Do family ties and structure matter for ethical and moral values?

Abstract

Purpose

This paper seeks to determine whether family ties and structure correlate with the ethical and moral values that are important underpinnings of economic activities.

Design/methodology/approach

The analysis uses data from the World Values Survey (WVS). Given the multilevel nature of the data in a cross-country setting, the paper utilizes a multilevel linear mixed-effects model with maximum likelihood estimation.

Findings

Families with strong ties and those with traditional family structures are less tolerant of unethical conduct and have more restrictive moral values than households where ties are weak and the household is not married. There also appears to be a bi-causal relationship in the data.

Originality/value

This paper considers a broad array of values in a cross-country setting and utilizes a multilevel modeling approach that has not been done in studies linking both family ties and structure.

Peer review

The peer review history for this article is available at: https://publons.com/publon/10.1108/IJSE-12-2021-0730.

Citation

James Jr, H.S. and Giwa-Daramola, D. (2023), "Do family ties and structure matter for ethical and moral values?", International Journal of Social Economics, Vol. 50 No. 4, pp. 491-508. https://doi.org/10.1108/IJSE-12-2021-0730

1. Introduction

There are numerous pathways by which economic institutions affect the economic performance of economies (Chang, 2011; North, 1990). One pathway is through social norms, which are the shared but unwritten beliefs and attitudes that people have about how to behave within societies. Of the many types of social norms, values—that is, the beliefs and judgments of people about what is important in life—are particularly important. Social norms and values are important because they can substitute for as well as complement more formal institutions and rules of behavior, including direct economic incentives, although social norms can also work against them (Pitlik and Rode, 2017). A second pathway is through marriage and the family (Doepke and Tertilt, 2016), including the relevance of family ties and structure (Alesina and Giuliano, 2014; Guttman and Voigt, 2022). Greif (2006, p. 308) notes that “family structure” is central to “one of the most fundamental institutional changes in history [because of] its growth-related implications” and therefore calls for a greater understanding of the relationship between the emergence of corporations and the “nuclear family structure”.

This paper explores the connection between these two pathways. The specific research question is: do family ties and structure correlate with the ethical and moral values that are underpinnings of economic activity? Previous research has explored how family characteristics and processes moderate or impact the morality and values of children and other family members (Albanese et al., 2016; Prioste et al., 2015). Research has also examined how family characteristics and strengths contribute to economic indicators and outcomes. For example, Alesina and Giuliano (2011, 2014) and Ermisch and Gambetta (2010) show that family ties negatively correlate with generalized trust in society. Other research shows how family relationships link to economic and cultural institutions and norms (Alesina and Giuliano, 2010, 2015; Enke, 2019; Guttman and Voigt, 2022). As significant as these studies are, few studies have examined the connection between family ties and structure and ethical and moral values, especially in a cross-country setting (see Alesina and Giuliano, 2014). In this context, family ties refer to “the strength and resilience of family loyalties, allegiances, and authority” that exist within family and household relationships (Reher, 1998, p. 203). Family structure refers to the organizational arrangement of family households, such as the marital status of parents and the presence and number of children and other family and non-family members.

To answer our research question, we use the sixth wave of the World Values Survey (WVS) in a cross-country study of 46 countries for which data are available. Dependent variables are indices of the ethical and moral values of the household head. Family ties and structure are constructed from variables provided in the WVS motivated by previous studies that have examined the characteristics of families (e.g. Alesina and Giuliano, 2014; Davis and Williamson, 2020). Controls include individual and country-level variables expected or previously shown to correlate with ethical and moral values. We also use a multilevel modeling approach that is similar to other cross-country studies examining correlates of ethical values of households (e.g. James, 2022; Martin et al., 2012).

This research is important because it supports Grief's (2006, p. 312) call to examine “the dynamic interplay between family structures and institutional development”. While it is not our intent to describe how ethical and moral values are taught or transmitted within households or to explain why some families are strong or function better than others, the paper seeks to determine if family ties and structure correlate with values, which in turn can help explain why such values might be weak or strong in society. Although studies have examined the correlation between family ties or structure on some values, such as collectivist versus individualist values (Davis and Williamson, 2020; Prioste et al., 2015), this paper focuses on a broader array of values that are relevant for economic performance.

2. Background

2.1 The relevance of values on economic performance

Values matter for economic performance because they guide, direct or constrain perspectives and behavior and form the basis for determining right and wrong. For example, North (1994, p. 58) says that some values, such as “honesty, integrity, and hard work” are “critical determinants of the costs of transacting in complex political and economic exchange”. Research on the reasons and means by which values affect economic performance has progressed along several lines. One subsumes values within a broader consideration of culture (e.g. Giavazzi et al., 2019; Nikolaev and Salahodjaev, 2017). Another considers “the extent to which key values are shared in society” (Beugelsdijk and Klasing, 2016, p. 523). Accordingly, organizations and societies characterized by shared values and consistency of value systems perform better than those in which values are diverse (e.g. Ren, 2010). One way values do this is by helping societal members “overcome the free-rider problem in the pursuit of socially valuable activities” (Guiso et al., 2011, p. 419). Another way is that shared values foster trust, whereas the diversity of values correlates with lower generalized trust (Beugelsdijk and Klasing, 2016).

An important literature considers the array of values within and across societies, including how they evolve and interact with micro-and macro-level constructs (see Halman and Gelissen, 2019). There are two types of values that are relevant to the economic performance of societies. First, ethical values are the standards and rules relating to the consumption, exchange, property rights and cooperation, including the support of trust and trustworthiness (e.g. Uslaner, 2002). Important ethical values include keeping promises, being trustworthy, protecting property rights and respecting the rule of law. Much has been written on the relationship between ethical values and the well-being of society, especially with respect to economic performance. Adam Smith and other classical political economists made this argument (Zouboulakis, 2010). Rose (2011) explained how ethical values rather than feelings of empathy and sympathy for others are a necessary underpinning to trust and the development of effective economic institutions. There is literature documenting the relationship between indicators of ethical values and economic outcomes (e.g. James, 2015; Nunner-Walker, 2007).

Second, moral values include a broader set of considerations, such as beliefs about life and death and sexual practices (Beugelsdijk and Klasing, 2016). Scholars have noted that economic activity can impact traditional moral values and beliefs (e.g. Bowles, 1998). However, moral values can also impact economic performance. One way is that changes in moral values affect social capital. For instance, Carden (2009) suggested that changes in sexual mores can produce social unrest, which then impacts the formation and development of social capital needed for economic development. A second way is that changes in beliefs about sexuality, abortion and other morally-relevant behaviors can affect the overall culture of societies (Giavazzi et al., 2019; Greenwood and Guner, 2010), which in turn impacts economic performance.

2.2 The role and relevance of family on economic performance

Family can affect economic performance directly and indirectly. The direct effects of the family on economic performance include the provision of labor and its purchases of goods and services to meet household needs. Marriage and cohabitation relationships create new households and, as such, foster independent economic units that interact with other economic agents within the economy. While the nature and measurement of household and union formation are changing and are a source of debate among scholars (Sassler and Lichter, 2020), research considers how various facets of marital and household structures affect economic and related indicators. For example, Guttman and Voigt (2022) test Todd's (1985) claim that societies dominated by nuclear families differ significantly from extended family relationships. The authors find that families characterized by extended relationships tend to exhibit greater attitudes toward racism and preferences for economic inequality, but also support a greater level of rule of law, among other findings (some of which are contrary to Todd's assertions).

The indirect effect of family on economic activity is by orienting and transforming children and other household members into economic agents and by socializing children into roles and teaching them the norms and values that matter to society, such as cooperating, respecting the property of others and exercising self-restraint. Scholars have attempted to link the characteristics of families with cultural and social norms and values. An important development here focuses on the role of family ties (Alesina and Giuliano, 2010, 2011, 2014). For example, Davis and Williamson (2020) report that family ties are negatively correlated with various indicators of individualism, while Alesina and Giuliano (2011, 2014) show that family ties and trust are negatively correlated, suggesting that family strength substitutes for rather than complements trust in society. In related work, Son and Feng (2019) show that networks with higher levels of family ties exhibit greater within-network trust than low tie networks, but the effect on generalized trust is inconclusive. However, Enke (2019) examines the relationship between family structures and ties and various cultural, social and moral systems across societies. He finds that tight kinship structures correlate positively with some features of a society's moral system, such as the relative importance of communal moral values, suggesting that improved moral systems could mediate the negative relationship between family ties and trust.

2.3 The link between family characteristics and ethical and moral values

Many studies view the family unit as the most vital source of moral influence (White and Matawie, 2004). During the early years of a child's life, parents play a crucial role in fostering or inhibiting the process of moral internalization (Hardy et al., 2008). In fact, parents who tend to be more ethical provide better moral ideas and reasoning than parents who are less ethical (Parikh, 1980). Given this context, how might family ties and family structure correlate with the ethical and moral values of family or household members? Insights from the literature do not suggest a clear answer. Given the importance of ethical and moral values generally and the relevance of family ties and structure in economic analyses, and because few studies examine these considerations empirically in a cross-country setting, we examine the relationship between family ties and structure and ethical and moral values.

3. Conceptual framing, methods and procedures

Data are from the sixth (2010–2014) wave of the WVS (Inglehart et al., 2014) [1]. The WVS compiles responses from face-to-face interviews with citizens in many countries around the world. Interview subjects were randomly selected and stratified by region and degree of urbanization. For this study, the sample consists of 65,218 adults from 46 countries aged 18 years and older.

There are no data on the values, beliefs and attitudes of all individual family or household members in the WVS. Therefore, our dependent variable is proxied by the values of the survey respondent or household head. Research suggests there is a strong connection between the values of parents and those of their children so that the values of a household head can proxy for those of family members (Albanese et al., 2016; Degner and Dalege, 2013; White and Matawie, 2004).

Table 1 shows the variables from the WVS used to construct indices of ethical and moral values, family ties and other variables used in the analysis. Our measures for values are based on previous research suggesting that morally debatable behaviors can be grouped into two distinct types—those relating to attitudes about legal and illegal behaviors—that is, ethical values—and those relating to attitudes about sexuality as well as life and death— that is, moral values (see Katz et al., 1994). Previous studies have used the WVS to construct indices of these two variables (e.g. Beugelsdijk and Klasing, 2016; Marozzi, 2021; Storm, 2016). Accordingly, we construct variables for ethical values and moral values from responses to this question: “Please tell me for each of the following actions whether you think it can always be justified, never be justified, or something in between”. For ethical values, five actions were presented to respondents that are presumed to reflect unethical conduct, such as stealing property and cheating on taxes. For moral values, five items presented to respondents deal with the justifiability of conduct relating to life and sexuality, such as abortion and sex before marriage. For both types of values, responses were reverse coded. The five components for ethical values were combined, and the five components for moral values were combined, respectively, by extracting the first principal component of all responses to those variables in the dataset. An increase in the variable representing ethical values indicates a greater intolerance of unethical behavior (see, for instance, James, 2022; Martin et al., 2012), while an increase in the moral values variable represents more restrictive rather than permissive beliefs about life, death and sexual behavior (see, for instance, Marozzi, 2021; Studlar and Burns, 2015) [2].

We follow Alesina and Giuliano (2014) and Davis and Williamson (2020) in constructing a measure of family ties using variables from the WVS that attempt to capture the strength of family relationships and connections. One of the three variables utilized by these authors is in the sixth wave of the WVS [3], which asks respondents to indicate how important or not important family is. In addition, three other variables in the WVS reflect the concept of family strength. The first asks respondents to state how much they trust family. The second asks how important tradition is to the respondent and how important it is to follow the customs handed down from one's family or religion. The third asks respondents to state how much they agree with the statement that an important goal in life is to make their parents proud. We combined the individual-level responses to these four variables by extracting the first principal component to create an index variable we label family ties.

Although there are different ways of operationalizing family structure, we utilize two methods. First, numerous studies proxy family structure by the marital status of the survey respondent (see, for instance, McLanahan and Percheski, 2008; Williams and Baker, 2021). Thus, three categories are identified: the respondent is married or widowed; the respondent is living together as married; or the respondent is divorced, separated, or single. Second, we distinguish between nuclear families (where children are emancipated from their parent's household) and extended family systems (where adults with children live in their parent's home) (see, for instance, Enke, 2019; Guttman and Voigt, 2022). Two variables in the WVS allow us to determine if the household consists of an extended relationship—a question asking if the respondent lives with their parents and a question asking about the number of children the respondent has. Respondents living with a parent and indicating they have one or more children are coded as being an extended household; otherwise, they are presumed to be a nuclear household.

Consistent with other studies on ethical judgments, individual control variables include the religiosity of the respondent, the number of children in the household, the degree to which respondents are happy, the respondent's age, gender, income level and level of education (e.g. Craft, 2013; Pan and Sparks, 2012). Observations with missing values were replaced with variable means, and country-level means were differenced from individual-level covariates. Country-level controls, also utilized by Martin et al. (2012) and James (2022), are created from the World Governance Indicators (WGI) developed by the World Bank [4] and the Corruption Perception Index (CPI) from Transparency International [5]. In order to avoid the problem of reverse causation, we use data from 2008 rather than later years. The WGI ranges from −2.5 (weak institutions) to 2.5 (strong institutions), while the CPI ranges from 0 (highly corrupt) to 10 (highly clean). Both indexes were rescaled to range from one to 10 and then averaged to create a country-level index of institutional quality.

Because of the multilevel structure of the data, in which both individual- and country-level effects can correlate with the dependent variables, we use a multilevel linear mixed-effects model with maximum likelihood estimation. The estimated equation takes the following form:

Yij=β0+β1familytiesij+β2familystructureij+β3Xij+B4Wj+eij+μj

where Yij is the dependent variable for ethical values or moral values for respondent i in country j; familytiesij and familystructureij are the variables for family ties or family structure (i.e. marital status and nuclear or extended family); Xij represent individual-level controls; Wj is a country-level predictor; and eij+μj are individual-level and country-level error terms, respectively (see Snijders and Bosker (2012), for additional details).

4. Results

4.1 Main analysis

Table 2 presents two sets of results analyzing the relationship between family ties and structure on ethical and moral values—one using ordinary least squares (OLS) with country controls and the other using the multilevel mixed methods technique with individual-level variables. The OLS results show that family ties have a positive and significant effect on both ethical values and moral values. Interestingly, the effect of family ties on moral values is larger than the effect on ethical values [6] Furthermore, households where the respondent is single or living in an unmarried relationship with another partner have lower ethical and moral values. The effect of extended family relationships is similarly negative, although its significance disappears once controls are added in the analysis of moral values (see OLS 3b).

Models 1a and 1b in Table 2 introduce the mixed-effects analysis and are presented with a random intercept and without covariates in order to calculate the intraclass correlation coefficient [7]. The percent of variability of values accounted for by country-level effects is 18.6% for ethical values and 37.7% for moral values, suggesting there is clustering in the data at the country level, especially in the case of moral values, thus confirming the need for a multilevel regression analysis. Models 2a and 2b show the results of the individual level analysis, which indicate that family ties are positively correlated with the ethical values and moral values of the household head, consistent with the OLS findings. In other words, households with stronger family ties are less accepting of unethical conduct and tend to hold more restrictive attitudes relating to conduct affecting life, death and sexuality than households with lower family ties.

Table 3 introduces the country-level control for institutional quality as well as interaction effects between family ties and family structure. The introduction of the country-level control does not affect the relationship between family ties and ethical and moral values. While the country-level control has a positive and substantive effect on ethical values (see Model 3a), it has a significantly negative effect on moral values (see Model 3b), suggesting that ethical values tend to be higher on average in countries with strong economic institutions while moral values tend to be lower as institutional quality increases. Similarly, including the institutional-level control does not change the relationship between the family structure variables (marital status and extended family relationships), which remain negative and significant except for the correlation between extended families and moral values. Furthermore, there appear to be important interaction effects. For example, the interaction between institutional quality and family ties is negative in the case of ethical values (Model 4a) but positive in the case of moral values (Model 4b), suggesting that improved institutional quality works against family ties in the ethical values model but works with family ties in the moral values model. Models 5a and 5b add interactions with the family structure variables. Institutional quality has a negative interaction with family households where the respondent is not married for both the ethical values and moral values models. The interaction between institutional quality and the presence of an extended household relationship is oppositive in both models; in the case of ethical values, the relationship is negative (Model 5a) but positive in the case of moral values (Model 5b).

Overall, the effect of the control variables is not surprising. Religiosity, happiness and age have a consistently positive effect on ethical and moral values, while income has a consistently negative effect on the dependent variables. Being female and having post-secondary education is positively correlated with ethical values but negatively correlated with moral values.

4.2 Robustness check on dependent variables

Previous cross-country studies examining correlates of attitudes toward morally debatable behavior typically construct the dependent variable as the average of individual components from the WVS (e.g. James, 2022; Martin et al., 2012). For comparison purposes, Table 4 provides the results of our multilevel analyses using this (averaged) method of constructing the dependent variables of ethical values and moral values (reported as Models 6a and 6b). However, Marozzi (2021) states that such studies may not be sufficiently rigorous if constructed indices are computed as unweighted averages of individual components. Thus, we provide an additional check of our results by calculating new measures for ethical values and moral values as weighted averages, using the procedure outlined in Marozzi (2021). Models 7a and 7b in Table 4 provide the results of this analysis. All models presented in Table 4 are generally consistent with those reported above. Family ties have a positive and significant effect on ethical values and moral values, and the results for marital status and extended family relationships follow similar patterns reported in Tables 2 and 3.

4.3 Bi-causality analysis

Although we have established a correlation between family ties and structure and ethical and moral values, it is useful to determine if there is causality among the explanatory and dependent variables. For example, do households with stronger family ties or who live as nuclear families cause ethical and moral values to be higher, or do individuals who are less tolerant of unethical conduct and who have more restrictive than permissive moral values have families with strong ties or avoid extended family relationships? Or is the causal connection more complicated? Some research suggests a more complicated relationship. For example, while families impact economic conditions, there is also literature showing how economic factors affect family structure and functioning (White and Rogers, 2000) and relevance (Guttman and Yacouel, 2007), including in cross-country settings (e.g. Majeed and Kanwal, 2019).

In order to tease out causal pathways, we present two analyses in Table 5. The first utilizes two-stage least squares techniques, and the second employs a three-square regression approach. Because our proxies for family structure are dummy variables, we consider only the potential bi-causal relationship between family ties and ethical and moral values. The results in Table 5 show that not only do family ties continue to have a positive and significant effect on ethical and moral values, but these values also correlate with higher levels of family ties. Thus, there appears to be a bi-causal relationship between family ties and ethical and moral values.

5. Discussion

This study has demonstrated that family ties correlate positively with the ethical and moral values of individuals in a cross-country setting, even after controlling for individual and country level effects. However, previous researchers have documented that family ties correlate negatively with generalized trust. These findings might seem counterintuitive given that Rose (2011) claims ethical and moral values are essential for the development of trust in society. How can family ties correlate positively with ethical and moral values while at the same time correlate negatively with trust, if ethical and moral values are in turn necessary for trust? There are two possible reasons. First, strong family ties foster an enclosed sense of security within households, so that outside the bounds of these relationships trust is jeopardized (Alesina and Giuliano, 2014; Ermisch and Gambetta, 2010). In contrast, weak family ties provide a stronger incentive for family members to get along with strangers in order to fulfill the needs that are not being met by family, thereby contributing to generalized trust. However, Enke (2019) concludes that improved moral systems could mediate the negative relationship between family ties and trust. If the family fosters within network security as well as socializes members on cooperative behavior outside the family through the transmission of ethical and moral values, such as those evaluated here, then the aggregate nature of cross-country studies may not fully capture the idiosyncratic dynamics of how values, family characteristics, and other factors interact at the household level.

Second, James (2015) shows that the generalized morality of citizens has a more pronounced effect on trust when economic institutions are weak than when they are strong. Thus, accounting for institutional quality may partly resolve the apparent counterintuitive relationships among generalized trust, family ties and ethical and moral values. There appears to be some empirical support for this conjecture. Table 6 reports results of a simple logistic regression with individual trust as the dependent variable for countries with low (e.g. Azerbaijan, Haiti), moderate (e.g. China, Thailand) and high (e.g. Australia, Sweden) quality of economic institutions [8]. Table 7 reports results where models 3a and 3b are re-estimated for countries with low, moderate and high institutional quality. The tables exhibit several interesting findings. First, the effect of family ties on trust is not consistently negative. Rather, it is neutral for countries with low institutional quality, negative for countries of moderate institutional quality and positive for countries with high institutional quality (see the model for family ties in Table 6). Second, ethical values have a negative effect on trust for low quality countries, no effect in the case of moderate countries and a positive effect in countries with good institutions. However, the effect of moral values on trust is negative for all levels of institutional quality. Third, consistent with the interactions shown in Models 4a, 4b, 5a and 5b in Table 3, improved institutional quality reduces the effect of family ties on ethical values and increases modestly the effect of family ties on moral values (see Table 7). Together these findings reveal that in countries with strong economic institutions, family ties can reinforce ethical values and trust, suggesting that family ties, generalized trust and ethical values complement economic institutions in some contexts. In contrast, ethical and moral values, family ties and generalized trust can interfere with each other when economic institutions are relatively weak, as implied by explanations for the negative relationship between family ties and trust, which might explain in part why economic institutions are weak. Given findings by Alesina and Giuliano (2014) that family ties negatively correlate with indicators of institutional quality, more work is clearly needed to understand better the extent to which family ties either complement or interfere with trust and institutional quality.

6. Conclusions

Do family ties and family structure correlate with the ethical and moral values that are important underpinnings of economic activity? We find evidence in the affirmative in our cross-country study. In particular, we find that households with strong family ties, married or widowed households and households characterized by a nuclear family structure are less tolerant of unethical conduct and have less permissive moral values when compared to single and living together households and households with weaker levels of family functioning. The results are generally robust after controlling for individual and country-level effects and a potential for bi-causality.

Our findings that family ties and structure correlate with ethical and moral values underscore the continuing importance of the family as an economic institution in society. The creation of strong families and the transmission of ethical and moral values within households are crucial and support the indirect way in which families impact the economic performance of societies—through the orientation and transformation of children into productive and ethical economic agents. This suggests that a strategy for strengthen the ethical and moral values of society generally and for encouraging the ethical judgments and behaviors of individuals should focus in part on strengthening traditional family units and the relationships among family members. Furthermore, if future scholarship confirms evidence that family characteristics correlate with values that promote economic activity, then a strategy for strengthening promoting economic performance should focus in part on strengthening traditional family units.

If family structure and ties matter for economic performance, then there might be a reverse effect whereby the breakdown of the nuclear family has real economic consequences. There appears to be evidence for this conjecture (e.g. Iceland, 2003). For example, Aguirre (2001) showed how rising divorce, illegitimacy and single and no parent households are linked to greater welfare costs, higher opportunity costs from money wasted in family courts, and deteriorating human capital development of children, each of which correlates with lower economic growth and development. Our paper points to an additional avenue by which a deterioration of the structure and a weakening of family ties can have adverse economic impacts on societies.

Table 1

Variable names, definitions and summary statistics

Variable	Definition (range in parentheses)	Mean (Std Dev)
Values variables
Claiming	How justifiable: “Claiming government benefits to which you are not entitled”(1 = always justifiable and 10 = never justifiable)	8.26 (2.50)
Avoiding	How justifiable: “Avoiding a fare on public transportation” (1 = always justifiable and 10 = never justifiable)	8.26 (2.46)
Stealing	How justifiable: “Stealing property” (1 = always justifiable and 10 = never justifiable)	9.15 (1.86)
Cheating	How justifiable: “Cheating on taxes if you have a chance” (1 = always justifiable and 10 = never justifiable)	8.75 (2.15)
Bribing	How justifiable: “Someone accepting a bribe in the course of their duties” (1 = always justifiable and 10 = never justifiable)	8.99 (1.98)
Ethical values	Extracted first principal component after combining the five previous “how justifiable” variables (−4.52 to 0.73)	0.00 (1.00)
Homosexuality	How justifiable: homosexuality (1 = always justifiable and 10 = never justifiable)	7.62 (3.00)
Abortion	How justifiable: abortion (1 = always justifiable and 10 = never justifiable)	7.66 (2.77)
Divorce	How justifiable: divorce (1 = always justifiable and 10 = never justifiable)	6.29 (3.07)
Sex before	How justifiable: sex before marriage (1 = always justifiable and 10 = never justifiable)	6.33 (3.28)
Suicide	How justifiable: suicide (1 = always justifiable and 10 = never justifiable)	8.70 (2.22)
Moral values	Extracted first principal component after combining the five previous “how justifiable” variables (−2.84 to 1.17)	0.00 (1.00)
Family structure variables
Living together	Dummy variable = 1 if living together as married; zero otherwise	0.08 (0.28)
Single	Dummy variable = 1 if divorced, separated, or single; zero otherwise	0.32 (0.47)
Married	Dummy variable = 1 if is or was married or widowed; zero otherwise	0.60 (0.49)
Extended	Dummy variable = 1 if living with parents and has at least one child; zero otherwise	0.10 (0.30)
Family ties variables
Family important	How important family is (1 = not at all important and 4 = very important)	3.91 (0.33)
Trust family	How much trust family (1 = not at all and 4 = completely)	3.78 (0.52)
Family tradition	How important tradition is and following customs handed down by family or religion (1 = not at all like me and 6 = very much like me)	4.42 (1.40)
Make parents proud	A main goal in life is to make parents proud (1 = strongly disagree and 4 = strongly agree)	3.31 (0.74)
Family ties	Extracted first principal component after combining the four previous variables (−8.18 to 1.16)	0.00 (1.00)
Individual-level controls
	How important religion is in respondent's life (1 = not at all important and 4 = very important)	3.06 (1.05)
	How often respondent attends religious services (1 = never or practically never and 7 = more than once a week)	3.94 (2.12)
	How often respondent prays (1 = never or practically never and 8 = several times a day)	5.37 (2.61)
Religiosity	Extracted first principal component after combining the three previous variables (−8.18 to 1.16)	0.00 (1.00)
Number of children	Number of children (0–8)	1.77 (1.72)
Happiness	Overall level of happiness (1 = not very happy and 4 = very happy)	3.17 (0.72)
Age	Respondent age, in years (18–93)	42.14 (16.65)
Income decile	Household income (1 = lowest decile and 10 = highest decile)	4.79 (2.09)
Female	Dummy variable = 1 if female; zero if male	0.53 (0.50)
Post-secondary education	Dummy variable = 1 if respondent had some or completed post-secondary (university) education; zero otherwise	0.26 (0.44)
Country-level control
Institutional quality	Average of institutional quality data from the WGI and CPI, from the year 2008 (1 = low quality institutions and 10 = high quality institutions; rescaled; range 2.63–9.00)	5.27 (1.74)
Other
Trust	Dummy variable = 1 if respondent reports most people can be trusted (in contrast to one needs to be very careful); zero otherwise	0.23 (0.42)
N		65,218

Table 2

Estimates from OLS and multilevel linear model of individual level variables on ethical values and moral values

Variables	Dep var: ethical values					Dep var: moral values
	OLS 1a	OLS 2a	OLS 3a	Model 1a	Model 2a	OLS 1b	OLS 2b	OLS 3b	Model 1b	Model 2b
Intercept	0.463*** (0.012)	0.505*** (0.012)	0.443*** (0.013)	0.059 (0.062)	0.056 (0.062)	−0.564*** (0.017)	−0.472*** (0.017)	−0.395*** (0.017)	−0.008 (0.092)	0.108 (0.091)
Family ties	0.069*** (0.002)		0.061*** (0.003)		0.102*** (0.004)	0.125*** (0.003)		0.084*** (0.003)		0.082*** (0.003)
Living together		−0.103*** (0.009)	−0.052*** (0.009)		−0.086*** (0.014)		−0.255*** (0.012)	−0.136*** (0.012)		−0.137*** (0.012)
Single		−0.101*** (0.005)	−0.046*** (0.006)		−0.056*** (0.009)		−0.209*** (0.007)	−0.068*** (0.008)		−0.072*** (0.008)
Extended		−0.040*** (0.007)	−0.020*** (0.008)		−0.026** (0.012)		−0.027*** (0.010)	−0.007 (0.010)		−0.008 (0.010)
Religiosity			0.012*** (0.003)		0.016*** (0.005)			0.212*** (0.004)		0.203*** (0.004)
Number of children			−0.006*** (0.002)		−0.008*** (0.003)			0.000 (0.002)		0.002 (0.002)
Happiness			0.012*** (0.003)		0.015*** (0.005)			0.018*** (0.004)		0.012*** (0.004)
Age			0.004*** (0.000)		0.004*** (0.000)			0.004*** (0.000)		0.004*** (0.000)
Income decile			−0.009*** (0.001)		−0.029*** (0.002)			−0.029*** (0.002)		−0.030*** (0.002)
Female			0.010** (0.005)		0.021*** (0.007)			−0.074*** (0.006)		−0.066*** (0.006)
Post-secondary education			0.032*** (0.006)		0.074*** (0.008)			−0.181*** (0.007)		−0.178*** (0.007)
Error variance
Intercept				0.178*** (0.037)	0.175*** (0.036)				0.391*** (0.082)	0.376*** (0.078)
Residual				0.779*** (0.004)	0.758*** (0.004)				0.647*** (0.004)	0.585*** (0.003)
Model fit
R-square	0.083	0.082	0.090			0.335	0.333	0.379
AIC	93,968	96,403	93,330	169,096	167,310	67,838	68,665	65,709	156,947	150,479

Note(s): OLS models estimated with individual country dummies. Models 2a and 2b estimated with random intercepts and fixed slopes. For marital status, married is the excluded category relative to living together and single. Standard errors in parentheses (with robust standard errors reported for the OLS regressions). N = 65,218. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Table 3

Estimates from multilevel linear model of individual and country-level variables on ethical values and moral values with interaction effects

Dep var: ethical values				Dep var: moral values
Variables	Model 3a	Model 4a	Model 5a	Model 3b	Model 4b	Model 5b
Intercept	−0.306 (0.190)	−0.305 (0.190)	−0.346* (0.191)	1.444*** (0.206)	1.444*** (0.206)	1.412*** (0.206)
Individual-level variables
Family ties	0.102*** (0.004)	0.131*** (0.012)	0.135*** (0.012)	0.082*** (0.003)	0.003 (0.010)	0.008 (0.010)
Living together	−0.086*** (0.014)	−0.086*** (0.014)	−0.077* (0.043)	−0.137*** (0.012)	−0.137*** (0.012)	−0.002 (0.038)
Single	−0.056*** (0.009)	−0.057*** (0.009)	0.039 (0.026)	−0.072*** (0.008)	−0.071*** (0.008)	0.007 (0.022)
Extended	−0.026** (0.011)	−0.026** (0.011)	0.073* (0.038)	−0.008 (0.010)	−0.008 (0.010)	−0.106*** (0.034)
Religiosity	0.016*** (0.004)	0.016*** (0.005)	0.016*** (0.005)	0.203*** (0.004)	0.201*** (0.004)	0.200*** (0.004)
Number of children	−0.008*** (0.003)	−0.007*** (0.003)	−0.007*** (0.003)	0.002 (0.002)	0.002 (0.002)	0.003 (0.003)
Happiness	0.015*** (0.005)	0.015*** (0.005)	0.015*** (0.005)	0.012*** (0.004)	0.012*** (0.004)	0.012*** (0.004)
Age	0.004*** (0.000)	0.005*** (0.000)	0.005*** (0.000)	0.004*** (0.000)	0.004*** (0.000)	0.004*** (0.000)
Income decile	−0.029*** (0.002)	−0.028*** (0.002)	−0.029*** (0.002)	−0.030*** (0.002)	−0.030*** (0.002)	−0.030*** (0.002)
Female	0.021*** (0.007)	0.022*** (0.007)	0.022** (0.007)	−0.066*** (0.006)	−0.066*** (0.006)	−0.066*** (0.006)
Post-secondary education	0.074*** (0.008)	0.073*** (0.008)	0.073*** (0.009)	−0.178*** (0.007)	−0.177*** (0.007)	−0.176*** (0.008)
Country-level variable
Institutional quality	0.068** (0.034)	0.068** (0.034)	0.076** (0.034)	−0.252*** (0.037)	−0.252*** (0.037)	−0.246*** (0.037)
Interactions
Family ties x Institutional quality		−0.005*** (0.002)	−0.006*** (0.002)		0.015*** (0.002)	0.014*** (0.002)
Living together x Institutional quality			−0.002 (0.008)			−0.024*** (0.007)
Single x Institutional quality			−0.018*** (0.004)			−0.015*** (0.004)
Extended x Institutional quality			−0.020*** (0.007)			0.020*** (0.006)
Error variance
Intercept	0.159*** (0.033)	0.160*** (0.033)	0.160*** (0.033)	0.187*** (0.039)	0.187*** (0.039)	0.186*** (0.039)
Residual	0.758*** (0.004)	0.758*** (0.004)	0.758*** (0.004)	0.585*** (0.003)	0.585*** (0.003)	0.584*** (0.003)
Model fit
AIC	167,308	167,303	167,287	150,449	150,387	150,361

Note(s): All models estimated with random intercept and fixed slopes. For marital status, married is the excluded category. Standard errors in parentheses. N = 65,218. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Table 4

Estimates from multilevel linear model of individual and country-level variables using raw and weighted measure of ethical values and moral values

Variables	Dep var: ethical values		Dep var: moral values
	Model 6a	Model 7a	Model 6b	Model 7b
Intercept	8.158*** (0.329)	0.000 (0.002)	10.603*** (0.471)	0.015*** (0.002)
Individual-level variables
Family ties	0.175*** (0.007)	0.018*** (0.000)	0.183*** (0.008)	0.017*** (0.001)
Living together	−0.154*** (0.024)	−0.010*** (0.002)	−0.316*** (0.027)	−0.019*** (0.002)
Single	−0.102*** (0.016)	−0.008*** (0.002)	−0.165*** (0.018)	−0.012*** (0.001)
Extended	−0.041** (0.020)	−0.004* (0.002)	−0.019 (0.023)	−0.002 (0.002)
Controls	Yes	Yes	Yes	Yes
Country-level variable
Institutional quality index	0.118** (0.059)	−0.000 (0.000)	−0.572*** (0.084)	0.001** (0.000)
Error variance
Intercept	0.477*** (0.100)	0	0.978*** (0.204)	0
Residual	2.259*** (0.013)	0.023*** (0.000)	2.982*** (0.017)	0.020*** (0.000)
Model fit
AIC	239,554	−61,565	256,652	−70,691

Note(s): All models estimated with random intercept and fixed slopes, including control variables used previously. In Models 6a and 6b, the dependent variables are the average of each of the five components comprising ethical values and moral values respectively. In Models 7a and 7b, the dependent variables are weighted using the technique described in the text. For marital status, married is the excluded category. Standard errors in parentheses. N = 65,218. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Table 5

Bicausality analysis using two-stage and three-stage least squares regressions

Variables	2SLS		3SLS
Ethical values	Dep: ethical values	Dep: family ties	Dep: ethical values	Dep: family ties
Intercept	0.184*** (0.022)	0.133*** (0.047)	0.208*** (0.021)	0.129*** (0.047)
Family ties	0.489*** (0.022)		0.481*** (0.017)
Living together	−0.089*** (0.015)		−0.126*** (0.007)
Single	−0.069*** (0.010)		−0.126*** (0.004)
Extended	−0.025** (0.012)		−0.022*** (0.006)
Ethical values		1.975*** (0.079)		1.923*** (0.065)
Individual level controls	Yes	Yes	Yes	Yes
Country dummies	Yes	Yes	Yes	Yes
R-square	0.215	0.013	0.224
Moral values	Dep: moral values	Dep: family ties	Dep: moral values	Dep: family ties
Intercept	−0.436*** (0.018)	−0.092*** (0.028)	−0.416*** (0.017)	−0.091*** (0.028)
Family ties	0.075*** (0.018)		0.302*** (0.015)
Living together	−0.144*** (0.012)		−0.179*** (0.010)
Single	−0.084*** (0.008)		−0.135*** (0.006)
Extended	−0.008 (0.010)		−0.006 (0.008)
Moral values		0.991*** (0.025)		0.976*** (0.024)
Individual level controls	Yes	Yes	Yes	Yes
Country dummies	Yes	Yes	Yes	Yes
R-square	0.412	0.035	0.334

Note(s): Instruments used in first-stage regressions include individual level controls as well as the following: for ethical values and moral values, whether respondent sees themself as part of the nation; for family ties, whether respondent trusts people they know personally. Analyses include country dummies. For marital status, married is the excluded category. Standard errors in parentheses. N = 65,218. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Table 6

Logistic regressions with trust as dependent variable for countries with low, moderate and high institutional quality

Variables	Model for family ties			Model for ethical values			Model for moral values
Institutional quality level	Low	Moderate	High	Low	Moderate	High	Low	Moderate	High
Family ties	−0.009 (0.024)	−0.048*** (0.015)	0.041** (0.018)
Ethical values				−0.063*** (0.022)	0.005 (0.015)	0.076*** (0.024)
Moral values							−0.138*** (0.029)	−0.055*** (0.017)	−0.139*** (0.020)
Other controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Model fit
Likelihood ratio	111.666***	230.962***	569.768***	119.719***	220.903***	574.565***	133.917***	230.819***	612.099***
AIC	13,777	35,404	18,421	13,769	35,414	18,416	13,755	35,404	18,379
% trust = 1	21.2	18.1	40.1	21.2	18.1	40.1	21.2	18.1	40.1

Note(s): Dependent variable is trust, which = 1 if respondent says others can be trusted; zero otherwise. Controls include marital status, age, income decile, female, and post-secondary education. Standard errors in parentheses. Countries with low (high) institutional quality had an index value that was at least one standard deviation below (above) the mean. N for low quality institutions is 13,418; N for moderate quality institutions is 37,713; N for high quality institutions is 14,087. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Table 7

Estimates from multilevel linear model of individual and country-level variables on ethical values and moral values for countries with low, moderate, and high institutional quality

Variables	Dep var: ethical values			Dep var: moral values
Institutional quality level	Low	Moderate	High	Low	Moderate	High
Family ties	0.116*** (0.009)	0.115*** (0.005)	0.053*** (0.007)	0.069*** (0.007)	0.084*** (0.004)	0.081*** (0.008)
Institutional quality	0.888 (0.602)	0.063 (0.078)	−0.077 (0.084)	0.041 (0.432)	−0.236*** (0.087)	−0.342* (0.193)
Other controls	Yes	Yes	Yes	Yes	Yes	Yes
Model fit
AIC	36,740	98,010	31,336	28,954	86,457	34,246

Note(s): All models estimated with random intercept and fixed slopes. Controls are the same as those listed in Table 3 models 3a and 3b. Standard errors in parentheses. Countries with low (high) institutional quality had an index value that was at least one standard deviation below (above) the mean. N for low quality institutions is 13,418; N for moderate quality institutions is 37,713; N for high quality institutions is 14,087. *** = p < 0.01, ** = p < 0.05, * = p < 0.10

Notes

1. Data are used from the sixth rather than the more recent seventh wave of the WVS because too many countries in the seventh wave had missing variables or data.

2. In stating how one might interpret what an increase or decrease in the moral values variable represents, we do not make or imply any judgment on the morality of any of the specific components used in constructing that variable.

3. The missing variables reflect the duties of parents and children and the love and respect for one's parents.

4. See http://info.worldbank.org/governance/wgi/#home

5. See https://www.transparency.org/research/cpi/cpi_2008/0

6. We note the low R-square in the OLS model for ethical values relative to the OLS model for moral values. James (2011, Table 3) reports a low R-square in regressions in which ethical values is the dependent variable (ranging from 0.02 to 0.09). As a comparison, Beugelsdijk and Klasing (2016, Table 2) use ethical and moral values as explanatory variables in models of trust, reporting that in the regression with ethical values, the adjusted R-square is 0.09 compared to 0.23 for the model with moral values.

7. The intraclass correlation coefficient is calculated by dividing the intercept error variance by the total error variance (Snijders and Bosker, 2012). The Akaike information criterion (AIC) is used to compare regression models. Because the absolute value of the AIC is meaningless, when comparing comparable models (e.g. Model 2a relative to Model 1a or Model 2b relative to Model 1b), the model with the lower AIC is the better model (see Baguley, 2012, p. 402).

8. Countries were identified as having low (high) quality economic institutions if their institutional quality index was at least one standard deviation below (above) the mean.

Conflicts of interest/Competing interests: The authors declare no conflicts of interest.