Assessing Two Dimensions of Interpersonal Trust: Other-Focused Trust and Propensity to Trust

Zhang, Ming

doi:10.3389/fpsyg.2021.654735

ORIGINAL RESEARCH article

Front. Psychol., 27 July 2021

Sec. Personality and Social Psychology

Volume 12 - 2021 | https://doi.org/10.3389/fpsyg.2021.654735

Assessing Two Dimensions of Interpersonal Trust: Other-Focused Trust and Propensity to Trust

$\r\nMing Zhang,*$ Ming Zhang^1,2*

¹CAS Key Laboratory of Behavioral Science, Institute of Psychology, Chinese Academy of Sciences, Beijing, China
²Department of Psychology, University of Exeter, Exeter, United Kingdom

One’s propensity to trust others and others’ trustworthiness are two important aspects of interpersonal trust. Both theory and research suggest that it is possible to distinguish between an individual’s propensity to trust (one’s “trustingness” or the extent to which one feels able to trust others) and their other-focused trust (the extent to which one feels that others are worthy of our trust). However, there is as yet no measure that distinguishes between these two components of trust. In three studies, we examined the psychometrics of a proposed two-dimensional measure of trust that encompasses propensity to trust and other-focused trust components. To test discriminant validity, we also administered measures of personality, personal self-esteem, social capital, propensity to like people, perceived social support, as well as general and personal beliefs in a just world. Factor analyses supported the proposed two-factor model for the new trust measure. Further analyses supported the difference between these measures.

Introduction

Trust is commonly defined as a confident expectation about a situation leading to a willingness to accept vulnerabilities that arise from uncertainty and risk (Dietz, 2011; Patent and Searle, 2019). Trust plays an important role in daily interpersonal interactions (Rotenberg et al., 2005), predicting individual behavior toward others (Colquitt and Salam, 2009; Yamagishi et al., 2015). Trust is also directly and strongly linked to individual well-being, health, and longevity (Young and McGrath, 2021).

Trust involves a willingness to be vulnerable to the actions of another person or people (Mayer et al., 1995; McEvily and Tortoriello, 2011). This implies that when trusting others, individuals often simultaneously bring together their own propensity to trust other people with their judgment of whether specific others deserve their trust (Colquitt et al., 2007; Lucas et al., 2011). Considering interpersonal interaction included the individual’s side and the other person’s side (Dietz and Den Hartog, 2006; Schoorman et al., 2007; Jones and Shah, 2016), it means the interpersonal trust depended on both two parties. Thus, interpersonal trust should be considered from two aspects: the general trustworthiness of others (i.e., a state-based trust, depends on other people) and one’s own general propensity to trust others (i.e., a trait-based trust, depends on oneself). There were series of studies focused on trustworthiness, propensity to trust, or both of them (Mayer and Davis, 1999; Frazier et al., 2013; Alarcon et al., 2018a,b; Patent and Searle, 2019). However, these two specific aspects of trust have not been clearly conceptualized or empirically differentiated by the items with clear face validity that described trust from the viewpoint of others and oneself (“people” vs. “I”).

The most common definitions of trust refer to the extent to which we think others are trustworthy (Mayer et al., 1995)—this is what we refer to as “other-focused trust.” This refers to the state-based general trustworthiness of others, that is, the judgment individuals make about whether or not the other(s) is (are) deserving of their trust, for example as a function of their perceived integrity and good character (Colquitt and Rodell, 2011). Individuals arrive at this judgment based on their observation of, or inference about, other people’s behaviors, or based on pre-existing (and often prejudiced) beliefs about them (Evans and Revelle, 2008). The extent to which we find specific others trustworthy can vary depending on who others are and our experiences with and beliefs about them, but people can also have more chronic tendencies to believe that particular others are worthy of trust, based for example on past experiences where trust was confirmed, or instead broken (Murray, 2015).

In addition to focusing on whether others are trustworthy, individuals can be (or feel) more or less trusting, or inclined to trust other people (Burke et al., 2007). For example, individuals who have undergone traumatic experiences might temporarily withdraw trust in others, irrespective of who they are (Jansen et al., 2015). That is, they might feel they cannot trust other people because of how they are feeling, rather than because of who the other is—this is what we designate here as “propensity to trust.” As defined here, propensity to trust is self-focused, less likely to vary depending on who others are and more likely to respond to variations in the individuals’ sense of safety and esteem (Mayer et al., 2009; Beldad et al., 2010). When high, propensity to trust reflects an optimistic conception of the world and generalized positive expectations about others (Zeffane, 2020). This also highlights that propensity to trust and other-focused trust—as two dimensions of interpersonal trust—are likely to be closely related, though possible to differentiate.

Two-dimensional models of trust have been proposed before (see Thielmann and Hilbig, 2015 for a review). The closest model to what we are proposing in this paper is by Yamagishi et al. (2015). These authors introduced a two-component trust scale that differentiates between “trust beliefs” and “trust preferences.” Trust beliefs refer to an individual’s estimate about the trustworthiness of others—a sample item is “Most people are trustworthy.” Though this subscale includes items that do not quite match our conceptualization of other-focused trust (e.g., “Generally, I trust others”), this is similar to what is assessed by most traditional measures of trust (e.g., Yamagishi and Yamagishi, 1994) and to what we designate here as “other-focused trust.” Trust preferences, by contrast, refer to an individual’s propensity to be vulnerable to others and reflect a particular sense of self as trusting in which “people derive personal satisfaction from being a trustful person.” Trust preference, by contrast, refers to an individual’s propensity to be vulnerable to others and reflects a particular sense of self (Yamagishi et al., 2015, p. 440).

Though this conceptualization by Yamagishi et al. (2015) is similar to what we propose here for “willingness to trust,” it differs from ours by explicitly incorporating a willingness to incur costs as a function of trusting others (e.g., “Even though I may sometimes suffer the consequences of trusting someone, I still prefer to trust than not to trust others”). Though this choice is likely to increase the empirical distinction between the two subscales, it ignored one situation. Specifically, there was a limitation in revealing as trusting those who are unlikely to habitually consider the consequences of their trust, but who if asked in this way might feel foolish to admit that they would be trusting if their trust were to be abused. This is also important because self-report measures are notorious for their vulnerability to impression management (Chan, 2009), and portraying oneself as open to abuse is not generally seen as desirable. As such, our aim was to develop a similar measure to that of Yamagishi et al. (2015), but which did not refer to the costs people are willing to incur for being trusting. This measure would allow researchers to, in the future, shed further light on the mechanisms underlying trust, or its consequences, by examining their differential drivers, how they might vary across contexts, and how they differentially guide behavior in social interactions.

As part of this development, we also aimed to examine whether propensity to trust and other-focused trust were differently associated with measures that have been previously shown to be related to trust. Previous research has shown that trust is associated with a range of personality traits (De Jong et al., 2016), as assessed by the Big Five Personality Inventory (Evans and Revelle, 2008), life satisfaction (Abbott et al., 2016), and personal self-esteem (Bao et al., 2016). For instance, perceived trustworthiness (other-focused trust) has been shown to be positively correlated with agreeableness and with conscientiousness (Evans and Revelle, 2008). People’s propensity to like others has also been shown to be positively related to trust, especially in the relationship between buyers and sellers (Nicholson et al., 2001). Trust is also closely related to social capital (Son and Feng, 2019) and perceived social support (Shin, 2013). For example, communities with higher levels of social capital tend to include citizens who are more trusting of one another and those who report receiving social support from others usually have more generous expectations of others, including greater trust (Ikeda, 2013). Trust is also likely to be related to an individual’s beliefs in a just world. The belief in a just world is defined as the perception that one lives in a world where people generally get what they deserve (Trost et al., 2014). People want to believe they live in a just world so that they can go about their daily lives with a sense of trust, hope, and confidence in their future so a high level of belief in a just world is likely to be related to a high level of trust (Schindler and Reinhard, 2015).

Although prior research has examined the relationship between trust and these other constructs, and distinguished different components of trust, there has been as yet no attempt to distinguish the correlates of propensity to trust and other-focused trust. Our goal is to develop a scale that allows for their separate measurement and to examine the differential associations between these two types of trust and personality variables, personal self-esteem, life satisfaction, social capital, social support, and belief in a just world.

Study 1: Item Selection

The key goal of this study was to develop items that have the face and construct validity as measures of propensity to trust and other-focused trust. First we selected a pool of items from existing scales based on their face validity regarding the two hypothesized components. Second, we conducted exploratory factor analyses to investigate whether the items could be divided into two sets, assessing propensity to trust and other-focused trust. Third, we explored the associations between the two subscales and other relevant variables previously shown to be associated with aspects of trust.

Materials and Methods

Participants

Participants were recruited through the online research crowdsourcing platform Prolific Academic. After excluding 11 participants who failed the attention check, a total of 491 adults (age: 35.12 ± 11.54 years) participated in this study, with a participants-to-variables ratio of 18:1, based on the criterion established by Worthington and Whittaker (2006). Table 1 reports the demographic information, education level, and nationality of participants. Due to the available distribution of gender and education, we examined the effect of gender and education on trust scores for all three studies (see Supplementary Material).

TABLE 1

Table 1. Participant characteristics.

Measures

An initial pool of 27 items was selected from three trust scales: Three items used in a study by Baltatescu (2009), 19 items from the Generalized Trust Scale (Couch et al., 1996), and 5 items from the General Trust Scale (Yamagishi and Yamagishi, 1994). The full wording of all items is provided in Table 2. All items were selected based on face validity regarding their potential to tap into the aforementioned definitions of propensity to trust and other-focused trust. Participants responded to each item on a Likert-type scale from 1 = strongly disagree to 5 = strongly agree. In order to explore possible differences between the two dimensions of interpersonal trust, participants also completed four individual difference measures including the Big Five Inventory (e.g., “I see myself as someone who is talkative”; John and Srivastava, 1999), the Liking People Scale (e.g., “Sometimes when people are talking to me, I find myself wishing that they would leave”; Filsinger, 1981), Rosenberg’s Personal Self-Esteem Scale (e.g., “On the whole, I am satisfied with myself”; Rosenberg, 1965), and the Satisfaction with Life Scale (e.g., “In most ways my life is close to my ideal”; Diener et al., 1985).

TABLE 2

Table 2. All 27 items used for testing in Study 1.

Procedure

The study was programmed using Qualtrics and distributed on the Prolific Academic crowdsourcing platform. The survey protocol was approved by the University of Exeter’s Psychology Ethics Committee. Participants were invited to participate in a study on trust and offered £1 in compensation, in line with compensation standards on this platform. After providing informed consent, participants provided basic demographic information, including gender, age, nationality, and education. Next, participants responded to the total pool of 27 items assessing interpersonal trust. Finally, participants responded to the 44-item Big Five Inventory, the 15-item Liking People Scale, the 10-item Personal Self-Esteem Scale, and the 5-item Satisfaction with Life Scale.

Results

Exploratory Factor Analyses

Principal components analysis was used to identify the items that would best distinguish between propensity to trust and other-focused trust. Oblimin rotation was used to interpret the factor loadings because propensity to trust and other-focused trust were expected to be correlated. The Kaiser-Meyer-Olkin measure of sampling adequacy (KMO = 0.94) and Bartlett’s test of sphericity, χ²(325) = 6470.04, p < 0.001, showed adequate fit. It was observed that 26 of the 27 items correlated at least 0.30 with at least one other item, suggesting reasonable factorability. The communalities for the 26 items were all above 0.30, further confirming that each item shared some common variance with other items. Given these overall indicators, factor analysis was deemed to be suitable.

Initial eigenvalues (see Table 3) in principal components analysis indicated that the first five factors explained 38.65, 7.17, 5.57, 4.66, and 4.24% of the variance in the items. Rotation sums of squared loadings indicated that the first five factors explained 6.24, 4.36, 7.28, 4.80, and 3.93% of the variance. Two criteria were used to determine the factor structure: (a) Retain items with a factor loading equal to or greater than 0.30 and (b) Items with double loadings were carefully considered and retained only where their content matched that of other items within a factor on which they loaded sufficiently. The first factor consisted merely of all reversed items and was therefore not theoretically meaningful. The second and fourth factors had a very low number of primary loadings. However, the third and fifth factors had a sufficient number of primary loadings that were theoretically meaningful and distinguished between the two theorized components. Therefore, the 12 items loading on these two factors were retained.

TABLE 3

Table 3. Factor loadings and communalities for the 27 items in Study 1.

A principal components factor analysis of the remaining 12 items was conducted using oblimin rotations. All items in this analysis had primary loadings on one factor over 0.50 except one item (item 18). After excluding this item, a principal components factor analysis of the remaining 11 items using oblimin rotations was conducted again. All items in this analysis had primary loadings over 0.50 on one of the two factors obtained. The factor loading matrix for this final solution is presented in Table 4. These 11 items were retained.

TABLE 4

Table 4. Factor loadings for the items in Study 1 and Study 2.

Internal Consistency

Internal consistency for each of the scales was examined using Cronbach’s alpha. Both subscales demonstrated good reliability, with a = 0.89 for the other-focused trust sub-scale (6 items) and a = 0.75 for the propensity to trust sub-scale (5 items). The average inter-item correlations, which were used in predictor and criterion constructs (Diamantopoulos et al., 2012), were r = 0.58 for the other-focused trust items and r = 0.38 for the propensity to trust items.

Associations With Criterion Measures

To examine convergent and discriminant validity, correlations between the two components of interpersonal trust and the individual difference variables were computed (see Table 5). Other-focused trust was moderately and positively correlated with agreeableness (r = 0.41, p < 0.001). Propensity to trust was highly positively correlated with agreeableness (r = 0.63, p < 0.001), and moderately positively correlated with extraversion (r = 0.44, p < 0.001). Other-focused trust and propensity to trust were both negatively associated with neuroticism (r = –0.27, p < 0.001; r = –0.30, p < 0.001, respectively) and liking people (r = –0.34, p < 0.001; r = –0.47, p < 0.001, respectively). Other-focused trust was positively correlated with propensity to trust (r = 0.50). We compared the correlations between the two types of trust and the criterion measures to check whether they were significantly different from each other (Table 5), by z-score for the difference between two correlations in the online calculator¹.

TABLE 5

Table 5. Comparison of correlation coefficients in Study 1.

z = \frac{(\frac{1}{2} ln \frac{1 + r 1}{1 - r 1}) - (\frac{1}{2} ln \frac{1 + r 2}{1 - r 2})}{\sqrt{(\frac{1}{n 1 - 3} + \frac{1}{n 2 - 3})}}

The results showed that propensity to trust had a significantly stronger correlation with extraversion, agreeableness, and liking people compared to other-focused trust.

Discussion

The results show that interpersonal trust can be empirically divided into two components: Propensity to trust and other-focused trust. This is consistent with what has already been reported by Yamagishi et al. (2015), but is now shown with subscales that do not also differ in the extent to which the consequences of one’s trust are explicitly considered. The factor analysis supported this distinction in the items we sampled and the resulting subscales proved to be internally consistent. Correlational analysis with other variables showed that propensity to trust and other-focused trust were both positively correlated with some of the measures included in this study (extraversion, agreeableness, conscientiousness, openness, self-esteem, and life satisfaction) and negatively correlated with neuroticism and liking of others. However, by comparing the magnitude of the correlations between the two types of trust and criterion measures, we found evidence that the two types of trust were distinguishable in some aspects. Specifically, propensity to trust was more strongly correlated with extraversion, agreeableness, and liking people than other-focused trust. This suggests that these facets of personality may reflect a broader interpersonal orientation that is more strongly reflected in propensity to trust than other-focused trust (which is less about the self). These divergences support the idea that the measure we developed taps into two forms of trust that can be empirically differentiated.

Study 2: Confirmatory Analyses

The key goal of Study 2 was to confirm the construct validity of the two subscales. We also aimed to extend the test of the scales’ discriminant validity by adding a social capital scale and a scale of perceived social support. We kept personal self-esteem in this study, to replicate the relationship found in Study 1.