Identifying ties that bridge generations : a quantitative social network analysis on intergenerational ties between non-kin relations

(1)

IDENTIFYING TIES THAT

BRIDGE GENERATIONS

A quantitative social network analysis on

intergenerational ties between non-kin relations

University of Amsterdam

General Sociology

Master thesis

1st supervisor: prof. dr. Völker

2nd supervisor: prof. dr. Leopold

07-07-2019

Author: Anne-Floor Bakker

Student number: 10759905

Word count: 13.193

(2)

Anne-Floor Bakker

Abstract

In our society, the old and the young live most of their lives segregated from each other. Age-segregation is a threat for social cohesion as it creates opportunity for negative sentiments to develop between different age groups. Intergenerational contact can counter the negative effects of age-segregation on both individual and societal level. The goal of this study was to gain insight into the conditions needed for intergeneration non-kin ties to develop and to assess the descriptive characteristics of these relationships in general. We used the data from the Bruggen over Scheidslijnen research (2019) to conduct a social network analysis and an ordered logistic regression. Our sample included 1925 respondents, which consisted of two age-groups: older adults (65+ year of age) and younger adults (18 to 35 years of age). The results showed different outcomes for the older and younger adults. For the older adults, working in an age-heterogenous work environment and doing voluntary work significantly increased the odds of having intergenerational non-kin ties, whereas no effect was found for the younger adults. Both the older and younger adults had increased odds of having intergenerational non-kin ties when intergenerational family ties were present in their social network. Additionally, having more social capital also increased the chances of having intergenerational ties for both age groups.

(3)

Anne-Floor Bakker

The many everyday boundaries between the old and the young are referred to as “age segregation” (Hagestad & Uhlenberg, 2005; Uhlenberg, 2000). The term “segregation” indicates that the distributional differences with regard to age are not merely a natural social occurrence but also a rather serious societal issue and a threat to social cohesion. Similar to averting other types of segregation (e.g., race or sex), substantial intergroup contact is important to prevent the formation of negative prejudices and stereotyping and thereby creates an opportunity for positive interactions and social capital exchange (Flap, 2004, p.13). Institutionalised age barriers (un)intentionally seem to discourage intergenerational contact, creating a potential breeding ground for mutual negative sentiments (Uhlenberg, 2000; Dykstra & Fleischmann, 2016). Feeding into this negativity, according to Lepianka (2015), is the media. She found that the media generally portrays both young and older adults in a negative manner. The media often depicts younger adults as being immoral and often relates them with adverse topics such as crime and social decay. In contrast, older adults are often characterised as dependent, incapable and in poor health (Lepianka, 2015, p.1097). Studies have shown that these negative attitudes can be internalised, especially as people age, creating negative consequences for health and self-conception (Harwood et al., 2005, p.395). Thus, by systematically separating different age groups and framing aging in an unfavourable way, negative sentiments such as prejudice and stereotypes can arise and eventually jeopardise solidarity and social cohesion within a society.

Social cohesion is important to ensure a safe and pleasant environment for all members of society. Forrest & Kearns (2001) define the following five domains of social cohesion: shared common values and civic culture; social order and social control (e.g., mutual respect and intergroup co-operation); social solidarity and reductions in wealth disparities (e.g., willingness to help each other); social networks and social capital (e.g., high levels of intergroup contact; collective action); place attachment and identity (pp.2128–2129). A society with low social cohesion would be mainly characterised by social disorder and conflict, as well as extreme social inequality and low levels of social interaction between and within communities (Forrest & Kearns, 2001, p.2128). Social cohesion is also an important factor in establishing intergenerational solidarity, which can be defined as the context of shared expectations and obligations regarding the ageing of individuals and the succession of generations (Bengtson & Oyama, 2007, p.3). A growing ageing population has produced concerns about the financial sustainability of the public welfare systems growing imbalances in the labour force. With too few younger cohorts to support (e.g., pension, healthcare) an

(5)

Anne-Floor Bakker

ever-increasing population of older people, research and policies have focused mainly on the economic dimension of intergenerational solidarity. However, it is important to note that intergenerational contact and solidarity cannot be reduced to a merely financial dimension. The ethical dimension—the willingness to help others, regardless of their contribution to society—is also an important component in the intergenerational contract. Intergenerational solidarity is important for an successful and sustainable social welfare state (Buffel et al., 2014, p.1786).

When discussing the everyday boundaries between age cohorts, the opportunity structure embedded within the social context is important to determine the chances of cross-age interaction (Riley & Riley, 2000, p.267). As many sociologists argue, the way our current society is structured seems to significantly limit the opportunity for individuals of different age cohorts to meet and interact with each other. Since the industrial revolution, Western societies have been organised by the notion of the tripartite life course, which chronologically divides life into three major phases: youth, in which we are educated; adulthood in which we join the labour force and start a family; and old age, which we spend in leisure (Hagestad & Uhlenberg, 2005, pp.346-347). Hagestad and Uhlenberg (2006) claim that the tripartite life course limits the age variety of the social context in which people live, work and socialise, thereby reducing the structural meeting chance between age groups (p.644). In a similar vein, Dykstra and Fleischmann (2016) found that the chance of intergenerational friendships between younger and older adults were indeed much higher when daily activities took place within environments that enabled regular intergenerational contact, such as working with elderly people (both paid and voluntarily) or attending religious meetings (pp.121–122). Noteworthy is that these contexts, which supposedly enable intergenerational non-kin contact, are rather specific and apply only to a small portion of the population.

There is however one social structure that successfully facilitates intergenerational contact like no other: the family. Multiple studies have found that positive family ties across generations have a mitigating effect on age-cohort segregation and ageism. These family ties can reduce negative stereotyping and prejudice, as well as prevent social isolation and institutionalisation (e.g., care facilities) in old age (Harwood et al., 2005; Hagestad & Uhlenberg, 2006). Because of the evident position the family has in facilitating intergenerational contact, this study will deliberately focus on contact between people of different generations who are not related to each other (i.e., non-kin). However, earlier studies have shown that having multigenerational family ties could increase the chances of having intergenerational ties with people outside of the family network as well. Dykstra and

Met opmaak: Engels (Groot-Brittannië)

(6)

Anne-Floor Bakker

Fleischmann (2016) argue that having strong and positive ties with family members much younger or older than themselves can cause people to be more positive and open-minded towards members of other age groups (Dykstra & Fleischmann, 2016, p.110). Consequently, this study will inquire into the role of intergenerational family ties as an indicator for having intergenerational non-kin ties.

When studying personal networks, the concept homophily is used to describe the phenomenon that people are more inclined to develop new ties to others who are similar to them in terms of, for example, age, sex, ethnicity, religion or social economic status (Lazarsfeld & Merton, 1954; Dykstra & Fleischmann, 2016). However, according to Granovetter (1973), there is much to gain from having connections to individuals who are less similar to oneself. Such individuals often possess different types of connections and resources because they roam in other social networks (p.1378). For example, a network member who is a retired lawyer can offer sound legal advice and might be able to help with writing an application letter; however, he or she might not be the right person to help fix one’s computer or to give advice on pursuing a job at a local gym. It pays to know people in different social spheres as this broadens the resources to which one has access. Bourdieu (1986) introduced the term “social capital” for this accumulation of actual or potential resources embedded in one’s social network (p.47). The volume of social capital acquired by a given actor depends on the number of network connections he or she can mobilise and the amount of capital (economic, cultural or symbolic) belonging to his network members. The network provides each member potential access to resources belonging to other members of the collective, and these potential profits form the basis of the solidarity which makes them possible (Bourdieu, 1986, p.51). Having ties to dissimilar others can connect two otherwise unrelated social networks to one another, thereby increasing the social capital of each network’s members (Granovetter, 1973, p.1378). Intergenerational ties can be seen as an example of ties to dissimilar others, making them a viable medium for social capital exchange.

Not much is currently known about the conditions necessary for intergeneration non-kin ties to develop (Riley & Riley, 2000, p.268). Intergenerational ties are important to study because networks can offer insight into the conditions necessary for intergenerational solidarity, as well as how they integrate people within society (Uhlenberg & Jong Gierveld, 2004; Dykstra & Fleischmann, 2016). Through quantitative social network analysis, we hope to gain insight into these conditions. The focus of this study is on two age groups at opposite ends of the age-spectrum: older adults 65 years of age or older and younger adults between the ages of 18 and 35. The central research question in study is: Under what conditions do

(7)

Anne-Floor Bakker

members of different generations have non-kin network ties with each other and what do these relationships look like? Does the theoretical notion of structural meeting chances and social capital explain these contacts?

2. Theoretical framework

This chapter will introduce, explain, and connect the theories central to this study and provide an overview of the relevant literature. It will also demonstrate the limits of this body of knowledge and the need for additional research. This study aims to contribute to knowledge on the prevalence and patterns of intergenerational contacts between non-kin.

The theoretical part is structured along two arguments: theories based on opportunity structures, including the role of family ties and theories relating to social capital. Based on the arguments presented in this theoretical framework, the related hypothesis will be introduced accordingly. This chapter will end with a schematic overview of the primary arguments and the related hypotheses. The following chapter will discuss the methods used to confirm or dismiss our hypothesis and hopefully answer the research question.

2.1. Opportunity structures

The saying “no mating without meeting” is central to studying interpersonal relationships (Dykstra & Fleischmann, 2016; Verbrugge, 1977). Sociologist argue that the social structure of society determines the opportunity for certain people to meet and develop relationships. People spend most of their time within their own age enclaves, sub-cultures, and age-related activities and this greatly reduces the opportunity for age-outgroup members to meet (Dykstra & Fleischmann, 2016; Blau & Schwartz, 1982). Within age-homogenous social spaces, the chance of developing relationships with others similar in age is much greater, while opportunities for intergenerational contact are rare (Dykstra & Fleischmann, 2016, p.108). Systematic segregation of certain societal groups from one another can create a breeding ground for negative sentiments and the deterioration of social cohesion (Uhlenberg, 2000; Dykstra & Fleischmann, 2016).

As previously mentioned, Hagestad and Uhlenberg (2005) believe that society is currently structured by the tripartite life course, which categorises people according to their age within three major chronological life phases. During the first phase “youth”, individuals together with their peers mainly take part in compulsory education that prepares them to become productive members of society. In second phase, “adulthood”, individuals are

(8)

Anne-Floor Bakker

generally expected to participate in the labour force, while simultaneously starting a family of their own. The third and final phase is dedicated to older adults, who are generally no longer active in the work force and enjoying this time of leisure. Individuals are assigned age-linked rights and obligations that are enforced by means of laws and institutional policies, which can be seen as formalisations of the differences between each group (Hagestad & Uhlenberg, 2005, pp.344-345). Clear examples of formalisations between age groups include laws making primary and secondary education compulsory for children, as well as the legally established minimum age for receiving retirement benefits. In short, the tripartite life course influences the socio-demographic composition of the context in which people live, work and socialise, and therefore comprise the opportunity structure to meet particular others (Mollenhorst, Volker & Flag, 2008, p.938). Currently, “spaces where young, middle-aged and older people from all walks of life can get to know each other enough to build mutual respect and develop cooperative relationships” are extremely rare (Braithwaite, 2002, p.332). Dykstra and Fleischmann’s (2016) recent study confirms that individuals who encounter people belonging to different age groups on a daily basis are more likely to have intergenerational friendships, provided that there is an opportunity for meaningful interaction (pp.121–122). As the opportunity to meet is key, we will examine the importance of different social contexts for forging age-heterogenous relationships.

Work is an institutional setting in which age is a central distributive principle, as the youngest and the elderly are generally excluded from the workforce. This means that adults in the workforce spend a large part of their day separated from both these age groups (Hagestad & Uhlenberg, 2006, p.641). However, when we disregard the two extremes (i.e., children and the elderly) the workplace does often provide the opportunity for people of different ages to meet. Studies show that older people have more opportunities to recruit network members of diverse ages when they are active in social contexts, such as workplaces that include younger adults. Dykstra and Fleischmann (2016) found that for both younger and older adults the odds of having intergenerational relationships rose when they worked with people much younger or older than themselves. However, being active in the workforce in itself was not enough to affect these odds (pp.121–122). Furthermore, because amount of time people spend in their workplace is substantial, there is less time available to spend in other social contexts. This can lead to limited opportunities to engage in social interaction, increasing the chances that people will engage in connections with colleagues less similar to themselves with respect to characteristics such as ethnicity, gender or age (Mollenhorst, Volker & Flag, 2008, p.943). It would be interesting to see whether being active in the workforce affects the chances of

(9)

Anne-Floor Bakker

having intergenerational non-kin ties for different age groups. The sample of this study is divided between two age groups who are generally underrepresented in the workforce, with the older adults in retirement and some of the younger adults still attending school. That being said, a substantial proportion of the younger adults will be entering into the workforce regardless. However, the relevance of the workplace for the attainment of intergenerational ties should not be underestimated for older adults, as workplace born friendships forged years earlier might have withstood the test of time. Based on these arguments the following hypothesis was formulated: individuals who are active in the workforce (=a) and work within a heterogenous organization (=b) are more likely to have age-heterogeneous ties than those who are not active in the workforce. The expected effect is stronger for individuals working within a heterogenous organization (work tie hypothesis).

Similarly, multiple studies found that participating in voluntary work also significantly increased the likelihood of having age-heterogenous friendships for both the older and the younger adults, as these contexts enable the recruitment of network members of diverse ages (Uhlenberg & Jong Gierveld, 2004; Dykstra & Fleischmann, 2016). This has led to the formulation of the following hypothesis: individuals who are active in a voluntary organisation are more likely to have age-heterogeneous ties than those who are not involved in voluntary organisations (voluntary tie hypothesis).

Studies have shown that the social context of the neighbourhood is especially important for older people and for youth. Although both groups spend large amounts of time close to home, older people and youth are less often likely to take part in the public sphere by being active in neighbourhood meetings or other formalised local initiatives (Buffel et al., 2014, p.1788). Keeping in mind the amount of time older people and youth spend in their neighbourhoods, it would be interesting to see whether the degree of urbanisation influences the chances of having intergenerational ties to non-kin. The degree of urbanisation of the living environment could be a factor influencing the likelihood of interacting with individuals of diverse ages, as in more urban living environments the population density is higher and therefore the likelihood of meeting others who are much older of younger than themselves would also be higher. The following hypothesis was formulated: individuals who live in more urbanised environments are more likely to have age-heterogeneous ties than those who live in less urbanised living environments (neighbourhood tie hypothesis).

(10)

Anne-Floor Bakker

2.2. Family ties across generations

The family seems to be the one social structure that truly facilitates substantial intergenerational contact (Uhlenberg, 2006, pp.649–650). In general, our societies are designed around the idea of the tripartite life course that organises its citizens chronologically into different structures based upon their age (Dykstra & Fleischmann, 2016, p.108). In contrast, the family structure is by definition based upon multiple generations and durable relationships (Harwood et al., 2005, p.395). Since the 1960s, there has been a significant decrease in de dominance of the nuclear family, with steep increases in the divorce rates and the number of single-parent and blended families growing (Bengtson, 2001, p.4). With the diversification of family structures over the last decades, Bengtson (2001) claims that family relationships across several generations are becoming increasingly important in our contemporary societies. He rejects the idea that the decline of the nuclear family equals a decline of the family in general (p.1). He argues that multigenerational family relations will become increasingly important, partly due to the longer years of shared lives between generations, as this creates more opportunities and needs for interaction and mutual influence. Due to the increasing heterogeneity of family forms, family relations now extend beyond purely biological links, meaning that just as one can be born into particular family bonds, one can also create family systems of their own (Bengtson, 2001, pp.1–2). Blended families are, in some cases, less able to provide stability, socialisation and support for their family members. Intergenerational family ties fulfil an increasingly important role in providing these essential family functions (Bengtson, 2001, p.5).

Intergroup contact is believed to be capable of changing attitudes towards outgroup members in general; however, decades of research suggest that mere contact is not enough to facilitate this change. It has been suggested that other conditions are necessary. For example, studies have shown that contact should be co-operative, in a close long-term relationship, include observations of shared values and be generally pleasant (Harwood et al., 2005, p.393). It has been suggested specifically that grandparent-grandchild relationships facilitate many of these conditions, as contact is often extensive, recurrent and plays out in a variety of contexts (Pettigrew, 1998, p.76). In line with these assumptions, some studies have found that positive grandparent-grandchild relationships result in positive attitudes towards that specific age-outgroup (Harwood et al., 2005, p.395). Similarly, Dykstra and Fleischman (2016) found that mutually warm, supportive and durable intergenerational family relationships increase the empathy people feel towards generational outgroups members. People who have strong multigenerational family relations therefore are more likely to develop positive

(11)

Anne-Floor Bakker

intergenerational ties outside of the family network (Dykstra & Fleischmann, 2016, p.110). Affirmatively, multiple studies have found that durable multigenerational family ties indeed serve to temper the negative effects of age-cohort segregation and ageism. Evidence shows that family relations are a key resource for knowledge about other age groups, creating mutual understanding and increasing empathy towards much older or much younger unknown others (Hagestad & Uhlenberg, 2006, p.649).

In summary, people who have a more positive attitude towards (kin) others who are much older or much younger than themselves are more likely to have intergenerational non-kin relations to others (Dykstra & Fleischmann, 2016; Uhlenberg & Hagestad, 2006; Harwood et al., 2005). Although direction of causality has not been determined, there seems to be a clear association for both older and younger adults. There are a number of possible explanations for this presumable interaction between having intergenerational family ties and having intergeneration ties outside the family network. First of all, as described earlier, the preference for a certain type of other could be based upon positive past experiences with family members of similar ages, making intergenerational ties between non-kin more common among those who have strong and positive intergenerational family ties as well. A second explanation stems from the opportunity that close intergenerational family ties create: when one has frequent contact with family members who are much older or younger than themselves, this could increase the chances of meeting this family member’s friends and acquaintances too, creating the opportunity to meet and forge intergenerational relationships outside of the family network. In this case the positivity of the relationship would not matter as much; merely having these ties would be sufficient, assuming the individuals meet on a regular basis.

These arguments have led to formulating the following hypothesis: individuals who have ties (=a) and have positive ties (=b) to family members who are much younger or much older than themselves are more likely to have age-heterogeneous relationships than those who do not have such ties. The expected effect is stronger for individuals who have positive family ties (family tie hypothesis).

2.3. Social capital

As discussed before, multiple explanations have been offered as to why people often have ties to others more similar to themselves (i.e., homophily, personal preferences and opportunity structures). However, relationships between dissimilar others (e.g., intergenerational ties), have long been overlooked in the literature. Not much is known about the motivations of the

(12)

Anne-Floor Bakker

actors and conditions necessary for these ties to develop (Dykstra & Fleischmann, 2016, pp.107-108). A possible motivation to develop and maintain ties to dissimilar others could be to expand one’s social capital.

According to Bourdieu (1985) social capital is defined as the cumulation of actual or potential resources embedded in an actor’s social network as a means for that actor to improve his or her living conditions (Bourdieu, 1985, p.48). Social networks therefore provide the necessary condition to access and use these embedded resources (i.e., social capital), which would otherwise be out of grasp (Lin, 2008, p.58). The volume of social capital acquired by a given actor depends on the size and strength of network connections he or she can mobilise and the amount of capital (economic, cultural or symbolic) belonging to his or her network members (Bourdieu, 1985; Flap, 2004). According to Granovetter (1973), ties to dissimilar others are particularly valuable, as individuals who are less similar to each other often possess different types of connections and resources because they roam in other social networks. It pays to know people from different social spheres to broaden the resources one has access to (p.1378). Benefits derived from these ties can occur in both material and symbolic form: a material profit being a service such as a job opportunity. A symbolic profit would be being associated with a prestigious social group (Bourdieu, 1985, p.52).

A growing body of knowledge supports the claim that social capital has a salutary effect on individuals as well as communities. It has been suggested that social capital can improve our health and make us wealthier and wiser (Putnam, 2000, p.287). Social capital, in general, provides people with the tools to transform aspirations into realities. Putnam (2000) argues that high levels of trust and citizen participation are fundamental forces that operate through a variety of mechanisms to produce socially desirable outcomes. These mechanisms are structured into three main arguments. First of all, social capital helps to resolve problems of collective action, as individuals trust each other to pay their dues for the public interest, knowing that by cooperating they all will be better off. Second, the trust underlying social capital eases interactions and transactions in everyday life. Assuming that fellow citizens have honourable intentions makes these affairs more pleasant and less costly for all parties involved. Last, social capital installs in us the awareness of the interconnectedness of our lives and fates. Active and trusting connections to others help us develop behaviour and characteristics which are beneficial for the group and for society, such as being more tolerant and empathetic as well as less cynical (Putnam, 2000, p.288). When taking these mechanisms underlying social cohesion into consideration for a well-functioning society, it is unsurprising

(13)

Anne-Floor Bakker

that people are generally alarmed by the reported erosion of social capital over the last three decades (Putnam, 2000, p.287).

Social networks and the social capital embedded within them can be viewed from an instrumental perspective, as individuals peruse their own interests by employing these resources. Actors with more social capital are better able to realise their goals and defend their interests. People are willing to invest in each other based upon the idea of trust and generalised reciprocity, knowing that in the future a service will be repaid them for their assistance to others now. In that sense, such actions can be seen as long-term investments (Flap, 2004, pp.4–5). However, it should be noted that not all social capital needs to be the result of a conscious investment decision. Social capital may also be inherited or obtained as a by-product of a social connection or interaction; not all social connections are aimed at maximizing social capital (Riedl & van Winden, 2004, pp.77–78).

The saying “it’s not what you know that counts, but who you know” bears some truth. Often, when people want to make things happen, they will call first on their family, friends and acquittances. Bypassing formal systems is often a less stressful and more-effective means of getting things done. For instance, when someone is looking for a capable car mechanic, a good dry cleaner or the best primary school, they will often first consult their network for reliable suggestions. Recommendations made by people within their network help them overcome information asymmetry and reduce the risk of getting involved with incompetent actors (Field, 2003, p.2). A similar trend can be found for jobseekers: various studies have shown that social contacts are a highly effective way of securing a new job or getting a promotion (Field, 2003, pp.50–51). Additionally, there is the role of the network in promoting knowledge exchange. Previous studies have found that when high-tech firms have a high turnover of youthful employees, firm-specific knowledge, which had been the preserve of old-guard workers, was unfortunately lost (Field, 2003, p.56). Similarly, young people can often benefit from the field-specific knowledge of more experienced people when it comes to doing taxes or writing a resume and/or application letter. Reversed, older adults can often benefit from the younger adult’s knowledge and experience regarding new technologies and societal movements, such as learning how to use a tablet or an online banking app (Hagestad & Uhlenberg, 2006, pp.646–647).

In short, the accumulation of social capital can have numerous advantages, both on an individual and a societal level. Based Granovetter’s (1973) arguments, we assume that high social capital implies connections to dissimilar others, as the social capital embedded in those ties are especially valuable. This has led to the formulation of the following hypothesis:

(14)

Anne-Floor Bakker

individuals with higher levels of social capital are more likely to have intergenerational ties than those who have lower levels of social capital (social capital hypothesis).

(15)

Anne-Floor Bakker Table 1. Schematic overview of central arguments & hypothesis

Central arguments Hypothesis 1.

Work-tie hypothesis

Because people spend a substantial amount time at work, it is more likely that they also engage in friendships in the workplace, even if their colleagues do not fully meet the preferred characteristics (e.g., age). So, when people work in age-heterogenous work environments, the chance of intergenerational friendships rises even further.

Individuals who are active in the work force (=a) and work within a heterogenous organization (=b) are more likely to have age-heterogeneous relationships than those who do not participate in the work force. The expected effect is greater for individuals working within a heterogenous organization.

2. Voluntary- tie hypothesis

People who do voluntary work have a greater chance of having intergenerational ties to non-kin, as these contexts often facilitate contact between people of different backgrounds and personal characteristics.

Individuals who are active in a voluntary organisation are more likely to have age-heterogeneous relationships than those who are not involved in similar organisations.

3. Neighbour-hood-tie hypothesis

In more urbanised neighbourhoods the likelihood of meeting others who are much older of younger than oneself is higher due to population density; therefore, the chance of having intergenerational friendships increases.

Individuals who live in more urbanised environments are more likely to have intergenerational ties than those who live in less urbanised environments.

4. Family-tie hypothesis

Having supportive and emphatic family relationships increases the chance of having intergenerational non-kin ties: more positive attitudes towards other age-groups and frequent contact with family increases the chance of meeting friends or acquaintances.

Individuals who have ties (=a) and positive ties (=b) to family members who are much younger or much older than themselves, are more likely to have age-heterogeneous ties to non-kin than those who have no such ties. The expected effect is stronger for individuals who have positive family ties.

5. Social capital hypothesis

Having high social capital implies having ties to dissimilar others bridging different networks together via a mutual connection. People with high social capital therefore have higher chances of having intergenerational non-kin ties.

Individuals with higher levels of social capital are more likely to have intergenerational ties than those who have lower levels of social capital.

(16)

Anne-Floor Bakker

3. Data and Measurements 3.1. The Sample

This study is part of a funded overarching research project: ‘Bruggen over Scheidlijnen: intergenerationele contacten en inclusiviteit – van wetenschap naar praktijk’. The funded project is a joint accord between researchers of the University of Utrecht (Veronique Schutjens and Gerald Mollenhorst) and the University of Amsterdam (Beate Volker). The project ‘Bruggen over Scheidslijnen’ is interested in the growing social distance between older adults and the young. The aim of the project is to investigate instances of intergenerational contact, as well as how these relationships are forged and in what context. In addition, based upon the findings of this study, current measures used to enable intergenerational contact will be evaluated and new implementations for encouraging intergenerational contact will be suggested. Data collection for Bruggen over Scheidslijnen was executed by an external research agency based in Amsterdam and took place between the end of April and beginning of May 2019. The questionnaire is quantitative in nature and is structured into five components: ‘background characteristics of the respondent’, ‘work and daily activities’, ‘relationships and social networks’, ‘membership, leisure and attitudes’ and ‘name interpretation’. This study will be the first to use the dataset from the Bruggen over Scheidslijnen project. The sample will consist out of 1925 respondents; which includes 1057 adults over 65 years of age and 868 adults between 18 and 35 years of age.

3.2. Network delineation

Throughout the survey name-generator methodology was used to identify significant network members in respondents’ personal social networks (Lin, 2008, pp.55–56). The respondents were asked to name the members of their household and any children no longer living at home. Next, respondents were asked to name their (living) parents and siblings. Subsequently, respondents were asked to name the network members they would ask for a favour in regard to a job around the house. Furthermore, the older adults were specifically asked to name network members who were substantially younger than themselves; younger adults were asked to name network members who were substantially older than themselves. Also, respondents were asked to name the network members they have discussed important personal matters with in the last six months. In addition, respondents were asked to name the network members they would rely on for help when they would fall ill (e.g. to do grocery shopping).

(17)

Anne-Floor Bakker

In the last part of the survey, called “name interpretation”, the respondents were asked to identify the names they had used earlier in the survey and provide some general information regarding these network members. The respondents were asked to label the network member as either “partner”, “ex-partner”, “parent”, “child”, “sibling”, “parents in law”, “brother/sister in law”, “grandchild”, “cousin”, “uncle/aunt”, “grandparent”, “other family member”, “friend”, “child of friend”, “parent of friend”, “colleague or boss”, “someone from the neighbourhood”, “direct neighbour”, “someone from the same hobby/sport/social club”, “student friend”, “acquaintance”, “professional caregiver” or “other, please specify”. Subsequently, respondents were asked identity the sex, age and educational level of the network members. Next, they were asked to describe their relationship to the network member in terms of frequency of contact, living proximity, duration of the relationship, context in which the relationship takes place, closeness and trust.

The data regarding the social network members amount to social network dataset. The general survey data and the social network data were compiled in two separate data sets due to their formats being incompatible. However, after the data handling the datasets were made compatible and merged together based upon the respondents’ identification numbers. We will come back to the analytical strategy later in this chapter.

3.3. Measurements

3.3.1 Dependent Variables

The dependent variable is intergenerational non-kin ties, this is an ordinal variable, as the possible values are divided into six categories ranging between zero and more than five. However, although these categories can be ranked from few to many intergenerational non-kin ties, the distances between the categories are not all equal (Bryman, 2012, p.335): having one intergenerational non-kin tie would increase the chances of having more, and so forth. Furthermore, the decision to measure the number of intergenerational non-kin ties was made after exploring the network data and noticing that many intergenerational ties would go to waste when operationalising them in a binary fashion, as many respondents reported multiple ties.

Subsequently, we will discuss the operationalising of the variable intergenerational non-kin ties. According to the Centraal Bureau voor de Statistiek (CBS) the age span of generations ranges generally between 10 and 15 years, as the “baby boomer generation” were born between 1940 and 1955, “generation X” between 1955 and 1970, the “pragmatic generation” between 1970 and1980 and “generation Y” born between 1980 and 1994 (van

(18)

Anne-Floor Bakker

Gaalen, 2014, pp.3–4). Based upon this generational division, we define intergenerational contact as contact between the respondent and a network member when they are at least 15 years apart in age. However, we choose not to define the direction of the age difference, as not many younger adults had network members younger than 18 years old. The same could be said for older adults, who also had a limited number of network members more than 15 years older than themselves. Furthermore, respondents were asked to identify the relationships in their social network. Based on these labels, a division was made between family and non-kin, family being: “parent”, “child”, “sibling”, “parents-in-law”, “brother- or sister-in-law”, “grandchild”, “grandparent” and “aunt or uncle”. Thus, an intergenerational non-kin tie is a connection between the respondent and a network member who are not related to each other and differ at least 15 years in age.

Additional descriptive information such as the frequency of contact, the duration of the contact, the initial meeting place, the context in which the contact generally takes place as well as the established level of closeness and trust, will be gathered regarding these intergenerational connections.

3.3.2 Independent Variables

In order to determine predictors for intergenerational ties between non-kin, the following topics will be included as independent variables to hopefully answer the hypothesis: participation in the labour force, participation in voluntary work, the degree of urbanisation of the neighbourhood, intergenerational family ties and social capital. The variables for each hypothesis will be discussed individually.

To test the work tie hypothesis, the variable work was created. The variable work was measured as a binary variable: a respondent is either working or not working. This variable is based upon the question in the survey, which inquires about the main daily activity of the respondent: “studying”, “employment”, “homemaker”, “retired”, etc. When respondents answered as being either “employed”, “self-employed”, “work(ing) for the family business”, “co-owner of a company” or “freelancer” they were labelled as working. Next, the second part of the working hypothesis is directed at the age-heterogeneity of their workplace. After the question regarding daily activities, the respondents were presented with two statements regarding the age heterogeneity of their (previous) work environment. The first one being, “I work(ed) in an organization where many people had a different age than myself” and the second one, “My direct colleagues differ(ed) more than 15 years from my own age”. The respondents were given a five-point Likert scale ranging from “strongly agree”, “agree”,

(19)

Anne-Floor Bakker

“neutral”, disagree”, “strongly disagree” (Bryman, 2012, p.166). When respondents scored this question with either “strongly agree” or “agree” they were coded as having a relatively age-heterogenous working environment.

The variable voluntary work was created to test the voluntary tie hypothesis and is again binominal in nature. Respondents are either active in a voluntary organization or they are not. This variable is based upon the question survey asking directly whether or not they do voluntary work, with “yes” or “no” as answer categories. If they do participate in voluntary work, the respondents are further asked to describe their activities as a volunteer and the amount of time per week/month they spend doing so.

To test the neighbourhood hypothesis, the variable degree of urbanism was created, based upon the postcode of the respondents. The degree of urbanism is ordinal in nature and is based upon data from the CBS determined by the four-digit postcode. The dataset from the CBS connecting all postcodes in the Netherlands to a level of urbanisation was downloaded and merged into the alter data file, enabling us to translate the postcodes of our respondents into a level of urbanism. The degree of urbanism is divided into five categories and ranges from “not urban”, “slightly urban”, “mildly urban”, “urban” and “strongly urban” (Leeuwen, Guldemond & Faqiri, 2017, pp. 30–31).

In efforts to test the intergenerational family tie hypothesis, the variable intergenerational family tie was created. As described with the intergenerational non-kin ties, respondents were asked to identify the relationships in their social network. Based upon these labels, a distinction was made between family and non-kin, non-kin being “friend”, “child of friend”, “parent of friend”, “colleague or boss”, “someone from the neighbourhood”, “direct neighbour”, “someone from the same hobby/sport/social club”, “student friend”, “acquaintance” and “professional caregiver”. However, for intergenerational family ties, we choose to use a substantially larger age difference of 30 years instead of 15 years, as intergeneration now applies to the familial definition of family generations succeeding each other instead of the societal definition of generations, which applies to birth cohorts experiencing similar historical events (van de Broek, Bronneman-Helmers & Veldheer, 2010, pp.11–12). Thus, an intergenerational family tie is between a respondent and a network member when the relationship is identified as family and when they differ in age by at least 30 years.

Next, for the second part of the intergenerational family tie hypothesis, the strength of the relationship needs to be determined as either positive or neutral/other; creating the positive intergenerational family tie variable. These labels will be assigned based upon the questions

(20)

Anne-Floor Bakker

regarding the frequency of face-to-face contact, the frequency of other forms of contact (telephone, social media, email, etc.), the level of closeness and the level of trust. The frequency of contact ranges from “every day”, “every week”, “every month”, “every three months”, “once or twice a year”, “less often”. The levels of closeness and trust were divided into five categories ranging from “very close”, “close”, “a little close”, “not particularly close”, “not close”, and “do trust strongly”, “do trust”, “do trust a little”, “do not particularly trust”, “do not trust”. A positive intergenerational family tie must meet the following requirements: both face-to-face contact and other forms of contact must be either on a daily, weekly or monthly basis; the level of closeness must be either “close” or “very close” and the level of trust must be either reported as “do trust” or “do trust strongly”. The variable positive intergenerational family tie is binary of nature, if the intergenerational tie does not meet the norm on one or more of these categories, the tie will be considered neutral/other.

To test the social capital hypothesis, the variable social capital was created. A respondent’s social capital will be determined on the basis of position-generating methodology. This method systematically samples a list of positions in a social hierarchy (e.g., the occupational status within a society). Each respondent is asked to indicate whether he or she knows anyone in that sampled position, then to name that person. The position-generating index can be used as a parameter for social capital (i.e., the pool of resources embedded in the respondent’s network) (Lin, 2008, pp.55–56). This variable will be operationalised into an ordinal variable, as the number of resources in one’s social network will be categorised as ranging between 0 and 10 ties, 11 and 20 ties, or 21 and 60 ties. This division between the different categories was made after exploring the data and based upon the frequency of occurrence. Therefore, the last category, between 21 and 60 ties, is a rather large range, however the frequency is much lower than within the other two categories. Therefore, it did not make sense to divide this last category in multiple smaller ranges, as the highest values were rather rare. Thus, the number of ties in the position generator are used as an indicator for the amount of social capital a respondent has access to; a higher number of ties equals a higher level of social capital.

3.3.3 Control Variables

In this study, the following control variables will be included: sex, educational level and relationship status. The year of birth will be transformed into years of age by subtracting the current year (2019) from the year of birth, creating the variable age. The variable sex is limited to two options: male and female. The educational level is defined by three main

(21)

Anne-Floor Bakker

categories based upon the Dutch educational system, according to the CBS (2017): lower (1), middle (2) and higher (3) educational level. The category ‘lower educational level’ consists of the following components: no education, primary education, ‘VMBO’ (middle management-oriented learning path) and ‘mulo’ (expanded lower education). The category ‘medium educational level’ included ‘havo’ (higher general secondary education), ‘vwo’ (pre-university education) and ‘mbo’ (senior secondary vocational education) diplomas. The category ‘higher educational level’ included ‘hbo’ (higher professional tertiary education) and (post) university level (CBS, 2016, p.10). The relationship status is operationalised as a binary variable having either a significant other or not, creating the variable significant other. Having a significant other includes being married, cohabiting or living apart together, whereas not having a significant other includes being divorced, widowed and single (lived together in the past / never lived together).

3.4. Analytical strategy

The data analysis was mainly conducted with SPSS version 25 for an Apple MacBook. In general, SPSS was used to do most of the data handling, such as the descriptive tables and the social network composition analysis. However, the ordered logistic regressions were performed in Stata. This is because Stata provides better systematic support when executing ordered logistic regressions in general. Subsequently, the related assumptions (i.e., proportional odds and multicollinearity) were also tested in Stata. Additionally, we also conducted two other types of regression to compare outcomes: a binominal ordered regression and a multiple regression analysis. However, the results were rather similar and did not fit our models better than the ordered logistic regression model.

As described earlier, the data used in this study exists out of two separate components: the alter survey data and the social network analysis data. Because of their different natures, these datasets are not initially compatible; however, both datasets are ordered based on the respondents’ identification number. Whereas the alter data contains just a single string of data regarding the respondent, the social network data includes multiple strings of data for each of the respondents’ network members. To be able to use both datasets in one regression, the data has to be made compatible by means of aggregating the social network data to the level of the respondent. This has to be done for each variable individually, and not all variables are suited for this approach. For example, aggregating the variable sex of the network members on the respondents’ level gives us the mean of the variable sex for all network members originating from that one respondent. The respondents’ identification number is the key variable used to

(22)

Anne-Floor Bakker

aggregate and eventually match both datasets to each other. Thus, by aggregating, we create one summarising value instead of multiple values (e.g., for each network member individually), making the transition from the network dataset to the alter dataset possible. We followed this procedure for other variables such as age (of the network members), educational level (of the network members) and the number of network members per respondent. Although, in the case of educational level, dummy variables had to be made for each educational level to summarise the data correctly.

However, when identifying the intergenerational non-kin ties, we want to compare each network members’ age to that of the respondent, as well as identify the nature of the relationship as being either family or non-kin. Based upon these requirements, each individual tie between the respondent and their network members is labelled either as an “intergenerational non-kin tie” or not. Subsequently, the number of intergenerational non-kin ties can be counted and summarised for each respondent, creating the variable compatible with the alter data and used in our analysis. We followed this same procedure for other variables such as intergenerational family tie and positive intergenerational family tie.

4. Results

4.1.1. Descriptive characteristics

In total, the sample existed out of 1925 respondents, of which 1057 respondents were labelled as older adults (65 years of age or older) and the other 868 respondents were labelled as younger adults (18 to 35 years of age). Due to the significant age difference between these two groups and the nature of this study, we will continue to approach these age groups separately and compare their outcomes. Next will be a discussion of the general descriptive characteristics of our respondents, as shown in Table 2. Starting with the older adults, half were male, their ages ranged between 66 and 89 with an average of 72 years of age. Overall, 26% of the participants were lower educated, 33% had a mid-level education and 40.5% had attained higher education. Secondly, the same characteristics for the younger adults showed the following outcomes: just over one-third of the respondents were male, their ages ranged between 19 and 35 with an average of age 29. The younger adults were generally more educated than the older adults, with only 3% being lower educated, 33.9% had received a mid-level education and 62.5% had attained a higher educational degree.

Next, we will briefly describe the descriptive characteristics of the social network members these on which respondents reported (Table 2). The older adults reported an average

(23)

Anne-Floor Bakker

of just under eight network members. These network members were in almost half of the cases male and their ages ranged between 16 and 97 with an average 57 years of age. The educational level of the network members of the older adults were as follows: 28.9% were lower educated, 22.9% had a mid-level education and 37.1% were higher educated. Secondly, the younger adults had slightly more than eight network members on average. These network members were in half of the cases male and their age ranges between 16 and 93, with an average of 42 years of age. The network members of the younger adults were in 16.7% of the cases lower educated, 26.5% had a mid-level education and 44.1% were higher educated.

Characteristics Older adults Frequencies (n) Younger adults Frequencies (n) Sex Male 49.8% (526) 285 (32.9%) Female 50.2% (530) 581 (67.1%)

Age (in years) Mean 72.4 29.0

(24)

Anne-Floor Bakker

Table 2. Characteristics for the older adults (n = 1056) and younger adults (n = 867).

Range 66–89 19–35

Educational level Low 26.0% (275) 3.0% (26) Medium 33.0% (349) 33.9% (294) High 40.5% (428) 62.5% (543) No answer 0.5% (5) 0.6% (5) Number of network members N 8106 6812 Mean 7.8 8.04 SD 4.8 3.66 Range 1–29 1–22 Total 8106 6812 Gender network members Male 47.5% 51% Female 52.5% 49%

Age (in years) network members Mean 57 42.2 SD 8.5 7.7 Range 16–97 16–93 Educational level of network members Low 28.9% (2342) 16.7% (1135) Medium 22.9% (1854) 26.5% (1805) High 37.1% (3009) 44.1% (3006) No answer 11.1% (901) 12.7% (866)

(25)

Anne-Floor Bakker 4.1.2. Descriptive network composition

The network composition of older adults and younger adults differs significantly for some types of network members. Our discussion will start with the family members, after which the non-kin members will be discussed for both age-groups (Table 3).

First, the older adults reported having slightly more often a significant other, as compared to the younger adults. Second, the younger adults reported significantly more often that both parents were still alive. This is in line the expectations due to the age of the older adults—the chance that parents would still be alive is rather slim, and this is reflected in the outcomes. Third, when it came to having children, older adults reported a substantially higher number of them compared to the younger adults. Again, this was in line with the expectations due to the age difference. Fourth, older adults reported having significantly more grandchildren, whereas younger adults reported having essentially none. Fifth, younger adults reported having still some grandparents who are alive, whereas older adults reported having virtually none. Sixth, both older and younger adults reported similar numbers of siblings. In the last family category, we combined less frequently mentioned network connections such as in-law family members, uncles, aunts and cousins under “other family members”. In this case the younger adults reported having slightly more other family members compared to the older adults.

With regard to the non-kin network members, first, younger adults reported having substantially more friends, as compared to the older adults. This could be due to difference in life phases. Younger adults are often not yet settled and still meet new people via education programmes or at work. Older adults will mostly have a stable group of friends, however due to increasing age, friends and family die more frequently. Second, younger adults also reported having more work-related connections. This difference is likely due to older adults retiring and exiting the workforce as the younger adults are just beginning their careers. Third, both older and younger adults reported the same amount of club friends (e.g., hobby, sports, social). Fourth, older respondent reported having more connections in their neighbourhood as compared to the younger adults. Fifth, both groups reported having virtually no connections to the children of friends, however it was more common among older adults. Sixth, similarly, both groups reported having almost no connections to parents of friends, however this was more common among younger adults. Seventh, both older and younger respondents reported the same number of acquaintances, on average. Similarly, the average number of professional caregivers was the same for both age groups.

(26)

Anne-Floor Bakker

Lastly, when looking further into the number of specifically intergenerational non-kin ties for both the older and younger adults (Table 4), it is striking that almost twice as many younger adults reported one or more intergenerational ties to non-kin (88,7%); then there are older adults who reported a similar tie (45,9%).

Table 3. Network composition for older and younger adults.

Table 4. Number of reported intergenerational non-kin ties for older and younger adults in relative group percentages (N).

Network member Older adults Mean (SD) Younger adults Mean (SD) Partner 0.73 (0.82) 0.65 (0.62) Parent(s) 0.12 (0.49) 1.77 (0.83) Child(ren) 1.73 (1.63) 0.33 (0.73) Grandchild(ren) 0.72 (0.37) 0.005 (0.08) Grandparent(s) 0.01 (0.08) 0.04 (0.20) Sibling(s) 1.63 (2.03) 1.57 (1.38) Other family members 0.31 (0.73) 0.53 (0.88) Friend(s) 1.06 (1.57) 1.55 (1.71) Work friend(s) 0.13 (0.54) 0.42 (0.94) Club friend(s) 0.27 (0.90) 0.27 (0.89) Neighbour(s) 0.88 (1.40) 0.38 (0.95) Child of friends 0.09 (0.42) 0.004 (0.07) Parent of friends 0.008 (0.11) 0.04 (0.26) Acquaintance 0.15 (0,54) 0.13 (0.50) Professional caregiver(s) 0.06 (0.31) 0.06 (0.36) Intergenerational non-kin ties Older adults (N=1057) 65+ years of age Younger adults (N=868) 18–35 years of age 0 54.1% (572) 11.3% (98) 1 19.5% (206) 6.5% (56) 2 10.5% (111) 31.7% (275) 3 6.1% (64) 17.2% (149) 4 4.1% (43) 15.3% (133)

(27)

Anne-Floor Bakker

5+ 5.8% (61) 18.1% (157)

(28)

Anne-Floor Bakker 4.2. Ordered logistic regression

The analyses were conducted separately for the two age groups, the older adults (65+ years of age) and the younger adults (18 to 35 years of age). Because of the ordinal nature of our dependable variable (number of intergenerational ties scaling 0–5+) we used an ordered logistic regression to calculate the chance that respondents have an intergenerational (15+ years of age difference) non-kin tie in their social network. Furthermore, we will test our hypothesis using three models. The first model starts with individual characteristics such as sex, educational level and having a significant other and is based upon the hypothesis linked to the opportunity structure theories, like being active in the workforce, working in an age-heterogenous environment, doing voluntary work and the degree of urbanisation of the living environment. Next, in the second model the variables for the family tie and positive family tie hypotheses were added. These variables were measured as having one or more family members who differ significantly in age (30+ years) from our respondent. The positivity of the family tie was measured by frequency of contact, levels of trust and likability. In the third and last model, the variable for the social capital hypothesis was included. This variable was measured by the number of ties to family, friends or acquaintances with a certain occupation and occupational status. The results are presented in odds ratios (OR). Odds ratios with a value smaller than 1 indicate a negative relation, whereas a value greater than 1 indicates a positive relation. One should be aware that the results should be interpreted given that all other variables are held constant.

The first model (Table 5) shows that for older adults, females have factor 1.3 higher odds of having intergenerational non-kin ties compared to males. Also, older adults with a high educational level have a factor 1.7 higher odds of having intergenerational non-kin ties compared to older adults with a lower educational attainment. However, having attained a mid-level education does not seem to influence the odds of having said ties compared to the lower educated reference group. Furthermore, older adults working in an age-heterogenous environment have a factor 1.3 higher odds of having intergenerational non-kin ties, compared to those who work in less age-heterogenous environments. This is in line with the findings of Dykstra and Fleischmann (2016), who found that being merely active in the work force is not enough to affect the odds of having intergenerational non-kin ties; the age diversity of the workplace was the deciding factor (pp.121–122). Lastly, older adults who are active in voluntary organisations have a factor 1.5 higher odds of having intergeneration non-kin ties, compared to those who are not active in similar organisations. These results support the early findings of Uhlenberg and Jong Gierveld (2004) and Dykstra and Fleischmann (2016), which

(29)

Anne-Floor Bakker

found that voluntary organisations accommodate the recruitment of age-heterogenous network members. However, neither having a job nor the degree of urbanisation of the living environment have a significant effect on having intergenerational non-kin ties for the older adults in the first model, resulting in no support for the neighbourhood tie hypothesis.

The results for younger adults differ substantially in the first model (Table 5). The educational level of the younger adults is of clear importance, as respondents with a mid-level education have a factor 2.1 higher odds of having intergenerational non-kin ties, compared to respondents with a lower educational level. This effect is even greater for the higher educated compared to the reference group, as they have a factor 2.4 higher odds of having said ties. Also, for younger adults, having a significant other increases the odds of having intergenerational non-kin ties by a factor 1.4, compared to respondents have who do not have a significant other. However, having a job, the age-heterogenous working environment, doing voluntary work or the degree of urbanisation are proven not to have a significant effect for the odds of having intergenerational ties to non-kin for the younger adults in the first model. Although it should be noted that younger adults are on average less frequently active in voluntary organisations compared to older adults; older adults are twice as often active in this type of organization.

In the second model (Table 5) the intergenerational family tie variables were added to the regression. The older adults who have an intergenerational family tie have a factor 1.8 higher odds of having intergenerational non-kin ties, compared to those who lack such a family tie. Our results counter those from the study of Dykstra and Fleischmann (2016), who claim that having strongly positive multigenerational ties further increases the chances of having intergenerational non-kin ties (p.110). However, we find that having a positive family tie does not seem to influence the effect even further. The remaining variables from the first model did change somewhat after including the variables of the second model. The effect of the variables sex, higher educational level, age-heterogenous work environment and voluntary work increased slightly, however the level of significance did not change for the older adults in the second model.

As Table 5 shows, the results for the second model are more extreme for the younger adults, as respondents who have an intergenerational family tie have a factor 11.6 higher odds of having intergenerational non-kin ties, compared to respondents who lack such a family tie. Again, this is in line with the results of Dykstra and Fleischmann (2016), who found that strong intergenerational family relations are a strong predictor for having non-kin intergenerational ties (p.110). The younger adults who reported having positive family ties

Identifying ties that bridge generations : a quantitative social network analysis on intergenerational ties between non-kin relations