Unstable couple relationships and high rates of repartnering have increased the share of U.S. families with stepkin. Yet data on stepfamily structure are from earlier periods, include only coresident stepkin, or cover only older adults. In this study, we use new data on family structure and transfers in the Panel Study of Income Dynamics (PSID) to describe the prevalence and numbers of stepparents and stepchildren for adults of all ages and to characterize the relationship between having stepkin and transfers of time and money between generations, regardless of whether the kin live together. We find that having stepparents and stepchildren is very common among U.S. households, especially younger households. Furthermore, stepkin substantially increase the typical household’s family size; stepparents and stepchildren increase a household’s number of parents and adult children by nearly 40 % for married/cohabiting couples with living parents and children. However, having stepkin is associated with fewer transfers, particularly time transfers between married women and their stepparents and stepchildren. The increase in the number of family members due to stepkin is insufficient to compensate for the lower likelihood of transfers in stepfamilies. Our findings suggest that recent cohorts with more stepkin may give less time assistance to adult children and receive less time assistance from children in old age than prior generations.
Family members often share the routine tasks in everyday life and provide more intense help in crises. How and the extent to which family members help each other depends on who is in the family and the strength of family ties, who may need assistance and who is able to provide it, and whether assistance is in the form of time or money. The availability of kin is a central element for describing the potential connections between family members. In demographic terms, kin availability indicates who is at risk of assisting or needing assistance from a family member: if the kin do not exist, no assistance can be given or received. The extent to which stepparents and stepchildren should be considered among the available kin is an important question for understanding the connections within U.S. families in light of the high rates of repartnering after a first childbearing union dissolves (Cherlin 2010).
In evaluating the importance of stepkin in U.S. families, we have two objectives. First, we aim to provide a demographic portrait of biological and step-relationships among parents and their adult children in contemporary American families and the nature of the ties across these generations as measured by the time and money they provide to one another. Using data on parent-child relationships and transfers between parents and adult children in the 2013 Panel Study of Income Dynamics (PSID), we examine the presence and numbers of parents and adult children and the prevalence of stepkin in both the older and younger adult generations. We also show how having stepparents and stepchildren is associated with manifest ties across generations.
The second objective is to investigate a question concerning stepkin and intergenerational ties that was first posed by Wachter (1997) but thus far remains unanswered. Using demographic microsimulation methods, Wachter forecasted a decline in the number of biological kin but an increase in stepkin during the twenty-first century. Speculating on the implications of his findings, Wachter asked, To what extent does the increase in number of family members due to stepkin compensate for the weaker ties between stepkin than biological kin? We explicitly address this question by taking account of the number of biological and stepkin in families and differences in the propensities of families with and without stepkin to transfer time and money to family members.
We find that nearly 30 % of American households have a stepkin tie in either the parent or adult child generation of their family and that stepkin ties are more common among households headed by younger adults. Moreover, stepkin dramatically increase the size of families. Among households whose heads and wives have living parents, stepparents increase the total number of parents by close to 20 %; among households headed by married couples who have adult children, stepchildren increase the total number of children by 66 %. In addressing the question Wachter (1997) posed, we find that having a larger number of family members is not systematically related to transfer behavior and that the effects of numbers of family members on transfers are mostly small and imprecisely estimated. However, we also find that family members are less likely to give time support in the presence of stepkin. Among married/cohabiting couples, the stepkin disadvantage is particularly large when wives rather than husbands have stepkin. Finally, combining the effects on transfers of family size and stepfamily structure, we find that the increased availability of kin due to stepfamily members does not compensate for the weaker bonds in stepfamilies.
The remainder of this article is organized as follows. In the next section, we consider previous research on family size and stepfamily differences in transfers. Then we describe the data, focusing on the 2013 Rosters and Transfers Module (2013 R & T) of the PSID. We go on to describe the methods that we use to analyze stepfamily structure and transfers. We present our portrait of the demographic availability of parents, stepparents, adult children, and adult stepchildren in today’s American families. We also examine intergenerational financial and time transfers within families, emphasizing how they differ by biological versus stepkin, stages of the life course, and gender. And, we estimate net associations between stepkin and transfers accounting for differences in the number of family members. We discuss our findings and then offer concluding observations about American families and their intergenerational ties.
Motivating the question of how the rise in the prevalence of stepfamilies will affect the ties between family members is the idea of competing forces: having more family members may raise the potential likelihood that an individual receives any transfers, but the potentially weaker ties with stepkin may reduce their prevalence. Although the question of how the increased prevalence of stepfamilies has reshaped family ties remains unanswered, research has examined how family size and step-relationships each relate to transfers between family members.
In principle, stepkin can increase the number of family members in both the parents’ and children’s generations, thereby increasing the potential number of providers and recipients of transfers across adult generations. Families with more members may have a higher incidence of transfers, even though some of the members of these families have weaker ties than others. Indeed, such is the prediction of some economic models of the provision of care across the generations within families. Larger families have the potential to provide more time or money than smaller ones if family members pool their incomes.1 However, other theoretical mechanisms suggest that individuals in large families may be less likely to provide help because more alternate family caregivers are available or because family size may alter individuals’ perceptions of their responsibilities if, for instance, responsibility is more diffuse in large families (van Gaalen and Dykstra 2006).2
Empirical evidence on the association between family size and transfers also is mixed. For instance, Checkovick and Stern (2002) found that elderly parents in large families are more likely to receive care from adult children than parents in smaller families, but Byrne et al. (2009) found that children with more siblings are less likely to provide care than those with fewer siblings. Some studies have found that adult children in large families are less likely than those in small families to receive financial support from parents (Emery 2013; McGarry and Schoeni 1995), but another study found that financial support may be greater in large families (Hurd et al. 2011: appendix tables B and C). Still other research has shown greater variability in intergenerational transfers, contact across generations, and quality of parent-child relationships in large families than in small families (McGarry 2016; Uhlenberg and Hammill 1998; Ward et al. 2009). Support from adult children also depends on the characteristics of their siblings, including whether they are biological or stepsiblings (Grigoryeva 2017; Pezzin et al. 2008). In short, the nature and quality of intergenerational interactions as a function of family size vary by relationship dimension, whether the analysis is conducted from the donor’s or recipient’s perspective, and by the relationship type and characteristics (e.g., parents vs. children; disabled parents vs. all parents).
Evidence on the relationship between stepfamilies and family transfers is more consistent than evidence on family size and transfer associations. High rates of union dissolution and repartnering mean that a significant share of U.S. families include stepkin (Furstenberg, Jr. 2014; Lin et al. 2017; Parker 2011; Yahirun et al. 2018). Previous research has suggested that ties between biological children and parents, and the incidence and amounts of time and money transfers between them, are stronger than between stepchildren and parents (Coleman and Ganong 2008; Eggebeen 1992; Kalmijn 2013; Pezzin and Schone 1999; Pezzin et al. 2008; Seltzer et al. 2013; White 1994). Theoretical explanations for these weaker ties of stepparents and stepchildren include the primacy of biological forces, social norms that place biological ties as central to intergenerational relationships, ambiguity about the rights and responsibilities of stepparents and stepchildren, and/or the lasting tensions from the disruption of biological parents’ union (Ganong and Coleman 2017). That relationships after widowhood are weaker in stepfamilies than in families with no stepkin suggests that challenges to familial solidarity other than through the legacy of divorce contribute to these weak ties (Seltzer et al. 2013). The related literature on the effects of divorce and repartnering on family ties has shown that adult children have less contact with their divorced parents compared with married parents (Albertini and Garriga 2011) and that children’s ties with divorced fathers are substantially weaker than ties with married fathers (Kalmijn 2013). However, the evidence that repartnering exacerbates these weaker ties is mixed. Kalmijn (2007, 2015) found that repartnering weakens ties between adult children and their fathers, but Cooney and Uhlenberg (1990) and White (1992) found no significant effects.
Contributions of Our Research
We extend past research in four ways. First, our portrait uses data from a single, population-representative data source. Previous conclusions about the stepfamily structure and composition of U.S. families required piecing together information from multiple sources—for instance, from samples of birth cohorts, combinations of data from birth and union histories, or reports from restricted age groups.
Second, we explicitly address Wachter’s question investigating in the same analysis the associations among family size, the presence of stepkin, and transfers. We improve on past research by using a contemporary, nationally representative sample rather than a sample restricted by age or disability. A further strength of our design is that we examine transfers between adult children and parents from each generation’s perspective to provide a more comprehensive assessment than in previous studies that looked only at transfers from parents to children or children to parents.
Third, we characterize the intergenerational structure of adults in extended families, which is possible because the PSID includes information on both coresident and noncoresident parents and adult offspring. To date, much of the research on parent-child relationships in the United States—and on step-relationships in particular—has focused on ties in childhood and adolescence (Bumpass and Lu 2000; Case and Paxson 2001; Ginther and Pollak 2004; Manning et al. 2014). The focus on younger families owes partly to the reliance on household surveys that typically provide limited information about family members who are not present in a household at the time of an interview.3 However, most U.S. parents and their adult offspring live in separate households—only 30 % of parents with adult children have a coresident adult child (authors’ calculations)—yet parents and children continue to help each other well after children leave their parents’ homes. The PSID data we use allow us to understand the availability of kin and material exchange between parents and children throughout adulthood, which is crucial in the context of the elongated transition to adulthood and caregiving in older age.
Finally, throughout our portrait, we pay attention to the intersection of gender and step-relationships. Hagestad (1986) described women as the family “kin keepers,” and evidence from research on caregiving has shown that daughters provide the majority of intergenerational care to aging parents and that grandmothers are more likely than grandfathers to provide childcare (Henretta et al. 1997; Hogan et al. 1993; Luo et al. 2012; McGarry 1998; Pillemer and Suitor 2006; Wolf and Soldo 1988). Women also are more likely than men to provide emotional support (Chesley and Poppie 2009). The gender difference in the socioemotional aspect of caregiving may manifest in a larger time disadvantage in help given to parents when women have stepparents compared with men. Understanding the intersecting dynamics of stepfamily ties and gendered caregiving roles is particularly important as younger cohorts with more exposure to stepkin reach older ages, when they may require care from stepdaughters.
Data, Measures, and Sample
We use data from the 2013 R & T of the PSID. The rosters identified and collected information about the adult biological and stepchildren (age 18 and older) and parents, stepparents, and in-laws of the head and of the spouse/partner,4 if present, of each PSID household.5 The roster data are uniquely suited to this analysis because of the explicit questions about stepparents and stepchildren of adults of all ages. The part of the module on transfers collected information on the incidence and amounts of time and money transfers to and from parents and adult children. In what follows, we briefly describe the overall design of the PSID and provide detail on how we use the data from the 2013 R & T.
The PSID began with a sample of approximately 18,000 people in 5,000 household units in 1968. The 2013 sample includes 24,952 people in 9,063 households, a product of increase in households due to children growing up and forming new households and decisions to reduce sample size. Weights are available to adjust for these factors. Individuals were interviewed annually until 1997, and subsequently every other year.
All individuals in households recruited into the PSID in 1968 are said to have the PSID “gene.” Individuals who are born to or adopted by someone with the PSID gene acquire the gene themselves and are recruited to become members of the PSID sample for the rest of their lives. This genealogical design implies that the study provides data on a sample of extended family members at each wave. Individuals without the PSID gene also are represented in the PSID as long as they live with a PSID sample member. Individuals without the gene are not followed if they stop living with a PSID sample member.
The 2013 Rosters and Transfers Module
In the 2013 Rosters and Transfers Module (2013 R & T) of the PSID, interviewers asked respondents to report the name and key characteristics of all living biological and adoptive parents as well as all living biological and adopted children at least 18 years old, for both the PSID head and spouse/partner, resulting in information about adult generations. These rosters include all parents and offspring regardless of whether they are PSID sample members (i.e., have the PSID gene). The roster questions identify each head and spouse/partner’s biological offspring and use this as the basis for distinguishing shared and stepchildren of the current union. The PSID used the fertility histories to preload children of each partner to improve coverage, and interviewers explicitly probed for information about children in the preloaded roster about whom the respondent did not spontaneously report. As a result, very little information is missing on the biological or step-relationships of offspring to the household head and spouse/partner. The roster questions also identified each respondent’s living biological parents and asked whether these biological parents are currently married to each other or someone else to identify stepparents. When available, preloaded information on parents also was used. The roster data are especially valuable because the data collection strategy obtains information about stepparents and stepchildren regardless of whether the respondent perceives them to be family members or part of the respondent’s support network. This improves coverage of step-relatives with weak ties, who might otherwise have been omitted from the roster.
The characteristics of respondents’ parents and adult children reported in the roster include current work status (working; temporarily laid off, sick, or on maternity leave; looking for work; retired; disabled; keeping house; or student), health (excellent, very good, good, fair, or poor), marital status (single or married/cohabiting), homeownership (owns, rents, other), number of children (only asked about respondents’ adult biological and stepchildren), and household income in four brackets (<$25,000; $25,000–$50,000; $50,001–$75,000; or >$75,000). In addition, respondents reported about the educational attainment of all adult children. The question about parents’ union status combined married and cohabiting unions, but the questions about offspring distinguished married from cohabiting relationships. Information on educational attainment of parents and parents-in-law was not collected in the 2013 R & T because it was collected elsewhere in the survey.
After completing the rosters of parents and offspring, interviewers asked respondents about transfers of time and money with each parent and adult child that occurred during the 2012 calendar year. Transfers of time include help with any activity such as “errands, rides, chores, babysitting, and hands-on care” and have no threshold of hours for reporting a transfer (i.e., respondents could report one hour). Financial transfers include “money, loans and gifts over $100.”6 The question asked about direct financial transfers rather than in-kind support. Individuals reported whether a transfer was given and how many hours or dollars were given. Transfers were reported for the household head and spouse/partner combined. For example, if a married woman gave time help to her parents but her husband did not give any time help to her parents, this would have been recorded as a transfer of time from the couple to the wife’s parents. Importantly for our analysis, transfers of time and money were collected for both coresident and noncoresident adult children and parents. For a detailed description of the 2013 R & T data, see Schoeni et al. (2015).
Analysis Sample and Unit of Observation
Our sample starts with the 9,063 households in the 2013 PSID. Elimination of households with missing information on the nature of the relationship with parents or children (biological vs. step) reduces the sample to 9,023 households.7 The unit of analysis in our study is the collective of the head and, if present, his spouse/partner of a PSID household unit. For convenience, we simply refer to this collective as the “household.” Each household is reported about by one respondent. That is, in a married-couple household, one respondent reports about both the head’s and spouse’s family members and their transfers. We also examine the subsamples of married/cohabiting couples, single household heads, and single men and women with living parents and/or adult children. The sample sizes of the entire sample and each subsample are listed in each table.
We analyze stepkin ties and transfers at the level of the household rather than at the level of the parent-child dyad for three reasons. First, our focus is on the family as a whole, which points toward using an aggregate measure of transfer activity and an aggregate measure of stepfamilies. Second, and perhaps most importantly, the description of stepparent and stepchild relationships using population-representative data for U.S. households of all ages is an important contribution of both the 2013 R & T data and this article. To our knowledge, the PSID is the only source of such data. Thus, our description of stepkin ties characterizes U.S. households in terms of the prevalence of stepkin and the extent to which stepkin increase the number of parent and adult child relationships. Analyzing transfers at the household level allows the description of stepkin to match the analysis of transfers. Third, the 2013 R & T data do not distinguish between transfers made by the husband and transfers made by the wife/cohabiting partner in couples for either time or money. This data limitation implies that we do not have adequate information on dyadic relationships with which to analyze parent-child dyads.
Measures of Stepkin Status and Transfers
We use two key measures in this article: the designation of stepchildren and stepparents and the definition of transfers. Stepchildren are identified by an explicit question about the relationship between the PSID household head and spouse/partner, and each adult child listed in the 2013 family roster. We treat a household as having a stepchild if any of the adult children on the roster is identified as a stepchild of either the head or the spouse/partner. Stepparents are identified by a set of questions on whether the biological/adoptive mother and father of the head and of the spouse/partner are currently married to each other and whether each parent is currently married to someone else or living with a romantic partner. By our definition, a household has a stepparent if the household head’s or spouse/partner’s biological/adoptive mother and father are not married to each other and at least one of his or her parents is currently married to someone else or living with a different romantic partner. When identifying both stepchild and stepparent relationships, we include those that arise through marriage or cohabitation; stepchildren may be the biological children of a current romantic partner, and stepparents may be the partners of biological parents.
We distinguish between whether it is the husband or wife/partner who has a stepparent or stepchild in our analyses of transfers for married/cohabiting couples. When we consider transfers to/from parents, we include two indicator variables about stepparents: the first is equal to 1 if the husband has a stepparent, and the second is equal to 1 if the wife/partner has a stepparent. When we consider transfers to/from adult children, we separate married/cohabiting couples into three mutually exclusive and exhaustive categories: (1) all children are the biological/adopted children of both the husband and wife/partner; (2) the wife/partner has at least one stepchild; and (3) the children are either all biological children of the wife/partner only (i.e., the children are all stepchildren of the husband) or a combination of joint children of the husband and wife/partner and biological children of the wife/partner only (stepchildren of the husband). This latter category includes all so-called blended families in which all of the stepchildren are the husband’s. These classifications allow us to examine gender differences in ties with stepkin.
The data exclude the stepchildren of respondents who do not have a spouse or partner and the stepparents of respondents without living biological or adoptive parents, who are sometimes called former stepchildren and former stepparents, respectively. This PSID data restriction implies that our estimates of the prevalence of stepkin are underestimates of the existence of stepchildren and stepparents acquired through prior unions.8 Nevertheless, by using reports about step-relationships through current unions, we rely on data from responses elicited by unambiguous questions and limit the extent to which only stepkin who are more connected to the family are enumerated.
We analyze the incidence of financial transfers that households give to and receive from parents (adult children) and of time transfers that households give to and receive from parents (adult children). We examine the transfers with all parents or all children combined, not transfers between parent-child dyads. That is, a household is considered to make a transfer to parents if they make a transfer to the parents of the head or the parents of the spouse/partner. Similarly, a household is considered to make a transfer to children if they make a transfer to any adult child. As noted earlier, all transfers to and from husbands and spouses/partners are combined because of the wording of the survey questions; that is, when either a household head or spouse/partner gives or receives a transfer to a parent (adult child), the household is considered to have engaged in a transfer.
Characteristics of Households, Parents, and Adult Children
In addition to the biological/step-relationship and transfer variables, we also construct measures of a range of characteristics of households, parents, and offspring that we incorporate in multivariate analyses (see the upcoming Methods section). Characteristics of the head/spouse and their household come from the 2013 core family and individual files and the 2013 R & T. Parent and parent-in-law characteristics come from the 2013 R & T, with the exception of parents’ education, which we obtain from the 2013 core family file. All characteristics of adult children come from the 2013 R & T. The rich array of characteristics of both parents and adult children is another advantage of the data we use.
We conduct our analysis in three stages. In the first, we describe the stepfamily structure of U.S. households and their demographic characteristics. To our knowledge, the PSID is the only data set that can provide contemporary, population-representative estimates of the availability of parents and adult children for U.S. households, including stepkin, and intergenerational transfers across all adult ages regardless of whether parents and offspring coreside.9 In the second stage, we analyze the financial and time transfers between generations and how they differ by stepfamily structure. Finally, in the third stage, we address whether the increase in kin due to stepfamily ties can offset any reductions in the incidence of transfers among households with stepkin. We describe the methods we use for each of these three stages in turn.
Structure of Stepfamilies
In the first stage of our analysis, we examine the distribution of biological and step-relationships for all U.S. households, both among the parent generation and among adult children. We also examine the distribution of these relationships among the subpopulations for whom step-relationships are possible. A household can have an adult stepchild only if a spouse or partner is present in the household and the head or spouse/partner has at least one adult child. Similarly, a household can have a stepparent only if the head (or spouse/partner, if present) has at least one living parent. Finally, for the households for which step-relationships are possible, we calculate the average increase in the size of the family as a result of step-relationships. We use PSID family weights in this analysis to produce estimates that are representative of U.S. households.
We distinguish families headed by someone younger than 55 years and families headed by someone age 55 or older; when considering subpopulations at risk, we distinguish between married/cohabiting couples and single household heads. We describe statistically significant age differences in the prevalence of intergenerational stepfamily structures based on t tests.
Transfers of Time and Money
In the second stage of our analysis, we examine the relationship between having a stepfamily member and transfers between parents and offspring. We present results based on two sets of regressions in which we first estimate the relationship between transfers with parents and whether at least one step-relationship exists with the parent generation. Next, we estimate the relationship between transfers with adult children and whether at least one step-relationship exists with the child generation. We distinguish among types of step-relationships—that is, whether the husband or the wife has stepparents or stepchildren.
In analyses of the relationship between transfers and stepfamily structure, we use regression methods to control for family characteristics that characterize the capacity to give transfers and need to receive transfers of each generation. The controls allow us to compare the incidence of transfers among families with similar characteristics—including a similar need for transfers and a similar capacity to give transfers—but who differ in whether they have stepkin ties. The method exploits the 2013 R & T data, which contain characteristics of both adult children and parents, allowing us to control for the potential recipient’s need for transfers and the potential donor’s capacity to help in terms of their financial and time resources. Insofar as the control variables hold family characteristics constant, our approach treats transfers of time and money as indicators of the willingness to help one another, conditional on the needs and capacities of each generation. We compare our main results, which include the full set of control variables, with those for the same models with controls only for family size and whether a step-relationship exists, the two primary variables of interest. Our conclusions are generally the same, as described later (see Table S1 in the online appendix).
The vector of characteristics of the head/spouse and their household included in the regression analyses includes marital status of the head/spouse, an indicator of whether the head or spouse is in poor health, the average age of head and spouse, the average years of education of head and spouse, whether the head is black, an indicator for homeownership, indicators of whether the head or spouse works and whether either is unemployed, total family income, the number of children under age 18 living in the household, the number of siblings, and whether the head or spouse has a sister. We also control for whether a parent is in the household in the analyses of transfers to/from parents and for whether an adult child is in the household in the analyses of transfers to/from children. Including coresidence does not alter our substantive conclusions.10 As an example of the role of the control variables, in the regression examining transfers of time to a household from their adult children, controls for marital status and health attempt to hold fixed the need for transfers. Similarly, in the regression examining financial transfers from a household to their adult children, family income and homeownership would control for the capacity to give transfers.
The vector of parent/in-law characteristics in the regressions includes the average age of parents; indicators of whether at least one parent is in poor health, retired, unemployed, or working; whether at least one parent has low income (<25,000), high income (>75,000), and missing income information; the average education of all parents; the total number of parents (including in-laws); and whether at least one parent is unmarried/unpartnered.
Finally, the vector of child characteristics consists of the number of adult children; the average age of adult children; the average education of adult children; and indicators for whether at least one adult child is a student, is unemployed, has low income (<25,000), has high income (>75,000), has missing income information, owns a home, is married, and has children of their own (grandchildren of the head/spouse).
The transfer analyses estimate stepfamily associations for the full sample and then stratified by age of the household head (younger than 55 years, and 55 years and older), as in the analyses of stepfamily structure. Because of the inclusion of many demographic and economic controls, our sample sizes are slightly smaller (by approximately 6 %) in the multivariate models because of missing covariates. When possible, we include indicator variables when a covariate is missing for an observation rather than deleting that observation. All multivariate results use unweighted data.
Does Having More Parents (Children) Due to Stepkin Compensate for Lower Rates of Transfers in Stepfamilies?
respectively, where and are the estimates of and from Eqs. (1) and (3), which characterize the effect of having stepparents (stepchildren) on the (adjusted) probability of a type m transfer, Tm; and are estimates of and from these same two equations and are the marginal effects of an extra parent (adult child) on the likelihood of such transfers. The latter marginal effects are multiplied by the average number of stepparents in households with living parents, and the average number of stepchildren in households with adult children, respectively. and provide quantitative estimates of whether the sheer number of step-relationships within families offset the effect of the presence in families of stepparents and stepchildren, respectively, on the transfers between generations.
We present two sets of estimates of and . One set is derived using estimates of , and from regressions that do not include our sets of control variables that attempt to account for the need for and capacity to give transfers. As such, they provide unadjusted measures of the net influence of these two forces on transfers that arise from the presence of stepkin. The other set of estimates of and use coefficient estimates for the regression specification that include controls and thus estimate the combined effects of stepkin on transfers, adjusting for differences in needs and capacities across families.
Despite its many strengths, the 2013 R & T design has at least two limitations relevant to an analysis of the influence of stepkin on transfers between generations of families. First, the design does not allow us to identify the source of age differences in stepfamily structure or transfers. That is, our estimated effects do not distinguish between differences due to age-related phenomena, differences in the historical periods that household heads and spouses experienced, and differences in their birth cohorts (e.g., differences between Baby Boomers or Millennials). In short, we cannot differentiate among age, period, and cohort effects or their interactions.
Second, the 2013 R & T includes information about children aged 18 and older but not younger children. The module was designed to focus on relationships between parents and all their adult children because such relationships are typically not covered in surveys that cover only members of the same households (i.e., members who live together). Our analyses based on the transfers between parents and their adult children complement past research focusing on minor children living with their parents. We do control for the number of minor children in the household in the regression-adjusted analyses.
Finally, we do not claim that the estimated coefficients measuring the effects of the presence of stepkin and numbers of kin on transfers in Eqs. (1)–(4), and thus our estimates of the combined consequences of the incidence and numbers of stepkin on transfers in Eq. (5), are causal, even after we control for observable characteristics that attempt to account for differences in needs and capacities of family members. We may have omitted dimensions of these needs and capacities that compromise our ability to fully account for their influence. In addition, the presence of stepkin in families and their numbers may themselves be partly endogenously determined by such omitted factors. Although one can consider employing methods to account for these potential sources of bias (e.g., instrumental variable methods) to isolate causal effects, these methods are not without considerable controversy, especially as they relate to identifying causal effects of the presence of stepkin on intergenerational transfers. This is especially true for identifying the influence of particular reasons for the presence of stepkin, such as divorce. Here, we focus on measuring associations between stepkin and transfers, leaving to future work the development of credible strategies to identify causal effects of stepkin on transfers. Measurement of these associations is a fruitful step in assessing the potential influence of stepkin on across-generation flows of transfers. In the Summary and Discussion section, we return to this issue to discuss the possible consequences of these omitted factors and/or sources of endogeneity for interpreting our results.
Stepparents and Stepchildren in U.S. Families
We begin by describing the availability of parents and adult children for PSID households and the presence of stepkin in both the parent and child generation, regardless of whether the parents and adult children coreside.
Table 1 shows the distribution of step- and biological kin for all U.S. households in the parent, the adult child, and both generations for the full sample and separately for younger and older households. The percentages show the availability of kin in the broad sense, without conditioning on survival (e.g., of parents for older households) to indicate who has the potential to engage in intergenerational transfers. The presence of stepkin is very common in both the parent and the child generation. Overall, 20 % of households have at least one stepparent, 47 % have only biological parents and parents-in-law, and 33 % do not have any living parents or parents-in-law. The prevalence of stepparents is more common among younger households (32 %) than among older ones (4 %). Stepkin also are very common among adult children. Eleven percent of households have at least one adult stepchild. Stepchildren are more common among older households (16 %) than among younger households (7 %). Finally, combining parents and adult children, 29 % of U.S. households have at least one stepparent or adult stepchild, 37 % of younger households and 19 % among older households. All age differences are statistically significant at the 5 % level.
Age differences in the presence of stepkin are partially obscured by the fact that older households may not have living parents and younger households may not yet have any adult children. To address this, we present in Table 2 estimates of the prevalence of stepparents and stepchildren among households that currently have living parents and/or adult children. We examine households with at least one living parent or parent-in-law and, among married/cohabiting couples, households with at least one adult child. We separate single household heads and married/cohabiting couples to examine stepparent ties. Single household heads cannot have stepchildren in the survey, so their counts are excluded in Table 2.
As shown in Table 2, conditional on having at least one living parent, 27 % of single-headed households and 32 % of married/cohabiting couples have at least one stepparent. Having stepparents is much more common among younger households, with 30 % of single household heads and 40 % of married/cohabiting couples having at least one stepparent compared with 8 % and 14 %, respectively, among older households. Conditional on having at least one adult child, 37 % of married/cohabiting couples have at least one stepchild. Although having adult stepchildren is more common among younger households (46 %), it is also very common among older ones (33 %). When a household’s parent and adult children are considered together, 18 % of single household heads and 52 % of married/cohabiting couples with at least one living parent or parent-in-law and at least one adult child have stepkin in one or more generations.
Not only are step-relationships among U.S. households highly prevalent, they also add considerably to the size of families. Table 3 shows the average number of biological parents and stepparents among single household heads and married/cohabiting couples with living parents or in-laws and the average number of biological children and stepchildren among married/cohabiting couples with adult children. (This table excludes counts of adult stepchildren of single household heads for the same reason stepchildren of single heads are excluded from Table 2.) Among married/cohabiting couples with at least one living parent, the presence of stepparents is associated with a 17 % increase in the total number of parents; the corresponding increase for single-headed households is 20 %. The presence of stepparents increases the number of parents by 19 % among young married/cohabiting couples compared with a smaller increase of 10 % among older married/cohabiting couples. Among younger, single household heads, stepparents increase the number of parents by 22 % versus only 7 % among older single household heads. All age differences are statistically significant.
The presence of stepchildren adds substantially to the total number of adult children of the head of a typical U.S. household. Among married/cohabiting couples with at least one adult child, stepchildren increase the number of adult children by 66 %—an addition, on average, of more than one child per household (1.07). As with stepparents, the inclusion of adult stepchildren constitutes a greater percentage increase of adult children among younger households (85 %) than among older households (60 %).11
Taken together, stepparents and stepchildren increase the total number of parent and adult child kin by nearly 40 % for married/cohabiting couples with living parents and children, and by 7 % for single household heads with living parents and adult children. Put differently, the demographic events of remarriage and repartnering significantly increase the availability of kin for both younger and older households.
Transfers Between Generations
The availability of parents and adult children, including those that arise from step-relationships, has the potential to change the incidence and nature of exchanges between generations in families. This section reports differences in the occurrence of actual transfers of time and money for those with and without step-relationships. We first describe transfers between households and their parents and then between households and their adult children. We find that the presence of step-relationships in U.S. households is associated with a reduction in the likelihood of material transfers within families. Furthermore, we show that this association between material exchanges and the presence of step-relationships differs by age, marital status, and gender (that is, whether the husband or wife has stepkin).
Transfers to Parents
Table 4 presents estimates of the net associations between having a stepparent and the incidence of each type of transfer, after controlling for household and parent characteristics; these estimates are presented for the whole sample, the sample of married/cohabiting couples, and the sample of single household heads combined, respectively, as well as for single household male and single household female heads. These associations are the estimates of the coefficients in Eqs. (1) and (2).
Having at least one stepparent does not significantly change the likelihood of a household giving money to or receiving money from their parents or receiving time from parents. However, having at least one stepparent reduces the likelihood of providing time transfers to parents by 4.32 percentage points.
Among married/cohabiting couples, having at least one stepparent does not produce statistically significant differences in the likelihood of engaging in transfers compared with not having a stepparent. However, when we examine separately whether wives and husbands have stepparents, we find that the likelihood of a time transfer is much lower when the wife has a stepparent than when the husband has one. Among married/cohabiting couples, the likelihood of providing a time transfer to any parent is 6.44 percentage points lower when the wife has a stepparent compared with couples with no stepparent, but we do not find a statistically significant difference when the husband has a stepparent. Moreover, this gender difference is statistically significant.
Time transfers to parents also are less likely in the presence of stepparents among single household heads. Overall, the likelihood of providing time to parents declines by 8.99 percentage points when single household heads have a stepparent. Among single household heads, the differences between single men and women in the association between stepparent relationships and time transfers are quite small; both single men and single women are less likely to provide time transfers to parents when they have at least one stepparent.
Table 5 presents estimates of regression-adjusted associations between the presence of stepparents and transfers separately for younger and older households. Table 5 has the same structure as Table 4. The results for the younger households are nearly identical to those in Table 4 for the full sample. An exception is that among younger, single household heads, having a stepparent also is associated with a 3.15 percentage point lower likelihood of receiving time transfers from parents. Among older households, having a stepparent is associated with a 12.10 percentage point lower likelihood of providing time transfers to parents; among younger households, the corresponding reduction is 3.80 percentage points. Among older households, having a stepparent also is associated with an 8.43 percentage point lower likelihood of receiving money from parents; among younger households, the presence of stepparents is not associated with the likelihood of receiving money.
With respect to differences in which member of a couple has a stepparent on the likelihood of transfers to parents displayed in Table 5, the patterns for younger couples are very similar to those for the overall sample. Among older couples, those in which the husbands, rather than wives, have a stepparent have a significantly lower likelihood of spending time with parents, although the association when the wife has a stepparent is still negative. However, the differences in the associations between transfers and which member of a couple has a stepparent are not statistically different from each other for older couples.
Transfers to Adult Children
Table 6 presents the net associations between having an adult stepchild and the incidence of each type of transfer: that is, , k = 1, . . . , 3 coefficients estimated from Eqs. (3) and (4), for the sample of married/cohabiting couples.12 The first panel shows the results for the presence of any stepchildren (Eq. (3)), and the second panel shows results that distinguish which member of the couple has stepchildren (Eq. (4)).
Having at least one stepchild is associated with a lower likelihood of making time transfers to or from adult children and receiving money transfers from adult children. We find no stepfamily disadvantage, however, in monetary transfers to adult children. Married/cohabiting couples who have at least one adult stepchild have a 3.69 percentage point lower likelihood of receiving money transfers from children, a 11.30 percentage point lower likelihood of giving time transfers to children, and a 13.30 percentage point lower likelihood of receiving time transfers from children compared with married/cohabiting couples who do not have stepchildren. All of these differences are statistically significant.
The estimates from Eq. (4) in the bottom panel of Table 6 indicate that the size and statistical significance of the negative association between couples having a stepchild and transfers with adult children depends on which couple member has stepkin. For all transfers but money to children, couples in which wives have adult stepchildren are less likely to engage in transfers compared with couples with only joint children, and all of these differences are statistically significant. Married couple households in which the wife has an adult stepchild are 4.43 percentage points less likely to receive money from their children, 14.10 percentage points less likely to provide a time transfer to children (such as caring for grandchildren), and 14.69 percentage points less likely to receive a time transfer from their children compared with couples in which all of their adult children are their own. Similarly, compared with couples in which all adult children are their biological or adopted children, couples in which all stepchildren are the husband’s (those that have only biological/adopted children of the wife or some such children in combination with joint children) are associated with lower likelihoods of all forms of transfers between parents and their children, although only the differences in time transfers to and from children are statistically significant (6.44 and 10.90 percentage points, respectively). Furthermore, the reduction in time transfers to children associated with having a stepchild are larger when the wife versus the husband has a stepchild, and these differences are statistically significant (see p values at the bottom of Table 6).
The gender differences in the results in Table 6 on the association between adult stepchildren and transfers mirror the results for the association between stepparents and transfers in the following sense. When married women have adult stepchildren, transfers between couples and their children are reduced by more than when married men have adult stepchildren, although both are reduced relative to having only joint biological/adopted children. Differences are larger for time transfers than for money transfers.
Table 7 contains estimates of stepfamily associations with transfers to and from adult children for younger and older households separately. We find a great deal of similarity in the signs, estimated magnitudes, and individual coefficient significance levels estimated separately for younger and older households compared with those in Table 6 that pool households of all ages. This is especially true for younger married/cohabiting couples, where the reduction in time transfers to children associated with having a stepchild are larger when the wife has a stepchild versus when all of the stepchildren are the husband’s stepchildren (see p values for the Age <55 panel in Table 7). In contrast, among older households, these same associations for the two configurations of stepchildren are not significantly different from each other (p values for Age 55+). That these more sizable and differentiated stepfamily effects hold only for younger families is consistent with the pattern of age differences in the stepparent effects for transfers to parents, especially time transfers.
To test whether our results are sensitive to the inclusion of covariates that control for the need for and capacity to give transfers, Table S1 in the online appendix shows the main results from Tables 4, 5, 6, and 7 with only the number of parents or adult children and the indicator variable for stepkin ties as covariates. The conclusions are similar. We think that including covariates for resources and needs is important to enable appropriate comparisons between families.
Does Having More Stepkin Compensate for the Lower Incidence of Transfers in Stepfamilies?
The results presented in Table 3 indicate that step-relationships—be they for parents or for adult children—increase the number of parents and adult children associated with U.S. households, with larger increases for younger households than for older ones. As the results presented in Tables 4, 5, 6, and 7 show, however, the presence of stepkin—both parents and adult children—is associated with reductions in the incidence of both time and money transfers, regardless of whether they flowed to or from these households. So, what is the net impact of stepkin on the incidence of transfers between generations? Do more stepkin compensate for the lower incidence of transfers in families with stepkin? Or does the presence of stepkin diminish the flow of transfers between generations despite adding additional kin to the family?
In Table 8, we present a series of estimates predicting the difference in the likelihood of transfers between those with and those without a stepparent (stepchild), after adjusting for the larger number of parents (children) in stepfamilies, using the formula in Eq. (5). Panel A shows the results with controls for the need for and capacity to give transfers, and panel B shows the results controlling for only the existence of stepkin ties and the number of adult children or parents. Consider, for example, the incidence of time transfers from married/cohabiting couples of all ages that have at least one adult child (leftmost cell in the bottom row of Table 8, panel A). Overall, married/cohabiting couples with an adult stepchild are 13.50 percentage points (–0.135) less likely to receive time from children than couples without any stepchildren. This net impact () is the sum of 1 = –0.133, the estimated effect of stepchildren on the probability of a time transfer from children (reported in Table 6), and , the estimated marginal effect of an extra adult child on the probability of receiving a time transfer (reported in Table S2, online appendix) weighted by the average number of stepchildren among married/cohabiting couples with at least one adult child (reported in Table 3). Table 8 presents estimates for these net impacts of stepparents and stepchildren on the likelihood of the various types of transfers for single household heads in addition to married/cohabitating couples and for younger (age <55) and older (age 55+) households. We use asterisks on the reported estimates to denote whether the net impacts we calculate are significantly significant.
The results in Table 8 indicate that transfers to and from parents and adult children are less likely for households with stepkin ties, even after adjusting for the larger family size of these households. Formal tests of the differences between the results in Table 8 (panel A) and the relevant estimates from results reported in Tables 4, 5, 6, and 7 show that only 5 of the 36 possible contrasts are statistically different (not shown but available upon request). That is, adjusting for the larger family sizes of households with stepkin does not alter the lower incidence of transfers in stepfamilies than families without stepkin. In addition, the direction of the difference between the estimates of stepfamily effects with and without adjustments for family size is inconsistent. Sometimes the associations that adjust for the increase in family size are larger than their counterparts in Tables 4, 5, 6, and 7, which hold family size constant, and other times the associations are smaller. This is because the coefficients on number of parents and number of adult children are not always positive. All else equal, having more parents or more adult children does not always increase the incidence of transfers. The coefficients for numbers of parents (adult children) are uniformly small and statistically significant in only one case. We report the coefficients in Table S2 of the online appendix for completeness.
In panel B of Table 8, we present results that correspond to those in panel A but use parameter estimates from Eqs. (1)–(4) without controls. For transfers with adult children, the reduced likelihood of transfers when stepchildren exist is reasonably similar regardless of whether controls are included. For transfers with parents, the point estimates in panel B tend to be larger and are more likely to be statistically significant than those in panel A.
Overall, the results in Table 8 show that the increase in the number of parents or children through stepkin does little to change the negative association between the likelihood of transfers and the presence of stepkin in U.S. households. As in Tables 4, 5, 6, and 7, the associations between stepkin and transfers, even after adjusting for the rather large increases in family size shown in Table 3, are nearly all negative, particularly among married/cohabiting couples with adult children.
Summary and Discussion
Using new data on the availability of kin, we show that stepparents and stepchildren are common in U.S. families. Some 20 % of households have at least one stepparent, more than 10 % have at least one adult stepchild, and nearly 30 % of households have a stepkin tie among either parents or adult children. Stepchildren increase the number of kin ties dramatically. Among married/partnered households, adult stepchildren increase the total number of adult children by two-thirds. We find age (or cohort) differences across households in how stepparents and stepchildren affect the availability of kin, with stepparents and stepchildren adding more kin ties to younger households than to older ones and adding more adult child ties than parent ties. These differences notwithstanding, step-relationships are common for both younger and older households.
Conditional on the availability of kin, we find that households with stepkin are less likely to engage in intergenerational transfers of time and money than households without stepkin. These differences are particularly large for time transfers. Households with stepparents are nearly 5 percentage points less likely to give time transfers to parents, and married/cohabiting couples with adult stepchildren are 11.30 percentage points less likely to give time to children and 13.30 percentage points less likely to receive time from children, even after controlling for a wide range of household and family characteristics associated with needs and capacities to transfer resources. Time transfers also are particularly sensitive to stepfamily composition, notably when the stepparent or child is a stepfamily member to the wife or female cohabiting partner. One interpretation of this finding draws on the idea that women are “kin keepers,” devoting more time to family caregiving and assistance than men do. When it is ambiguous who is in the family or when relationships are less emotionally close, as may occur in stepfamilies, women are less likely to provide this family assistance.
Our estimates that adjust for the increase in family size from stepkin suggest that the negative stepfamily-transfer association outweighs the larger size of kin network that results from the presence of stepfamily relationships in U.S. households, especially for transfers with adult children. Finally, our finding that the negative association between stepfamilies and intergenerational transfers dominates increases in the number of kin associated with greater numbers of stepfamily relationships in U.S. households answers Wachter’s (1997) speculation about whether stepkin would compensate for the declining availability of biological kin in the future.
Some limitations to our analyses point to directions for future research. As we noted in the Methods section, we do not attempt to estimate causal relationships, so the associations between stepkin and transfers that we report are descriptive in nature. Although we control for an extensive set of covariates to remove the effect of the need for and capacity to give transfers, omitted observable and unobservable characteristics are likely associated with both transfers and the existence of stepkin. For example, throughout we find that the reduction in the incidence of money transfers in the presence of stepkin is smaller than the reduction in the incidence of time transfers. One possible explanation may be that time transfers—more than monetary transfers—are associated with emotional closeness and that, on average, people are less emotionally close to their stepchildren and stepparents. In this sense, it is not having stepkin per se that reduces the incidence of transfers, but rather having kin with whom one is less emotionally close. We are unable to distinguish among motivations for transfers.
Our results also do not distinguish among potential reasons for why transfers are less likely in the presence of stepkin. Perhaps transfers are less likely because people are less emotionally close to their stepkin and are thus less likely to engage in exchanges of material support. Another possibility is that transfers are less likely in the presence of stepkin because divorce and repartnering weaken family ties between biological parents and children. For example, fathers’ disengagement from biological children after divorce may explain why we find that households with adult stepchildren are even less likely to engage in transfers when the adult children are the husband’s biological children from a past relationship (i.e., the wife’s stepchildren) than when the children are from the wife’s past relationship (the husband’s stepchildren).
But this legacy-of-divorce explanation is more difficult to apply to our findings of a gender difference in transfers between households and their parents. Again, we find that transfers are less likely when the wife has the step-relationship than when the husband does. We think it is unlikely that the gender difference is driven by differences between daughters and sons in the effects of divorce and repartnering on parent-child relationships. Although gender differences may exist in responses to the dissolution of parents’ unions and new partnerships, we think that the women’s roles as kin keepers and the gendered division of family labor are more likely explanations for our findings. The role of divorce in explaining weaker ties with stepkin and the reasons behind weaker ties in the presence of stepkin are extremely important questions that we leave to future work.
Finally, our finding of smaller gender differences in the relationship between stepkin and transfers for older than younger households may occur because of the increase in men’s involvement with family in their retirement years (Kahn et al. 2011) or the potentially longer duration of stepkin ties in older households. But because we cannot distinguish between age and cohort effects, the smaller gender differences in the stepkin-transfer association may also be due to cohort differences in norms.
More generally, our analysis highlights the need for further work examining the latent support network, or family safety net, and how this differs between families with and without stepkin. Vignettes suggest that both the norms of family obligation and relationship quality affect the willingness of family members to provide help to one another and that step-relationships affect both norms and the strength of ties (Ganong and Coleman 2006; Seltzer et al. 2012). More work using vignettes may further illuminate the motivations behind intergenerational transfers and how these motivations are affected by the presence of stepkin. The PSID design also provides another way to examine support from stepfamily members during an emergency, when the safety net may be activated. The design allows researchers to observe short-term financial and time support that families provide when someone in the family experiences an emergency, such as an unanticipated health crisis (Dalton and LaFave 2017), or the financial support or support through coresidence that family members provide during times of sudden economic hardship, which may differ for families with and without stepkin. Finally, new data on the parents of respondents in the National Survey of Adolescent to Adult Health (Add Health) embeds explicit questions about whom family members would rely on in an emergency in a module on intergenerational transfers. These new data will allow comparisons of latent connections between families with stepchildren and those with biological children.
Our research describes contemporary American families, but our findings have implications for the future. The weaker intergenerational connections between families with step-relationships combined with the greater likelihood that younger people have step-ties raises concerns about the availability of family support in future generations. The greater prevalence of stepfamily members points to the importance of understanding the factors that shape whether stepkin are considered to be among the available kin both in everyday life and in times of crisis, and which of these factors might help to mitigate the negative association between stepkin and transfers found in this study. Previous studies of attitudes about obligations suggest that stepparents who helped raise their stepchild may be more likely to receive assistance from that child later in life (Coleman et al. 2005). Furthermore, previous research has found that children were much more likely to live with stepfathers than stepmothers during childhood (Kreider 2008). Such findings suggest that in future generations, stepfathers may fare better than stepmothers in terms of receiving support in later life to the extent that ties with stepchildren have been established earlier in life (Seltzer and Bianchi 2013). More generally, unraveling how stepfamilies shape family connections sheds new light on how the changing structure of American families will affect the help that family members provide to each other in the future.
This article was presented at the 2015 annual meeting of the Population Association of America in San Diego, CA. We thank Shelly Lundberg for comments on that earlier draft. The data from the 2013 PSID Rosters and Transfers Module used in this article, as well as the analyses presented, were funded by NIA Grant P01 AG029409, and the construction of the weights was supported by the Alfred P. Sloan Foundation (Grant No. 2011-6-24). The project was also supported in part by the California Center for Population Research at UCLA (CCPR), which receives core support (P2C-HD041022) from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), and by the Duke Population Research Institute (DuPRI), which receives core support (P30AG034424) from the National Institute on Aging. We thank Sung Park and Joshua Rasmussen for their research assistance in preparing this article. Sadly, Suzanne Bianchi passed away before this article was published, but the article could not have been written without her contributions.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
See Hoerger et al. (1997) for an economic model in which the care of elderly relatives is a function of the pooled income of its family members. In theoretical models with income pooling, larger family size increases the capacity of families to help finance the care of elderly parents. This theoretical prediction of resource pooling applies, in principle, to the time that particular family members in one generation may devote to the care of those in the other generation.
For example, in noncooperative, game-theoretic economic models of adult children’s provision of care to parents, adult children view caregiving as costly, and siblings’ caregiving substitutes for one’s own provision of care. This creates a “free-rider” problem in which the presence of one or more siblings reduces the incentive to provide care to parents (Byrne et al. 2009; Checkovich and Stern 2002; Engers and Stern 2002; Hiedemann and Stern 1999).
An exception is Lin et al. (2017), which provided a national portrait of later-life stepfamilies in the United States.
We use the term spouse/partner to refer to what PSID calls wife or “wife,” which includes legal wives and cohabiting partners of at least one year. Heads in the PSID are men except in single female–headed households and households in which the PSID sample member is a woman who has been cohabiting with her partner for less than one year.
We use the term household to refer to what PSID calls a family unit, which consists of individuals who live together and are related by blood, marriage, or adoption, or who are not related but share income and expenses.
The low bound of $100 is much more likely to capture financial transfers in poor families than the higher bound of $500 currently used by the Health and Retirement Study (HRS), thereby enhancing our ability to compare transfers across households that differ in family structure and economic resources (McGarry and Schoeni 1995).
We retain in the sample all persons who report at least one child or parent record with a valid relationship code. We eliminate households whose head or spouse has children or parents with invalid relationship codes for every such relationship. For example, a head may report two children but identify their relationship as “other,” “don’t know,” or “refuse,” rather than “biological” or “adopted.” We also eliminate households in which the heads (and spouses, if present) report that they do not know, or refuse to answer, whether their biological parents and the biological parents of their spouse, if present, are living. For all heads and spouses with a valid report of whether at least one parent is living, we assume that parents about whom they do not know or refuse to answer are not living. For example, we code heads who report that their mother is dead and they do not know whether their father is alive as having no living parents.
Attitude data suggest that respondents are more likely to report about the existence of former stepparents when relationships are closer and more enduring than when ties with former stepkin are weaker (Coleman et al. 2005; Schmeeckle et al. 2006). How to improve the quality of data on step-relationships from previous unions is an important topic for new research.
We define adult ages as 18 and older, consistent with the definition of adult offspring used in the 2013 R & T. In practice, PSID heads and their spouses are almost universally older than 18 years. In 2013, three heads and one spouse were under age 18. We include these younger heads in our sample for completeness.
We thank reviewers for the suggestion to include a control for coresidence. Controlling for coresidence accounts for any differences in how respondents report about transfers with coresident kin versus noncoresident kin. Our substantive conclusions are not affected by controlling for coresidence. The estimates in Tables 4 and 5 on transfers with parents are nearly identical with and without controls for coresidence. The estimates in Tables 6 and 7 on the association between the presence of stepchildren and the incidence of transfers with adult children are approximately 10 % smaller when we control for coresidence.
These percentage increases are high because biological children of married couples include only the biological children of both the husband and the wife, and stepchildren include the stepchild of either the head or the wife. The increases also are high because offspring from previous unions are more likely to be age 18 or older than offspring of the current union.