a few times a week, and 8 = daily. The average frequency with which older parents provided and received any type of support from midlife children. ^gRated 1 (Coresided with any child) and 0 (Did not coresident with any child). ^hNumber of young grandchildren aged under 18. ⁱChildren with any of the 10 life problems (e.g., health concerns, financial difficulties, emotional problems, divorce). ^jBeliefs that parents should help adult children, rated 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = always.

*p < .05. **p < .01. ***p < .001.

Discussion

This study uses a typology approach to understand the persistent support that parents provide to their children even in old age. Research has primarily examined the frequency of parental help (Fingerman et al., 2011; Huo et al., 2019; Suitor et al., 2007) but we contribute to the literature by asking about parents’ beliefs regarding who initiates such help. These beliefs may influence how older parents feel about helping. Findings exhibit variation in the initiation of help: some older parents tend to volunteer to help midlife children (initiators), some parents tend to step in per child’s request (responders), some parents do both (initiators/responders), and other parents rarely help (uninvolved). Older parents who are better educated, had more children with problems, and felt more obligated to help are more involved with their midlife children by both initiating help and assisting per request. Among the vast majority of older parents who help children, parents who both volunteer and help per request enjoy their helping behaviors more than parents who primarily initiate help. Yet, helping midlife children can still be stressful, even when parents volunteer to do so.

Factors Associated with Who Initiates Help

We offer additional evidence to the Multidimensional Intergenerational Support Model (Fingerman et al., 2013) by examining older parents’ beliefs about who initiates parental help to midlife offspring. We consider the context in which parent help occurs (i.e., resources, needs) and individual beliefs that drive such help.

Parent resources

Findings are in line with the tenets of social capital theory (Furstenberg & Hughes, 1995; Furstenberg & Kaplan, 2004), which is prominent in sociological accounts of family exchanges. We observe that upper SES parents are more active in helping midlife children, as they are more likely to initiate help and/or help per request. This study suggests that having greater resources may also equip older parents with better ability to cope with their children’s requests. These results may also set up for research to further test social exchange theory (Kohli & Künemund, 2003). Data on parents’ beliefs of who initiates the help they receive from children are needed to determine how parents’ resources may influence support exchanges with adult children.

Child problems

Findings regarding child problems add to an understanding of contingency theory and altruism theory (Eggebeen & Davey, 1998; Kohli & Künemund, 2003). A rich literature documents that parents offer more help to their adult children suffering problems (Fingerman et al., 2009; Gilligan et al., 2017; Huo et al., 2019). Yet, little research tests what triggers such help—whether midlife children directly seek parental help (Iida et al., 2008) or whether parents provide unrequested support. We find that older parents who have a greater number of children with problems tend to help in either way, through requests or their own inition. Even in midlife, children still turn to their older parents with requests for help and parents also step in and volunteer to help when children incur difficulties.

Parent obligation

We also break new ground by investigating how older parents’ sense of obligation influences why they help midlife children. Findings suggest that more obligated older parents both initiate help to benefit these children and also actively respond to requests from offspring. This sense of parental obligation may be a lifelong belief or result from earlier experience (Levitzki, 2009; Umberson et al., 2010). Parents could obtain obligation beliefs as a result of a history of relationships with a particular child who has needed help throughout life. This study does not draw on longitudinal data but future studies may look into how relationship history influences parental initiation of help and feelings of obligation to assist midlife children.

Implications of Who Initiates Older Parents’ Help

We also examine whether profiles of help initiation are associated with the frequency and older parents’ evaluation (e.g., rewards and stresses) of helping. Interestingly, older parents offer the same amount of advice or practical help regardless of who initiates their help. With regard to emotional support, older parents who both instigate help and respond to requests offer more frequent emotional support than parents who primarily help per request. The latter may have emotionally detached ties to their children and only step in when children incur troubles. We do not observe a link between help initiation and financial support; offering money may be more dependent on parent financial resources. Findings need to be interpreted with caution, though. We acknowledge the initiation of parental help may vary by situations, but here we assess parents’ general beliefs regarding who initiates their help. The initiation of parental help is not specific to a particular child or the type of support parents actually provide.

In addition, these profiles of help initiation influence how older parents view their helping behaviors. Consistent with social exchange theory (Kohli & Künemund, 2003), we find that parents who are initiators/responders view helping as more rewarding than parents who primarily initiate help. Indeed, the older parents who believe they are also responding to child’s requests may enjoy the child’s reliance on them, as shown in prior research (Byers, Levy, Allore, Bruce, & Kasl, 2008). The literature has extensively documented the benefits of helping in later life (Kahana, Bhatta, Lovegreen, Kahana, & Midlarsky, 2013; Thomas, 2010). Yet, we also find that older parents who are initiators/responders view helping as more stressful than parents who are uninvolved. It seems likely that these parents who engage in active and reciprocal relations with children also experience greater stress owing to a greater feeling of obligation. Regardless of their resources and obligation, older parents may still expect children to be competent in adulthood and they suffer when these children fail (Fingerman, Cheng, Wesselmann, et al., 2012; Suitor et al., 2007). Although helping children with problems may alleviate parents’ distress over these children on a daily basis (Huo et al., 2019), this may not hold true in the long run (Ross, 2017; Thoits, 2011).

This study examines parents’ general style of initiating help, and does not specify whether the help parent initiates and the help child requests are given to the same child. Initiating help to one child may co-occur with helping another child incurring trouble because the child requests it. Prior research suggests that having at least one child with problems generates poor well-being for the parent, and such stresses cannot be buffered with having successful children (Fingerman, Cheng, Birditt et al., 2012). Likewise, the reward of initiating help to one child may be overridden by the stress of having to cope with other children’s requests simultaneously.

Limitations and Implications

Several limitations to this study warrant consideration. This study relies on older parents’ reports. It can be difficult to discern who initiates parental help using one party’s reports, especially in families where helping is common and reciprocal. We also do not know the extent to which these situations represent all initiations or requests for help or only successful ones. Parents may overestimate their investment in children (Giarrusso, Feng, & Bengtson, 2005), which leads to a bias in their reports. Researchers should also look into consequences for how the child evaluates the help they receive. Indeed, from the child’s perspective, parental help could be perceived as intrusive (Fingerman, Cheng, Cichy, Birditt, & Zarit, 2013). Future research may also explore the initiation of family help with qualitative data to elucidate some of these nuances (Napolitano, Furstenberg, & Fingerman, 2019). This study does not include parents with severe physical or cognitive impairments. Future research should explore more directly how these impairments influence the amount and type of parental support among parents with ill-health and disability. Moreover, research suggests more religious adult children feel more obligated to help older parents (Gans, Silverstein, & Lowenstein, 2009; King, Ledwell, & Pearce-Morris, 2013), which may hold true for older parents helping children. We did not measure religiosity but future research may pursue this issue.

We are also limited by considering a maximum of three children per parent. We selected a random child, a child who gives most to parent and a child who gives least, which should be representative of the range of support for that family. We only examine everyday support but not in-vivo transfers or bequests which are also common in late life parent–child ties and have to come from the parent. Research on bequests suggests discrepancies between parent intention and child expectation (Kim, Eggebeen, Zarit, Birditt, & Fingerman, 2013). Future research may explicitly ask whether profiles of help initiation hold in in-vivo transfers or bequests.

This study utilizes a typology approach to advance an understanding of older parents’ help to midlife children. Findings reveal variation in the initiation of parental help in late life, which has been largely overlooked in the literature regarding parent-child support exchanges. Scholars have linked structural (e.g., resources, needs) or psychological factors, (e.g., obligation) to family help (Fingerman et al., 2013), but we offer empirical evidence regarding how these factors are associated with who initiates such help. Moreover, the initiation of parental help influences how older parents feel about their helping behaviors. We find that helping midlife children can be accompanied by stress, which persists regardless of parents’ resources and willingness to help. This study offers some caveats and suggestions to family therapists that possibly could give some guidance to improving family help and relationships. There are costs and benefits to helping children in later life that may add up differently for families depending on parental resources, beliefs, and the persistence of children’s needs or demands.

Funding

This study was supported by the National Institute on Aging (NIA, R01AG027769 to K. L. Fingerman); the John D. and Catherine T. MacArthur Foundation; and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD, 5 R24 HD042849 to the Population Research Center at the University of Texas at Austin).

Conflict of Interest

The authors certify that they have NO affiliations with or involvement in any organization or entity with any financial interest or nonfinancial interest in the subject matter or materials discussed in this manuscript.

Author Contributions

M. Huo designed the study, performed statistical analyses, and wrote the manuscript. L. Napolitano offered critical revisions and rewrites for this manuscript. F. F. Furstenberg assisted in conceptualizing and rewriting drafts of this manuscript. K. L. Fingerman was the Principal Investigator on the grant that funded the study; she designed the study, oversaw data collection and statistical analyses, and helped writing drafts of the manuscript.