The Structure of Individualized and Generalized Political Trust: A Network Analysis of Data from Two Representative-Sample Studies

Abstract

This study examined the relationship between generalized (trust in the political system) and individualized (trust in a specific political actor) political trust in two surveys with representative samples from Latvia (total N = 2,001). The results of a regularized partial correlation network analysis of the data from both surveys revealed that variables associated with each level of trust formed separate clusters, which were virtually unrelated in Study 1, and moderately related in Study 2. Across both datasets, political cynicism was the only variable robustly (negatively) related to both levels of trust. External political efficacy was strongly related to generalized political trust, whereas internal political efficacy was unrelated to either level of political trust. Generalized interpersonal trust was only weakly related to generalized political trust and was unrelated to individualized political trust. The findings show that trust in the political system and trust in specific political actors form separate, although related, network structures and are likely determined by a different set of variables, representing different kinds of psychological reasoning. Generalized political trust can be seen as part of a general attitude towards one’s country, whereas individualized trust is based on more situational factors and shaped through the mechanisms of person perception and group perception.

Plain language summary

The purpose of this study was to examine how people form their judgments of trust (or distrust) in politicians, political parties, and in political system as a whole, and how trust in politicians and parties (or individualized political trust) is different from the overall trust in the system (generalized political trust). The data for this study were collected in two surveys from representative population samples from Latvia, a country characterized by low levels of political trust. The survey participants answered several evaluative questions about trusted and distrusted politicians and parties, and about the political system of Latvia. Network analysis, a method allowing to visualize the relationships among the participants’ responses, was used in the data analysis. The results showed that trust in the political system and trust in specific political actors should be considered as separate, although related, psychological concepts, and are likely determined by a different, although overlapping, set of factors representing different kinds of psychological reasoning. Generalized political trust can be seen as part of a general attitude towards one’s country, whereas individualized trust is based on more on situational factors and shaped through the mechanisms of person perception and group perception.

Individualized and Generalized Political Trust

Political trust has been defined as “a summary judgment that the [political] system is responsive, and will do what is right even in the absence of constant scrutiny” (Miller & Listhaug, 1990, p. 358), or “granting other political actors discretionary powers over the use of collective goods while recognizing that this judgment comes at some risk to oneself” (Fisher et al., 2010, p. 162). One can distinguish between two levels of political trust: trust in the system with its institutions and procedures, which can be defined as macro-level, generalized, or diffuse political trust, and trust in the specific people who are part of the system at the given point in time, which can be defined as micro-level, individual, or specific political trust (Norris, 2017). The more general, diffuse level of political trust constitutes a relatively stable, durable, and abstract feeling toward the political system and its operating principles as a whole; the specific political trust represents the attitudes towards the performance of specific institutions and office-holders (Easton, 1975). The objective of the two studies reported in this paper is to examine the relationship between both levels of trust and their facets/ components. In addition, we examine the association of both trust levels with several political trust-related constructs reviewed below.

The previous literature on political trust has taken two basic approaches to address the relationships among both trust levels. One approach regards political trust as a general evaluation of the political system as a whole, and the evaluation of specific political actors and institutions is dependent on this general attitude toward the system (Catterberg & Moreno, 2006; Christensen & Lægreid, 2005; Hetherington & Husser, 2012; Hooghe, 2011). The other approach sees the trust judgment of a political actor as determined by the perception of specific qualities of the political actor or institution, together with situational circumstances and contextual factors affecting the trustworthiness of the actor, leading to differentiated forms and implementations of trust judgments on the two different levels (Fisher et al., 2010; Proszowska et al., 2021; Ruelens & Nicaise, 2020). In fact, the two levels of political trust mentioned here do not represent distinct categories; rather, they can be seen as end-poles of a continuum of political trust-related judgments and attitudes, ranging from national identity and general support of principles and values of a political system at the most general, or diffuse level, through evaluations of regime performance and trust in political institutions at the intermediate level, to evaluations of particular political actors at the most individualized, specific level (Norris, 2017). One can expect that the performance of specific incumbents will affect people’s confidence in institutions, which, in turn, may affect the overall evaluation of the political system, and vice versa—the overall evaluation of the political system should color the evaluation of its components. In addition, the perceived qualities, specific actions, and past overall performance of specific political actors can be associated with actual or potential gains and losses for the citizens, leading to much more differentiated trust judgments than would be the case for generalized political trust (Hetherington & Globetti, 2002; Sears & Funk, 1990). These findings suggest that individual political trust may have a separate set of correlates, with implications for the theoretical conceptualization of trust and the practical understanding of factors that increase or decrease trust in specific politicians. Based on the previous findings, we predicted that generalized and individualized trust will be positively correlated but will not form a single construct.

Elements of Political Trust

Previous research has suggested that three central characteristics of the trustee play a crucial role in establishing and maintaining trust (Burke et al., 2007; Mayer et al., 1995): the perceived ability or competence of the trustee (the collection of skills, competencies, knowledge and other characteristics that enable the trustee to have influence within a specific domain), perceived benevolence (the extent to which a trustee is seen as willing to do good to the trustor, free from any egocentric motives), and the perceived integrity (adherence to several moral principles that the trustor finds acceptable or preferable). The same model has been successfully applied to examine the contents and determinants of political trust (Gaina et al., 2020; Grimmelikhuijsen & Knies, 2015; Hamm et al., 2019). In the context of political trust, benevolence represents the extent to which a citizen perceives a political actor as caring about the welfare of the public and to be motivated to act in the interests of the society; ability can be seen as the extent to which a political actor is perceived as capable, professional, and effective; integrity reflects the extent to which a citizen perceives the political actor to be sincere, tell the truth, and fulfil promises (Grimmelikhuijsen & Knies, 2015). As applied in conceptualizing political trust, the three factors have been shown to form separate but highly intercorrelated variables that predict the general attitudes towards government willingness to follow government recommendations and endorse government activities (Hamm et al., 2019). In addition, political actors strive for and often exert a certain degree of power and control over citizens, and there is always a possibility that this power will not be used appropriately, resulting in a necessity for risk assessment for the citizens when they judge the trustworthiness of politicians and political parties (Grimmelikhuijsen & Knies, 2015; Slovic, 1993). These four components of trust—ability, benevolence, integrity, and risk—constitute the basis of the theoretical model used in this study. In addition, elements for generalized political trust may include the overall perceived effectiveness of the political institutions (Christensen & Lægreid, 2005; Rothstein & Stolle, 2008) and the perceived fairness and transparency of the functioning of the political system (Levi, 1998; Mishler & Rose, 2001).

Constructs Related to Political Trust

Political Efficacy

One construct that is closely related to political trust is political efficacy, simply defined as “citizens’ perceptions of powerfulness (or powerlessness) in the political realm” (Morrell, 2003, p. 589). It is a feeling that the individual can influence the political process in a given society and that political engagement and activity can bring about meaningful results and political, economic, or social change in society. Political efficacy has been shown to contain two separate components—internal efficacy, referring to beliefs about the person’s competence to understand the political process and to participate effectively in politics, and external efficacy, referring to beliefs about the responsiveness of authorities and institutions to citizen influence (Acock & Clarke, 1990; Craig et al., 1990). External political efficacy is thus conceptually closely linked to generalized political trust and can be seen as one of its components; it should be strongly positively related to generalized political trust because it represents the evaluation of one element of the political system (its responsiveness) and thus is likely to be a significant contributor to the evaluation of the system as a whole. Internal political efficacy, on the other hand, is a more stand-alone concept that is theoretically more independent of political trust and can be seen as an individual difference variable. We included measures of both components of efficacy in our study to test whether internal and external efficacy are differently related to both levels of political trust.

Political Cynicism

Another construct that is directly negatively related to political trust is political cynicism, “the extent to which people hold politicians and politics in disrepute, the extent to which these words symbolize something negative rather than something positive” (Agger et al., 1961, p. 477), or “an orientation toward politicians that attributes their actions to self-serving motives” (Citrin & Stoker, 2018, p. 50). Although political cynicism can be largely seen as the opposite concept to political trust (c.f. Erber & Lau, 1990; Litt, 1963), the former is conceptually and empirically distinguishable from the latter (Dancey, 2012; Pattyn et al., 2012). Political cynicism represents a negative perception of politicians as a social category; politicians are seen as a negatively stereotyped and derogated outgroup. Cynicism permeates the different levels of political trust but cannot be equaled to political trust, because there are multiple other factors contributing to the trust judgment on both general and individualized levels of trust. In the current studies, political cynicism was included as a separate measure, in addition to measures of political trust.

Perception of the Socio-Economic Situation

A common theoretical perspective relates political trust to the satisfaction with the performance of political institutions (Catterberg & Moreno, 2006; Christensen & Lægreid, 2005; Mishler & Rose, 2001). When faced with the need to evaluate the performance of political authorities and institutions, people often use their subjective well-being as an evaluative criterion. Citizens form a judgment whether they feel happy enough to grant their support to the authorities because they hold the authorities accountable for honoring their part of a psychological “happiness contract”—granting conditions in a society that allow people to reach a satisfactory level of subjective well-being (Esaiasson et al., 2019). In addition, direct evaluations of the economic situation within the country and in one’s household are also significant predictors of political trust (Mishler & Rose, 2001). An important factor in reducing satisfaction with institutional performance is the perceived corruption in society (Uslaner, 2013). Corruption also lowers life satisfaction by decreasing political trust as a mediator (Ciziceno & Travaglino, 2018).

Generalized Trust

Generalized trust can be defined as a cognitive tendency to trust others, including people about whose trustworthiness there is not enough information. Theoretically, generalized trust should be positively related to political trust, but findings regarding the relationship between generalized trust and political trust so far have been inconclusive. Some studies have found a strong positive relationship between both types of trust (Schiffman et al., 2010; Schyns & Koop, 2010; Zmerli & Newton, 2008), while others have demonstrated only weak links or have found both types of trust to be unrelated (Kaase, 1999; Mishler & Rose, 2001).

The Current Research

The two survey studies reported below examine the relationship among the two levels of political trust, their structure, and their association with the reviewed trust-related constructs with representative population samples from Latvia. Latvia, a Northern-European democracy that regained independence in 1991 after 51 years of Soviet occupation, is characterized by one of the lowest political trust levels in Europe, and this low-trust dynamic has been relatively stable over the last couple of decades (ESS Round 9: European Social Survey Round 9 Data, 2018; Eurobarometer, 2023). In this way, Latvia fits a general pattern observed among the new democracies of post-communist societies in Eastern and Central Europe, which display lower levels of political trust in comparison with more established democracies (Rose & Mishler, 2011; Zavecz, 2017), which makes Latvia an interesting and relevant case for in-depth study of political trust and its mechanisms.

The primary focus of the current study is the relationship between both trust levels, that is, how well both types of trust can be differentiated as separate constructs with corresponding facets/ components, and how strongly both types of trust are related in people’s judgments about specific political actors and the political system as a whole. We also include several additional variables—political efficacy, political cynicism, generalized interpersonal trust, as well as life satisfaction, perception of the economic situation, and perceived corruption—to better understand and illustrate the differences between both types of trust judgments and to test whether both types of trust differentially relate to these additional variables. We examine individualized trust towards two kinds of political actors—individual politicians (Study 1) and political parties (Study 1 and Study 2).

Method

Participants

Study 1

A nationally representative sample of Latvian adults was surveyed via face-to-face computer-assisted interviews in December 2019 (N = 1,001; 51.8% women and 48.2% men; with the following age distribution: 18–24 years—9.9%, 25–34 years—20.2%, 35–44 years—19.5%, 45–54 years—18.8%, 55–64 years—16.3%, and 65–74 years—15.4%). Most respondents were ethnic Latvians (59.0%) and 41.0% were representatives of other ethnicities; 88.1% were citizens of Latvia and 11.9% were non-citizens. The respondents were approximately equally distributed among the capital city Riga (32.5%), other cities (37%), and rural areas (30.5%). Additional information on the regional distribution of respondents is available in the Supplementary Material. The respondents were surveyed at their residences by professional interviewers from a market research company. The participants were selected using a stratified random sampling method (random route procedure). The response rate (RR1) for the survey was 44.9% (contact rate = 68.3%, cooperation rate = 65.8%). The participants completed the interview in Latvian or Russian, based on their language preference.

Study 2

A nationally representative sample of Latvian adults was surveyed via face-to-face computer-assisted interviews in August 2021 (N = 1,000; 55.2% women and 44.8% men; with the following age distribution: 18–24 years—9.7%, 25–34 years—17.8%, 35–44 years—18.7%, 45–54 years—19.6%, 55–64 years—19.2%, and 65–74 years—15.0%). Most respondents were ethnic Latvians (66.6%) and 33.4% were representatives of other ethnicities; 94.3% were citizens of Latvia and 5.7% were non-citizens. Most respondents were from urban areas (Riga 34%, other cities 37%, rural areas 29%). The respondents were surveyed at their residences by professional interviewers from a market research company. The participants were selected using a stratified random sampling method (random route procedure). The response rate (RR1) for the survey was 34.4% (contact rate = 71.4%, cooperation rate = 48.4%). The participants completed the interview in Latvian or Russian.

Measures and Procedure

Study 1

The participants completed two survey versions, administered in a between-groups design. In the trust version of the survey, the participants (N = 503) were first asked to name a political party in Latvia that they trusted most. The participants could select the party from a list of 10 major parties, or name another option. Then, the participants were asked to name a politician whom they trusted most. In the distrust version of the survey, the respondents (N = 498) in an identical format were asked to name a party and a politician whom they distrusted most. Asking the participants to come up with their own responses and using two versions of the survey allowed us to maximize the variation of the individualized trust ratings, at the same time ensuring that the respondents were rating a party or politician they were familiar with and about whom they were more likely to be able to form a trust judgment.

Following these introductory questions, the respondents were reminded of the party they had mentioned and asked to answer several questions about it. The overall trust was measured by a single item To what extent do you generally trust or distrust the politicians of this party? with the response scale ranging from 1 (do not trust at all) to 10 (trust completely). The participants then rated the perceived benevolence, ability, and integrity of the party, as well as the risk associated with the party. The perceived benevolence was measured with two items (Spearman–Brown coefficient = .95): The politicians of this party really care about the well-being of the people of Latvia and Politicians of this party are ready to work selflessly for the benefit of Latvian voters. The perceived ability was measured with two items (Spearman–Brown coefficient = .95): The politicians of this party skillfully cope with their job responsibilities and The politicians of this party are competent (they have the necessary skills and knowledge for the job). The perceived integrity was measured with two items (Spearman–Brown coefficient = .92): The politicians of this party are honest and The politicians of this party are consistent in their words and deeds. All items were rated on a Likert-type scale ranging from 1 (do not agree at all) to 7 (completely agree). The perceived risk associated with the party was measured with 2 items (Spearman–Brown coefficient = .77): The activities of this party pose a risk, a feeling of instability, for you and your family, with a response scale from 1 (do not agree at all) to 7 (completely agree), and In general, do you think that the activities of this party endanger Latvian society or, on the contrary, improve its security? with a response scale from 1 (endangers to a very large extent) to 7 (improve its security to a very large extent), with the scale midpoint 4 labelled “neither endangers nor improves its security” The latter item was reverse-scored later when calculating the perceived risk index.

The expected utility associated with the party was measured with a single item How would you rate the expected material benefits /losses for yourself and your family (e.g., improvement/deterioration of your financial situation, favorable/unfavorable changes in legislation, etc.) from the activities of this party while it is in power? with the response scale ranging from 1 (would expect significant losses) to 7 (would expect significant gains), with the scale midpoint 4 labelled “situation would not change either for better or worse.”

Following the questions about the party, the respondents were reminded of the politician they had mentioned at the beginning of the survey and asked to answer the same questions (that they had just answered about the party) about this politician. The questions were reformulated to be answered about the single individual politician, but otherwise, they were all identical to the questions asked about the party. The Spearman–Brown coefficients for the measures of politician’s benevolence, competence, integrity, and associated risk were .96, .95, .94, and .81, respectively.

The remaining questions in the survey were identical for all respondents. Political cynicism was measured by three items (Cronbach’s alpha = .76): These people engage in politics, mostly driven by personal interests, These people engage in politics, mainly to pursue the interests of certain groups, which may not coincide with the public interest in general, and These people engage in politics, thinking primarily of the public interest as a whole (reverse-scored), with a 7-point Likert-type response scale ranging from “do not agree at all” (1) to “completely agree” (7). Generalized trust was measured with a single item Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people, with the response scale ranging from 1 (you can’t be too careful) to 10 (most people can be trusted). Life satisfaction was measured with a single item All things considered, how satisfied are you with your life as a whole nowadays, with the response scale ranging from 1 (very dissatisfied) to 10 (very satisfied).

External political efficacy (system responsiveness) was measured with two items (Spearman–Brown coefficient = .90): The political system in Latvia allows people like you to have a say in what the government does, and The political system in Latvia allows people like you to have an influence on the political processes, with the response scale from “do not agree at all” (1) to “completely agree” (7). Internal political efficacy (subjective competence) was measured with two items (Spearman–Brown coefficient = .89): To what extent do you feel competent to engage in political activities and organizations? (response scale from “I do not feel competent at all” (1) to “I feel fully competent” (7)), and How confident are you about your ability to actively participate in Latvia’s political processes? (response scale from “not confident at all” (1) to “fully confident” (7)).

The perceived effectiveness of the political system was measured with a single item The public administration system of the Republic of Latvia as a set of institutions operates effectively (performs its functions). The perceived transparency of the political system was measured with a single item The operation of the public administration system of the Republic of Latvia is sufficiently open and transparent. The perceived justness of the political system was measured with a single item The public administration system of the Republic of Latvia ensures the observance of the rights of all citizens and equal treatment of all. These items had the same response scale ranging from “do not agree at all” (1) to “completely agree” (7). The overall trust in the political system was measured with a single item To what extent do you trust or do not trust the public administration system of the Republic of Latvia as a whole, with the response scale from “do not trust at all” (1) to “trust completely” (10). Perceived corruption was measured with a single item In your opinion, how widespread is corruption (for example, bribery, use of taxpayers’ money for personal purposes, etc.) in public administration institutions of the Republic of Latvia, with the response scale from “not widespread at all” (1) to “very widespread” (10).

Finally, the respondents rated the current economic situation in Latvia on a scale ranging from “very bad” (−5) to “very good” (5), with “average” as the scale midpoint (0), and the current material situation of their family, using the same response scale.

Study 2

Study 2 used selected measures from Study 1 with very little modification. Half of the participants (N = 502) completed the trust version of the survey, and the other half (N = 498) completed the distrust version.

The participants first rated their overall trust in the political system on the 10-point single item used in Study 2 (To what extent do you trust or do not trust the public administration system of the Republic of Latvia as a whole?). The participants then indicated their agreement to several statements about the political system, taken from Study 1 in identical or adapted format. External efficacy (system responsiveness) was measured with the same 2 items as in Study 1 (Spearman–Brown coefficient = .93). The perceived effectiveness of the political system was measured with the same single item as in Study 1. The perceived benevolence, ability, and integrity of the political system were measured with one item each, The Latvian public administration system really cares about the well-being of the Latvian residents, Employees of the public administration system skillfully cope with their job responsibilities, and The employees of our public administration system are honest, respectively. The perceived risk associated with the political system was also measured with a single item, The operation of the public administration system endangers Latvian society. All of these items had the same response scale from “do not agree at all” (1) to “completely agree” (7).

After rating the political system, the participants were asked about a political party that they trust (in the trust version of the survey) or distrust (in the distrust version), using the same procedure as in Study 1. The participants then rated the perceived benevolence, ability, and integrity of the perceived political party, as well as the perceived risk with one item each. These items were the same used for rating the political system, with their exact formulations coming from the respective measures of the party in Study 1. All of these items had a response scale from “do not agree at all” (1) to “completely agree” (7). The participants also rated the expected utility associated with the party on the same single item used in Study 1.

Political cynicism was measured with the first two items used in Study 1 and Study 2, dropping the third, reverse-scored item (Spearman–Brown coefficient = .74). Generalized trust was measured with a new single item Overall, do you think that most people can be trusted, regardless of whether they have proven their trustworthiness, with the response scale from “do not agree at all” (1) to “completely agree” (7). Internal political efficacy was measured with the same items as in Study 2 (Spearman–Brown coefficient = .95).

Data Analytic Approach

To visualize the relationships among the trust-related constructs in both studies, we used a network analysis of the variables included in the models. Network analysis allows visualizing the relationships among multiple interrelated variables, and it is an especially appropriate method for presenting data where the directionality of causal links among the variables is difficult to establish, as is true in the case of political trust and its correlates. This data analytic approach treats the correlations among constructs as a network where network nodes represent variables in a data set, and edges represent pairwise conditional associations between variables in the data while conditioning on the remaining variables (Borsboom et al., 2021). The network analysis was based on regularized partial correlations using graphical LASSO (least absolute shrinkage and selection operator) with a tuning parameter chosen using the Extended Bayesian Information Criterium (EBIC). The analysis was done in the JASP software (0.14.1) Network module, which runs on the R package “bootnet” for calculations and the R package “qgraph” for network visualization. Bootstrapping with 1,000 resamples was used to evaluate the network robustness and estimate the stability of the network centrality measures. Each network analysis was repeated twice. The first analysis included separate components of each trust level, and the second analysis used indices calculated for individualized and generalized political trust. The selection of measures for the calculation of indices was based on exploratory factor analysis of the items included in the network analysis. The parallel analysis method with Promax rotation was used to run the analyses in the JASP software (0.14.1) Factor module. The main results of the network analysis (visualizations of the network structures and network centrality measures) are presented below, whereas the technical details and more in-depth information—the results of the network bootstrapping, the factor analyses, as well as the weights matrices for the calculated networks—are available as electronic Supplementary Materials.

Results

Study 1

One hundred and ten participants did not name either a party or a politician they trusted, and 99 participants did not name either a distrusted party or distrusted politician. In addition, three participants named a trusted politician but did not name a trusted party, and three participants named a distrusted politician but did not name a distrusted party; 27 participants named a trusted party, but not a trusted politician, and 32 participants named a distrusted party, but not a politician. Data analysis for political party ratings was thus based on 786 observations, and data analysis for politician ratings was based on 733 observations in the results reported here. The analysis of differences between the trust and distrust conditions is beyond the scope of this paper, and the results are reported across both conditions, but in all the analyses reported below the robustness of the results was checked between both sub-samples to make sure that the findings did not differ significantly depending on the trust versus distrust version of the survey.

The first network analysis included indices for the perceived benevolence, competence, and integrity of the political party, the risk associated with the political party, the single-item measure of utility associated with the party, the single item measure of trust in the party, the indices for cynicism, internal and external efficacy, the single item measures of the perceived corruption, life satisfaction, perceived economic situation in the country and the perceived economic situation in one’s family, generalized trust, the single item measures of the effectiveness, transparency, and justness of the political system, as well as the single item measure of trust in the political system. The resulting network is depicted in Figure 1a and the network centrality measures for all included variables can be seen in Figure 2a. The precise regression weights among the variables and their confidence intervals can be found in the Supplementary Material. The second network analysis for Study 1 data included all the same variables, except the measures for the individualized political trust were about the politician, not the political party. The resulting network can be seen in Figure 3a, its centrality measures in Figure 4a, and the weights matrix and confidence intervals can be found in the Supplementary Material.

Figure 1.

Networks of political trust-related constructs in Study 1 for individualized trust in a political party, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Figure 2.

Centrality measures for networks of political trust-related constructs in Study 1 for individualized trust in a political party, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Figure 3.

Networks of political trust-related constructs in Study 1 for individualized trust in a politician, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Figure 4.

Centrality measures for networks of political trust-related constructs in Study 1 for individualized trust in a politician, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Based on the results of the exploratory factor analyses (included in the Supplementary Material), we used the similarly scaled items loading onto the same factor to calculate an individualized political trust index for the party (Cronbach’s alpha = .94) and for the politician (Cronbach’s alpha = .96) by calculating the average score of the measures of benevolence, competence, integrity, utility, and (reverse-scored) risk, and for generalized political trust (Cronbach’s alpha = .88) by taking the average of the measures of system effectiveness, fairness, transparency, and external efficacy (system responsiveness), and repeated the network analysis with these indices instead of the separate measures (see Figures 1b and 3b) for the network structure and Figures 2b and 4b for centrality measures). A visual examination of the network structure in Figures 1a and 3a clearly shows that the variables related to trust in the political system on the one hand and variables related to trust in the political party and the individual politician on the other, in each analysis form two separate, distinct “network communities” with much stronger links among the variables related to the same trustee and weaker links among variables related to different trustees. For both networks calculated with trust indices in Figures 1b and 3b, there are no direct non-zero links between both trust levels. There were very weak correlations between trust in the party and trust in the system (Spearman’s rho = .10, p = .009) and between trust in the politician and trust in the system (Spearman’s rho = .09, p = .016), but both correlations became non-significant when partialized on cynicism. The correlations between the single-item measures of trust in the party and trust in the politician on the one hand, and trust in the system on the other were even weaker (.08, p = .03, and .06, n.s., respectively), the former becoming non-significant after partializing on cynicism. Taken together, these results offer partial support for the main prediction of our study: both trust levels are clearly distinct constructs, but there was no direct relationship between both trust levels, and both trust judgments were related only through their common negative relationship to political cynicism.

We found that external efficacy was strongly related to all three generalized trust measures (system effectiveness, fairness, and transparency) and unrelated to any measures of individualized political trust. Network analysis showed that internal efficacy was positively related to trust in the political system index (generalized political trust), but unrelated to trust in parties or politicians (individualized political trust). However, it should be noted that when the external efficacy measure was excluded from the calculation of trust in the system index, the confidence intervals for the edge weights between internal efficacy and trust in the system index in both analyses included 0. A look at the more detailed-level network analysis (Figures 1a and 3a) shows that considering the items forming the generalized trust index, internal efficacy is only clearly positively related to external efficacy, but either unrelated or even negatively related to others.

There were no significant relationships between life satisfaction and trust in the party, trust in the politician, or trust in the political system. Evaluation of the economic situation in the country was positively related to trust in the political system, but unrelated to both individualized trust measures. Evaluation of the economic situation in one’s family was unrelated to both individualized trust measures, and trust in the political system. There was a significant negative relation between perceived corruption and trust in the political system, but no relationship between perceived corruption and trust in party or politician. With a similar pattern, the network analysis showed a weak but significant positive relation between generalized interpersonal trust and trust in the political system index, but no relation between generalized interpersonal trust and either of the individualized political trust indices.

A look at centrality measures for the more detailed network analyses (Figures 2a and 4a) shows no single high-centrality node, with most measures of individualized and generalized political trust displaying relatively high strength (or degree, the sum of the absolute edge values connected to a given node, see (Dalege et al., 2017). The centrality analysis of the networks with political trust indices (Figures 2b and 4b) reveals that trust in the political system has the highest strength (degree) of all nodes.

Study 2

Thirty-three participants did not name a political party they trusted, and 26 participants did not name a distrusted political party. The analysis is thus based on 941 observations, where the participants provided the relevant responses for the target of individualized political trust. Like in Study 1, the data were analyzed across the trust and distrust subsamples.

In the first network analysis of Study 2 data, we entered indices for internal and external efficacy, as well as for political cynicism, and single-item measures of generalized trust, life satisfaction, the perceived benevolence, competence, integrity, risk, and utility of the political party, the single-item measure of trust in the political party, and the perceived benevolence, competence, integrity, risk, and utility of the political system, along with the single-item measure of trust in the political system. The resulting network is depicted in Figure 5a. The network centrality measures for all included variables can be seen in Figure 6a, and the regression weights among the variables and their confidence intervals are provided in the Supplementary Material.

Figure 5.

Networks of political trust-related constructs in Study 2 for individualized trust in a political party, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Figure 6.

Centrality measures for networks of political trust-related constructs in Study 2 for individualized trust in a political party, with network analysis for: (a) separate components of political trust and (b) indices of political trust.

Based on the results of the exploratory factor analyses (included in the Supplementary Material), we used the similarly scaled items loading onto the same factor to calculate an individualized political trust index for the party (Cronbach’s alpha = .94) by calculating the average score of the measures of benevolence, competence, integrity, utility, and (reverse-scored) risk, and for generalized political trust (Cronbach’s alpha = .90) by taking the average of the measures of system effectiveness, benevolence, competence, integrity, (reverse-scored) risk, and external efficacy (system responsiveness), and repeated the network analysis with these indices instead of the separate measures (see Figure 5b for the network structure and Figure 6b for centrality measures). A visual examination of the network structure in Figure 5a reveals that the variables related to trust in the political system on the one hand and variables related to trust in the political party on the other form two separate, distinct “network communities” with much stronger links among the variables related to the same trustee, and weaker links among variables related to different trustees.

The network edges in Figure 5 show a different pattern than those in Figures 1 and 3, suggesting that the results of Study 1 and Study 2 differ in terms of relations between both trust levels. There are many non-zero links between the generalized and individualized trust nodes in Figure 5a; more importantly, in Figure 5b, there is a strong positive link between trust in the system and trust in party indices. These results show that contrary to Study 1 results, Study 2 fully supports the main prediction of our study, revealing a positive relationship between both trust levels. Political cynicism is strongly negatively related both to trust in the political system, and to trust in the political party, although the latter relationship is weaker than the former one (but still significant).

We found that external efficacy was strongly related to most generalized political trust measures (except the perceived system competence) in the network analysis and loaded on the same factor with all other generalized political trust items in the factor analysis. It was unrelated to any measures of individualized political trust. Network analysis showed that internal efficacy was positively related to trust in the political system index (generalized political trust), but very weakly related to trust in the political party (individualized political trust). The confidence interval for the corresponding network edge included zero if the external efficacy measure was excluded from the calculation of trust in the system index. There was no robust relationship between life satisfaction and trust in the party, or trust in the political system. Also, we did not find a relationship between generalized interpersonal trust and both levels of political trust, as confidence intervals for the corresponding network edges included zero.

Centrality measures for the more detailed network analyses (Figure 6a) show a similar pattern to Study 1 data, with multiple measures of individualized and generalized political trust displaying relatively high strength, and in the network with political trust indices (Figure 6b), trust in the political system has the highest strength (degree) of all nodes.

Discussion

The two studies reported here aimed primarily at examining the relationships between generalized and individualized political trust. The results are somewhat inconclusive in the sense that Study 1 showed no significant relationship between generalized (trust in a political system as a whole) and individualized (trust in specific parties and trust in individual politicians) political trust judgments, whereas Study 2 found a positive relationship between judgments of political trust at both levels (political system and political parties). Study 1 results suggest that trust in the political system with its institutions and procedures is psychologically distinct and clearly differentiated from trust in the specific people who are part of the system at the given time. The only variable related to both levels of trust was political cynicism, which is fully in line with the theoretical predictions based on the very essence of this construct as a negative view of politicians as a category (Agger et al., 1961; Citrin & Stoker, 2018). Study 2 results showed that both types of political trust judgments are related; nevertheless, it should be noted that, despite the positive relationship between both constructs, the two levels of trust emerged as two distinct latent variables in the factor analysis and can be seen forming two separate “network communities” in the network analysis with much stronger links among the variables related to the same trust level, and weaker links among variables related to the two different trust levels.

Taken together, our results offer support to the theoretical reasoning that conceptualizes the two trust levels as different implementations of trust judgment, affected by the perception of specific qualities of the political actors, institutions, or the system as a whole at a particular moment in time, reflecting the situational circumstances and contextual factors affecting the trustworthiness of the specific trustee (Fisher et al., 2010; Proszowska et al., 2021; Ruelens & Nicaise, 2020). The differences between the results of both studies fit this theoretical reasoning. The two studies used different measures of trust in the political system; the items used to measure trust in the system in Study 2 were much more similar to the items measuring trust in the political party than the items used to measure trust in the system in Study 1. The similar formulation of items in Study 2, leading the participants to evaluate both types of trustees on the same criteria, may have evoked more aligned judgments of trust than was the case in Study 1 when the participants evaluated the trustworthiness of the political system and individual political actors on different criteria. This interpretation suggests that the relationship among different political trust judgments may be at least partially dependent on the measurements employed in each specific instance of research. Another possible explanation for the difference between the study results rests in the timing of both studies. The data for Study 1 were collected in late 2019, before the outbreak of the COVID-19 pandemic; the data for Study 2 were collected in the summer of 2021, at the height of the COVID-19 crisis. It is also possible that the more aligned judgments of trust in the system and specific political parties in Study 2 reflect a change in people’s processing of relevant political information. For example, such a change may represent increased attention to the performance of institutions and specific political incumbents in tackling the crisis or increased general support for the political system as a whole and its constituents in a “rally-around-the-flag” effect, or, quite the opposite, a generally negative response to pandemic restrictions imposed by the government. Such an interpretation suggests that the relationship among different political trust judgments depends on the specific social, economic, or political context. Given these contextual effects, it is evident that no single study can provide definitive evidence of the relationship between both levels of trust, indicating the necessity for further research.

The findings reported here point to several promising research directions. One is the dynamic relationship between both levels of trust, for example, the contributing role of individualized trust judgments in the formation of generalized political trust, or the reversed mechanism of how generalized trust in the political system affects the individualized trust judgments about specific political actors. Such research would require more data-intensive designs, measuring individualized trust in multiple political actors, and perhaps using experimental manipulations and/or process-tracing techniques for studying the formation of trust judgments. Another interesting and relevant line of research is the mediating and/or moderating role of political orientation and party affiliation in both types of political trust judgments. The relationship between political partisanship and political trust has been well-established in the literature (Christensen & Lægreid, 2005; Hetherington & Husser, 2012; Popp & Rudolph, 2011), but differentiating between both levels of trust might yield new insights about the relationship between trust and ideological beliefs, past voting behavior and future commitments, various political identities, and performance evaluation of political actors and political systems.

The results also provide evidence that the perceived benevolence, competence, and integrity of the trustee form a robust core structure of political trust both on the individualized and generalized trust level, confirming previous findings both in general trust literature (Burke et al., 2007) and political trust research (Grimmelikhuijsen & Knies, 2015; Hamm et al., 2019). Our results also indicate that the perceived risk associated with the trustee can be considered another essential component of political trust that could be added to the three-component model and included in internally consistent political trust measures. In addition, we also showed that the personal utility (expected gains vs losses) associated with the trustee is another inherent component of political trust. Although utility may be considered a default component of a rational judgment (Fischhoff & Broomell, 2020) and thus implicitly represented in the evaluation of all the previously mentioned trust components, in many circumstances conceptualizing and operationalizing it as a separate construct can provide valuable information about how people judge the trustworthiness of political actors and institutions, as indicated by the significance of personal utility in the evaluation of specific policy initiatives and stances (Harring, 2013; Hetherington & Globetti, 2002).

For the generalized political trust, our results clearly show that external political efficacy is an integral part of the evaluation of the trustworthiness of the political system, more closely related to other components of trust in the system than internal political efficacy. Internal efficacy was unrelated to individualized political trust and related to generalized political trust only through its relation to external efficacy. Our results show that internal efficacy plays a limited role in political trust judgments, both on the individualized and generalized levels. Political cynicism was robustly negatively related to generalized and individualized political trust but did not load on the same factor as political trust. Our findings clearly show that political cynicism and political trust are related but separate constructs, contributing to the literature drawing a conceptual and empirical distinction between both (Dancey, 2012; Pattyn et al., 2012). Generalized interpersonal trust was only weakly related to generalized political trust. This result is in line with previous findings where simple, single-item measures had been used to measure political trust and generalized interpersonal trust (c.f. Zmerli & Newton, 2008). To clarify the relationship between political and interpersonal trust, more refined, multi-item measures of interpersonal trust should be used.

Our studies provide a concise method for measuring political trust (for various trustees), covering the components of the perceived benevolence, competence, integrity, and associated risk of the trustee, which can be used either in an 8-item form (tested in Study 1) with two items per component, or in a short 4-item form (tested in Study 2). Both versions of the scale show good internal consistency reliability, and we encourage other researchers to use them in various political contexts and cultures as a simple and straightforward measure of political trust.

Our findings also contribute to the literature on the relation between political trust and perception of socio-economic circumstances. Our results showed that life satisfaction was unrelated either to individualized or generalized political trust, offering little support for the “happiness contract” hypothesis (Esaiasson et al., 2019). The results of Study 1 suggest some indirect link between both constructs, though, where life satisfaction is related to the perception of the economic situation in one’s family, which in turn is related to the perception of the economic situation in the country, which in turn is related to generalized political trust. However, this relation, and the possible causal links among the constructs involved, require further research. The perceived corruption was negatively related to generalized political trust, but unrelated to individualized political trust, suggesting that the perceived corruption may be negatively associated with the trust in the political system. However, this association does not automatically translate into distrust in specific politicians or parties.

Taken together, our findings suggest that trust in the political system and trust in specific political actors form separate, although related, network structures, and are likely determined by a different, although overlapping, set of variables representing different kinds of psychological reasoning. Generalized political trust can be seen as part of a general attitude towards one’s country, whereas individualized trust is based on more situational factors and shaped through the mechanisms of person perception and group perception. Both types of trust thus should be considered as separate, although related, constructs, which may differentially predict political judgments and behaviors. In addition, the relationships between both levels of political trust and their correlates are related to contextual factors, such as political or socio-economic circumstances, and the method of measurement used. Further examination of these contextual factors should provide more insight into the psychological mechanisms and determinants of political trust.

Supplemental Material

sj-docx-1-sgo-10.1177_21582440241281541 – Supplemental material for The Structure of Individualized and Generalized Political Trust: A Network Analysis of Data from Two Representative-Sample Studies

Supplemental material, sj-docx-1-sgo-10.1177_21582440241281541 for The Structure of Individualized and Generalized Political Trust: A Network Analysis of Data from Two Representative-Sample Studies by Girts Dimdins, Martins Priedols, Ivars Austers, Viktorija Gaina and Veronika Leja in SAGE Open