The research is about political identity and trust. Political identity, and how it affects our trust in the economy, is now a very popular topic in the U.S. and elsewhere. Our approach is an experimental one — in a sense we run an experimental survey on a representative sample of the U.S. and we measure whether the identity of the partners in a transaction that involves trust affects the beliefs of the trustworthiness of people. So, for example, if we know whether the subject is a Democrat or a Republican, we want to see whether they would think a Republican is more trustworthy or less trustworthy than a fellow Democrat. That was our main question.

The paper has many results. I think that the most interesting result is that, all in all, regardless of your partisan identification, people are pessimistic about the trustworthiness of the others. These perceptions of other people’s beliefs are pessimistic compared to the actual behavior of people. For example, in one particular game the decision to trust entailed giving you the chance to choose for me some significant amount of money — which would possibly lead to a higher gain — or giving you the choice to just take the money. So we found that overall, around 80 to 90 percent of the people decided to share the money, to choose the cooperative outcome, but everybody — both Republicans and Democrats — thought that only around 50 to 60 percent would have done this. So there is a 30 percent gap between what people actually did and what people believed. These pessimistic beliefs were an interesting finding, so we ran another experiment with the same population. However, this time, we wanted to see whether the information about the actual high rates of trustworthiness would have an effect on the beliefs on these new individuals. So we told our subjects that the particular individuals they were matched with, whether Republican or a Democrat, cooperated roughly 90 percent of the time. We wanted to see whether the beliefs actually changed with objective information. The fact of the matter is that only Democrats who were matched with another Democrat updated their beliefs and became more optimistic. All the other combinations did not update their beliefs based on objective information. Our explanation for our results is that mostly, especially when you are matched to an outgroup member, you believe that that person was not trustworthy.

One of the implications of these things is, it’s really hard to change beliefs, even with information. Further, the political identity does matter for both your behavior and your beliefs. There are stereotypes that are very strong, and perhaps that’s not that surprising, but what is surprising is that these stereotypes are very hard to change. Even if changing them would have entailed higher gains for both parties.

I did not expect that, but we have to take this result with a grain of salt because it is not statistically significant. Nevertheless, it’s interesting that at least qualitatively there is a little bit more trust and belief in the trustworthiness of Democrats, for both Republicans and Democrats.

There is one theory that I really like and apply to my paper: it is a view that humans are goal-oriented information processors, meaning that we all acquire information based on what we want to make out of that information. In the particular case of the presidential election, basically the example was the stark contrast between Hillary Clinton’s probability of winning, as featured in the New York Times, versus Trump’s, which was 85 versus 5 percent in the last two weeks before the election. So that’s striking. One could say that these estimates have a percentage of error, but another explanation might be that actually we are not willing to go and look for sources that are more objective. Academics, perhaps, are biased towards reading media that use our language, so we ourselves are goal-targeting information processors. We, in a sense, neglected objective information from other sources. And I’m included in that because I tend to trust what the New York Times or other similar media said.

One thing I’m particularly interested is the way that we look for sources of information. We talked about this: we choose where are we going to get our information from, we choose our friends, our professors, we choose which books we are going to read. So we choose where to learn from. That makes us learn less, because if I’m learning from somebody who’s very similar to me, probably this person has the same information that I have. So it only reinforces my previous knowledge. In the ideal case in economics, to maximize efficiency, we would like to have as much information from very diverse sources so that we can use all the information. But we tend to get the information from the sources that are closer to us and that doesn’t allow us to learn much. I’m pursuing some research in that direction to understand that problem. That problem has a name in economics: correlational neglect. And while people have worked on that, I would also like to explore more of this, not only for the population in general but also for academics; we are perhaps also biased towards our own prior knowledge.

Yes, I would agree with that. And it’s not bad per se to get information from people like you, but it would be better if we were aware that we are getting information from sources that we tend to agree with.