Does the nuclear family affect social trust? Longitudinal evidence from Germany

ABSTRACT

While research indicates that social trust might benefit societies’ political and economic development, the sources of social trust are subject to debate. This article investigates a less investigated factor in the development of social trust: how far the nuclear family – that is, partnerships and parenthood – affects trust towards other people. The data are from three waves of the German Socio-Economic Panel study collected between 2003 and 2013. Longitudinal estimates indicate that having any number of children increases trust towards others while being partnered has a negative effect on social trust. Both effects are near-linear over the 19-year life course period investigated. The positive effect of parenthood is much stronger than the negative effect of partnership. These results are consistent across genders and ages. They indicate that social trust can change, but that it changes slowly during an adult’s life. The results are discussed in relation to the social roles of adult members of nuclear families, including their activities in various social arenas.

KEYWORDS:

• Social trust
• nuclear family
• partnership status
• parenthood

1. Introduction

Social trust is considered important for well-functioning societies, including their economic development (Dincer and Uslaner 2010; Horváth 2013). Hence, a growing body of research has investigated the potential sources of social trust (e.g. Delhey and Newton 2003; Stolle and Hooghe 2004): what makes some individuals more trustful than others? Two perspectives dominate current debates. One argues that social trust is a stable characteristic of individuals that may reflect genetic factors (Sturgis et al. 2010) or early childhood experiences (Collins and Read 1990; Uslaner 2002). The other argues that social trust is malleable due to interactions with other people and institutions (Coleman 1990; Hardin 2002).

Social capital theory argues that social trust is an outcome of social participation in various arenas, including voluntary associations, public services and neighbourhoods (Kumlin and Rothstein 2005; Li et al. 2005; Paxton 2007; Stolle and Hooghe 2004). While cross-sectional data have often supported these claims (Coleman 1990; Gelissen et al. 2012; Putnam 1993, 2000), longitudinal data provide less conclusive evidence (Gross et al. 2004; van Ingen and Bekkers 2015). The absence of longitudinal findings has led some observers to argue that social trust is most likely an unchanging characteristic over the adult life course (Nannestad 2008; Sturgis et al. 2010).

This article investigates the role of a less considered factor in the development of social trust: the nuclear family, in the sense of entering partnership and parenthood. The theoretical arguments here parallel to other institutional explanations, including social capital theory, but with one additional argument: the nuclear family is probably more important in most people’s lives than the other institutions emphasised in the literature on social trust, such as voluntary organisations, neighbourhoods or the welfare state. The empirical analysis investigates how social trust changes after entering the nuclear family in panel data from Germany. The results show that parenthood has a strong and consistent effect on social trust across age and gender groups.

2. Social trust

Much research on generalised social trust can be traced back to Rosenberg (1956), who asked: ‘generally speaking, do you believe that most people can be trusted, or that you can’t be too careful in dealing with people?’ This and similar items have since been included in many large-scale surveys, including the World Values Survey and the European Social Survey. The item has been interpreted as a measure of the trustworthiness of the average person we encounter (Coleman 1990; Glanville and Paxton 2007). Research shows that responses to this item are more strongly related to people we do not know than to people we do know (Freitag and Traunmüller 2009; Uslaner 2002, 52–6), certainly in wealthy European countries such as Germany (Delhey et al. 2011).

Social capital theory considers social trust to be important for well-functioning societies because it facilitates cooperation between people and is thus expected to be beneficial for various outcomes including politics and economic prosperity (Coleman 1990; Putnam 1995; Rothstein and Stolle 2008). Empirical research supports these arguments in as far social trust is correlated with a range of beneficial factors at the levels of both individuals (e.g. happiness, health and income) and societies (e.g. economic development, civic participation) (Alvarez et al. 2017; Bjørnskov 2008; Dincer and Uslaner 2010; Horváth 2013; Paxton 2007).

3. The nuclear family

Several theorists have argued that the nuclear family, a group consisting of two parents and one or more children, is important for the development of social trust. Much of this theory assumes that social trust is transmitted from parents to children or an outcome of parent–child interaction. This article investigates, in contrast, how the nuclear family affects social trust in its adult members. It develops theoretical expectations based on the relatively small literature on social trust in nuclear families, on the larger literature on social network and social relation of nuclear families, as well as from parallel explanations related to other social institutions.

Theories about social trust sometimes focus on the institutional settings that are supposed to facilitate the development of social trust (i.e. the nuclear family versus other social arenas), typically in sociology and political science, and sometimes on the mechanisms that build trust toward people we do not know, typically in social psychology. Attachment theory is one example (of the latter) which argues that social trust is an outcome of the interaction between parents and children. If parents are accessible and responsive during times of need, the child will feel safe and secure and is also likely to develop confidence and trust in other people. (e.g. Collins and Read 1990).

The social learning perspective argues that similar mechanisms also apply to adult individuals, who may also develop social trust based on past experiences. If our experiences from previous social encounters are positive, we will develop positive expectations about how the new people we meet will treat us. However, if these experiences are negative, we become insecure and less trustful (e.g. Hardin 2002; Glanville and Paxton 2007).

4. Social capital

Much sociological theory regarding the relationship between the nuclear family and social trust is part of more general theories about social capital. Much of this theory is about the development of social trust and social capital in children (Edwards 2004). Social capital theory also includes scattered arguments that the nuclear family may build social trust in its adult members, even if other arenas are more emphasised in this literature (e.g. voluntary organisation, public services). Social capital is not a unified concept, and social trust is sometimes classified as an aspect of social capital and sometimes an outcome of social capital (e.g. Portes 1998; Son and Feng 2019). For simplicity and conceptual clarity, this article uses the narrower definition of social capital as social networks and social relations that may build social trust.

Influential social capital theorists have expressed concerns about a declining role of social institutions such as the nuclear family (Coleman 1990; Putnam 1993). Putnam (1995) used the decline of traditional families as one example of what he saw as a decline in social capital in the USA. He supported this claim with empirical findings indicating that marriages that include children are statistically associated with higher social trust and civic engagement. However, he also noted that the direction of causality is complicated since it is conceivable that loners and paranoid people are hard to live with.

The first argument (the nuclear family affects social trust) is about social causation, the latter argument (social trust affects marital partnerships) is about social selection. Longitudinal (e.g. panel) data offer one method for resolving such issues of causal or (more accurately) temporal order because longitudinal data allow for estimating how entering partnerships and parenthoods are associated with either an abrupt or a slow change in social trust.

5. Social networks

Much social capital theory sees social trust as an outcome of personal networks and social relations (Alecu 2021; Granovetter 1985; Li et al. 2005; Glanville et al. 2013). Putnam (1993, 1995) argued that the more we connect with other people on a face-to-face basis, the more we trust them. This general (the more, the better) argument is supported by longitudinal data at the level of individuals (Botzen 2015; Glanville et al. 2013; Li et al. 2005).

Social capital theory also argues that different types of social network and social relations facilitate social trust differently. One line of arguments can be traced back to Granovetter (1973) who argued that our infrequent social contacts with acquaintances (weak ties) may work as bridges between individuals and wider society. Following these arguments, social network and social capital analysts started to distinguish between bonding and bridging social networks or social capital (Claridge 2018). For example, Putnam (2000) argued that social trust is more likely to develop if we connect with people who are dissimilar from ourselves (bridging social capital) than if we only connect with people who are similar or have similar interests (bonding social capital), such as friends and family members. Relevant (dis)similarities may include religion, ethnic group and political preference (Stolle and Hooghe 2004).

Putnam (1993) argued that voluntary organisations are important for the development of trust between people, including trust towards fellow citizens. One reason is that these organisations bring together agents of equivalent status and power. Putnam (1993) also argued that one requirement is the existence of civic virtues such as tolerance and equality, which excludes discriminatory organisations such as the Ku Klux Klan. Further, associations that produce public goods, such as parent-teacher associations, are more likely to generate social trust than associations that produce private goods, such as choirs and bowling teams.

Several authors have discussed the role of neighbourhood interactions and the extent to which these networks constitute bonding or bridging social capital. Most social network analysts tend to see neighbourhoods as bonding social capital (e.g. Bowen et al. 2000; Claridge 2018). However, this may depend on the characteristics of the neighbourhood (i.e. the degree of geographical segregation), and perhaps also the alternative bonding networks to which people may have access. Research indicates that different social groups develop social trust from different social networks (Li et al. 2005).

Empirical research does not always support such standard classifications of which types of network are important for social trust. For example, a study of British data shows that social trust is more strongly correlated with neighbourhood interaction than with civic participation, while social networks are in-between (Li et al. 2005). A longitudinal study of data from the USA indicates that social trust is correlated with any change in informal ties including those with friends, relatives and neighbours (Glanville et al. 2013).

6. Partnerships and parenthood

There is not much empirical research on how marital partnerships and parenthood might affect social trust beyond cross-sectional correlations (e.g. Putnam 1995). There is more research on how marital partnerships and parenthood affect people’s activities in various social settings (e.g. Furstenberg 2005). If we accept the argument that participation in social networks leads to more social trust, then we can investigate how social networks change following transitions into marital partnerships and parenthood.

The social withdrawal hypothesis argues that personal networks become smaller when people cohabit and marry (Kalmijn 2003, 2012; Rözer et al. 2015; Slater 1963). While some studies support this hypothesis (e.g. Kalmijn 2003), the empirical support is not very clear, and it may depend on how far one’s partner is classified as part of one’s social network after one enters cohabitation or marriage (Rözer et al. 2015). One longitudinal study indicates that social networks tend to decrease temporarily after one enters a marriage or cohabitation but return to previous levels after a couple of years (Rözer et al. 2015). Longitudinal findings also indicate that partnerships may affect the nature of stronger ties, with fewer friends and more family relations, but not weaker ties (Kalmijn 2012), which might be more important for social trust (Putnam 2000).

Parenthood is associated with several changes in activities and social networks. It is less clear how far these social networks build social trust or lead to more bonding or bridging social capital. On the bridging side, parents may access more diverse social networks than non-parents because they will typically connect with their children’s friends as well as with other parents (Song 2012). Longitudinal findings indicate that social integration increases with parenthood (Nomaguchi and Milkie 2003). Parents also become acquainted with the providers of formal services for children, such as nursery schools and sports clubs, as well as with informal services such as babysitting or shared transport to children’s activities (Ambert 2001; McDonald and Mair 2010).

There is also evidence that parents are more involved in community activities than non-parents, and longitudinal research indicates that parenthood is associated with more voluntary work (Lancee and Radi 2014) and with a shift in social relations towards more local ties (Kalmijn 2012; Nomaguchi and Milkie 2003). After one becomes a parent, one’s friends and acquaintances become less important and neighbours become more important (Kalmijn 2012). Finally, time-use data indicate that having children leads to a change in participation, from informal activities at home (e.g. visits, visitors, co-residents) to informal activities outside the home (cafes, clubs, parties, events), and that this effect is particularly pronounced when the children reach school age (van Ingen 2008). All these changes might indicate that parenting should lead to more social trust.

There are counterarguments as well. The above changes in social relations might mean that parents develop fewer weak ties with acquaintances and more strong ties with neighbours and family members (Kalmijn 2012). If so, this might also mean that they substitute bridging relations, which supposedly mean more social trust, with bonding relations, which supposedly mean less social trust. Further, parenting can be a time-consuming and exhausting activity that can drain one’s resources for participation in external networks that supposedly build social trust (Nomaguchi and Milkie 2003).

7. Hypotheses

Both theory and research provide stronger arguments that parenthood leads to more rather than less social integration with non-family members and in arenas outside the home. Hence, we hypothesise (H1) that parenthood leads to more social trust. Theory and research provide fewer clues about how marital partnerships affect social trust. Nevertheless, both theory and research suggest less rather than more involvement with non-family members and in arenas outside the home. Hence, we hypothesise (H2) that partnerships lead to less social trust.

We also investigate how far any significant findings vary by gender and age. Because women tend to be more involved with children and family relations than men (e.g. Craig and Mullan 2010), we hypothesise (H3) that the effects of parenthood and partnerships are stronger on women than on men. Some theorists argue that social learning effects are stronger at younger rather than older ages (see Stolle and Hooghe 2004). Hence, we also hypothesise (H4) that life-course transitions that occur at a young age affect social trust more strongly than those that occur at later ages.

8. Method

The nuclear family is made up of two components: marital partnerships and children. How these transitions affect social trust (or any other outcome) can only be investigated as changes along the life-course of individuals in longitudinal (typically panel) data. When analysing longitudinal/panel data the impact of both transitions can be investigated as an abrupt change or a slow change in an outcome variable, in our case social trust. Abrupt changes are investigated by dummy variables (when/after entering), while slow changes are investigated by linear terms of the number of years being partnered or being a parent. The current analysis does not distinguish among partnerships’ legal statuses as cohabiting, married or both.

The German Socio-Economic Panel (SOEP) study collected information about our dependent variable, generalised trust in 2003, 2008 and 2013. The questionnaire included three related items, but only the first two are sufficiently highly correlated (r = 0.49) to warrant the construction of an index. Both items were presented as rather general statements: (1) ‘people can generally be trusted’; (2) ‘nowadays you can’t rely on anyone’. There were four response categories, ranging from ‘strongly agree’ to ‘strongly disagree’. Both the index (calculated as the mean of the two items) and its two items are investigated, with values varying from one (low trust) to four (high trust).

The statistical analysis investigates how social trust changed between the three waves of data, using linear regression models with fixed effects for individuals. Hence, only those participating in two or three waves are investigated. The fixed effects regression model has the advantage of controlling for all time-invariant (fixed) characteristics of the individuals (e.g. gender, education level, religion, country of origin), and models only how far a change in a situational factor (e.g. entering parenthood, year partnered) is associated with a change in the outcome (social trust) variable. Hence, it also controls for selections effects, including the possibility that people high in social trust are more likely to enter nuclear families or remain partnered ‒ a persistent problem in any cross-sectional analysis. All models also control for period effects, using dummy variables. The analysis includes 17,817 individuals: 8472 participating in two waves, and 9345 participating in three waves, giving a total of 44,979 wave-level observations (8472*2 + 9345*3).

Changes in trust are investigated along with changes in three demographic characteristics: ageing, entering marital partnerships and parenthood, and the durations of marital partnerships and parenthood. Ages vary from 17 to 99 years (ages above 100 are excluded), with a mean of 51 years (Table 1). Age is investigated using three linear slopes separated at 40 and 60 years, meaning that social trust can change with different ages slopes below 40 years, between 40 and 60 years and above 60 years.

Table 1. Descriptive statistics of the data at the level of 44,979 observations for 17,817 respondents.

	Mean	SD	Low	High
Social trust index	2.6	0.6	1	4
Social trust item 1	2.6	0.7	1	4
Social trust item 2	2.6	0.8	1	4
Current age	50.6	16.8	17	99
Partnership	73.3%		0	1
Parenthood	68.6%		0	1
Years partnered	6.9	6.8	0	19
Years parent	10.6	8.2	0	19
Number of children	1.4	1.2	0	5

The durations of the marital partnerships are collected from a marital history file from the SOEP. The duration of current partnerships can be traced back as far as 1984, when the SOEP started, giving a maximum of 19 years in the first wave in 2003. Therefore, durations are set to a maximum of 19 years. The maximum duration of parenthood is also set to 19 years for a firstborn child, to facilitate comparison between the two nuclear family effects, and because 19 years typically represents a transition point from childhood to adulthood. The mean values of (0–19) years are seven years for partnerships and 11 years for parenthood (Table 1). Individual ages and the durations of partnerships and parenthood are investigated through the number of years divided by ten, avoiding one decimal in the presentation of the regression results. Altogether, 283 combinations of individuals and years ended a partnership (separation, divorce, widowhood) during an interview year. These observations have been removed because partnership dissolution is not investigated in this paper, and because other research indicates significant changes in quality of life around the time of a partnership dissolution.

9. Results

A first analysis investigates how far the development of social trust is associated with abrupt or slow changes in the formation of the nuclear family. The results (Table 2) show that social trust changes slowly, also when related to the nuclear family. Abrupt changes (dummy variables, model 1 and 3) are non-significant, while the slow changes (number of years in each state, model 2 and 3) are significant. The effects of being partnered and being a parent point in different directions, however. Years partnered is negatively correlated with social trust. The estimate (from model 2 without abrupt changes) indicates that marital partnerships lasting 19 years or more are associated with a 0.04 (0.022*1.9) reduction in generalised trust, corresponding to 7% of its standard deviation (0.6, Table 1).

Table 2. Regression results from the analysis of social trust (1–4) as related to entering partnerships and parenthood and years being partnered and a parent plus ageing.

	1		2		3
Partnership ^a	−0.009	(.012)			−0.010	(.014)
Years partnered ^b			−0.022*	(.010)	−0.025*	(.011)
Parenthood ^a	0.013	(.021)			−0.015	(.021)
Years parent ^b			0.063**	(.018)	0.066**	(.019)
Age <40 ^b	0.049**	(.014)	0.024	(.015)	0.026	(.016)
Age 40–60 ^b	0.014	(.010)	0.012	(.012)	0.014	(.012)
Age >60 ^b	−0.006	(.011)	0.005	(.012)	0.007	(.012)
Fixed effects	Yes		Yes		Yes
Period effects ^a	Yes		Yes		Yes

Notes: ^a 0 or 1; ^b years/10; * p <0.05; ** p <0.01.

Years of being a parent are positively associated with generalised trust. Because the positive effect of parenthood is much (2.9 times) stronger (in model 2) than the negative effect of being partnered, it more than outweighs the negative effect of marital partnership. The estimate (from model 2) indicates that parenthood leads to a 0.12 (0.063*1.9) increase in social trust after 19 years, corresponding to 19% of its standard deviation.

A typical pattern of family formation starts with partnership and leads to parenthood later. The combined effects of these two aspects of family formation can be estimated to produce an 0.08 increase in social trust after 19 years in both states (test of difference: t = 2.0), corresponding to 13% of the standard deviation of social trust. Hence, we can conclude that the nuclear family builds trust towards other people, even if marital partnerships on their own tend to reduce social trust.

Table 3 tests alternative specifications of the family formation variables in Table 2 (model 2 without abrupt changes). Model 4 investigates the number of children in addition to the durations of partnership and parenthood. It appears that social trust is related to parenthood as such and not to the number of children in a family.

Table 3. More detailed analyses of family formation variables in relation to social trust (1–4) by adding dummy variables for entering partnerships and parenthood (model 4) and splitting the duration variables (years partnered and being a parent at nine years, model 5).

	4		5
Years partnered ^b	−0.022*	(.010)
Years parent ^b	0.061**	(.019)
Number of children	0.004	(.012)
Partnered <9 years ^b			−0.023	(.013)
Partnered >9 years ^b			−0.019	(.018)
Parent <9 years ^b			0.039	(.029)
Parent >9 years ^b			0.077**	(.024)
Age <40 ^b	0.022	(.016)	0.028	(.016)
Age 40–60 ^b	0.013	(.012)	0.010	(.012)
Age >60 ^b	0.005	(.012)	0.004	(.012)
Fixed effects	Yes		Yes
Period effects ^a	Yes		Yes

Notes: ^a 0 or 1; ^b years/10; * p <0.05; ** p <0.01.

Model 5 splits the two durations – of partnership and parenthood – at nine years, allowing the linear slopes to be different before and after nine years. The partnership slopes are of similar magnitude before and after nine years’ duration, while the parenthood slope is slightly steeper after nine years (age of oldest child) than before nine years in the data. The latter difference is non-significant, however (p > 30%). Hence, we assume that they are most likely linear.

We know that parenthood affects the daily activities of mothers and fathers differently because mothers tend to be more involved with children, leading to a more traditional division of labour between men and women (e.g. Craig and Mullan 2010). Model 6 (Table 4) tests whether the nuclear family (i.e. partnership and parenthood) affects women’s social trust more strongly than men’s. No such differences are statistically significant. Nevertheless, there are slight tendencies in the data for social trust to be more positively affected by being partnered and having children in women than in men.

Table 4. Introducing interaction terms between gender and age versus family formation variables in the analysis of social trust (1–4).

	6		7
Years partnered ^b	−0.031*	(.014)	−0.013	(.035)
Years parent ^b	0.057*	(.025)	0.022	(.044)
Age <40 ^b	0.023	(.016)	0.024	(.016)
Age 40–60 ^b	0.013	(.012)	0.001	(.017)
Age >60 ^b	0.006	(.014)	−0.005	(.018)
Years partn.*female	0.017	(.017)
Years parent*female	0.013	(.031)
Years partn.*age			−0.002	(.006)
Years parent*age			0.009	(.009)
Fixed effects	Yes		Yes
Period effects ^a	Yes		Yes

Notes: ^a 0 or 1; ^b years/10; * p <0.05.

There has also been speculation that life-course events might affect people’s trust more strongly during the early years of adulthood than at later ages (e.g. Stolle and Hooghe 2004). Model 7 (Table 4) tests this hypothesis. There are no such tendencies in the data. There is a slight but non-significant tendency for the onset of parenthood at a higher age to be more strongly associated with a rise in social trust than when parenthood starts at a younger age. Because this tendency in the data is the opposite of our hypothesis, we presume that the effects of years partnered and years of parenthood are of similar magnitude over the ages of adult life.

The analyses above have investigated generalised social trust as the mean of its two items: (1) ‘people can generally be trusted’ and (2) ‘nowadays you can’t rely on anyone’. Table 5 investigates these two items separately, both with high values (four) for high levels of trust. No differences between the two items are statistically significant. Nevertheless, years partnered are slightly more strongly correlated with the statement ‘you can’t rely on anyone’ than with the statement ‘people can generally be trusted’. Because some other data sources (e.g. World Values Survey) tend to rely on item 1 only (‘people can generally be trusted’), there is the possibility that some other analyses may not reproduce similar results.

Table 5. Separate analyses of the two items (1 and 2) of social trust (1–4) in models that otherwise correspond to model 2 in Table 2.

	Item 1: ‘People can be trusted’		Item 2: ‘You can’t rely on anyone’
Years partnered ^b	−0.011	(.010)	−0.033**	(.013)
Years parent ^b	0.055**	(.021)	0.070**	(.024)
Age <40 ^b	0.031	(.018)	0.017	(.020)
Age 40–60 ^b	−0.001	(.013)	0.026	(.016)
Age >60 ^b	−0.005	(.013)	0.014	(.016)
Fixed effects	Yes		Yes
Period effects ^a	Yes		Yes

Notes: ^a 0 or 1; ^b years/10; ** p <0.01.

10. Discussion

There are two main findings from our research. First, the nuclear family builds social trust insofar as it includes children. Second, any change in social trust associated with the nuclear family occurs slowly, over many years. Given the robustness of these results, it is astonishing that this finding has gone unnoticed by previous research. One possible explanation might be the methods. Social scientists have only relatively recently started to investigate longitudinal data on topics such as social trust. Moreover, even when such data have been investigated, slow-changing processes have often been overlooked. For example, some empirical investigations have assumed that social trust will change whenever people enter or leave a voluntary organisation (Bekkers 2012; van Ingen and Bekkers 2015). However, our data on social trust following family transitions show no such abrupt changes. Further, such modelling assumptions are not consistent with classical contributions to social capital theory, which argue for long historical changes in social trust at the level of (Italian) regions (Banfield 1958; Putnam 1993).

Social capital and social network analysts have argued that social trust is important for well-functioning societies (e.g. Granovetter 1985; Putnam 1993), and much theoretical and empirical research has discussed potential sources of social trust (Delhey and Newton 2003; Rothstein and Stolle 2008; Stolle and Hooghe 2004; Uslaner 2002). Nevertheless, most analysts have overlooked the institution of the family, albeit with some exceptions (e.g. Putnam 1995). One explanation might be that social capital theory has been dominated by political scientists and political sociologists, to whom the nuclear family may not have been sufficiently visible as a social institution, perhaps because it has been conflated with individuals.

Much research has investigated the role of voluntary organisations in building social trust (e.g. Putnam 1993). There are some important differences between nuclear families and voluntary organisations. While it can be difficult to say how important voluntary organisations are in the lives of their members, there can be little doubt about the importance of one’s children and marital partner. While having children, and perhaps also having a marital partner, is a lifelong commitment, people may enter and leave voluntary organisations at short notice. The nuclear family affects the daily lives, identities and futures of its members much more than do voluntary organisations or other institutions investigated in social capital literature such as public services and neighbourhoods.

Voluntary organisations likely have more heterogeneous effects on social trust because they can aim to produce very different goods or outcomes (e.g. Putnam 1993; Rothstein and Stolle 2008). Our study indicates that the nuclear family has similar effects on social trust when we split the data by age and gender. Still, other research indicates variation in how the nuclear family affects social trust (Ermisch and Gambetta 2010). Hence, the results regarding both parenthood and partnerships are probably average effects with some variation.

When social capital analysts have investigated the role of the family, the assumption has typically been that social capital and social trust are passed down from parents to children via socialisation processes (e.g. Coleman 1990). There are some scattered arguments for the opposite flow of effects as well, including Putnam’s (1995) argument that families with children build social trust in their adult members. Somewhat similarly, the sociology of childhood has highlighted the role of children as agents in creating networks with each other and with adults, both within and outside the family (Leonard 2005).

It is difficult to say how far the results of the current study are confined to Germany or if they would be reproduced in data from other countries as well. Germany has been classified as a typical European country with regard to the development of the nuclear family (Pfau-Effinger 2004; van de Kaa 1987). A somewhat related issue is whether ongoing changes in the nuclear family might affect the results of the current study, including those associated with the second demographic transition (van de Kaa 1987). The fact that the current study does not reveal any different results for age and gender (i.e. mothers and fathers) groups provide few clues such direction, even though motherhood is associated with a stronger change in daily activities than fatherhood (Craig and Mullan 2010).

This analysis has hypothesised that parenthood builds social trust because it leads to more social connections outside the home. These mediating factors are supported by social networks and social capital research. Unfortunately, SOEP does not provide much information about participation in relevant social networks. SOEP has information about the frequency of nine types of leisure activities, however, of which seven tend to rise (statistically significantly) with parenthood when modelled as outcome variables similar to model 2 in Table 2; the other two items are unrelated to parenthood (results not shown in tables). The strongest associations are with visits to cinemas and music events, followed by participation in voluntary organisations.

There may also be other mediating mechanisms beyond social connections outside the home. One such additional mechanism might be that parents are subject to stricter social control than non-parents. Other adults, including one’s partner and other family members, as well as children, tend to have strong expectations regarding how parents, as opposed to childless adults, should behave, especially in the presence of children. This additional social control might make parents more trustworthy, or appear to be more trustworthy, than non-parents. If so, they might also more easily engage cooperatively with other people in ways that ultimately build trust.

A further mechanism might be that parents are more conscious of the issue of trusting others because the behaviour of other people can affect their children. This consciousness may lead to (more or less) constant monitoring and evaluation of the trustworthiness of other people who are physically and emotionally close to one’s children. If parents feel that their child is safe and secure, they might develop confidence and trust in other people as well. This mechanism is consistent with the social learning perspective regarding how social trust is affected by previous social encounters (e.g. Hardin 2002; Glanville and Paxton 2007).

Our research indicates that most children benefit their parents by making them more trustful towards other people. This additional social trust may lead to better-functioning societies (Dincer and Uslaner 2010; Horváth 2013). One implication of these arguments is that children may be a public good that benefits not only their parents but also their whole society insofar as they contribute, in one way or another, to trust and cooperation between its members.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Morten Blekesaune

Morten Blekesaune has been Professor of Sociology at the University of Agder, Norway, since 2010, coming from Norwegian Social Research in Oslo. He has conducted quantitative research on longitudinal and comparative data on various individual level outcomes including well-being and behaviour.