Trust and confidence
Frédérique Six, December 28, 2015
In the last decade or so the interest within the mainstream trust literature in more impersonal forms of trust, such as system trust, organizational trust, institutional trust, has surged (e.g., Bachmann & Inkpen, 2011; Searle et al, 2010; Kroeger, 2011; Möllering, 206; Sydow, 2006). Interestingly, these are all called trust, rather than confidence.
I recently came across a trust model – Trust Confidence and Cooperation model (Earle etal 2007) – that is not cited in this mainstream trust literature and that uses confidence and trust as explicitly separate concepts in a way that struck me as odd at first, given that I also hold on to the (implicit) assumptions dominant in the mainstream trust literature. I am now not sure how I should adapt my thinking. So I present you with my puzzles, in the hope that a further conversation may develop that may help clarify these conceptual puzzles. I start with a brief presentation of the Trust Confidence Cooperation Model (TCC) by Earle et al (2007) that is used often in the domain of risk management. I then briefly revisit Luhmann (1998) before presenting my thoughts and struggles.
Trust Confidence and Cooperation model (TCC)
The TCC model is based on the premise that there are two principal pathways to cooperation. One path is via social trust and is based on ‘morality information’, while the second path is via confidence and is based on performance information (Figure 1).
“Trust is social and relational; confidence is instrumental and calculative. We define trust as the willingness, in the expectation of beneficial outcomes, to make oneself vulnerable to another based on a judgment of similarity of intentions or values. Confidence is the belief, based on experience or evidence (e.g., past performance), that certain future events will occur as expected” (Earle, 2009, p. 786).
Figure 1: the TCC model
Thus the basis for trust is shared values that can be measured in many different ways.
”In empirical studies, trust has been indicated variously by measures of in-group membership, morality, benevolence, integrity, inferred traits and intentions, fairness, and caring” (Earle, 2009, p. 786).
Earle et al (2007) stress that trust is placed in the freedom of others and is experienced as affect or emotion. It requires that the trustor has a choice among alternatives. It can be forward looking and creative. They also claim that general trust [trust among strangers] is “a very demanding sociality to maintain over time. General trust cannot supply the foundation for daily life” (p. 6). The target of trust is (treated as) an agent, in other words agent-based trust (p. 12).
The basis for confidence, on the other hand, is past performance ,
“or institutions/procedures designed to constrain future performance.[…] Past performance and performance-controlling institutions/procedures can be measured in a variety of ways: by indicators of evidence, regulations, rules/procedures, contracts record keeping/accounting, social roles, ability, experience, control, competence, and standards “ (Earle, 2009, p. 786).
Confidence thus requires others to act in specific expected ways. It is less demanding than trust as it is the normal code of operation (Earle et al 2007). Confidence involves exposure to an external system: an individual expresses confidence in “an entity by relying on it to properly control his/her external exposure” (Earle et al, 2009, p. 5). Confidence can only look back and is based on familiarity and is therefore an undemanding form of sociality to maintain over time. The target of confidence is (treated as) an object in other words (object-based confidence) (p. 13).
Trust and confidence interacting
Earle et al (2007) see the two concepts as connected. The bases for confidence “are justified accepted only within the group or community of trust that generated them. Also, any judgment of trust presupposes judgements of confidence” (Earle et al, 2009, p. 5). Or “loss of confidence makes trust necessary for establishing a new record of experience or past performance, allowing the replacement of trust by confidence” (p. 5). They claim that trust entails risk while confidence is occasioned by danger.
People are said to divide information in two types.
“Morality information is equivalent to actions reflecting the values of an agent, and performance information is simply behavior of an object. The agent’s values define a relationship of trust, and it is within that relationship that the performance information and the confidence to which it leads are judged” (p. 7).
The implications are, they claim, that in times of low social uncertainty, when morality information is not relevant, social trust does not play the dominant role in cooperation. Social trust becomes more important in times of uncertainty, when morality information is relevant.
They refer to a long list of often cited sources to claim the “ubiquity if [their] two core concepts: trust […] and confidence” (p.11), even though they do not use the concept of confidence, but only trust (Figure 2).
Earle et al (2007) use Ring’s concepts to disagree with the saying “trust comes on foot but leaves on horseback”, by replying with “trust generally tends to be resilient, while confidence generally tends to be fragile” (p.13).
Figure 2: Earle et al’s (2007) reinterpretation of trust literature
|Source||Original categorization||Earle et al’s reinterpretation||My response (not complete)|
|Confidence in persons
Trust based on social relations
Confidence in institutions
|Zucker (1986)||Process trust
|Confidence in persons
Trust based on social relations
Confidence in institutions
|Lewicki and Bunker (1995)||Calculus-based trust
|Confidence based on cost-benefit analysis
Confidence based on predictability
Trust based on social relations
|Identification-trust is very rare in work situations and not always desirable, suggesting trust is rare, and there is mainly confidence, even in dyads|
|Sitkin and Roth (1993)||Trust
|Confidence based on performance
Trust based on shared values/distrust based on lack of share values
So distrust is only about lack of shared values, not lack of competence?
|Das and Teng (1998)||Goodwill trust
|Das and Teng define level of confidence … as level of certainty …
Not sure about competence always being about confidence.
|Ring (1996)||Fragile trust
|Hardin (1998)||Trust in persons
Trust in institutions (but that cannot work)
|Levi (1998)||“trust depends on confidence in institutions that back up the trustee” (p.79)|
|Tonkiss and Passey (1999)||Trust based on shared values
Confidence based on institutional and contractual forms
|Rousseau et al (1998) and Kramer (1999)||Instrumental/calculative trust
|Kramer (1999)||History based trust
|Interpersonal trust based on interaction, combining both trust and confidence
Social trust based on social relations
Confidence in persons
Confidence in institutions
|Does this show limits of distinction?
But what about morality of profession?
Revisiting Luhmann (1988)
Luhmann (1988) states that confidence is the normal everyday situation: “you are confident that your expectations will not be disappointed” (p.97). You not only disregard the possibility of disappointment because you think it is unlikely, but also “because you do not know what else to do”, i.e. you experience a situation where you cannot see an alternative. Trust, on the other hand, “presupposes a situation of risk”(p. 97) and you can avoid taking the risk. “you do not depend on trusting relations in the same way you depend on confidence, but trust too can be a matter of routine and normal behavior” (p. 97). Importantly, according to Luhmann the distinction is not black and white, it depends on perception and attribution. If you do not consider alternatives, then you are in a situation of confidence, while if you do consider alternatives and actively choose one over the other, then you see the situation as one of trust. When defining the situation as one of confidence, then you respond to disappointment with external attribution, whereas in the case of trust , you have to consider internal attributions. And Luhmann adds an interesting requirement for trust “trust is only required if a bad outcome would make you regret your action” (p.98), because you had a choice between alternatives and would with hindsight choose another alternative. Because of his emphasis on having a choice between alternatives, Luhmann proposes that “a relation of confidence may turn into one of trust if it becomes possible (or is seen to be possible) to avoid that relation” (p.98) as is the possibility of trust turning to confidence if you perceive a situation as no longer open to influence.
My thoughts and struggles
I can see the benefit of distinguishing more carefully between trust and confidence than most researchers studying impersonal forms of trust, including myself!, seem to have done recently, but I wonder whether Earle et al in their TCC model may be pushing it too far.
- The distinction between perceiving to have a choice or not, seems relevant, which is what Luhmann appears to stress. Although this not only may differ by person, but also over time for one person. So how useful is it both conceptually and empirically?
- What about all the research into automatic and deliberate decision-making (see Kahnemann’s bestseller Thinking, fast and slow, 2011) that shows that we make far fewer decisions consciously than we think? That would suggest that most of our lives we have confidence rather than face decisions about trust, also in interpersonal interactions. And also that we tend to overestimate the degree to which we trust compared to which we actually only have confidence, as we tend to overestimate our deliberate decision-making. Or maybe that is exactly the crucial point to make!
- How about trust in government? You do not have a choice in whether you want to interact with government or not, only possibly whether you want to emigrate or not. Does it matter whether citizens are asked about how much they trust in government versus how much confidence they have in government? Currently both are used in surveys.
- How about research into trust and regulation? Is that all about confidence, rather than trust? But Braithwaite & Makkai (1994) showed how the degree to which care home directors perceived that the inspectors trusted them, led to higher levels of compliance during the next inspection. That surely is not just about confidence? Also research by Heimer & Gazley ( 2012) showed the importance of the actual interpersonal dimension and interaction between inspectors and employees of the inspected organization.
- The distinction between being vulnerable to the actions of other people versus large, impersonal systems, also seems to make sense, as we align the concepts with the distinction between structure and agency, but we know all the conceptual complications that come with this duality. Furthermore, persons also have concrete interactions with individuals as they interact with systems, cf Giddens’ facework (1990). So should we be talking about trust in that individual and confidence in the institutions as the two components of what we now call trust in organizations, institutions or systems (e.g. Kroeger 2013; Bachmann & Inkpen, 2011; Sydow, 2006)?
- I personally can work best with the distinction between intentions and competence as the main dimensions of trustworthiness. Intentions then encompass both goodwill/benevolence (which may be achieved without sharing values), and integrity in terms of both adherence to and acceptability of principles (values).
- As Dietz (2011) showed, you usually base your assessment about the other party’s trustworthiness on different sources of information. So you assess the past performance of the other party for clues about both intentions and competence. You also interpret any other source of information, such as third party testimonials, external regulation for that other party etcetera, for such clues. Past performance not only gives clues about competence, as Earle et al seems to suggest as they reinterpret competence trust into confidence (see Figure 2), but also clues about goodwill and norm/values congruence, through relational signals (cf Six, 2005).
- And therefore I find it difficult to work with Earle et al’s distinction between morality information versus performance information. What kind of information is morality information if it is not part of performance? Much external regulation has very different impact on regulatees compliance depending on the regulatee’s dominant motivational posture or decision frame, so (cf V. Braithwaite, 2009; van Wijk & Six, 2014).
- So several sources of information are interpreted for clues about the intentions of the trustee – is s/he willing to do what I need him to do – and clues about their competence –is s/he able to do what I need him to do?
- Lots of research shows the relevance of distinguishing between intentions and competence/ability because of the attributions that follow a breach of trust. Competence-based trust violations require different remedies than integrity-based trust violations (e.g., Ferrin et al 2007; Janowicz-Panjaitan & Krishnan 2009). In trouble situations, when the flow of positive expectation in trust relations is disrupted, where lack of competence was deemed the cause are more likely to trigger external attributions and not lower trust as that are seen as less controllable, while perceived lack of good intentions would more likely trigger attributions of lack of effort and have more serious consequences for the relation as that is seen as more controllable (Weiner 1995; Anderson et al 1996).
- Although not exactly the same, I can see how a distinction between cognitive and affective elements can also be helpful, both conceptually and empirically, but with more complications. As McAllister (1995) pointed out, it is difficult to separate precisely the affective from the cognitive.
- Where I think that Earle et al (2007) push it too far, is that all performance is about confidence and that trust is only about shared values.
- Whether the other person can actually perform the action that you trust him to perform, is not relevant. If that is the case, then there is always a combination of trust and confidence needed, even in interpersonal situations.
- All control measures (rules, performance, etcetera) only contribute to confidence and not trust, so trust and control are not complements any longer.
- Both confidence and trust are only possible when values are shared if I interpret their model correctly, since bases for confidence only work within the community of trust in which the measures (regulations, rules/procedures, contracts record keeping/accounting, social roles, ability, experience, control, competence, and standards) are agreed. What happens when there are insufficient shared values to establish such a community?
 When I refer to “the mainstream trust literature” is use this term to refer to publications by the group of people active within the First International Network on Trust. This network started within the domain of management, organization studies, OB, business research. As a network it aims to be inclusive for all perspectives, but the original domains are still the largest.