A Double Multi-Actor Perspective on Adult Intergenerational Ties: The Role of Biological Relatedness Vis-à-Vis Shared Residence

Ruben van Gaalen, University of Amsterdam
Suzanne de Leeuw, University of Amsterdam
Maaike Hornstra, University of Amsterdam
Katya Ivanova, University of Amsterdam
Kirsten Van Houdt, University of Amsterdam
Matthijs Kalmijn, University of Amsterdam

We examine the effects of biological relatedness vis-à-vis shared residence on the strength of adult intergerenational ties using a new multi-actor survey where respondents raised in complex families were systematically oversampled. A unique feature of the survey is that it allows for a double multiactor design. We sampled children 25-45 years old, called ‘anchors’ (N = 6,485). All parent figures of these children – called ‘alters’ – were found via the registers and also participated in the study (N = 9,325). The anchor data make it possible to apply a multiple parent design where the children are the higher-level units (level 2) and the parents are the lower-level units (level 1). We asked the alters about the relationship they have to all their adult children, yielding a multiple child design, where parents are the level-2 units of analysis and adult children are he level-1 units. Using fixed-effects models, we then make direct comparisons within the level-2 units. In the multiple parent design, this implies making comparisons among different types of parents within adult children. For example, we compare how much contact a daughter has with her divorced father to how much contact she has with her stepfather. Implicitly, this approach controls for the influence of all unobservable child characteristics. In the multiple child design, the fixed-effects model implies that we compare different types of children within each parent. For example, we compare the closeness a father feels to his current biological children to the closeness he feels to his stepchildren. This approach controls for all unobservable parent characteristics. By combining the two within-family designs in one study, the goal is to get closer to the causal nature of the effects of biology vis-à-vis shared residence.

Abstract

Due to the rise in divorce and remarriage, an increasing number of children grow up with multiple ‘types’ of parents: resident biological parents, stepparents, non-resident biological parents, and the possible partners of these parents. Similarly, more and more parents have multiple types of children: children from a present union, children from a former union who do or do not share the household, and possibly stepchildren. These trends point to an increase in the complexity of parent-child configurations (Thomson 2014) and raise important and fundamental questions about the role of biology, shared residence, and marriage (Hofferth and Anderson 2003, Kalmijn 2013).

Past research on the relevance of biological relatedness for later parent-child relationships has often pointed the ‘primacy of biology.’ Ties between biological parents and children are believed to be more intimate, more supportive, more harmonious, and more stable than ties between stepparents and stepchildren. Unfortunately, such conclusions often depend on simple dichotomies between stepparents and biological parents (Daly and Wilson 1996). Moreover, past studies often compare two types of children or two types of parents in parallel samples (e.g., a group of stepparents vis-à-vis a group of biological parents), without sufficiently controlling for confounding factors. Effects of ‘biology’ have also been complicated by other sources of heterogeneity. A direct comparison between stepparents and married biological parents is ‘unfair’ to stepparents since the latter have not been able to invest as long in the child. To solve this, variation is needed in the duration of shared residence but this requires the inclusion of divorced parents. Such an analysis would be complicated, however, by the potentially disrupting effects of parental divorce, in particular on father-child ties.

We re-examine the effects of biological relatedness vis-à-vis shared residence on the strength of adult intergerenational ties using a new multi-actor survey where respondents raised in complex families were systematically oversampled via the population registers, the OKiN survey (Kalmijn et al. 2017). A unique feature of the OKiN is that it allows for a so-called double multiactor design. We started with a sample of children who were 25-45 years old, called the ‘anchors’ (N = 6,485). The parents of these children and the partners of these parents – called the ‘alters’ – were found via the registers and also participated in the study (N = 9,325). The anchor data make it possible to apply a multiple parent design where the children are the higher-level units (level 2) and the parents are the lower-level units (level 1). Each adult child can have multiple parents in the data and this number can vary from child to child. We also asked the alters about the relationship they have to all their adult children, regardless of the type of child. This yields a multiple child design where parents are the level-2 units of analysis and each parent can have a variable number of adult children (the level-1 units).

We use fixed-effects models to make direct comparisons within the level-2 units. In the multiple parent design, this implies making comparisons among different types of parents within adult children. For example, we compare how much contact a daughter has with her divorced father to how much contact she has with her stepfather. Implicitly, this approach controls for the influence of all unobservable child characteristics. In the multiple child design, the fixed-effects model implies that we compare different types of children within each parent. For example, we compare the closeness a father feels to his current biological children to the closeness he feels to the children from his previous marriage. This approach controls for all unobservable parent characteristics. By combining the two within-family designs in one study, the goal is to get closer to the causal nature of the effects of biology vis-à-vis shared residence. While studies have used within-family comparisons before (Arranz Becker et al. 2013, Evenhouse and Reilly 2004), no studies that we know of have applied a double multi-actor design.

We focus on three outcome variables: contact, closeness, and conflict. Contact is important because it is a form of social support and is a requirement for giving other forms of support (Silverstein et al. 2010). Closeness is an important subjective measure of the strength of parent-child ties (Pillemer et al. 2010). Conflict is important because positive and negative aspects of intergenerational relationships are partly orthogonal (Luescher 2002). On the independent side, we have extensive information on biological relatedness, the parents’ marital history, the child’s residential history when growing up, and parental involvement, also for non-resident parents. The table below shows which types of parental figures there are in the anchor data and which types of children there are in the alter data. The data collection has recently been completed and the codebook has been written.

Presented in Session 1126: Ageing and Intergenerational Relations