Family change and community life

Family change and community life

Exploring the links

Jody Hughes, Wendy Stone

Research Paper – May 2003

Go to:

You are in an archived section of the AIFS website. Archived publications may be of interest for historical reasons. Because of their age, they may not reflect current research data or AIFS' current research methodologies.

Read the publication

Executive summary

In the social capital literature, “the family” and “the community” are frequently linked. Generally, it is assumed that the family plays a key role in creating social capital in the community. This is also the case in current Australian public policy. For example, the Commonwealth Government’s Stronger Families and Communities Strategy promotes the family alongside the community as traditional institutions which provide the most effective source and form of social support, and relies heavily on the assumption that these institutions strengthen each other.

It is therefore not surprising that recent concern about community decline has been linked to widespread changes that have occurred within family life. It is argued that family “breakdown” – including increased rates of de facto marriages, divorce and sole-parent families as well as increased rates of female labour force participation and individualism within relationships – are in part responsible for low levels of trust and engagement in community life.

Yet while the idea that changes in family life are a source of community decline features in the work of influential social capital thinkers (Putnam 1995; Bourdieu 1993; Fukuyama 1999), and is common among politicians and policy makers and in the media, the relationship between family change and community life has rarely been the focus of empirical scrutiny.

Using a national random sample of 1500 Australians from the Australian Institute of Family Studies Families, Social Capital and Citizenship survey (2001), this paper begins to explore this link. These data are used to explore whether there is a link between changes in family life and community social capital, and the nature of this link.

To explore whether there is a link between family changes and community life, the relationship between family characteristics commonly associated with family change (including marital status, the presence of children, female labour force participation, and individualism within family life) and levels of community social capital is examined. Two key measures of community social capital are used. These are norms of generalised trust and reciprocity, and levels of community group membership.

Then, two main ideas about the nature of this link are explored. The first is that these changes in family life have led to a breakdown of family social capital – a breakdown of family ties and norms of trust and reciprocity within families – and that this in turn leads to lower levels of community social capital. The second is that these changes in family life have had an impact on the other resources available to families such as human and financial capital, and that this in turn leads to lower levels of community social capital.

The paper presents a series of detailed findings, comparing men and women in a range of family circumstances, with varied levels of resources. The paper concludes with several key findings, each of which has important theoretical and policy implications. Among these are the following:

  • The relationship between family life and the two key measures of community social capital differs: there is a stronger link between family circumstances and levels of generalised trust and reciprocity than there is between family circumstances and levels of group membership and involvement – where other factors are more important.

In terms of the link between family life and norms of trust and reciprocity within the community:

  • Findings show some support for the idea that family change is related to lower levels of community social capital for men. Married men have higher levels of generalised trust and reciprocity than unmarried men, consistent with previous research; and this appears to be in part because married men have higher levels of family social capital than unmarried men. However, men who are divorced or separated and men who are in de facto relationships have significantly lower levels of generalised trust and reciprocity than married men even when levels of family social capital and other resources are taken into account.
  • Working full-time is associated with low levels of generalised trust and reciprocity among women with dependent children. There is no evidence that this is related to the impact of maternal employment on the quality of family life. It is likely that long hours spent in paid work are not conducive to community social capital where one also has primary or sole responsibility for the care of young children, because of the time constraints and pressures associated with parenting and full-time work.
  • High levels of household social capital are associated with low levels of generalised trust and reciprocity for lone mothers. This may be because mothers who face the type of time and resource constraints that lone mothers do also face a trade-off between engaging in the household and engaging in the broader community.
  • In contrast with the idea that family change is related to low levels of community social capital, non-traditional attitudes to relationships and gender roles appear to relate to community life in positive ways. For example, among women, support for maternal employment is associated with high levels of generalised trust and reciprocity. And among men, support for independence within intimate relationships is associated with high levels of family social capital and trust and reciprocity in the community.

In terms of the link between family life and community group membership and involvement:

  • The key characteristics of families associated with family change – such as marital status, family structure and household employment – are less relevant to group membership than they are to levels of generalised trust and reciprocity.
  • Where relationships do exist between these variables they are only evident for men, and are generally consistent with those found for generalised trust and reciprocity: men who are divorced or separated have significantly lower levels of group membership than other men (consistent with the idea that family change is related to community decline), but men who support the idea of independence in intimate relationships report higher levels of group membership than other men (inconsistent with the idea that family change is related to community decline).
  • Engagement in community groups and organisations is more strongly related to the health, educational and financial resources of individuals and their families, than it is to family circumstances like marital status, family structure or household employment.

In sum, our findings show a complex set of relationships between family life, social capital and the other resources available to families, as well as the gendered nature of these relationships. There is evidence to support some aspects of the idea that family change has led to community decline, but also several important variations and qualifications.

The implications of these results include the following:

  • Future research could explore how marital status and household circumstances are linked to men’s broader family and community connections – and could inform a policy focus on helping men to build and/or maintain strong family and community relationships regardless of their marital or household circumstances.
  • If we want to foster high levels of community engagement and trust we need to pay attention to issues of work and family balance, particularly for women. It seems likely that supporting men and women to flexibly organise and balance their work and family responsibilities will foster these community outcomes.
  • While some forms of family social capital are positively associated with community social capital in some circumstances, in other circumstances strong family norms and obligations appear to limit possibilities for broader community engagement and activity in a way that is characteristic of “familism”. These findings counter the assumption that high levels of family social capital will always translate into high levels of community social capital.
  • Other factors are also importantly related to community social capital, including the economic and human capital resources of individuals, their families and their local communities. Hence, community connections and trust may be enhanced through things such as money, education, public health and neighbourhood resources.
  • It is also likely that other factors are also important predictors of community social capital – factors we have not examined here. These may include individual personality characteristics, as well as other demographic factors such as cultural background or ethnicity. Future avenues of research could explore these links.
  • A final point to note is that while this paper has focused on “family” ties, it is likely that networks of friends, neighbours and/or work-mates (which are often used as alternative measures of community social capital) may also be important for building generalised norms of trust and reciprocity and broader forms of community engagement. The changing nature and function of these relationships, and the extent to which they provide links to broader forms of community engagement, may be a fruitful source of future research.
Introduction

Recently, concern about the “decline of community” has emerged in academic and political discourse and in the popular media. This has led to an emphasis on “social capital” – thought to be an important ingredient of strong and sustainable communities.

In this context, there has also been a focus on the changes that have occurred in family life. This is because the quality of family life is often linked to the quality of community life. One of the most pervasive assumptions underlying the social capital literature is that families are a key source of social capital in communities (Putnam 1995; Bourdieu 1993; Fukuyama 1999). This assumption is also prominent in public policy frameworks which position strong families as the cornerstone of strong communities (Howard and Newman 2000).

It follows that changes in family life have been seen as responsible for the changes that have occurred in community life, including declining levels of social capital in communities. In debates about community decline, the family based changes of most concern include the increased number of marriages ending in separation and divorce, the growth of lone-parent families, and the increased workfore participation of women. It is argued that these changes in family life constitute a loss of social capital within families – representing a weakening of family bonds and norms of trust and reciprocity (Fukuyama 1999). This is in turn thought to lead to declining levels of community attachment and trust of people generally – which are both seen as key dimensions of social capital at the community level (Putnam 1995).

While these ideas feature in the work of influential social capital thinkers (Putnam 1995; Bourdieu 1993; Fukuyama 1999) and are common among politicians and policy makers and the media (The Age 2001), they have rarely been the focus of empirical scrutiny. This paper draws on data from the Australian Institute of Family Studies Families Social Capital and Citizenship Project survey to begin to test these ideas.

The over-arching question we examine in this paper is how changes in family life relate to community social capital. In doing so, we focus on the key characteristics of families most commonly associated with family change and “breakdown”, including family type, household employment, and attitudes to relationships and gender roles. We explore whether there is any relationship between these variables and levels of community social capital, measured through levels of generalised trust and reciprocity, and rates of community group membership. In doing so, we also examine the extent to which social capital within families, and other family resources such as human and financial capital, can be understood as mechanisms by which these key changes in family life are linked to community life.

The remainder of the paper first introduces literature on social capital and family life, and sets up in detail the hypotheses tested. Following a description of the survey data and the analytic approach, the results are presented. Discussion and conclusions are presented in the final section.

The literature

Defining social capital in family and community realms

Social capital can be defined as networks of social relations characterised by norms of trust and reciprocity (Bourdieu 1993; Coleman 1988; Putnam 1993; see Winter 2000 for discussion). It is social relations of this quality and nature that are thought to affect the capacity of people to come together collectively to resolve problems they face in common (Stewart-Weeks and Richardson 1998: 2). These social relations can exist in a multitude of different types of networks, which exist at different social scales. These range from household and family level ties, to other informal networks (including friends and work-mates), to community based or “societal” relations people have with people they do not know personally (Stone 2001).

Family level social capital has been the focus of social capital researchers such as James Coleman (1988). This research points to the importance of social capital in families to a host of positive outcomes for family members, particularly the education and employment outcomes of youth (Coleman 1988; Furstenburg and Hughes 1995). This will come as no surprise to family researchers who may see family social capital as corresponding closely with dimensions of family cohesion. When we talk about family social capital we are talking about family bonds which are trusting and mutually supportive. While family cohesion is a broader concept, family social capital is closely related to particular qualities of family cohesion such as bonding, affective involvement, support and helpfulness (Baer 2002: 668).

While some attention has been paid to social capital in families, social capital in the broader community has more often been the focus of the social capital literature. In Coleman’s work for example, family relationships were most effective for reinforcing group norms and sanctioning non reciprocal behaviour when the family was embedded within a network of dense community ties. Social capital theorists such as Putnam (1993, 1995, 1996) and Fukuyama (1996, 1999) have been concerned with the extensiveness of networks and the level of association – particularly voluntary association – beyond the family. These writers argue that the number of groups and organisations in a particular community (including, for example, sporting groups, religious groups and neighbourhood groups) is a key measure of the social capital of a community, representing the capacity for unrelated adults to come together and cooperate in respect to interests they share in common. These various memberships are thought to sustain trust and an ethos of reciprocity and cooperation, which are also thought to be key elements of community social capital (Inglehart 1997; Knack and Keefer 1997; Hughes, Bellamy and Black 1999). This is the sort of trust and reciprocity that is extended to strangers, often on the basis of expectations of behaviour or a sense of shared values. This type of social capital is seen as a public good, thought to generate positive outcomes for whole communities (Portes 1998; Putzel 1997; and see Cox 2000 for discussion and alternative view).

Relationship between family and community social capital

In the social capital literature, “the family” and “the community” are frequently presented as paired, or linked, institutions. This is also the case in current Australian public policy. For example, the Commonwealth Government’s Stronger Families and Communities Strategy promotes “the family”, alongside “the community”, as traditional institutions which provide the most effective source and form of social support – and relies heavily on the assumption that these institutions strengthen each other, As Howard and Newman (2000: 2) state: “It is strong family relationships that are the vital building blocks of strong communities. In turn, it is only strong communities that have the capacity to truly engage families in economic and community life.” Restated in social capital terms, the policy assumes that families rich in social capital will nourish stocks of social capital within their local communities, and that by living in localities rich in social capital, families will prosper.

Perhaps the most pervasive assumption about the link between family and community in the social capital literature is about the role of the family in creating social capital in the community. In a review of social capital and family life, Winter (2000: 5) shows the assertion that “family life is a bed-rock of social capital” features in the work of Putnam (1995), Bourdieu (1993) and Fukuyama (1999) – some of the most influential social capital thinkers. As described by Winter (2000: 5-6): “There is a sense in which the role of the family has been idealised ‘as the most productive site of social capital and therefore a pillar of civic virtue and democracy’ (Putzel 1997: 945). The nature of this idealisation is that it is assumed that families will provide models of good relationships and civic virtues (Cox 1995: 28-29).”

While the mechanisms by which families generate social capital in communities are rarely articulated, they are thought to include the role of families in the creation of social networks and in the transmission of behavioural norms (Winter 2000). For example, it is likely that where children experience good quality connections with community and civil society through their families, this will lead to an increased propensity for those children to become engaged and active citizens in adulthood. More generally, where children are exposed to cooperative behaviour early in life they are more likely to become cooperative adults (Mark 2002).

It is also widely believed that family relationships are essential for the development of basic trust (King 2002). As quoted in Mistzal (1996: 157): “It is an everyday and valued conception in our society that the family is the primordial source and location of trust. The expression, ‘If you can not trust your family, then whom can you trust?’, has a large value and ideological appeal for us (Barber 1983: 26).”

Broadly speaking, the family can be seen as an intermediate institution through which individuals are tied to society (Waite 1995). Relationship and family formation can be important life transitions in this respect. Relationship formation not only bonds two individuals but typically also their extended family and friendship networks (Glezer 1997: 8). It can also represent the beginning of the type of family and community building associated with child bearing and home ownership – both of which are key predictors of neighbourhood attachment and involvement (Glezer 1997: 9).

The extent to which children are a conduit to social capital in the community has also been explored in previous research, but findings have been mixed (Winter 2000: 11; Hughes and Black 2002). Children appear to be a conduit to community participation through such things as parent and children groups, and interest in the provision of local infrastructure such as parks for children’s sport, but a barrier to other forms of participation (such as in parents’ own hobby or interest groups) due to the time and resource limitations associated with child rearing.

In sum, it is common to assume that families and communities strengthen each other, and that families are a key source of social capital in communities. While there is some empirical evidence to support these ideas, it is scant, and the ideas themselves are often not even directly expressed or articulated. Nonetheless, it is likely that, in practice, the nature of family norms and connections will affect the extent to which individuals can engage in aspects of civic life, and the extent to which they are trusting of persons beyond the family unit.

Changes in family life

In the past four decades families have been transformed by high rates of separation and divorce, increasing popularity of cohabitation, later ages at marriage, and falling levels of fertility (Ravanera et al. 2002: 2). This has led to dramatic changes in family form, with a substantial increase in the number of lone-parent families and lone-person households, and family households now being considerably smaller. For example, in Australia between 1986 and 2001, the proportion of all types of families that were lone-parent families increased from 7.8 per cent to 10.7 per cent; and the proportion of lone-person households increased from 19 per cent to 24 per cent. There has also been a dramatic increase in the workforce participation of women with dependent children. This has been accompanied by an increase in the proportion of families in which two adults are employed (from 35.7 per cent in 1979 to 45.1 per cent in 1998) and an increase in families in which no adult is employed (from 11% in 1979 to 18% in 1998).(Gregory 1999).

The norms governing family life are also undergoing change from being primarily obligational to negotiated. Increased individualism and the changing roles of men and women have meant greater freedom and flexibility of entry to and exit from relationships, in defining relationships, and in organising work and family responsibilities, at least in some families. Giddens (1992) argues that there has been a shift towards “the pure relationship”, by which he means: “a social relationship which is entered into for its own sake, for what can be derived by each person from a sustained association with another; and which is continued only in so far as it is thought by both parties to deliver enough satisfactions for each individual to stay in it” (Giddens 1992: 58). What this means in practice is that relationships are increasingly based upon negotiated normative frameworks and are highly individualised – people decide upon how they want to live together and how they wish to collaborate and communicate (Smart and Neale 1999: 8). And negotiated intimacy is not restricted to couples. The realm of kinship relations is also increasingly negotiated rather than governed by strict rules and normative expectations, being based on empathy and affinity rather than obligation and duty (Finch and Mason 1993).

Implications for community social capital

While few dispute the significance of the family changes outlined above, there are varying interpretations of what they mean for community life. The dominant interpretation is one of family and community decline. This interpretation, which we will call the “family decline thesis”, assumes that these changes in family life will lead to declining levels of community social capital (The Age 2001; Fukuyama 1999; Putnam 1995). The main explanation provided is that changes in family life constitute a weakening of social capital within families, and social capital within families is a key source of social capital in communities. Thus, by damaging family relations, changes in family life will lead to declining levels of community social capital.

To elaborate, the family decline thesis sees increased rates of divorce and separation, the increased number of lone-parent families, the increased workforce participation of women, and increased individualism in relationships, as constituting a weakening of family bonds and norms of trust and reciprocity (social capital) within family life. Divorce and separation are seen to amount to the breakdown of ties, trust and reciprocity within families. Similarly, women’s increased workforce participation is seen as a threat to social capital within families, as increased hours in paid work mean that women have less time to devote to their children and partner, and women’s increased financial self sufficiency means they have less need to form or maintain cooperative relationships with men (Misztal 1996).

The family decline thesis also assumes that if family social capital is weakened, the family will be less effective at generating community social capital. If marriage and children are a conduit to community engagement and attachment, and intimate relationships an important source of trust and security, it follows that falling marriage and fertility rates, and increased rates of divorce and separation may lead to lower levels of community engagement and attachment, and a decreased trust of people generally, including strangers. Where the family is seen as the key site for the development of behavioural norms, it follows that if people don’t experience cooperative relationships in their family life, they are less likely to have cooperative relationships with others in the community. Similarly, if individuals don’t learn how to take responsibility for others within the family – say as parents or providers – it will block their transition into responsible adulthood (Smart and Neale 1999: 4-5, Misztal 1996).

While the family decline thesis represents the dominant interpretation of changes in family life and what they mean for community, there are of course also alternative possible interpretations. It is possible that changes in family life will be associated with declining levels of community social capital, consistent with the family decline thesis, but not for the reasons the decline thesis emphasises. Another plausible explanation for why we might expect these changes in family life to be linked to declining levels of community social capital is not because of their impact on family social capital per se, but because of their impact on the other resources available to families, such as human and financial capital.1 Which may in turn relate to community social capital. For example, divorce is associated with financial strain and poor health, and these factors in turn restrict an individual’s capacity to engage in various community activities (Gallie and Paugam 2000). Another example, women’s increased workforce participation may be linked to declining community social capital because the time constraints associated with paid employment may make other forms of community participation and voluntary activity more difficult.

Previous research provides some support for this explanation. It shows that lone mothers have lower levels of education and financial resources than couple mothers (Gray et al. 2002) and that these resources are an important predictor of community social capital (Hogan and Owen 2000; Hughes, Bellamy and Black
2000; Putnam 1996: 4; Knack and Keefer 1997; Onyx and Bullen 2000; Sampson et al. 1997). As well, Putnam (1996: 6) found that women working part-time were the most trusting and civically engaged – more so than women who were working full-time and women who were not working at all outside the home. This suggests that women’s paid employment is not a problem per se, but that long hours in paid employment may not be conducive to high levels of trust and community participation among women. However, Putnam (1996: 6) also found that for men, longer hours in paid employment were linked to more community engagement. Thus the question of whether or not women’s increased workforce participation is associated with lower levels of community social capital, and why, requires further exploration, taking into account family circumstances, and the employment status and hours of partners, where present.

It is also possible that the above changes in family life are in fact unrelated, or positively related, to community social capital. Family theorists who interpret changes in family life in a positive light point out that while some family ties may have weakened, others may have become stronger (Misztal 1996). For example, while women may spend more time in paid work, men have become more involved in the upbringing of their children. And while divorce may involve the weakening of some bonds, it may strengthen relationships between other family members, and open up opportunities for new relationships to be formed.

Furthermore, Giddens (1992) argues that trust and reciprocity are defining characteristics of modern relationships, which can no longer be held together by law or tradition. And rather than reflecting an amoral individualism, the increased level of negotiation required in modern relationships can be seen to reflect the rise of democracy within intimate life, and this should play through to wider debates about morality and ethics (Smart and Neale 1999: 11). According to Gross and Simmons (2002: 539), Giddens’ “pure relationship” “takes seriously the values of autonomy and equality”. Thus the rise of the pure relationship should coincide with “the solidification of democratic ideals in the polity . . . As individuals experience the joys of egalitarian social arrangements in their most intimate relationships, they may carry with them, when they participate in the public sphere, a sense of the importance of these very values” (Gross and Simmons 2002: 539).

Finally, an alternative perspective on the relationship between family and community level social capital highlights the possible tensions between the two. From this perspective the family can in fact be oppositional to social capital in the community. In circumstances of “familism”, family and kinship ties and obligations are elevated above other sorts of social ties and obligations and allegiance to the family “crowds out” the weaker ties of community (Winter 2000). This was a key concern of Fukuyama (1996, 1999), and relates to social capital’s “dark side” – one of the key themes in social capital discourse. This is the idea that the same strong ties that bring benefits to members of a group can also exclude others from those benefits, a theme that is also developed in the work of Portes (1998), Putzel (1997) and in Australia by Eva Cox (see Cox and Caldwell 2000).

This idea can also be linked to the recent distinction in social capital theory made between “bonding” and “bridging” social capital, and the importance of both (for discussion see Putnam 1998; Narayan 1999; Woolcock 2000). Bonding social capital tends to involve trust and reciprocity in dense or closed networks, or close personal ties, and helps the process of “getting by” in life on a daily basis. Family relationships are seen as the primary example of bonding social capital. “Bridging social capital” tends to involve trust and reciprocity in overlapping networks; or more distant, weak ties outside one’s family and close friends. Bridging social capital is thought to help the process of “getting ahead” in life, and to make accessible the resources and opportunities that exist in one network to a member of another. In addition, these weaker ties tend more often to be with people different from oneself, which is argued to enhance their capacity to bridge social cleavages. People’s connections through such things as community groups and organisations are typically seen as an example of bridging social capital.2

In circumstances of familism, we would expect to find that strong bonds of trust and reciprocity inside the family co-exist with weak bonds of trust and reciprocity outside the family. This can therefore also be seen an example of too much bonding social capital, where norms of trust and reciprocity tend to be restricted to members of one’s own family grouping. In these circumstances the “breakdown of the family” may have an upside in respect to community social capital. It may lead to greater levels of association, trust and reciprocity outside the family. This is recognised as a potential positive spin off of family break down by Fukuyama (1999), despite his view that changes in family life represent a loss of social capital in itself.

In sum, while the dominant interpretation of change in family life and its implications for community social capital is one of family and community decline, there is very little empirical evidence about how family life relates to community social capital. We do not know whether the experience of divorce and separation, life in lone-parent family households, the increased employment of women with children, or individualism in intimate relationships, are associated with low levels of community social capital. Nor do we know how this might occur. As outlined earlier, in this paper data from the Families Social Capital and Citizenship Project is used to explore these links.

Hypotheses

The over-arching question we explore in this paper is how family life relates to community social capital. In doing so, we set out specifically to test the family decline thesis, which generates three key hypotheses.
The first hypothesis is that several key family characteristics that have become increasingly common will be associated with lower levels of community social capital. In particular:

  • community attachment and trust will be lower in lone-parent households than couple family households;
  • community attachment and trust will be highest among those who are married and lowest among those who are separated or divorced;
  • community attachment and trust will be lower in families where women work full-time than in families where women work part-time or do not work at all outside the home; and
  • community attachment and trust will be lower among people who hold non- traditional family values.

In addition, according to the family decline thesis, the main reason these family characteristics will be associated with lower levels of community social capital is because they are associated with a decline of social capital within families, and family social capital is a key source of community social capital. Consistent with the decline thesis we would expect to find that:

  • community attachment and trust will be lower among those with low levels of family social capital; and
  • the relationship between community social capital and the above family characteristics will be weaker when levels of family social capital are taken into account.

Alternatively, the above family characteristics may be related to community social capital because they are related to the other resources available to families, such as human and financial capital. If this were the case we would expect to find:

  • community attachment and trust will be lower among those with a lack of other resources such as economic and human capital; and
  • the relationship between community social capital and the above family characteristics will be weaker when these other resources are taken into account.

In exploring these three key hypotheses we will also be able to examine the alternative hypotheses discussed above, including the possibility that:

  • the above family characteristics of concern to family decline theorists are not directly associated with community social capital, or are in fact positively related to community social capital;
  • family social capital is not associated with community social capital, or in fact has a negative relationship with community social capital; and
  • other resources such as human and financial capital are not associated with community social capital, or in fact have a negative relationship with community social capital.

1 Although family decline theorists emphasise the negative impact of changes in family norms and behaviours, some recognise that decreased resources associated with family decline may also relate to declining community engagement (see for example Fukuyama 1999). However, economic resources are not prioritised in the explanations provided by the decline thesis generally.

2 While there is not necessarily a neat match between bonding and bridging and levels of family and civic relationships, since some family relationships may be incredibly diverse and have the function of "bridging" social cleavages, and some civic ties may be with heterogenous groups and have the function "bonding" connections rather than "bridging", we can nonetheless assume that in general, family ties will be those most likely to offer close supports, and civic relationships are those most likely to put people in touch with different resources and with people different from themselves.

Methodology

Data and measures

The survey data used in this paper were collected in 2000–2001 as part of the Families, Social Capital and Citizenship Project being undertaken by the Australian Institute of Family Studies. A total of 1506 respondents participated in the survey. These were drawn from a national random sample of households (with at least one person aged 18 years or older).3

We examine two measures of community social capital: first, the number of community groups and organisations respondents belong to; and respondents’ level of generalised trust and reciprocity. Respondents were asked whether they belonged to or had been involved with a range of different types of community groups and organisations in the last 12 months. These ranged from parent teacher groups to sporting groups to professional groups and organisations.4 Memberships of each type of group and organisation were combined to compute the total number of group memberships.5Generalised trust and reciprocity is measured on a scale of 0 to 10 and is based on respondent’s level of agreement with the following two statements: “Generally speaking, most people can be trusted” and “Generally speaking, most of the time people try to be helpful” (see Appendix B for alpha scores).

Conceptually, the measure of group membership and involvement is included as an indicator of the extent of a person’s connections at the civic level, or more formal community ties and attachments (while this is clearly related to civic participation, it is used here as a measure of connectedness). The measure of “generalised” trust and reciprocity is thought to capture the extent to which trust is extended to people generally in the community, including strangers, and the extent to which individuals perceive norms of reciprocity to exist among people generally in the community. This is a direct measure of community social capital.

Figure 1

Figure 1, described in text

At the family level, three different sets of explanatory variables are used to explore three different types of relationships, consistent with the hypotheses set out above. These three sets of explanatory variables, along with our two measures of community social capital, are presented in Figure 1.

The first set of explanatory variables we examine include the key family characteristics associated with “family change” and of most concern to family decline theorists. These variables are straight forward. They include family type, marital status, household employment, and attitudes about relationships and gender roles (about the workforce participation of mothers and the value placed on individualism in relationships). Including these variables enables us to examine, for example, how lone-parent families compare with couple families in terms of community connections and norms. Measures of attitudes to independence in family and work life are included as indicators of changing family values and norms.

Second, we include a set of “family social capital” items. This includes measures of various dimensions of social capital in three different types of family relationships: relationships between family members who live in the same household; relationships with kin beyond the household, and norms that operated within the respondent’s family of origin (the household they grew up in). Household social capital is a composite item, and includes such things as network density (the extent to which family members know one another’s close friends), feelings of trust and reciprocity, and the extent to which family members feel close to one another, and share interests and hobbies. For kinship networks we include a measure of network size (the number of relatives and in-laws respondent’s have), and norms of trust and reciprocity that exist in these relationships. The item about norms of community engagement within one’s family of origin is included as another way of examining the transmission of norms among family members.

Third, we include measures of other family resources, including the human and financial capital available to families (which includes variables such as education and financial wellbeing), as well as to the social and economic circumstances of the localities in which families live.

While the measures of key family characteristics and other resources are fairly straight forward, the measures of social capital at both the family level and community level warrant more explanation. In previous work we have discussed the measurement of social capital in terms of network structure, and norms of trust and reciprocity that can exist in a range of types of relationships (see Stone 2001; Stone and Hughes 2002). To the extent that it is possible, aspects of both network structure and the norms governing relationships have been included in the analysis presented here at both the family and community level.

Given the broad array of ways in which social capital has been measured and conceptualised, as well as the fact that social capital can exist in many different types of networks and at different social scales, it is worth noting that several different measures could be used as measures of either family level or community level social capital. This is particularly true of measures of community level social capital, an area of social capital research that has been most well developed and contested. For example, while we use the extent of civic group involvement to measure community connections, other approaches could include examining a dichotomous measure of group membership or membership in different types of groups and organisations.

As well, in line with other authors, most notably Putnam, levels of active involvement in groups and organisations could be used (although we have argued elsewhere this is an outcome of social capital, rather than a measure of social capital per se). In addition, civic group membership may not tell us much about other aspects of community life. Another line of inquiry would be to examine the relationship between family life (as defined here) and community social capital but with a focus on the informal relationships that can exist among community members (say, with friends or neighbours). These are perhaps questions for future research.

Analytic strategy

Despite these possibilities, the measures included in this paper relate to both network structure and norms of trust and reciprocity within both family life, and at the community level, and provide scope for a thorough test of the hypotheses presented above. The overarching question of how family life relates to community social capital is explored using a series of hierarchical regression models for each measure of community social capital (sometimes referred to as sequential regression or nested regression). Hierarchical regression allows us to examine the unique contribution of each predictor variable, controlling for the effects of the other variables in the model; and introduce predictor variables into the analysis in a series of steps, at each step assessing what the new set of variable adds to the model, and the effect it has on the relationships already observed in the model (Tabachnik and Fiddel 1996: 149-150).

In the first step, the key family characteristics associated with family change, and of most concern to family decline theorists (marital status, family type, household employment etc.), are included as predictor variables. Thus we establish first whether there is a relationship between community social capital and these characteristics of family life.

In the second step, family social capital measures are included as additional predictor variables. The aim in this step is to establish whether higher levels of community social capital are associated with higher levels of family social capital, consistent with the family decline thesis; and whether any relationships evident in step one are modified by the inclusion of the family social capital variables. If family social capital moderates the relationship between community social capital and the above family characteristics in a way that is consistent with the family decline thesis, we expect the relationships observed in step one (represented by regression coefficients) to weaken (or become smaller) in step two.

In the third step measures of other resources are included as additional predictor variables. At this step we establish whether high levels of community social capital are associated with high levels of economic and human capital and related individual and neighbourhood resources. We are also able to see whether these resources moderate any relationships that were evident in earlier steps between community social capital and other family characteristics.

At each step other variables are controlled for which are not of specific interest in this paper but which the literature suggests might also be related to community social capital, including respondent’s age, whether the respondent speaks a language other than English at home or with their family, and locality type (rural and remote area indicator). A description of the key independent variables in the models is provided in Appendix A and summary statistics for each variable are provided in Appendix B.

Model specifications and sample restrictions

We use three samples in the analysis. Each model is run for men and women separately as it is expected that family life, and the impact of family life on one’s experience of community, will vary by gender. All the men in the survey are included6 except the 15 men living in lone-parent households, as there are too few such men to make reliable comparisons with men in other family types. This provides a sample of 586 men.

Two samples of women are used in analysis. One set of models includes all the women in the survey (n=700), and a second set includes the sub-sample of women with dependent children (n=283). Thus we present two sets of results for women. This is because we have a specific interest in how household social capital relates to community social capital, and our measure of household social capital is constructed from a series of questions that were only asked of people living with dependent children. Thus the first set of models includes all women, and the second set of models includes only the sub-sample of women with dependent children and includes our measure of household social capital as an additional predictor variable.7 These additional analyses could not be conducted for men because the number of cases would be too small to ensure a reasonable case to independent variable ratio.

In comparison to the total sample of women, the sub-sample of women with dependent children contains a higher proportion of women who are married, in paid work, in excellent or very good health, and who own or are purchasing their own home. The women in the sub-sample also have a slightly larger number of relatives on average and a slightly larger number of group memberships. These differences between the total sample and the smaller sub- sample can be observed in Appendix B where summary statistics are provided.

Before running any of the regression models we examined the bivariate relationship each of our two measures of community social capital had with each of the independent variables, as well as correlations between all the predictor variables, to avoid problems of multi-collinearity. These analyses are drawn on in interpreting the results of the regression models.

Finally, it is important to note that we are limited in the extent to which we can explore questions of change with the use of cross sectional data. Although the context of this paper is one of family and community change, the analysis we present here is point in time only.

Methodological issues

The data contain a number of biases. First, as the data were collected via the telephone, those who are highly mobile, or who do not have a telephone or have a silent number, as well as those who “vet” calls via answering machine, are under-represented in this study. As well, the survey was only conducted in English due to limited resources. Where householders were unable to undertake the survey in English, these households were excluded from the study.

Another bias relates to those who chose to participate in the survey. The survey over-represents women (69.2 per cent compared with 31.8 per cent men) and those who are tertiary educated (both because they are more likely to be contacted via the telephone and/or more likely to agree to be interviewed). The survey data have been weighted by sex and education for the purposes of analysis, such that men and women as well as people with all levels of educational qualifications are represented in the data in proportions consistent with national statistics8.

A further source of bias of specific relevance for analysis of social capital should be mentioned. It is possible that those people who responded to the telephone survey are those most likely either to be trusting of organisations and institutions, or to engage in civic or community life. The evidence is mixed. Some social capital research indicates that there is a difference in the social capital of people who take part in social surveys and those who do not (see for example Bullen 2002), while, other research finds no such difference (Keeter et al. 2000). As a result, it is difficult to know if our survey overstates levels of trust and participation in Australian life, (if results are generalised to the whole community). Whether there are any differences in levels of trust and connectedness between those who agreed to participate in the survey and those who did not remains unknown.

3 Interviews were conducted via the telephone using the Institute’s Computer Assisted Telephone Interviewing (CATI) system , and averaged 32 minutes duration. For more information on survey methodology and fieldwork outcomes see the Families Social Capital and Citizenship Fieldwork Report (Hughes and Stone 2002) available on the Australian Institute of Family Studies website at http://www.aifs.gov.au.

4 The different types of groups included groups related to children and parenting, sports/recreation and hobby groups, trade unions, professional and technical associations, political parties, environmental, human rights, community or welfare groups, arts, culture or educational groups and self help or support groups.

5 Eighty five per cent of respondents belonged to at least one group or organisation, and respondents belonged to four groups on average (This figure may be higher than in other studies because many different types of civic and community groups/organisations were prompted for. On the other hand, other studies may under-represent the extent of civic involvement for the same reason). Sporting groups were the most common, followed by community/welfare groups, arts/educational groups, child related groups and professional groups/trade unions.

6 Providing they have given complete information on the variables of interest.

7 There are fewer other independent variables in the second set of models because when we limit the sample to women with children in the household we no longer need to include the child related variables and we also need to collapse categories on a few predictor variables because the smaller number of cases overall means there are too few cases in some categories to make reliable statistical comparisons.

8 For more details see the Families, Social Capital and Citizenship Fieldwork Report (Hughes and Stone 2002) available on the Australian Institute of Family Studies website at http://www.aifs.gov.au.

Family life and community norms

This section presents the results of our analyses of the relationship between family life and norms of generalised trust and reciprocity for men, then women, then women with dependent children. It concludes with a summary of findings on generalised trust and reciprocity.

Relationship between family life and generalised trust and reciprocity for men

Hierarchical regression analyses were used to predict levels of generalised trust and reciprocity. Table 1 presents results from these analyses for men. Column one contains the results from step one, where only the key characteristics of families associated with family change, along with three control variables were included. Column two contains the results from step two, including measures of family social capital as additional predictor variables. Column three contains the results from the third and final step, where measures of other resources available to respondents and their families are included as additional predictor variables.

By focusing first on the results in column one we can examine how variables relating to family type, household employment, and non-traditional family values are related to levels of generalised trust and reciprocity among men. Together these variables, along with the three control variables (respondent age, ethnicity and locality type) accounted for 13.7 per cent of the total variability in levels of generalised trust and reciprocity. We also estimated the model with just the three control variables, and the inclusion of the set of family characteristics increased the level of variance explained by 12.6 per cent. Thus our set of key family characteristics have considerable explanatory power in predicting levels of generalised trust and reciprocity among men.

As can be seen from the coefficients in the first column of the table, many of the individual variables were significantly related to levels of trust and reciprocity among men, but not all of these relationships were consistent with the family decline thesis. Consistent with the decline thesis, and with previous social capital research (Putnam 1996; Hogan and Owen 2000), we found levels of trust and reciprocity to be highest among married men, and lowest among men who were divorced or separated. As can be seen from the coefficient in column one, men who were divorced or separated were estimated to have levels of trust and reciprocity almost 1.5 percentage points lower, on a scale of 0 to 10, than married men (the comparison group) when the other variables in the model were taken into account. This is substantial given a standard deviation of 1.7 on this scale for men. Men in de facto relationships were estimated to have levels of trust and reciprocity about 1 point lower on the scale than married men, and men who were single and had never married were estimated to have levels of trust and reciprocity about half a point lower on the scale.

In his highly influential research on social capital, Putnam (1996) found that trust and civic engagement were highest among those who are both married and have children. Putnam’s research has lent support to the view that marriage and family life are a key source of community social capital. While, as outlined above, we found trust and reciprocity was highest among married men, in contrast to Putnam (1996) we found that men with children in this study had lower levels of generalised trust and reciprocity than men without children (as can be seen from the negative coefficients for children in Table 1). This includes men who had children under 18 years of age living at home and men who had children over 18 and/or living elsewhere only. This effect is net of the respondent’s own age, which is controlled for.

While this finding is inconsistent with Putnam’s (1996) research, more recent Australian research found trust to be lower among people with children. Hughes and Black (2003) found trust of local people to be lower among those with preschool aged children, and trust of “most Australians” to be lower among those with preschool and primary school aged children. They interpret this as suggesting that the vulnerability of young children may make parents more cautious about, or wary of, strangers. There are also other possible interpretations of this finding. For example, research on subjective wellbeing has shown that high levels of trust are associated with high levels of subjective wellbeing – for example, self-reported happiness and satisfaction with life (De Neve and Cooper 1998) – and that men may experience a dip in levels of subjective wellbeing when they have children (which return to higher levels as children get older) (Headey and Wearing 1992: 81). It is also possible that low levels of generalised trust and reciprocity among men with children outside the household are related to the experience of divorce or separation from ones children and former partner. It is plausible that men in these circumstances may be less trusting of people generally.

The other concern of family decline theorists is the increased labour force participation of women. The decline thesis suggests that trusting and reciprocal relationships are fostered by the exigencies of interdependence which in the family realm have traditionally been brought about by financial necessity and the specialisation of roles. However, as can be seen in Table 1, we found no evidence that having a partner in paid employment, either full-time or part- time, was related to generalised trust and reciprocity among men. As the coefficient for “support for independence in relationships” shows, we also found that men who were supportive of the idea of independence in intimate relationships had higher levels of trust and reciprocity than men who had more traditional attitudes to relationships. While this survey question did not refer to financial independence specifically, and captures values and ideals rather than the practice of independence, in conjunction these findings suggest that high levels of personal independence in work and family relationships are not inconsistent with strong norms of trust and reciprocity.

Men’s own employment status was related to their level of trust and reciprocity. Men who were unemployed or not in the labour force were estimated to have levels of trust and reciprocity that were significantly lower than other men when the other variables in the model were taken into account (as can be seen by the negative coefficients for the other labour force categories in column one). This may be because men who are unemployed or not in the labour force have fewer financial resources than other men, measures of which we include as predictor variables in step three. It is also noteworthy however that there was no apparent difference in levels of trust and reciprocity between those who worked 50 or more hours per week, those who worked under 50 hours per week9, and those who were retired, when the other variables in the model were taken into account.

Step two included measures of family social capital. As a result, the level of variance in generalised trust and reciprocity explained improved significantly, rising by 4 per cent. The coefficients for the individual variables show that for men, two measures of family social capital were associated with levels of generalised trust and reciprocity in a way that is consistent with the family decline thesis. First, having a large kin network is associated with higher levels of generalised trust and reciprocity. Second, having high levels of trust and reciprocity in kin networks is associated with higher levels of generalised trust and reciprocity. These findings could be seen to provide support for the view that family social capital – at least that which is kin based – is a possible source of community social capital, and therefore changes that threaten kin based social capital may also threaten community. Although it should be noted that those who have low levels of trust in people generally may as a result have poorer quality relationships with relatives.

Including family social capital in the model also modified the relationships apparent in step one. The coefficient for de facto relationships dropped by about 14 per cent (from -1.05 to -0.9), the coefficient for divorced or separated dropped by about 17 per cent (from –1.46 to –1.21), and the coefficient for never married decreased in size and was no longer statistically significant. Thus it appears that one of the reasons unmarried men have lower levels of generalised trust and reciprocity may be that they have lower levels of family social capital – the main explanation provided by family decline theorists. However, as the coefficients for de facto and divorced/separated still remain fairly large and statistically significant in step two, access to family social capital can not be the only reason that marriage and divorce are associated with lower levels of generalised trust and reciprocity.

The other notable change was that respondents’ attitude to independence in relationships was no longer significant when family social capital variables were included. This suggests that men who support independence in intimate relationships have higher levels of generalised trust and reciprocity because they have higher levels of kin based social capital. This finding is interesting given the concern among family decline theorists that women’s increased independence is a threat to family life (even though it is not a direct indicator of the practice of independence).

Table 1. Hierarchical regression analyses predicting levels of generalised trust and reciprocity among men
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics
live with kid/s (< 18 yrs) -0.85* -0.71* -0.47
older kid/s in house or kids out house only -0.77* -0.77* -0.54*
defacto -1.05* -0.90* -0.82*
single and never married or widowed -0.56* -0.42 -0.15
single and separated or divorced -1.46* -1.21* -0.75*
in paid work 50 or more hrs per week 0.74* 0.66* 0.12
in paid work <50 hrs per week 0.81* 0.81* 0.23
retired 0.78* 0.86* 0.39
has partner in paid work 35+hrs per week -0.27 -0.26 -0.22
has partner in paid work <35 hrs per week  0.29 0.21 0.12
support for working mothers i 0.03 0.02 -0.01
support for independence in relationships i 0.12* 0.08 0.09
Family social capital      
number of relatives i   0.01* 0.01
trust & reciprocity among relatives i   0.24* 0.19*
low civic activity in family of origin   -0.12 0.001
high civic activity in family of origin   0.15 0.24
Other resources
incomplete secondary school     -0.004
tertiary qualified     -0.02
financially comfortable     0.08
financially struggling     -0.41
home owner     0.11
excellent or very good health     0.01
poor health     -0.66*
advantaged area (highest quintile)     0.18
disadvantaged area (lowest quintile)     0.15
safety of neighbourhood i     0.19*
Control variables
age i 0.02* 0.02* 0.02*
language other than English -0.18 -0.2 -0.31
rural or remote area 0.36* 0.33* 0.348
Constant 5.02 2.96 2.67
N 581 581 581
R Square 0.16 0.2 0.28
Adjusted  R Square 0.14 0.18 0.24
Multiple R 0.4 0.45 0.53
Df 15, 565 19, 561 29, 551
F 7.14 7.54 7.24
R Square change 0.13# 0.04 0.07
F change 7.04* 7.76* 5.51*

Notes: 1. * indicates that underlying coefficient is significant at the 5 per cent confidence level.
2. i=interval (or continuous) variable.
3. Omitted categories are: no kids; married; unemployed or not in the labour force; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good health; no language other than English spoken at home or with family; live in a capital city or other metropolitan area; live in an area of average socio-economic advantage.
4. #=change in R Square when family change variables are added to model subsequent to the three control variables.

In the third step measures of other resources were included. This improved the overall explanatory power of the model, increasing the level of variance explained by about 7 per cent. Of the set of resource variables, respondent health and perceptions of neighbourhood safety were the strongest individual predictors.10 As can be seen from the coefficient in column three, men who were in poor health were estimated to have levels of trust and reciprocity almost 0.66 points lower (on a scale of 0 to 10) than men who were in good health (the comparison group). Men who felt their neighbourhoods were safe places to live reported substantially higher levels of generalised trust and reciprocity than other men, with each 1 point increase in sense of safety (on a scale of 0 to 10) corresponding with an increase of 0.19 in levels of generalised trust and reciprocity (on the same scale).

Including these resource measures also modified the other relationships in the model. The coefficients for having children aged less than 18 years at home and for respondent employment status both dropped in size and were no longer significant. This suggests that the reason men with children at home and men who are unemployed have lower levels of generalised trust and reciprocity than other men may be that men with these characteristics are relatively poor in respect to these other resources. That is, unemployment may be associated with low levels of generalised trust and reciprocity for men because unemployment is associated with financial strain and poor health (physical or psychological) for men.11 Similarly, living with children may be associated with low levels of generalised trust and reciprocity for men because living with children is associated with financial strain, which is in turn related to low levels of generalised trust and reciprocity. However, caution must be taken in drawing such conclusions because we can not determine the direction of causality.

The coefficients for the relationship status variables also changed when other resources were taken into account in the model. The coefficient for de facto dropped by about 10 per cent (from -0.9 to -0.82), and the coefficient for divorced or separated dropped by more than 35 per cent (from –1.21 to -0.75). Thus men who are married may have high levels of generalised trust and reciprocity in part because they have higher levels of family social capital, which is in turn positively associated with generalised trust and reciprocity, and in part because they have greater access to other resources like human and financial capital, which are also associated with generalised trust and reciprocity. Lack of access to economic resources appears particularly important in explaining the exceptionally low levels of trust and reciprocity reported by men who are separated or divorced. Bivariate analyses confirm that men who were divorced or separated had below average levels of education, health, home ownership and self-reported financial wellbeing, and were more likely to live in socio- economically disadvantaged and unsafe areas.

However, neither family social capital nor other resources can fully explain the relationship between marital status and levels of generalised trust and reciprocity, as marital status was still the strongest predictor of generalised trust and reciprocity in the final model, when all of the other variables were taken into account. Other factors not measured here that might explain the difference include personality factors that make men both more likely to be married, and more likely to be trusting (De Neve and Cooper 1998).

Finally, the coefficient for kin based trust and reciprocity dropped by about 20 per cent (from -0.24 to -0.19) and the coefficient for kin network size decreased in size and was no longer statistically significant when these other resources were taken into account. This suggests that high levels of trust and reciprocity in kinship networks may be associated with high levels of generalised trust and reciprocity for men in part because high levels of trust and reciprocity in kinship networks are associated with high levels of financial and human capital for men. As well, the size of men’s kinship networks may not matter in itself. Rather, it may be something about the other resources of men who have large kin networks that leads to, or is associated with, high levels of trust and reciprocity for men.

Relationship between family life and generalised trust and reciprocity for women

Table 2 presents results from regression analyses for the total sample of women. Column one shows that key family characteristics (family type, household employment, and family values) had less bearing on levels of trust and reciprocity among women than they did among men.12 Together these variables, along with the three control variables, accounted for 8.5 per cent of the total variability in levels of generalised trust and reciprocity among women. When we estimated the model with just the three control variables and then added the family characteristics in a second step the set of family characteristics increased the level of variance explained by only 2.8 per cent.

The one variable in this set that was significantly related to levels of generalised trust and reciprocity among women was attitudes to the workforce participation of women with dependent children. Women who agreed with the statement “Women who work can have just as good a relationship with their children as women who do not work” had higher levels of generalised trust and reciprocity than women who disagreed with this statement, regardless of whether they were actually in paid work or had children. Thus, as with men, it appears that non- traditional attitudes to relationships and gender roles are associated with high levels of generalised trust and reciprocity among women.

None of the other core family characteristics entered in step one were related to generalised trust and reciprocity. Interestingly, generalised trust and reciprocity was not related to marital status nor the presence of children, unlike for men. Thus we found no support for the family decline thesis in the first stage of analysis with this sample of women.

By including the family social capital variables in the second step, the level of variance explained increased by 4.1 per cent. As in the men’s model above, generalised trust and reciprocity was higher among those with large kin networks and those with high levels of trust and reciprocity in kin networks.

Including family social capital did not alter the relationship between generalised trust and reciprocity and attitudes to women’s employment – women who supported the idea of mothers combining paid work and parenting had higher levels of generalised trust and reciprocity than other women, regardless of the level of social capital within their kin networks. Thus the only assumption of the family decline thesis that could be supported by these analyses for women was that social capital in kin networks is positively associated with community social capital. There was no evidence among this sample of women that high levels of kin based social capital in turn relates to family type or marital status, unlike for the sample of men, or to any of the other characteristics commonly associated with family change such as women’s employment and attitudes to relationships and gender roles.

Finally, when measures of other resources were included, the level of variance explained rose by 12.8 per cent. Thus this set of variables appeared to explain most of the variation in levels of generalised trust and reciprocity among women. As with men, respondent health and perceptions of neighbourhood safety were the strongest individual predictors. Those in poor health were estimated to have levels of trust and reciprocity about five per cent lower than those in good health, and every one point increase in feelings of safety in one’s neighbourhood (on a scale of 0 to 10) was estimated to increase levels of generalised trust and reciprocity by .25 (on a scale of 0 to 10).

The addition of other resources in the model also moderated the relationships observed in earlier steps. First, the coefficient for support for working mothers dropped by more than 35 per cent (from .07 to 0.48). This suggests that one of the reasons support for working mothers is associated with high levels of generalised trust and reciprocity among women is that support for working mothers is more common among women with high levels of economic and human capital. However, as the coefficients remained statistically significant this is not likely to be the only reason support for working mothers is positively related to generalised trust and reciprocity among women.

Table 2. Hierarchical regression analyses predicting levels of generalised trust and reciprocity among women
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics
live with kid/s (< 18 yrs) 0.08 0.15 0.15
older kid/s in house or kids out house only 0.02 -0.01 0.17
defacto -0.25 -0.21 -0.16
single and never married or widowed -0.19 -0.16 -0.08
single and separated or divorced -0.26 -0.14 0.09
in paid work 35 or more hours per week -0.12 -0.16 -0.13
in paid work <35 hours per week 0.08 -0.02 0.01
unemployed -0.06 0.02 0.04
retired -0.33 -0.41 -0.27
support for working mothers i 0.06* 0.07* 0.05*
support for independence in relationsh i 0.03 0.02 0.02
Family social capital      
number of relatives i   0.01* 0.01
trust & reciprocity among relativesi   0.20* 0.13*
low civic activity in family of origin   0.08 0.03
high civic activity in family of origin   0.14 0.09
Other resources
incomplete secondary school     0.27*
tertiary qualified     0.26
financially comfortable     0.13
financially struggling     -0.01
home owner     0.02
excellent or very good health     0.07
poor health     -0.51*
advantaged area (highest quintile)     -0.11
disadvantaged area (lowest quintile)     0.12
safety of neighbourhoodi     0.25*
Control variables
agei 0.03* 0.04* 0.02*
language other than English 0.02 0 -0.1
rural or remote area 0.12 0.06 -0.19
Constant 4.76 2.79 2.04
N 700 700 700
R Square 0.08 0.13 0.25
Adjusted  R Square 0.07 0.1 0.22
Multiple R 0.29 0.35 0.5
Df 14, 685 18, 681 28, 671
F 4.48 5.4 8.1
R Square change 0.03# 0.04 0.13
F change 1.91* 7.98* 11.46*

Notes: 1. * indicates that underlying coefficient is significant at the 5 per cent confidence level.
2. i=interval (or continuous) variable.
3. Omitted categories are: no kids; married; not in the labour force; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good health; live in an area of average socio-economic advantage; no language other than English spoken at home or with family; live in a capital city or other metropolitan area.
4. #=change in R Square when family change variables are added to model subsequent to the three control variables.

The coefficients for the family social capital variables also changed, with the coefficient for trust and reciprocity in kinship networks dropping by about 35 per cent (from 0.20 to 0.13), and the coefficient for the size of kinship networks dropping and becoming non-significant. Thus, as we found with men, it appears that part of the reason low levels of family social capital are associated with low levels of generalised trust and reciprocity among women is that women with low levels of family social capital also have low levels of economic and human capital.13

Relationship between family life and generalised trust and reciprocity for women with dependent children

Table 3 presents results from regression analyses for the sub-sample of women with dependent children. As noted in the analytic approach outlined above, these additional analyses were conducted for women primarily so as to be able to examine the relationship that norms of generalised trust and reciprocity have with household social capital.14 However, as can be seen in Table 3, these analyses revealed that several predictor variables of interest in this paper had a different relationship with levels of generalised trust and reciprocity for the sub-set of women with dependent children than they did for the entire sample of women.

Column one shows that among women with dependent children, those who were working full-time in paid employment were estimated to have levels of generalised trust and reciprocity lower than those who were working part-time or not at all. As can be seen from the coefficients in column one, while there was not a statistically significant difference between mothers who were not working and those who were working part-time, mothers working more than 30 hours a week were estimated to have levels of generalised trust and reciprocity at least 0.5 percentage points lower than mothers who were not working at all in a paid job (on a scale of 0 to 10). This is consistent with Putnam’s (1996) finding that women working part-time had the highest levels of trust.

Otherwise the results in the first step of the analysis were similar to those derived for the total sample of women: levels of generalised trust and reciprocity did not differ by relationship status but were higher among women who supported the idea of mothers’ combining paid work with child rearing. The former finding is worth highlighting with this sub-sample of women because it suggests there is no difference between lone mothers and couple mothers in respect to levels of generalised trust and reciprocity, although lone-parent families are a core concern of family decline theorists.

In total, the block of key family characteristics improved the level of variance explained over and above the control variables by 5.8 per cent. Thus, although it is not standard to compare these statistics across different samples and models, on the face of it this block of key family characteristics appeared to explain more of the overall variance in levels of generalised trust and reciprocity among women with dependent children than it did among the total sample of women.This presumably reflects the significance of hours spent in paid employment to women with dependent children.

As with the previous models, family social capital variables were entered in the second step, but this time household social capital was included as an additional predictor variable. Again the set of family social capital variables significantly improved the level of variance explained by the model, which rose by 7.4 per cent. As for the effect of individual predictors we found the impact of kin based social capital was consistent with results for the total sample of women and the sample of men: high levels of kin trust and reciprocity were associated with high levels of generalised trust and reciprocity. However, in respect to household social capital we found the opposite pattern – we found that high levels of household social capital were associated with low levels of generalised trust and reciprocity. Thus high levels of trust, reciprocity and engagement within the household appeared to be at odds with high levels of community level trust and reciprocity, at least among women with dependent children.

Thus we found no evidence to support the presumption that strong relationships with those with whom you live necessarily translate into higher levels of trust and reciprocity within the broader community. Rather, this finding in respect to household social capital could be seen as providing support for the view that a strong emphasis on family can be detrimental for community – at least for women with dependent children.

Table 3. Hierarchical regression analyses predicting levels of generalised trust and reciprocity among women with dependent children
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics      
defacto -0.38 -0.38 -0.59
lone parent -0.4 -0.4 -0.19
in paid work 35 or more hours per week -0.55* -0.59* -0.49*
in paid work <35 hours per week  -0.12 -0.27 -0.25
support for working mothers i 0.08* 0.08* 0.06*
support for independence in relationships i 0.13 0.12 0.08
Family social capital      
household social capital i   -0.14* -0.15*
number of relatives i   0.001 0.001
trust & reciprocity among relatives i   0.22* 0.14*
low civic activity in family of origin   0.33 0.09
high civic activity in family of origin   0.32 0.14
Other resources      
incomplete secondary school     0.29
tertiary qualified     0.28
financially comfortable     0.16
financially struggling     0.12
home owner     -0.14
excellent or very good health     0.24
advantaged area (highest quintile)     0.04
disadvantaged area (lowest quintile)     0.08
safety of neighbourhood i     0.28*
Control variables      
age i 0.02 0.03* 0.02
language other than English -0.03 -0.07 -0.13
rural or remote area -0.19 -0.22 -0.35
Constant 5.07 4.23 3.45
N 283 283 283
R Square 0.07 0.14 0.3
Adjusted  R Square 0.04 0.1 0.24
Multiple R 0.26 0.38 0.55
Df 9,273 14,268 23,259
F 2.21 3.16 4.89
R Square change 0.06# 0.07 0.16
F change 2.81* 4.61* 6.65*

Notes: 1. Sample excludes 7 lone mothers with children under 5 years only.
2. * indicates that underlying coefficient is significant at the 5 per cent confidence level,
3. i=interval (or continuous) variable.
4. Omitted categories are: married; not in paid employment; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good or poor health; live in an area of average socio-economic
5. #=change in R square when family change variables are added to the model subsequant to the three control variables.

One possible explanation for this pattern is that women with dependent children are faced with a trade-off between family and community engagement as a result of constraints of time associated with child rearing – a trade-off that is not as pertinent to men or to women without children. However, it is not possible to draw conclusions about the direction of causality on the basis of these analyses. It is equally plausible that mothers who have low levels of trust in people generally (including strangers), and/or who perceive people generally to be unhelpful will be more likely to focus on developing close relationships within the family household.

We explored this relationship further in supplementary analyses using the sub- sample of mothers in couple families, i.e. those with a partner and children at home (n=250). In this sub-sample we found that household social capital was not significantly related to generalised trust and reciprocity.15 Thus household social capital only had a negative relationship with generalised trust and reciprocity for women in lone-parent families. This was confirmed in bivariate analyses that showed a negative correlation between household social capital and generalised norms for lone mothers and a positive but not significant correlation between these two variables for couple mothers. Therefore if women with dependent children are faced with a trade-off between family and community engagement as a result of constraints of time associated with child rearing, this appears to occur only for mothers without the support of a partner.

As can be observed in column two, including our measures of family social capital did not significantly affect the other relationships. Thus the lower levels of generalised trust and reciprocity among mothers who work full-time are not a result of women working full-time having low levels of family social capital.
Measures of other resources were included in the model in the third step and results were similar as those for the total sample of women. The main thing to note in this sample is that when these resources were taken into account the coefficient for full-time paid employment dropped by about 20 per cent (from .59 to .49) but remained relatively substantial in size and statistically significant. Thus women working full-time were estimated to have lower levels of generalised trust and reciprocity even when other resources were taken into account.

Interestingly also, the coefficient for household social capital actually increased when these other resources were taken into account (by about 10 per cent). Thus there is no evidence that access to human and financial capital explains the negative relationship between household social capital and community social capital. Nonetheless, it is likely that there is a complex interaction between family type, household social capital, community social capital, and the other resources available to individuals and families.

Summary of findings on generalised trust and reciprocity

To summarise our findings in respect to generalised trust and reciprocity, we found some support for the family decline thesis, but more among men than among women. We found both marital status and the presence of children were important predictors of generalised trust and reciprocity among men, with marital status being perhaps the strongest predictor of all the variables examined. Men who were divorced or separated and men who were in de facto relationships were estimated to have levels of trust and reciprocity significantly lower then men who were married, consistent with previous research (Putnam 1996; Hogan and Owen 2000). This appeared to be in part explained by the higher levels of trust and reciprocity in the kin networks of married men, consistent with the assumptions of the family decline thesis, and in part by the other resources of married men, including high levels of financial and human capital.

The core family characteristics associated with family change appeared to have less bearing on levels of generalised trust and reciprocity among women. However, this does not mean that family type does not matter for women. One interesting finding was that for women in lone-parent families high levels of household social capital were associated with low levels of generalised trust and reciprocity. This relationship was not apparent for couple mothers. One possible interpretation of this finding is that some kind of trade-off exists between household social capital and generalised norms of trust and reciprocity for women in lone-parent families because of the time and resource constraints they face. However, it is also possible that having low levels of trust in people generally leads mothers without partners to focus more on relationships within the family household.

Another conclusion to draw from these findings is that we can not assume that high levels of family social capital will always translate into high levels of community social capital – an assumption that underlies the family decline thesis and is common among social capital theorists and policy makers. While strong norms of trust and reciprocity in kin based relations were related to trust of people more generally in the community, including strangers, and a perception that people generally try to be helpful, high levels of household social capital appear to be associated, if anything, with low levels of generalised trust and reciprocity, at least for women in lone-parent families.

We also found that working full-time was associated with lower levels of generalised trust and reciprocity among women with dependent children, consistent with previous social capital research (Putnam 1996 found the highest levels of trust among women working part-time). On the face of it this may be seen as further support for the family decline thesis. However, inconsistent with the main explanation provided by the family decline thesis, low levels of generalised trust and reciprocity among women working full-time did not appear to be explained by lower levels of family social capital.
One possible interpretation of this finding is that mothers working full-time have less time to engage in other activities, all else being constant, that build trust and a sense of cooperation in the community. However, it is possible that the experience of being a parent and working full-time lowers womens levels of trust and reciprocity by some other machanism, or reflects some other unmeasured characteristic of mothers working full-time.

The interpretation of this finding as a resource issue, rather than a question of values, is reinforced by our findings in respect to attitudes to women’s employment. We found levels of generalised trust and reciprocity were higher among women who support maternal employment than among women with more traditional attitudes. In addition, support for independence in intimate relationships was associated with high levels of generalised trust and reciprocity among men, and this appears to be at least in part because support for independence in intimate relationships among men is associated with high levels of family based social capital. Thus overall non-traditional attitudes to relationships and gender roles were associated with high levels of generalised trust and reciprocity.

We now turn attention to the question of whether these same factors predict membership in community and civic groups and organisations, a second measure of community social capital.

9 We constructed the time in paid work variable as greater or less than 50 hours per week for men as this was the median number of hours worked among men in paid employment.

10 We also re-estimated step three of the model excluding health and neighbourhood safety and found that living in a socio-economically advantaged area was associated with higher levels of trust and reciprocity.

11 This interpretation is consistent with bivariate analyses.

12 We constructed the time in paid work variable as greater or less than 35 hours per week for women as this was the median number of hours worked for women in paid employment.

13 In bivariate analyses high levels of trust and reciprocity in kin based networks were positively correlated with self-reported financial wellbeing, home ownership, health and sense of safety in ones neighbourhood.

14 Our measure of household social capital was constructed from a set of items that were only asked of respondents living in households with dependent children. We were unable to conduct these additional analyses for men as the sub-sample of men living in households with dependent children was too small to ensure a reasonable case to independent variable ratio.

15 Full results available from the authors on request.

Family life and group membership

This section presents findings on the relationship between family life and group membership for men and women in the survey. Again, the section is broken into four parts: findings for the sample of men; findings for the total sample of women; findings for the sub-sample of women with dependent children; and a drawing together of the results.

Preliminary bivariate analyses showed there was little difference in the overall number of groups men and women belong to16, but levels of group membership were higher in the sub-sample of women with dependent children than in the total sample of women.

Relationship between family life and group membership for men

Table 4 presents results of the analysis of group membership for men. Results from the first step of analysis are presented in column one. In step one we examined how variables relating to family type, household employment, and non-traditional family values related to levels of group membership. These variables, along with the three control variables (respondent age, ethnicity and locality type), accounted for only 4 per cent of the variability in levels of group membership among men. This core set of family characteristics thus appeared to have less influence on group membership for men than it did on their levels of generalised trust and reciprocity.

However, two of these variables were related to group membership in the model and in a similar way as they were to norms of generalised trust and reciprocity. First, levels of group membership were significantly lower among men who were divorced or separated than among married men (the comparison group), consistent with the decline thesis, and with other Australian research in this area (Hughes and Black 2003: 8). As can be seen from the coefficient in column one, men who were divorced or separated were estimated to belong to 1.14 fewer groups than married men (when the other variables in the model were taken into account). This is a considerable difference given a mean and standard deviation of only 3.8 (see Appendix B). However, unlike for generalised trust and reciprocity, there was no difference between men who were in de facto relationships and men who were married in respect to levels of group membership.

Also consistent with the results for generalised trust and reciprocity and inconsistent with the family decline thesis, men who had non-traditional attitudes towards gender roles had higher levels of group membership. Specifically, men who valued some degree of independence in intimate relationships were estimated to have higher levels of group membership, when other variables in the model were taken into account.
When we entered measures of family social capital into the model (step two) the level of variance in levels of group membership explained rose by 4.5 per cent. As can be seen in column two of Table 4, men who said that civic activity was common in the family they grew up in were estimated to belong to a larger number of groups than other men. Thus the likelihood of men becoming members of civic groups depends in part on the nature of the norms and values within one’s family of origin, and the extent to which they include norms of civic engagement.

Including measures of family social capital in the model did little to alter the relationships apparent in step one. The coefficient for being divorced or separated and the coefficient for support for independence in relationships actually increased in size somewhat. Thus these relationships appeared at least as strong, if not stronger, after controlling for levels of family social capital.

In the third and final step, measures of other resources were added to the model. As Table 4 shows, these other resources were also important predictors of group membership for men, increasing the level of variance explained by the model by about 5 per cent. Of the set of resource variables included, home ownership, respondent health and the level of socio-economic disadvantage of the area were each significant individual predictors.

As can be seen in column three, men who owned or were purchasing their own home were estimated to belong to about 1.1 more groups than men who were renting or in other forms of accommodation, men with excellent or very good health were estimated to belong to 0.6 more groups than men who were in good health (the comparison group), and men living in socio-economically disadvantaged neighbourhoods were estimated to belong to 1.2 fewer groups on average than men living in areas of average socio-economic status.

Table 4. Hierarchical regression analyses predicting levels of group membership for men
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics
live with kid/s (< 18 yrs) 0.62 0.55 0.31
older kid/s in house or kids out house only 0.61 0.48 0.64
defacto 0.05 -0.11 0.03
single and never married or widowed -0.4 -0.41 -0.31
single and separated or divorced -1.14* -1.17* -0.6
in paid work 50 or more hrs per week 0.4 -0.01 -0.88
in paid work <50 hrs per week 0.47 0.12 -0.71
retired -0.3 -0.38 -0.74
has partner in paid work 35 or more hrs per week 0.56 0.68 0.64
has partner in paid work <35 hrs per week  -0.15 -0.08 -0.21
support for working mothersi 0.05 0.07 0.07
support for independence in relationships i 0.21* 0.24* 0.27*
Family social capital
number of relatives i   0.01 0.01
trust & reciprocity among relatives i   -0.2 -0.29*
low civic activity in family of origin   0.13 0.1
high civic activity in family of origin   1.40* 1.38*
Other resources
incomplete secondary school     -0.01
tertiary qualified     0.57
financially comfortable     0.3
financially struggling     0.27
home owner     1.069*
excellent or very good health     0.65*
poor health     -0.61
advantaged area (highest quintile)     -0.19
disadvantaged area (lowest quintile)     -1.19*
safety of neighbourhood i     0.03
Control variables
age i -0.01 -0.01 -0.03
language other than English 0.16 0.18 0.1
rural or remote area -0.38 -0.29 -0.21
Constant 2.59 3.83 4.37
N 579 579 579
R Square 0.07 0.11 0.16
Adjusted  R Square 0.04 0.08 0.12
Multiple R 0.25 0.33 0.4
df 15, 563 19, 559 29, 549
F 2.59 3.62 3.66
R Square change 0.05# 0.05 0.05
F change 2.70* 7.06* 3.44*

Notes: 1. * indicates that underlying coefficient is significant at the 5 per cent confidence level.
2. i=interval (or continuous) variable.
3. Omitted categories are: no kids; married; unemployed or not in the labour force; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good health; no language other than English spoken at home or with family; live in a capital city or other metropolitan area; live in an area of average socio-economic advantage.
4. #=change in R Square when family change variables are added to model subsequent to the three control variables.

Including these resource variables in the model affected the relationships evident in step one. The coefficient for separated or divorced became smaller in size and was no longer statistically significant, which suggests that the reason men who are divorced or separated belong to fewer groups than married men is because they have fewer economic and human capital resources. However, the coefficient for the attitudinal item again rose in the final model, suggesting that placing value on independence in intimate relationships is even more strongly associated with group membership when other resources are taken into account.

Finally, including these resource variables in the model also affected the coefficients for family social capital. The coefficient for trust and reciprocity in kin networks actually increased by more than 30 per cent (from -0.20 to -0.29) and became statistically significant. Interestingly also the coefficient was negative, which indicates that men with higher levels of trust and reciprocity in kinship networks have lower levels of group membership than men with lower levels of trust and reciprocity in kinship networks. This suggests that there may be something of a trade-off occurring for these men, whereby investment is made in either kinship relationships or civic life, but not strongly in both, on average.

Table 5. Hierarchical regression analyses predicting levels of group membership for women
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics
live with kid/s (< 18 yrs) 0.14 0.18 0.36
older kid/s in house or kids out house only -0.81 -0.84 -0.53
defacto -0.47 -0.24 -0.08
single and never married or widowed -0.25 -0.3 0.04
single and separated or divorced 0.08 0.24 0.55
in paid work 35 or more hours per week 0.1 0.15 -0.38
in paid work <35 hours per week 0.48 0.42 0.13
unemployed 0.66 0.68 0.83
retired -0.07 -0.02 -0.15
support for working mothers i 0.05 0.06 0.05
support for independence in relationships i -0.003 0.01 -0.02
Family social capital
number of relatives i   0.01 0.01
trust & reciprocity among relatives i   0.04 -0.01
low civic activity in family of origin   0.39 0.22
high civic activity in family of origin   1.58* 1.38*
Other resources
incomplete secondary school     -0.25
tertiary qualified     1.40*
financially comfortable     0.34
financially struggling     -0.95*
home owner     0.38
excellent or very good health     0.27
poor health     -0.68
advantaged area (highest quintile)     0.37
disadvantaged area (lowest quintile)     0.23
safety of neighbourhood i     -0.02
Control variables
age i 0.03 0.03* 0.03*
language other than English -0.15 -0.17 -0.13
rural or remote area -0.08 -0.1 0.25
Constant 2.18 0.79 0.7
N 700 700 700
R Square 0.02 0.07 0.13
Adjusted  R Square 0.002 0.05 0.09
Multiple R 0.15 0.27 0.36
Df 14, 685 18, 681 28, 671
F 1.1 3.03 3.58
R Square change 0.02# 0.05 0.06
F change 1.37 9.61* 4.31*

Notes: 1. * indicates that underlying coefficient is significant at the 5 per cent confidence level.
2. i=interval (or continuous) variable.
3. Omitted categories are: no kids; married; not in the labour force; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good health; live in an area of average socio-economic advantage; no language other than English spoken at home or with family; live in a capital city or other metropolitan area.
4. #=change in R Square when family change variables are added to model subsequent to the three control variables.

Relationship between family life and group membership for women

Table 5 presents results of group membership for the total sample of women. Column one shows that none of the individual family characteristics examined in step one were significantly related to overall levels of group membership for women. Together these variables (along with respondent age, ethnicity and locality type) did not explain a significant amount of variation in the number of groups and organisations women belonged to. Thus, again we found that these key family characteristics have less relevance for women than men.
When family social capital was entered into the model (in step two) the variance in levels of group membership explained rose by around five per cent. As for men above, high levels of civic activity in one’s family of origin were associated with high levels of group membership in this sample of women. As can be seen in column two this is in fact the only family social capital variable associated with group membership among women in their adult life.

Including measures of other resources in the third step again improved the predictive power of the model, increasing the level of variance explained by about six per cent. Of the set of resource variables included, education and self- reported financial wellbeing were the strongest individual predictors for women. As can be seen from the coefficient in column three, women who had tertiary levels of education were estimated to belong to about 1.4 more groups than women who had average levels of education, and women who said that they were struggling financially were estimated to belong to 1 less group than women who said they were getting by financially. Thus, while we found health and neighbourhood safety to be the strongest predictors of generalised trust and reciprocity among the total set of resource variables included, more direct measures of human and financial capital – education and financial wellbeing – appear to be the strongest predictors of community group membership.

Including these resource variables in the model also affected the coefficients for family social capital. The coefficient for high civic activity in one’s family of origin dropped by about 15 per cent (from 1.58 to 1.38). Thus one of the reasons that high levels of civic participation in one’s family of origin were associated with high levels of group membership is that respondents who had high levels of civic participation in their family of origin also had relatively high levels of human and economic capital.

Relationship between family life and group membership for women with dependent children

Table 6 presents results from the models of group membership for the sub- sample of women with dependent children. Column one shows that, as for the entire sample of women, none of the core family characteristics entered in step one are significantly related to levels of group membership for women with dependent children. That is, variables representing family change do not appear to be directly related to levels of community group membership for women.

When measures of family social capital were entered into the model in step two, the predictive power of the model improved 6.5 per cent. Interestingly, those who had high levels of household social capital were estimated to belong to a larger number of groups than those with lower levels of household social capital. Thus while high levels of household social capital were found to be related to low levels of generalised trust and reciprocity in the previous section, high levels of household social capital appear to be conducive to community group membership in this sample of women.

Given our finding in the previous section that the relationship between household social capital and generalised trust and reciprocity differed for lone-parent and couple mothers, we also examined the relationship between household social capital and group membership separately for women in lone-parent and couple family households (in supplementary analyses17). What we found is that it is only women in couple families for whom there is a positive correlation between household social capital and group membership, and when the analysis is restricted to couple mothers, the coefficients for household social capital are substantially bigger (0.46 in step two and 0.49 in the final model). In contrast to couple mothers, for women in lone-parent families there is a negative but non- significant relationship between the two variables. Therefore, consistent with our analyses in the previous section, we have found the relationship between household social capital and community social capital varies by family type.

Table 6. Hierarchical regression analyses predicting levels of group membership for women with dependent children
Variable Step 1
Family change variables
Step 2
Family social capital
Step 3
Other resources
Key family characteristics
defacto -1.09 -0.81 -0.58
lone parent 0.14 0.43 0.9
in paid work 35 or more hours per week -0.24 -0.07 -0.45
in paid work <35 hours per week  0.52 0.52 0.39
support for working mothers i 0.03 -0.004 -0.01
support for independence in relationships i -0.02 -0.01 -0.03
Family social capital  
household social capital i   0.31* 0.32*
number of relatives i   0 0
trust & reciprocity among relatives i   -0.03 -0.09
low civic activity in family of origin   -0.28 -0.31
high civic activity in family of origin   1.54* 1.41*
Other resources
incomplete secondary school     0.06
tertiary qualified     1.57*
financially comfortable     0.27
financially struggling     -0.83
home owner     0.66
excellent or very good health     0.13
advantaged area (highest quintile)     0.41
disadvantaged area (lowest quintile)     0.32
safety of neighbourhood i     -0.03
Control variables
age i 0.003 0.02 0.003
language other than English 0.35 0.28 0.14
rural or remote area -0.62 -0.56 -0.36
Constant 3.78 0.48 0.52
N 283 283 283
R Square 0.03 0.09 0.14
Adjusted  R Square -0.01 0.04 0.06
Multiple R 0.16 0.3 0.37
df 9, 273 14, 268 23, 259
F 0.81 1.92 1.81
R Square change 0.02# 0.07 0.05
F change 0.86 3.84* 1.59

Notes: 1. Sample excludes 7 lone mothers with children under 5 years only.
2. * indicates that underlying coefficient is significant at the 5 per cent confidence level.
3. i=interval (or continuous) variable.
4. Omitted categories are: married; not in paid employment; average level of civic activity in family of origin; yr 12 certificate, trade or apprenticeship qualification; average financial wellbeing; non home owner; good or poor health; live in an area of average socio-economic advantage; no language other than English spoken at home or with family; live in a capital city or other metropolitan area.
5. #=change in R Square when family change variables are added to model subsequent to the three control variables.

When measures of other resources were included in the third step, results were similar to those for the total sample of women. The main thing to note is that when other resources were taken into account the coefficient for household social capital increased by about 4 per cent. This also happened in respect to generalised trust and reciprocity, and suggests that the relationship between household social capital and community social capital is even stronger after controlling for resources such as human and financial capital.

Summary of findings on group membership

To summarise our findings in this section, the family characteristics most commonly associated with family change by family decline theorists had less impact on group membership than they did on levels of generalised trust and reciprocity. Where relationships did exist they applied only to men, and were generally consistent with those found for generalised trust and reciprocity: men who were divorced or separated had significantly lower levels of group membership than other men (consistent with the decline thesis), but men who supported some degree of independence in intimate relationships reported higher levels of group membership than other men (inconsistent with the decline thesis).

There was also evidence to suggest that the main reason men who are divorced or separated have lower levels of group membership than other men is that they have fewer financial and human capital resources. In fact, one of the key conclusions we can draw in respect to group membership is that engagement in community groups and organisations is more strongly related to the health, educational and financial resources of individuals and their families, as well as liberal attitudes and more substantive norms of civic engagement (which are themselves associated with these resources), than it is to characteristics like family type. These findings suggest that engagement in community groups and organisations is essentially a middle class phenomenon.

While we did not find that family type, marital status or the presence of children were directly related to group membership for women, we did again find that the relationship between family and community life was moderated by family type for women with dependent children. High levels of household social capital were associated with high levels of group membership for partnered mothers, but household social capital was unrelated to community group membership for lone mothers. This may be because the level of household social capital reported by women in couple families is likely to reflect the quality of their relationship with their husband or partner, and be positively related to their engagement in the community because the more cooperative their relationship with their partner the better able they are to pursue interests outside the home.

In contrast, the level of household social capital reported by lone mothers is likely to reflect the quality of their relationships with their children and between their children (the level of closeness, trust, reciprocity, shared activities and knowledge of each others’ friends), and a focus on the quality of these relationships may result in something of a trade-off with community engagement as a result of pressures of time and other resources.

An additional finding was that high levels of trust and reciprocity among kin were associated with low levels of group membership for men. This differed from our results above, where we found a positive relationship between kin based social capital and levels of generalised trust and reciprocity. While strong trust and reciprocity among kin may be associated with trust of people generally in the community, it appears there may be a trade-off for men between either investing in kinship relationships or civic life. Men don’t appear to be strong in both, relatively speaking, when other variables are taken into account

It is also notable that we did not find that the number of hours spent in paid work was a factor in explaining number of group memberships. It may be that this is because these analyses have focused on the total number of group memberships, without making a distinction between different types of groups and organisations, whereas many of the family characteristics examined above may be related to membership in particular types of groups and organizations.18

16 Both men and women belonged to four groups on average and the variance among men was the same as the variance among women.

17 Full results available from the authors on request.

18 In supplementary analyses we examined how hours spent in paid employment relates to membership of different types of groups and organisations and found that the relationship varies considerably by type of group, as well as gender. Not surprisingly, we found that membership of trade unions, professional and technical groups and organisations increases with the number of hours worked, for men and women. For women, membership of child related groups as well as sporting, hobby and/or recreation groups is highest among those working part-time, while women's membership of the more "civically minded" groups (such as environmental, human rights, community and/or welfare groups, arts, cultural and/or educational groups, and political groups or parties) did not appear to be related in any clear way to hours spent in paid employment. For men, membership of most of the other types of groups and organisations was related more clearly to whether or not they were in paid employment than the number of hours worked for those in paid employment. However, membership of environmental, human rights, community and/or welfare groups - the more civic minded groups and organisations - was higher among men who are working part-time than among men who are working full-time or not at all.

Discussion and conclusion

This paper has examined how family life relates to community social capital. In doing so, it has specifically set out to test the “family decline thesis” that posits that changes in family life have led to declining levels of community social capital. Drawing on data from the Australian Institute of Family Studies Families, Social Capital and Citizenship Project, we used hierarchical regression to examine how the family characteristics of most concern to family decline theorists, including family type, marital status, household employment and attitudes to relationships and gender roles, relate to two key measures of community social capital: levels of generalised trust and reciprocity, and the number of groups and organisations people belong to.

Our analyses also explored two possible explanations for why changes in family life may relate to diminished community social capital. The first explanation was that changes in family life constitute a weakening of social capital within families, which in turn weakens community social capital. This is the main explanation provided by family decline theorists. The second explanation was that changes in family life are associated with decreased access to other resources such as human and financial capital, and that these resources are in turn related to community social capital. We explored these arguments by examining whether these additional variables (family social capital and other resources) were themselves related to community social capital, and whether including these variables in the models moderated the relationships observed between community social capital and our key family characteristics.

Our findings have shown a complex set of relationships between family life, social capital and the other resources available to families, as well as the gendered nature of these relationships. We have found evidence to support some aspects of the family decline thesis, but also several important variations and qualifications to the thesis that need to be highlighted.

We conclude by summarising the key findings of the study, and the implications of each for research and policy.

First key finding

Our first main finding is that marital status is one of the most powerful predictors of community social capital for men. What our analyses show is that married men have higher levels of community social capital than unmarried men; and that this is likely to be at least in part because married men have higher levels of family social capital than unmarried men. In contrast, community social capital was not associated with marital status for women. This may be because women have traditionally placed more emphasis on fostering and/or maintaining family relationships than have men, regardless of their relationship status, age or stage in the life course (Finch and Mason 1993), and that these family based relations can be an important source of community connection and trust. However, this cannot be the only explanation – married men have higher levels of community social capital than unmarried men even when levels of family social capital, and other resources, are taken into account.

Future research could explore how marital status and household circumstances are linked to men’s broader family and community connections – or the processes by which these variables are entwined. This could inform a policy focus on helping men to build and/or maintain strong family and community relationships regardless of their marital or household circumstances.

Second key finding

Second, we found that working full-time is associated with low levels of generalised trust and reciprocity among women with dependent children. While this finding may also be seen as providing support for the family decline thesis, this finding did not appear to be explained by the impact of women’s work on the quality of family relations. It is likely that long hours spent in paid work are not conducive to community social capital where one also has primary or sole responsibility for the care of young children, because of the time constraints and pressures associated with parenting and full-time work. However, it is also possible that this finding reflects some unmeasured characteristic of mothers working full-time.

Our results suggest that if we want to foster high levels of community engagement and trust we need to pay attention to issues of work and family balance, particularly for women. It is unrealistic to expect high levels of community engagement and workforce participation from those who have primary responsibility for the care of young children. Also, if respect for personal autonomy in work and family life is associated with high levels of community engagement, trust and reciprocity, it seems likely that supporting men and women to flexibly organise and balance their work and family responsibilities will foster these community outcomes.

Third key finding

This draws attention to our third main finding which is that non-traditional attitudes to relationships and gender roles – including support for maternal employment among women and support for independence in intimate relationships among men – are associated with high levels of generalised trust and community group membership. Rather than providing support for the (rather pessimistic) family decline thesis, these findings give credence to Giddens’s (1996) view that the increased respect for personal autonomy and equality that exists in modern relationships should carry through to public life, fostering broader democratic practices and ideals.

However, it should also be acknowledged that causality may operate in the opposite direction – that the experience of being involved in community groups and organisations, and/or a general tendency to think the best of others (that people are trustworthy and generally try to be helpful) may foster the development of non-traditional attitudes to relationships and gender roles, or counter conservative ones.

Fourth key finding

Fourth, we found that strong family norms of trust and reciprocity do not always translate into broader forms of trust and engagement in communities. While some forms of family social capital were positively associated with community social capital in some circumstances, in other circumstances strong family norms and obligations appeared to limit possibilities for broader community engagement and activity, in a way that is characteristic of “familism”.

These findings counter the assumption that high levels of family social capital will always translate into high levels of community social capital – an assumption that underlies the family decline thesis and is common among social capital theorists and policy makers.

Fifth key finding

Our fifth main finding was that financial and human capital resources are important – and powerful – predictors of community social capital, and in part account for the decline thesis findings. The low levels of generalised trust and reciprocity observed among un-married men were explained in part by low levels of family social capital, but also in part because being married is associated with other forms of socio-economic advantage. In addition, the high levels of generalised trust and reciprocity observed among women with non traditional attitudes to gender roles, and among men with high levels of family social capital, may also in part be accounted for by the fact that women and men with these characteristics have high levels of financial and human capital.

Hence, an important policy consideration is that at least a minimum level of financial and human capital resources as well as time are likely to be necessary for the translation of family social capital into community social capital. And that more broadly speaking, social connections and trust between families and their communities may be enhanced through access to other resources such as money, education, public health and resourced neighbourhoods.

Sixth key finding

Sixth, we found that these models explain only some of the variation in levels of group membership and generalised trust and reciprocity. This means that other factors are also important predictors of community social capital – factors we have not examined here. These are likely to include individual personality characteristics which we have mentioned but not examined in the analyses in this paper, as well as other demographic factors such as cultural background or ethnicity, which we have only partially controlled for here. Future avenues of research could explore these links.

Summary

In sum, the analyses in this paper make clear that a focus upon “maintaining” family relationships alone will not always result in increased outcomes for the community. We have found that resources are equally as important, if not more important predictors of community social capital than the structure or quality of family relations. In addition, we have found that these variables explain only some of the variation in levels of social capital at the community level.

A final point to note is that while this paper has focused on “family” ties, networks of friends, neighbours and/or work-mates may also be important for building generalised norms of trust and reciprocity and broader forms of community engagement. The move away from traditional patterns of partnering, marriage and family formation can be understood as a decentring of family relationships and as such may be associated with the elevation in importance of informal ties outside the family (Pahl 2000; Budgeon and Roseneil 2002; Monti 2002).

While networks of friends, neighbours and work-mates are often seen as alternative measures of community social capital, it is possible that some of these non-family relationships substitute for, or take on the function of, family relationships as theorised at the beginning of this paper – providing trust and support and a basis for building broader community ties and values. The changing nature and function of these relationships, and the extent to which they provide links to broader forms of community engagement, may be a fruitful source of future research.

References
  • The Age (2001), Editorial, Saturday 25th August.
  • Baer, Judith (2002), “Is social capital within the household a risk or protective factor during adolescent development?”, Journal of Marriage and the Family, vol. 64, August, pp. 668-675.
  • Black, A. & Hughes, P. (2000), “Social capital in urban life”, Paper presented to Quality of Life Conference, Singapore, February.
  • Bourdieu, P. (1993), Sociology in Question, Sage, London.
  • Budgeon, S. & Roseneil, S. (2002), “Cultures of intimacy and care beyond ‘the family’: Friendship and sexual/love relationships in the twenty-first century”, Paper presented to the XV World Congress of Sociology, Brisbane, Australia, July.
  • Coleman, J. (1988), “Social capital in the creation of human capital”, American Journal of Sociology, vol. 94, S95-120.
  • Cox, E. & Caldwell, P. (2000), “Making policy social”, in I. Winter (ed.) Social Capital and Public Policy in Australia, Australian Institute of Family Studies, Melbourne. Cox, E. (1995), A Truly Civil Society, ABC Books, Sydney.
  • Dasgupta, P. (1988), “Trust as a commodity”, in D. Gambetta (ed.) Trust, Basil Blackford, Oxford.
  • De Neve, K. M. & Cooper, H. (1998), “The happy personality: A meta-analysis of 137 personality traits and subjective wellbeing”, Psychological Bulletin, vol. 124, no. 2, pp..197-229.
  • de Vaus & Wolcott (eds) (1997), Australian Family Profiles: Social and Demographic Patterns, Australian Institute of Family Studies, Melbourne.
  • Finch, J. & Mason, J. (1993), Negotiating Family Responsibilities, Routledge, London. Fukuyama, F. (1996), Trust the Social Virtues and the Creation of Prosperity, Penguin, London. Fukuyama, F. (1999), The Great Disruption: Human Nature and the Reconstitution of Social Order, Free Press, New York.
  • Furstenberg, F. & Hughes, M. E. (1995), “Social capital and successful development among at-risk youth”, Journal of Marriage and the Family, vol. 57, pp. 580-592.
  • Gallie, D. & Paugam, S. (eds) (2000) Welfare Regimes and the Experience of Unemployment in Europe, Oxford University Press, Oxford.
  • Giddens, A. (1992), The Transformation of Intimacy: Sexuality, Love and Eroticism in Modern Societies, Polity Press, Oxford.
  • Gilding, M. (2001), “Changing families in Australia 1901-2001”, Family Matters, no.60 Spring /Summer, pp. 6-11.
  • Glezer, H. (1997), “Cohabitation and marriage relationships in the 1990s”, Family Matters, no. 47, pp. 5-9.
  • Gray, M., Qu, L, de Vaus, D. & Millward, C. (2002), Determinants of Australian Mothers’ Employment: An Analysis of Lone and Couple Mothers, Australian Institute of Family Studies, Melbourne.
  • Gregory, R. G. (1999), “Children and the changing labour market: Joblessness in families with dependent children”, CEPR Discussion Paper no. 406, Research School of Social Sciences, Australian National University, Canberra.
  • Gross, N. & Simmons, S. (2002), “Intimacy as a double-edged phenomenon? An empircal test of Giddens”, Social Forces, vol. 81, no. 2. pp. 531-555.
  • Headey, B.& Wearing. A. (1992), Understanding Happiness: A Theory of Subjective Wellbeing, Longman Cheshire, Melbourne.
  • Hogan, D. & Owen, D. (2000), “Social capital, active citizenship and political equality in Australia,” in I. Winter (ed.) Social Capital and Public Policy in Australia, Australian Institute of Family Studies, Melbourne.
  • Howard, J. & Newman, J. (2000), Stronger Families and Communities Strategy, Department of Family and Community Services, Canberra.
  • Hughes, J. & Stone, W. (2002), Familes, Social Capital and Citizenship Fieldwork Report, Australian Institute of Family Studies, Melbourne. Online: www.aifs.gov.au/institute/ research/soccapfieldwork.pdf
  • Hughes, P. & Black, A. (2003), “Social capital and family life: Evidence from the Australian community survey”, Paper presented to the 8th Australian Institute of Family Studies Conference, Melbourne, 12-14 February.
  • Hughes, P. & Black, A. (2002), “The impact of various personal and social characteristics on volunteering, Volunteering Australia, vol. 7, no. 2.
  • Hughes, P., Bellamy, J. & Black, A. (1999), “Social trust: Locally and across Australia”, Third Sector Review, vol. 5, no. 1, pp. 5-24.
  • Hughes, P., Bellamy, J. & Black, A. (2000), “Building social trust through education”, in I. Winter (ed.) Social Capital and Public Policy in Australia, Australian Institute of Family Studies, Melbourne.
  • Inglehart, R. (1997), Modernization and Postmodernization: Cultural, Economic and Political Change in 43 Societies, Princeton University Press, Princeton NJ.
  • Keeter. S., Miller, C., Kohut. A., Groves, R. M. & Presser, S. (2000), “Consequences of reducing non-response in a national telephone survey”, Public Opinion Quarterly, vol. 64, pp. 125-148.
  • King, V. (2002), “Parental divorce and interpersonal trust in adult offspring”, Journal of Marriage and the Family, vol. 64, August, pp. 642-656.
  • Knack, S. & Keefer, P. (1997), “Does social capital have an economic payoff? A cross-country investigation”, The Quarterly Journal of Economics, November, pp. 1251-1288.
  • Mark, N. P. (2002), “Cultural transmission, disproportionate prior exposure, and the evolution of cooperation”, American Sociological Review, vol. 67, pp. 323-344.
  • Misztal, B. A. (1996), Trust in Modern Societies: The Search for the Bases of Social Order, Polity Press, Cambridge, Mass.
  • Monti, D., Butler, C., Curley, A. & Tilney, K. (2002), “Social ties and civic attachments in late 20th century America”, Paper presented to the XV World Congress of Sociology, Brisbane, Australia, July.
  • Narayan, D. (1999), Bonds and Bridges: Social Capital and Poverty, Word Bank, Washington DC.
  • Onyx, J. & Bullen, P. (2000), “Sources of social capital”, in I. Winter (ed.) Social Capital and Public Policy in Australia, Australian Institute of Family Studies, Melbourne.
  • Pahl, R. E. (2000), On Friendship, Polity, Cambridge.
  • Portes, A. (1998), “Social capital: Its origins and applications in modern sociology”, Annual Review of Sociology, vol. 24, pp. 1-24.
  • Putnam, R. D. with Leonardi, R. & Nanetti, R. Y. (1993), Making Democracy Work: Civic Traditions in Modern Italy, Princeton University Press, Princeton, NJ.
  • Putnam, R. D. (1995), “Bowling alone: America’s declining social capital”, Journal of Democracy, vol. 6, no. 1, pp. 65-78.
  • Putnam, R. D. (1996), “The strange disappearance of civic America”, Policy, Autumn, pp. 3-15. Putzel, J. (1997), “Accounting for the ‘dark side’ of social capital: Reading Robert Putnam on democracy”, Journal of International Development, vol. 9, no. 7, pp. 939-949.
  • Qu, L. & Weston, R. (2001), “Starting out together: Cohabitation or marriage”, Family Matters, no. 60 Spring/Summer, pp. 76-79.
  • Ravanera, Z. R., Beaujot, R. & Rajulton, F. (2002), “The family and political dimension of social cohesion: Analysing the link using the 2000 national survey on giving, volunteering and participating”, Discussion Paper no. 02-07, Population Studies Centre, University of Western Ontario, Canada.
  • Smart, C. & Neale, B. (1999), Family Fragments? Polity Press, Malden, Mass.
  • Research Paper No. 32, April 2003 Australian Institute of Family Studies 33
  • Stone, W. (2001), Measuring Social Capital, Research Paper No. 24, Australian Institute of Family Studies, Melbourne.
  • Tabachnik, T. B. & Fiddell, L. S. (1996), Using Multivariate Statistics, 3rd edition, Harper Collins, Northridge.
  • Toth, J, F., Brown, R. B. & Xu, X. (2002), “Separate family and community realities? An urban-rural comparison of the association between family life satisfaction and community satisfaction”, Community, Work and Family, vol. 5, no. 2, pp. 181-202.
  • Uslaner, E. M. (1999), “Trust and consequences”, Paper presented to the Communitarian Summit, February, Arlington, VA. Online: www.bsos.umd.edu/gvpt/uslaner/working.htm
  • Waite, L. (1995), “Does marriage matter?”, Demography, vol. 32, no. 4, November, pp. 483-507. Weston, R., Stanton, D., Qu, L. & Soriano, G. (2001), “Australian families in transition”, Family Matters, no. 60, Spring/Summer, pp. 12-23.
  • Winter, I. (2000), Family Life and Social Capital: Towards a Theorised Understanding, Working Paper No. 21, Australian Institute of Family Studies, Melbourne.
  • Winter, I. (ed.) (2000), Social Capital and Public Policy in Australia, Australian Institute of Family Studies, Melbourne.
Appendices

Appendix A: Variables used in regression models

A.1 Key characteristics of families

Live with children under 18 years is set to one if at least one child aged under 15 years living in the household and set to zero otherwise.

Live with children over 18 or have children living elsewhere only is set to one if respondent has children but none of those children are living in the household and/or children living in the household are aged over 18 years of age; and set to zero otherwise.

De facto is set to one if the respondent has a partner who they share a house with but they are not married to this person; and set to zero otherwise.

Single and never married or widowed is set to one if the respondent is not currently in a relationship and is either widowed or has never been married; and set to zero otherwise.

Single and separated or divorced is set to one if the respondent is not currently in a relationship and is separated or divorced; and set to zero otherwise.

Lone parent is set to one if the respondent is single and living with at least one child aged under 15 years; and set to zero otherwise.

In paid work 50 or more hours per week is set to one if the respondent is in paid work and works 50 or more hours per week; and set to zero otherwise.

In paid work < 50 hours per week is set to one if the respondent is in paid work and works less than 50 hours per week; and set to zero otherwise.

In paid work 35 or more hours per week is set to one if the respondent is in paid work and works 35 or more hours per week; and set to zero otherwise.

In paid work < 35 hours per week is set to one if the respondent is in paid work and works less than 35 hours per week; and set to zero otherwise.

Retired is set to one if the respondent is retired; and set to zero otherwise.

Has partner in paid work 35 or more hours per week is set to one if the respondent has a partner in paid work 35 or more hours per week; and set to zero otherwise.

Has partner in paid work < 35 hours per week is set to one if the respondent has a partner who is in paid work but works less than 35 hours per week; and set to zero otherwise.

Support for working mothers measures the extent to which respondents agreed (on a scale of 0 to 10) that “A working mother can establish just as good a relationship with her children as a mother who does not work outside the home”.

Individualism in intimate relationships measures the extent to which respondents agreed (on a scale of 0 to 10) that “Maintaining a sense of independence is important in any intimate relationship”.

A.2 Family social capital

Social capital in the family household is a composite measure of the level of social capital within the family household. The measure is based on five survey items that measure different dimensions of social capital in the family, each of which took the form of likert type scales. To measure trust in the family respondents were asked “On a scale of 0 to 10, where 0 equals not at all and 10 equals completely, how much would you say you trust the family you live with to act in your best interests?”. To assess norms of reciprocity, respondents were asked on the same scale “how much would you say family members are willing to help each other out?”. To measure the density of family friendship networks, following Coleman (1998) and Furstenburg and Hughes (1995), respondents were asked “On a scale of 1 to 5, to what extent is it true to say that family members know each other’s close friends”. To assess the extent of family engagement and closeness, respondents were asked on the same scale how much it would be true to say that “Family members share interests and hobbies with one another”, and that “Family members feel very close to one another”. These five measures were combined to produce a single composite measure of the quality of family relations, or social capital, within the family household.

The size of kinship networks is a composite measure of the total number of parents, children (including own and/or partners’ children), siblings and other relatives and in-laws the respondent has living outside the household.

Norms of trust and reciprocity in kinship networks is a composite measure of the level of trust and the level of reciprocity in extended family networks. To measure trust in extended family networks respondents were asked “On a scale of 0 to 10, where 0 equals not at all and 10 equals completely, how much would you say you trust your relatives to act in your best interests?” To assess norms of reciprocity, respondents were asked “On the same scale of 0 to 10, where 0 equals not at all and 10 equals completely, how much would you say your relatives are willing to help each other out?” (See Stone and Hughes (2002) for discussion of validity and reliability of this approach).

High civic activity in family of origin is set to one if the respondent reported that members of the family they grew up with were involved with some kind of helping, voluntary work, political or community activity “a lot” or “quite a lot” rather than “not at all” or “just a little”; and set to zero otherwise.

Low civic activity in family of origin is set to one if the respondent reported that members of the family they grew up with were involved with some kind of helping, voluntary work, political or community activity “not at all” or “just a little” rather than “a lot” or “quite a lot”; and set to zero otherwise.

A.3 Other resources

Incomplete secondary school is set to one if the respondent has a highest level of educational attainment of less than Year 12; and set to zero otherwise.

Tertiary qualified is set to one if the respondent has a degree level qualification; and set to zero otherwise.

Financially comfortable is set to one if the respondent reported that they were “living comfortably” or “doing alright” financially, rather than “just getting by”, “finding it quite difficult” or “finding it very difficult”; and set to zero otherwise.

Financially struggling is set to one if the respondent reported that they were “finding it quite difficult” or “finding it very difficult” financially rather than “living comfortably”, “doing alright” or “just getting by”; and set to zero otherwise.

Home owner is set to one if the respondent said they owned or were purchasing the home they lived in; and set to zero otherwise.

Excellent or very good health is set to one if the respondent reported being in excellent or very good health rather than good or poor health, and set to zero otherwise.

Good health is set to one if the respondent reported being in good health rather than excellent, very good or poor health; and set to zero otherwise.

Poor health is set to one if the respondent reported being in poor health rather than excellent, very good or good health; and set to zero otherwise.

Neighbourhood in highest percentile of socio-economic advantage takes the value of one if the respondent lives in a postcode which is ranked in the 75 to 100 percentile of postcodes according to the ABS SEIFA index of relative socio- economic disadvantage and zero otherwise.

Neighbourhood in lowest percentile of socio-economic advantage takes the value of one if postcode is ranked in the 0 to 25 percentile of postcodes and zero otherwise.

Safety of neighbourhood measures the extent to which respondents agreed (on a scale of 0 to 10) that they are satisfied with the safety of their neighbourhood.

A.4 Control variables

Age Age of the respondent in years.

Language other than English is set to one if the respondent or their family speaks a language other than English at home or with their family; and set to zero otherwise.

Rural or remote area is set to one if the respondent lives in a rural or remote area and zero if respondent lives in a capital city or other metropolitan area; and set to zero otherwise.

Appendix B: Summary statistics

Appendix B: Summary statistics

Appendix B is only available in PDF format. If you require an accessible version of this appendix please contact us and we will endeavour to provide the content you need in a format you can use.

Acknowledgements

The authors would like to thank the numerous colleagues and peers who have contributed to the ideas upon which this paper is based, including participants at both the Australian Social Policy Conference 2001 and the International Sociological Association World Congress 2002, where earlier versions of this work were presented.

The paper has benefited considerably from the interest and helpful comments of colleagues at the Australian Institute of Family Studies, particularly Matthew Gray, David de Vaus and Ann Sanson, to whom we are grateful.

 

We are also indebted to Janeen Baxter, Barbara Misztal and Ian Winter, whose valuable comments have improved this paper. Any remaining shortcomings are, of course, the responsibility of the authors.

Publication details

Research Paper
Published by the Australian Institute of Family Studies, May 2003
50 pp.
Suggested citation:

Hughes, J., & Stone, W. (2003). Family change and community life: Exploring the links (Research Paper No. 32). Melbourne: Australian Institute of Family Studies.

 

 

Download Publication