Trust, mistrust and distrust

Trust is important for a well-functioning society and democracy. If we cannot trust each other, politicians and political institutions, we are in deep trouble. For that reason, I am happy to see that scientists aim to understand not only what can explain trust, but also how we should understand trust in the first place.

Specifically, what does it mean to have low trust? Is the opposite of trust mistrust or distrust? Apparently, neither of the two is the opposite of trust. Low trust is different from mistrust and distrust. And mistrust and distrust are not the same. This is the core argument in a series of academic publications.

To take on example, consider the working paper “Exploring Trust, Mistrust and Distrust“. The paper describes how trust, distrust and mistrust are all part of the same trust family with different manifestations, evaluative triggers, associated attitudes and behavioural consequences. Trust is, for example, associated with confidence, whereas distrust is associated with insecurity and mistrust is associated with caution (see Table 1 in the paper for an overview).

My initial thought upon reading about these different notions of trust was one of trust in their empirical relevance (it is quite obvious that trust, mistrust and distrust are different things), but the more I thought about it the more mistrusting I became until I started distrusting the trust family. I would even go as far as saying that trust in politicians can be understood as an unidimensional concept where ‘trust’ are the high values on the scale, ‘distrust’ are the low values on the scale – and ‘mistrust’ are between the two.

I am not contesting the fact that it is possible to conduct a factor analysis that will return some Eigenvalues that are good enough (i.e., above 1) to make you sleep well at night, but this is not sufficient if you want to develop a new scale. Specifically, there are several potential problems and psychometric properties to consider, before you can – with any level of confidence – conclude that you have demonstrated different dimensions of trust.

To illustrate one potential issue, consider the set of survey items designed to measure trust, mistrust and distrust (from Table 2 in the paper):

Branch Question
Trust The government has good intentions
The government understands the needs of my community
Politicians often put the country above their own interests
Most politicians are honest and truthful
In general, the government usually does the right thing
Distrust The government acts unfairly towards people like me
Politicians usually ignore people like me
Politicians don’t respect people like me
Politicians are often incompetent and ineffective
Mistrust People in the government often show poor judgement
It is best to be cautious about trusting the government
Information provided by the government is generally unreliable
In general, politicians are open about their decisions
I am usually cautious about trusting politicians
I am unsure whether to believe most politicians

One of my main concerns here is related to acquiescence bias, i.e., that respondents are more likely to agree than disagree with statements. This is a concern as all survey items with a positive interpretation (good intentions, honest and truthful, etc.) are in one branch of the trust family. Similarly, all of the items in the distrust category are negative (“ignore people”, “incompetent and ineffective”, etc.). My concern is that these items are tapping into the same underlying concept and the differentiating factor is the acquiescence bias.

Actually, there is one survey item in the ‘Mistrust’ branch that has a positive direction, namely “In general, politicians are open about their decisions”. Unsurprisingly, the result of a factor analysis (provided in Table 6 in the paper), shows that this specific item is not loading well with the ‘Mistrust’ family – but the ‘Trust’ family. This is definitely not a good sign.

These issues also show up in other papers using these survey items to measure three branches of trust. First, consider the paper “How trust, mistrust and distrust shape the governance of the COVID-19 crisis“. Here, you can see that “In general, politicians are open about their decisions” shows no factor loading with the ‘Mistrust’ branch – but, again, it fits nicely within the ‘Trust’ branch.

Second, consider the paper “Trust, Mistrust and Distrust: A Gendered Perspective on Meanings and Measurements“. This paper is interested in gender differences related to the three branches of trust. However, one core problem is that, at least for men, there is no empirical evidence for three dimensions. Or as the authors write in the paper: “Surprisingly, however, the men’s EFA identifies only two latent concepts: trust and distrust, meaning mistrust appears not to exist for male respondents.”

This is not to say that we should not consider and work with different notions of trust, but in order to do so will require much more work on the scale development than I have seen in the literature so far. Of course, I might have missed some instrumental work within the literature, and in that case I hope future studies will do a better job in citing this work.

I do not have access to any of the data used in the papers mentioned above. However, some of the fun empirical questions to ask are whether you can have high trust, high distrust and high mistrust at the same time and whether any discriminant validity tests would confirm that we are working with three distinct branches of trust. I believe such insights are needed before we capture any practical variation that – at the end of the day – improves our understanding of how trust matters for contemporary politics.