The degree to which one trusts (i.e., has confidence or faith in) others is related to both physical and psychological well-being (Helliwell, Huang, & Wang, 2018), and longer life expectancy (Giordano, Mewes, & Miething, 2019). This is presumably because higher degrees of trust are related to higher support-seeking from others, and because trust facilitates cooperation with other people (Kawachi, 2018). Trust – both in people in general and in political institutions – has also been linked to favorable societal functioning, such as low levels of corruption (Rothstein & Uslaner, 2005), and the expansion of social services through the welfare state (Svendsen & Bjørnskov, 2013).
From a developmental perspective, attachment theory posits that trust is developed through experiences in early close relationships, and that mental models of these experiences impact trust in future relationships as well as in the world at large (Bowlby, 1973). Attachment orientations might also impact how people function in ways that have bearing at the societal level, including support-seeking from societal institutions. Previous research, however, has not linked these two strands of literature together – trust beyond interpersonal relationships on the one hand, and attachment research on the other.
In the present study we therefore tested whether attachment orientations, defined as attachment-related avoidance and anxiety, are related to trust in other people in general and in political institutions. More specifically, we focus on political institutions that are designed to provide support and care, namely welfare state institutions.
Attachment theory was originally developed to describe and understand the relationship between the infant and their caregiver(s) (Bowlby, 1969/1982), but has later been expanded to also understand interpersonal relationships in adulthood (Mikulincer & Shaver, 2016). The theory posits that safety and security are acquired through proximity with attachment figures, and that experiences in attachment relationships lead to the formation of characteristic mental models (“internal working models”) of oneself, others, and the world at large (Bowlby, 1973). Experiences of sensitive and responsive responses of attachment figures presumably lead to feelings of security within the specific relationship, to higher trust in one’s abilities, and to more positive representations of other people (i.e., secure attachment). In contrast, experiences with attachment figures who reject one’s emotional needs or respond unpredictably instead lead to insecure working models, where one ultimately finds alternative strategies to acquire felt security (Mikulincer & Shaver, 2016). Typically, this is done either by relying on oneself instead of others (i.e., avoidant attachment), or trying to stay as close as possible to the attachment figure to avoid being abandoned (i.e., anxious attachment).
In adulthood, differences in attachment strategies can be captured by two orthogonal dimensions (Brennan, Clark, & Shaver, 1998): attachment-related avoidance and anxiety. Attachment-related avoidance reflects a higher degree of discomfort with intimacy and dependency in relationships, and is characterized by negative mental models of other people (Mikulincer & Shaver, 2016). In contrast, attachment anxiety reflects fear of not being loved and extensive worries of abandonment, and is characterized by negative mental models of oneself. Although attachment research has historically been dominated by a typological approach (i.e., secure, avoidant, anxious, fearful as distinct categories), later research has indicated that attachment orientations are more accurately conceptualized and measured using a dimensional model than a typological one (Fraley et al., 2015). Low levels of both attachment-related avoidance and anxiety thus indicate a relatively secure attachment orientation.
Not surprisingly, attachment security is positively linked to trust in interpersonal relationships. For example, people with a secure attachment orientation report higher levels of trust in their partners (Karantzas, Feeney, Goncalves, & McCabe, 2014), have higher access to trust-related memories and thoughts, and achieve faster conflict resolution in trust-threatening situations (Mikulincer, 1998). In contrast, attachment insecurity, and especially avoidance, is linked to lower trust in close relationships, with one consequence being relationship dissatisfaction for both oneself and one’s partner (Fitzpatrick & Lafontaine, 2017).
Attachment orientations are also related to representations of and behavioral inclinations in relation to other objects in addition to people. For example, people with higher attachment anxiety experience more stress when separated from inanimate objects such as their cellphones (Keefer, Landau, Rothschild, & Sullivan, 2012), and religious individuals’ image of God generally corresponds to their attachment orientation (Granqvist, 2020). Presumably, attachment working models do not only reflect representations of other people, but also manifest in representations of other objects and of the world at large (Bowlby, 1973). Similarly, attachment orientations may not only be related to trust in close interpersonal relationships (Fitzpatrick & Lafontaine, 2017) but also in other sources that could provide support, such as welfare state institutions. Nevertheless, research on attachment orientation in relation to trust beyond close relationships is scant (Bradshaw, Kent, Henderson, & Setar, 2019; Gillath, Ai, Branicky, Keshmiri, Davison, & Spaulding, 2021), and in welfare-state institutions, non-existent. One aim of the present study was to help fill this important knowledge gap in the literature.
Social and Political Trust
Social trust (i.e., trust in people in general) is conceived as a rather stable psychological disposition (Dawson, 2019) that begins to form early in life (Uslaner, 2002), and is socially transmitted between parent and child (Dohmen, Falk, Huffman, & Sunde, 2012). On an aggregated level, both social and political trust are fairly stable in cultures and societies, though political trust displays greater variation than social trust (Listhaug & Jakobsen, 2018). Nonetheless, events, such as changes in democracy (Listhaug & Jakobsen, 2018), influence political trust, and social trust can be eroded by traumatic experiences (Sturgis, Read, & Allum, 2010). Also, a socioeconomic background of higher education and income is linked to higher trust (Alesina & La Ferrara, 2002).
Social and political trust are positively correlated with one another (Schyns & Koop, 2010), such that higher degrees of political trust seem to predict later social trust (Sønderskov & Dinesen, 2016). Also, countries displaying more expansive welfare states tend to display higher social and political trust (Svendsen & Bjørnskov, 2013).
Trust and the Welfare State
A welfare state refers to a set of legislations and social policies provided by the state through the means of taxation, including healthcare, education and social benefits, creating a safety net that provides support and care for the individual. The organization and size of the welfare state differ to a large extent among countries regarding whom welfare services are directed to: the larger population or marginalized groups (Esping Andersen, 1990). Experiences of welfare services that are designed based on universal principles (i.e., aimed at the larger population) rather than on means testing (i.e., targeted at marginalized groups) have been linked to higher political and/or social trust, for example among Swedes (Kumlin & Rothstein, 2005) and Americans (Mettler & Stonecash, 2008) alike. Presumably, experiences from welfare state services can be seen as representative of the political system in general (Kumlin, 2004). Perceptions of “fair” treatment from welfare state services may, accordingly, be generalized to one’s perception of the political system (Rothstein & Uslaner, 2005), as well as to one’s perception of the trustworthiness of people in general (Rothstein & Stolle, 2008). However, most previous research has focused on trust in political institutions in general, and across countries with different kinds of welfare systems, but have not examined trust specifically in relation to welfare state institutions.
Following this strand of research it seems likely that universal welfare states, whose political institutions are perceived as more fair, may lead people to become more trusting. However, the matter of causality has been disputed. The welfare state could be portrayed as a large-scale social dilemma, where individuals need to trust both the performance of political institutions and that other people contribute to the system (i.e., pay taxes, do not freeride) to the same extent as oneself (Rothstein, 2001). In this view, building and maintaining an expansive welfare state presupposes a fair share of both political and social trust, and indeed historical levels of social trust predict the current size of the welfare state (Svendsen & Bjørnskov, 2013). Hence, trust could also be seen as a prerequisite to build an expansive welfare state.
The Present Study
The main purpose of the present study was to examine whether attachment orientations are related to trust in political institutions, more specifically welfare state institutions. As the size and universality versus targeted nature of the welfare state differ among countries, data from two different national contexts with different size and organization of the welfare state were collected: the United States and Sweden. The United States has a smaller, targeted welfare system that is primarily based on means-testing and directed at marginalized groups. Sweden, on the other hand, has one of the most expansive welfare states in the world, and is based on universal principles to a larger extent.
Since previous studies have primarily examined trust towards other types of political institutions, we developed a new scale to measure trust specifically in welfare state institutions. Also, because attachment orientation has not been studied in relation to general social trust, we included measurements of social trust in the study. This also gave us the opportunity to explore whether social trust statistically mediates the presumed link between attachment orientation and trust in welfare state institutions, or, conversely, whether welfare state trust statistically mediates a link between attachment orientation and social trust.
Because interpersonal attachment orientation has not been studied in relation to social or welfare state trust previously we refrained from making specific predictions, and instead treated the research questions as exploratory (see pre-registration in the Supplementary Materials). However, based on theoretical assumptions and research on dyadic trust in interpersonal relationships, we tentatively expected that attachment-related avoidance (i.e., negative model of others) would be negatively related to social trust and, perhaps by extension, to welfare state trust as well. It is less straightforward to anticipate results for attachment anxiety, which is not marked by a negative model of others but a negative model of self.
Additionally, the welfare state and religion have previously been argued to function as competing sources of security, because religiosity is negatively correlated with both size of the welfare state and preference for welfare spending (Scheve & Stasavage, 2006). Religion can also fill a compensatory function for insecure attachment (Granqvist, 2020). We therefore controlled for the possible influence of religious beliefs and behavior in our final analyses. Finally, because left-right political orientation may coincide with different attitudes toward the welfare state, we also controlled for such attitudes in the analyses. Common demographic variables (age, gender, and educational level), and experiences of welfare state services were also controlled in the analyses, because these variables could impact trust in the welfare state (Kumlin, 2004) and/or be linked to attachment orientations (Mikulincer & Shaver, 2016).
The participants were recruited through the participant pool of Qualtrics and partook voluntarily in exchange of a small monetary reward. To minimize demographic differences between the two contexts (the US and Sweden) and to guarantee a spread of demographic variation within the samples, identical quotas in the two contexts were used based on age, gender, and socioeconomic background (educational level; for further details regarding the quotas see Supplementary Materials). To ensure that participants answered carefully, attention checks were included in the form of control questions with one correct answer. A failure on the attention checks, or if time spent on survey was under a threshold of three minutes, led to exclusion of the participant from the survey.
The final participants included a total of 564 people. The US sample consisted of 284 participants (146 women) with the mean age of 45.56 (SD = 16.64), and the Swedish sample of 280 participants (143 women, 1 non-response) with the mean age of 44.88 (SD = 16.40). Ethical guidelines from the American Psychological Association and Stockholm University were followed.
The study was conducted online. First, the participants were asked background questions (age, gender and education level), which were used for the quotas, and then questions about their attachment orientation. Secondly, participants were randomly assigned to one of three priming conditions where they read a short paragraph about either public welfare, God, or the planet Pluto (control condition). These conditions were used to test for possible priming effects on social and political trust. Because no consistent effects of priming were found participants from the three conditions were analyzed jointly in the analyses that follow (for pre-registrations and analyses see Supplementary Materials). Thirdly, participants marked a set of questionnaires regarding their social and political trust and their own experiences of welfare services, their level of religiosity, and left-right political opinions. Next, we describe the questionnaires included in parts one and three.
Attachment orientation was measured with a short version of the Experiences in Close Relationships scale (ECR, Lafontaine et al., 2016; for the original version see Brennan et al., 1998). This is a 12-item questionnaire measuring attachment orientations on two dimensions: attachment-related avoidance (e.g., “I don't feel comfortable opening up to romantic partners”, 6 items), and attachment anxiety (e.g., “I worry about being alone”, 6 items). This short version of ECR has shown comparable psychometric properties as the original scale (e.g., Fitzpatrick & Lafontaine, 2017). Reliability was also good in the present study for both attachment-related avoidance (α = .84 in Sweden, and .81 in the US) and anxiety (α = .85 in both Sweden and the US). For descriptive details on the scales used, see Table 1.
|Measurement||Items||Scale||US (n = 284)
||Sweden (n = 280)
|Welfare state trust||8||1-7||3.51||1.34||.89||3.82||1.37||.89|
|Aggregated social trusta||14||–||-0.04||0.76||.93||0.04||0.68||.92|
|Left-right political orientationb||6||1-7||4.46||1.09||.61||4.11||1.11||.63|
aAggregation of the standardized items of Generalized Social Trust, Faith, Trust Scale and WVS items of Social Trust. bLower values equals left.
Religiosity and Political Opinions
Religiosity was measured with six items about religious beliefs and behaviors (e.g., “Religion is important in my daily life”, “I regularly pray to God”, [Granqvist, 1998], scale 1 [strongly disagree] to 7 [strongly agree]. Political orientation was measured with six items taken from the World Value Survey (WVS) asking about economic attitudes in favor of equality and redistribution or economic liberty and competitive markets (e.g., 1 [left] = ”Incomes should be made more equal”; 7 [right] = “We need larger income differences as incentives for individual effort”.
General social trust was measured with four sets of questions. First, a version of the standard general social trust question, “People in general are to be/are not to be trusted” (scale 0 to 10, Lundmark, Gilljam, & Dahlberg, 2016). Second, a single question regarding faith in other people (“Do you think most people would try to take advantage of you if they got a chance, or would they try to be fair?”, scale 0 to 10). Third, the general trust scale, comprising six items (e.g. “Most people are basically honest”, scale 1 [Do not agree at all] to 7 [Completely agree], αSweden = .90, αUS = .89 [Yamagishi & Yamagishi, 1994]). Finally, six questions gauging trust in particular groups taken from the WVS (e.g., “Your close family”, “People of another nationality”; scale 1 [Do not trust at all] to 7 [Completely trust]). These different measurements of social trust were highly interrelated and yielded virtually the same results in subsequent analyses. For data reduction purposes, all items (standardized) were therefore aggregated to one scale for overall social trust.
Trust in the Welfare State
To tap into welfare state trust we created a new scale asking about trust in six different domains of the welfare state: elderly care, social benefits, healthcare, education, unemployment benefits, and paid sick leave. The domains included were based on a pilot study investigating what words people associated with the welfare state in both Sweden and the US [Gruneau Brulin, Hill, Laurin, Mikulincer, & Granqvist, 2018]. Four to five questions were created for each domain (e.g., “I trust publicly provided elder care in…”), as well as four questions about trust in the welfare state in general (e.g., “I trust the social safety net”). The participants were asked to rate each statement on a seven-point scale ranging from strongly disagree to strongly agree. After factor analysis (see below), the scale could be reduced to eight items. This eight-item version was used in subsequent analyses. For a presentation of the final scale, see Appendix.
To achieve a preliminary validation check of the welfare state trust scale and to test if our new scale differentiated from other kinds of political trust, we included three sets of questions from the Swedish Welfare State Survey (Svallfors, 2011): (1) Performance of the welfare state (e.g., “To what extent do you think the public sector in Sweden/the United States manages to . . .” e.g., “Provide the sick with the health care they need?”); (2) sustenance problems (“How likely do you think it is that you…” e.g., “Will not get the health care you need in case you get ill?”); and (3) abuse of welfare policies (e.g. “How common do you think it is that social benefits and services are used by people who don’t really need them?”). In addition, we used five single items taken from the WVS regarding trust in political institutions (the Government, the Police, the Legal System, Politicians, Political Parties).
Finally, to control for participants’ experiences of welfare state services due to adverse life events we included four items from the Swedish Welfare State Survey (Svallfors, 2011) asking whether the participants had been hospitalized, unemployed, on sick leave, or received social benefits during the previous three years. A majority of participants (Sweden, 62%, the US 72%) answered yes to at least one of these questions. Subsequent analyses used an aggregated variable for these items (ranging from 0 to 4).
The main purpose of this study was to test if attachment orientation (i.e., attachment-related avoidance and anxiety) was related to welfare state trust and to social trust. To enable these tests, we created a new scale for trust in the welfare state. We start by presenting the construction (and preliminary validation) of the welfare state trust scale, and then proceed with the main results on attachment orientation in relation to social and welfare state trust.
Construction of the Welfare State Trust Scale
Structure of the Scale
To explore the structure of the welfare state trust scale we performed an exploratory factor analysis in each subsample (Sweden and the US). Because we had no preconception about whether or how the underlying factors might relate to each other we used orthogonal (varimax) rotation. No clear differences in solutions between the Swedish and US samples were found (for analyses in the two samples separately, including factor loadings and scree plots, see Supplementary Materials). The two samples were thus analyzed together. The analysis revealed that the scale was best represented by one factor, and that the number of items could be reduced to eight without losing psychometric quality. Specifically, the first factor explained more than 50% of the total variance in both samples (eigenvalues [EVs] > 10), compared to less than six percent for the remaining factors. In other words, participants who showed trust in one domain of the welfare state tended to trust other domains too, in both samples. Further, an inter-item reliability check revealed high correlations among the different items within the same domains (rs > .60), indicating that additional questions are not needed. We therefore removed all but one item in each domain, together with two items regarding the welfare state in general. Thus, the final scale consisted of eight items in total. A test of internal consistency revealed strong reliability in both samples (αs = .89 in both Sweden and the US). Finally, all subsequent analyses were performed with both the full scale (26 items) and with the final eight-item scale. Results were virtually identical.
The preliminary validation check of the welfare state trust scale revealed that welfare state trust scale correlated strongly with both WVS political trust (.58 < r < .64), and perceived performance of the welfare state (.69 < r < .80). However, an oblique factor analysis based on all single items from the WVS political trust, perceived performance, and welfare state trust scales revealed that the items clustered in accordance with the original scales, suggesting conceptual non-equivalence between the scales (for full presentation, see Supplementary Materials). Finally, due to the more expansive welfare state of Sweden, higher ratings of trust in the welfare state would be expected in the Swedish compared to the US sample. This was also confirmed (for descriptive statistics, see Table 1; t(280, 284) = 2.70, p = .007, Cohen’s d = .23). In conclusion, these findings lend preliminary support for the construct validity of the Welfare State Trust Scale.
Attachment and Trust
To examine the relation between attachment orientations on the one hand and welfare state and social trust on the other, we performed two OLS regressions using attachment-related avoidance and anxiety as predictors and the collapsed social trust scale and welfare state trust scale as outcomes. The analyses were first conducted separately in the two samples (Sweden and the US). However, the results were virtually identical so we treated the samples as one, but still controlled for nation (dummy-coded). To test for possible confounding influences, we also included age, gender, socioeconomic background (educational level), religiosity, political orientation, and experiences of welfare services as predictors. The regressions were performed in blocks. The first block included the potential confound variables (Block 1), and the second block added attachment-related avoidance and anxiety (Block 2). Additionally, a third block was added to test if controlling for welfare state trust and social trust, respectively, had an impact on the presumed effects of attachment orientation (Block 3). For bivariate associations of the variables included in the regressions, see Table 2.
|1. Welfare state trust||–|
|2. Social trust||.43***||–|
|7. Left-right political orient.a||.09*||.09*||.20***||.07||.02||.14***||–|
|8. Welfare experiences||-.11*||-.18***||-.07||-.01||-.20***||.04||-.11*||–|
|9. Sweden sample||.11**||.06||-.02||.00||.06||-.46***||-.16***||-.10*||–|
|10. Attachment avoidance||-.18***||-.30***||.03||-.07||-.02||-.02||-.03||.02||-.09*||–|
|11. Attachment anxiety||-.00||-.18***||-.34***||-.10*||-.09*||.02||-.20***||.26***||-.07||.09*|
aLower values equal left.
*p < .05. **p < .01. ***p < .001.
Attachment-related avoidance was negatively related to social trust (ß = -0.28, ΔR2 = .08, p < .001). However, despite a negative bivariate correlation between attachment anxiety and social trust (r = -.18, p < .001), no unique relation appeared in the regression model when the potential confounds were controlled (ß = -0.05, p = .28). Also, older age, female sex, higher religiosity, higher education, and less experiences of welfare state services were related to higher degrees of social trust (see Table 3 for details).
|B||SE||CI (95%)||ß||B||SE||CI (95%)||ß||B||SE||CI (95%)||ß|
|Age||0.01***||0.00||[0.00, 0.01]||0.15||0.01***||0.00||[0.00, 0.01]||0.15||0.01***||0.00||[0.00, 0.01]||0.19|
|Female||0.10||0.06||[-0.02, 0.21]||0.07||0.06||0.06||[-0.05, 0.17]||0.04||-0.03||0.05||[-0.13, 0.08]||-0.02|
|Religiosity||0.05**||0.02||[0.02, 0.08]||0.14||0.04*||0.02||[0.01, 0.07]||0.11||0.02||0.01||[-0.01, 0.05]||0.05|
|Left-right pol. orient.a||0.02||0.03||[-0.03, 0.07]||0.03||0.01||0.03||[-0.04, 0.06]||0.01||-0.01||0.02||[-0.06, 0.03]||-0.02|
|Education||0.08***||0.02||[0.04, 0.12]||0.16||0.07***||0.02||[0.04, 0.11]||0.15||0.05**||0.02||[0.01, 0.08]||0.10|
|Welfare experiences||-0.08**||0.03||[-0.13, -0.03]||-0.13||-0.07**||0.03||[-0.12, -0.02]||-0.11||-0.06*||0.03||[-0.10, -0.01]||-0.09|
|Sweden sample||0.15*||0.07||[0.02, 0.28]||0.11||0.09||0.06||[-0.03, 0.22]||0.06||-0.00||0.06||[-0.12, 0.11]||-0.00|
|Avoidance||-0.15***||0.02||[-0.19, -0.11]||-0.28||-0.12***||0.02||[-0.16, -0.09]||-0.23|
|Anxiety||-0.02||0.02||[-0.06, 0.02]||-0.05||-0.03||0.02||[-0.07, 0.01]||-0.07|
|Welfare State Trust||0.20***||0.02||[0.16, 0.24]||0.38|
|Intercept||-0.83**||0.17||[-1.17, -0.49]||-0.12||0.21||[-1.17, -0.49]||-0.67||0.20||[-1.06, -0.27]|
Note. Values in square brackets indicate the 95% confidence interval for each coefficient.
aLower values equal left.
*p < .05. ** p < .01. ***p < .001.
We also included an interaction effect of sample (Sweden vs. the US) and attachment orientations but no such effects were found (ß = 0.00, p = .94, for both avoidance and anxiety). Additionally, the third block, in which welfare state trust was entered had negligible impact on the relation between attachment-related avoidance and social trust (ß = -.23, p < .001, compared to ß = -.28, p < .001). We thus conclude that trust in the welfare state cannot statistically explain the relation between attachment-related avoidance and social trust.
Welfare State Trust
The regression analysis revealed that attachment-related avoidance was also negatively related to welfare state trust (ß = -0.14, p < .001, ΔR2 = .02), but attachment anxiety was not (ß = 0.05, p = .26). Hence attachment-related avoidance, but not anxiety, is related to both trust in other people and in welfare state institutions. Additionally, being female, higher levels of religiosity, right-wing political opinions, higher education, lower age, and being in the Swedish sample all predicted higher ratings of welfare state trust (see Table 4 for details). Lastly, we tested for possible interaction effects between sample and attachment orientations on welfare state trust, but no effects were found (avoidance: ß = -0.03, p = .48, anxiety: ß = 0.00, p = .95). Thus, as with social trust, the relation between attachment orientation and welfare state trust was not dependent on cultural context.
|B||SE||CI (95%)||β||B||SE||CI (95%)||β||B||SE||CI (95%)||β|
|Age||-0.01**||0.00||[-0.02, -0.00]||-0.12||-0.01*||0.00||[-0.02, -0.00]||-0.10||-0.01***||0.00||[-0.02, -0.01]||-0.16|
|Female||0.43***||0.11||[0.21, 0.65]||0.16||0.41***||0.11||[0.19, 0.63]||0.15||0.37***||0.10||[0.17, 0.57]||0.13|
|Religiosity||0.11***||0.03||[0.05, 0.17]||0.17||0.10***||0.03||[0.05, 0.16]||0.16||0.07**||0.08||[0.02, 0.13]||0.11|
|Left-right pol. orient.a||0.12*||0.05||[0.02, 0.22]||0.10||0.12*||0.05||[0.02, 0.22]||0.10||0.11*||0.05||[0.02, 0.20]||0.09|
|Education||0.14***||0.04||[0.07, 0.22]||0.15||0.14***||0.04||[0.06, 0.21]||0.15||0.08*||0.04||[0.01, 0.15]||0.09|
|Welfare experiences||-0.07||0.05||[-0.17, 0.03]||-0.06||-0.08||0.05||[-0.18, 0.02]||-0.07||-0.02||0.05||[-0.12, 0.07]||-0.02|
|Sweden sample||0.51***||0.13||[0.27, 0.76]||0.19||0.47***||0.16||[0.22, 0.72]||0.17||0.40***||0.12||[0.17, 0.62]||0.15|
|Avoidance||-0.15***||0.04||[-0.23, -0.07]||-0.14||-0.03||0.04||[-0.10, 0.05]||-0.03|
|Anxiety||0.04||0.04||[-0.03, 0.12]||0.05||0.06||0.04||[-0.01, 0.13]||0.07|
|Social trust||0.78***||0.08||[0.63, 0.94]||0.41|
|Intercept||2.39***||0.33||[1.75, 3.03]||2.70***||0.41||[1.89, 3.51]||2.80***||0.38||[2.05, 3.54]|
Note. Values in square brackets indicate the 95% confidence interval for each coefficient.
aLower values equal left.
*p < .05. **p < .01. ***p < .001.
When social trust was added to the regression (Block 3) the effect of attachment-related avoidance on welfare state trust disappeared, indicating that social trust might statistically mediate the relation between attachment-related avoidance and welfare state trust. To explore the indirect and direct effect further, we performed a path analysis with 10000 bootstrapping (Hayes, 2013, using ‘boot’ package in RStudio, version 1.3-28). As can be seen in Figure 1, there was an indirect effect of attachment avoidance on welfare state trust through social trust (ß = -0.12, 95% CI [-0.16, -0.07]). For consistency, we also explored an alternative model by testing whether attachment-related avoidance had an indirect effect on social trust through welfare state trust (but recall that controlling for welfare state trust only marginally affected the regression coefficient for avoidance on social trust). Although significant, the indirect effect of avoidance on social trust (via trust in the welfare state) was indeed smaller (ß = -0.05, 95% CI [-0.09, -0.02]), compared to the indirect effect of avoidance on welfare state trust via social trust (again, ß = -0.12). Hence, we suggest that the model using social trust as a statistical mediator between attachment avoidance and welfare state trust better represented the data in the present study. Theoretically, it is also more plausible that attachment-related avoidance has an impact on social trust, because social trust regards other human beings, and that this form of trust generalizes to the welfare state, rather than the opposite pattern. We return to this matter in the discussion below.
In summary, attachment-related avoidance, but not attachment anxiety, was uniquely inversely related to both social trust and to welfare state trust. Also, the relation between attachment avoidance and welfare state trust may be mediated by social trust.
The aim of the present study was to examine whether attachment orientations are related to trust in welfare state institutions as well as social trust. To be able to measure trust in the welfare state we developed a new questionnaire focused on welfare state trust specifically. Across two different nations – Sweden, a country with an expansive universal welfare state, and the United States, a country with a small welfare state targeted at marginalized groups – attachment-related avoidance was uniquely negatively related to both social trust and trust in welfare state institutions. Also, the relation between attachment-related avoidance and welfare state trust was statistically mediated by social trust. These findings suggest that attachment-related mental models have implications beyond human relationships; they may also extend to how people relate to, in this case trust, societal institutions. Thus, political trust, captured in this study as trust in the welfare state, may to some extent be shaped by one’s experiences in close relationships.
These results are in line with Bowlby’s (1973) contention that attachment-related working models are not restricted to self and others in close relationships but also have implications for our models of the world at large. We emphasize, therefore, that the link we observed between attachment-related avoidance (negative model of others) and low welfare state trust (cf., model of the world) was not theoretically indiscriminate but was indirect and occurred via low trust in other humans (i.e., social trust). In a similar vein, previous studies have demonstrated that attachment orientation predicts anxiety when separated from inanimate objects such as smartphones (Keefer et al., 2012), and religious believers’ representations of God (Granqvist, 2020). However, this is, as far as we know, the first study to indicate that attachment orientation (presumably through related mental models) also relates to how societal institutions are represented by people. After this study was concluded, our results regarding the relation between attachment-related avoidance and social and political trust, as well as the mediation findings, have been replicated in a new sample (Gruneau Brulin, Shaver, Mikulincer, & Granqvist, 2022).
Unlike avoidance, attachment-related anxiety (i.e., positive model of others, negative model of self) was unrelated to welfare state trust. We speculate that this may be because individuals high in attachment anxiety do not possess negative models of others (Mikulincer & Shaver, 2016). Indeed, some research indicates that such individuals may in fact be especially prone to treat others – whether interpersonal ones, smart-phones, or gods – as external rescuers to aid their “helpless” selves (Granqvist, 2020). In other words, it is not distrust in others but rather in themselves and in their own agentic abilities that tends to be a characteristic problem for these individuals. To post-hoc explore whether attachment anxiety was related to some facet of political mistrust that might instead express a negative model of self, we ran regressions between attachment orientations on the one hand and the political trust items that were used to preliminarily validate our welfare trust scale on the other. Interestingly attachment anxiety, but not avoidance, was related to higher levels of perceived sustenance problems in relation to the welfare state, for example that one “won’t manage a month’s sick leave without serious economic consequences?” (ß for anxiety = 0.10, p = .02, see Supplementary Materials for full model). Hence, attachment anxiety appears not only to be related to experiences of being “helpless” in relation to other humans, but also in relation to the welfare state. This finding further strengthens the notion that attachment mental models have implications beyond interpersonal relationships, and in relation to the welfare state. However, because these analyses were performed post-hoc, the results should be interpreted with caution and need to be corroborated by future research.
Further, attachment avoidance was related to social trust, and more strongly so than to welfare state trust (8% versus 2% explained variance). A path analysis also showed that the relationship between attachment-related avoidance and welfare state trust was indirect through social trust. Previous studies have found mixed results regarding the direction of causality between social and political trust. On the one hand, experiences of political institutions, thus political trust, predict future social trust (Sønderskov & Dinesen, 2016). On the other hand, social trust begins to shape early in life (Uslaner, 2002), is transmitted between parents and children (Dohmen et al., 2012) and remains fairly stable throughout life (Dawson, 2019). As attachment working models also begin to form very early in life and are related to trust, we suggest that attachment theory may contribute to understanding people’s development of trust, both to close relationship partners, to people in general, and – by extension – to societal institutions. Furthermore, there may be two possible developmental routes for this. First, and as we have highlighted in the present article, relational experiences with attachment figures are internalized and may affect trust both towards other people and societal institutions, such as welfare state institutions. Second, attachment relationships function as contexts for social learning and cultural transmission (Granqvist, 2021), which could thus help to understand how trust is transmitted from parents to children (Dohmen et al., 2012). Because social and political trust are important for societal well-being, our results also suggest the importance of creating conditions for the development of secure (or non-avoidant) attachment, for the benefit of society. Our results suggest that creating conditions for secure attachment could also mean that citizens develop higher trust in the welfare state, and may also give more support for it. Note however that future research with prospective longitudinal designs is needed to sort out causality. Additionally, even though this perspective emphasizes the development of trust early in life, by no means does it preclude the possibility that later experiences with, for instance, political institutions could alter people’s trust (cf. Sønderskov & Dinesen, 2016). We hasten to add, also, that insecure, including avoidant, attachment may fill other adaptive functions (e.g., Ein-Dor & Hirschberger, 2016).
For the purpose of this study we also developed a new scale to specifically measure trust in the welfare state. Although the included items were directed at different domains of the welfare state, our analyses showed that they formed one homogenous factor. Thus, if trust is high in one domain of the welfare state, for example elderly care, trust also tends to be high in other domains, such as the healthcare system. Relatedly, in a preliminary validation of the scale, welfare state trust was strongly correlated with trust in other political institutions as well, further indicating that political trust overall is fairly consistent within a person. Welfare state trust was also strongly correlated with perceived performance of the welfare state, suggesting that trust in the welfare state is dependent to a large extent on how well one perceives that the welfare state carries out its functions. Despite high correlations among welfare state trust, political trust, and perceived performance of the welfare state, these three scales formed three different factors, indicating that, despite substantial interrelatedness they do capture different aspects. A limitation of this study, however, is that the preliminary validation of the scale was conducted in the same sample used for scale construction and for examining associations with attachment. Hence, for more definitive validation purposes, the welfare trust scale should be evaluated in an independent sample, and preferably in relation to behavioral outcomes related to welfare state services.
Finally, welfare state trust was higher in the Swedish sample compared to the US sample. This is probably a reflection of the more expansive welfare state in Sweden. It is also in line with previous findings (Rothstein & Uslaner, 2005) indicating that other types of political trust are higher in countries with larger welfare states.
Additional Strengths, Limitations, and Future Directions
The present study breaks new ground in highlighting that attachment orientations may influence how people relate to, in this case trust, societal institutions, and that attachment orientation may influence trust outside close relationship and also influence trust in people in general. There is considerable conceptual distance between the concepts studied (Fiedler, 2014), especially between attachment and welfare state trust. Despite this, these concepts were empirically related in a theoretically meaningful way. In addition, the study was conducted in two large samples, from two different nations representing markedly different ways of organizing the welfare state. Our key results remained similar after statistically controlling for possible confounds. There are some important limitations to keep in mind, however.
First, because the study was based on cross-sectional data, interpretations of causality and process-direction are ambiguous. Hence, although we have examined possible mediational links and expressed theoretically informed interpretations about process direction, these links and interpretations require empirical substantiation in future prospective longitudinal research. Second, the results are based on self-reports, which are sensitive to social desirability and limited self-awareness, and have been questioned for how they translate to actual behavior (e.g., Glaeser, Laibson, Scheinkman, & Soutter, 2000; but see Sapienza, Toldra-Simats, & Zingales, 2013). Hence, future research is needed to show that attachment orientations also predict real-life trust behavior in relation to the welfare state. Third, the overall explained variance in the regression models (11-26%), as well as the unique explanatory power of attachment orientations on trust (2-8%), were generally modest. Possibly other psychological factors, such as personality (Müller & Schwieren, 2020) or self-regulation (Keller, Mayo, Greifeneder, & Pfattheicher, 2015) could contribute to understanding what may affect trust in the welfare state. Finally, to corroborate and extend our findings, we encourage future research to include attachment measurements when studying social and political trust. Future research should also employ longitudinal or experimental designs to disambiguate causality/process direction. Finally, we call for the use of behavioral measurements to verify a predictive relation between attachment orientation and political/welfare trust.
The present study is, as far as we know, the first to show a negative relation between attachment-related avoidance on the one hand and generalized social and political trust, in this case welfare trust, on the other. Despite the importance of social and political trust for individual and societal wellbeing, psychological research has given surprisingly little attention to trust outside close interpersonal relationships. Instead, trust has primarily been studied within political science and economy, which has in turn led to a neglect of pursuing psychological explanations and processes underlying trust in favor of cultural, political, and societal processes. The present paper thus contributes to understanding both how attachment orientations and related mental models may be expressed in relation to non-human entities and the psychological underpinnings of social and political trust.