PsyFi Search

Wednesday 10 March 2010

Trust Is In the Eye of the Beholder

Is Beauty More Than Skin Deep?

There’s been quite a bit of debate down the years about so-called beauty biases. Various bits of research have suggested that us humans have a bit of a soft spot for other humans who are easy on the eye, to the extent that we tend to attribute abilities to them that they haven’t got. Merely looking attractive is enough to help you get on in life, it seems.

However, there are usually potentially confounding problems with these experiments, in particular the possibility that lovely looking people might actually be better at stuff than the rest of us ugly has-beens. Fortunately a combination of the banking crisis and the proliferation of the internet has offered a glorious opportunity for field experimenters to conduct a bit of real-world research. Even better, it’s suggesting that automating risk management removes a vital layer of protection for lenders and that scammers can play on these traits to defraud us. So are the beautiful also brighter?

Confidence is the Key

Of course, we might look askance at the idea that someone might actually be better at stuff simply because of appearances. However, this might not be quite as stupid a suggestion as it seems. Although it does seem unlikely that the handsome amongst us are born smarter, at least on the evidence of celebrity reality shows, there’s a distinct possibility that they could learn to be so. There’s rather a lot of research that shows that attractive looking children tend to receive more praise, are forgiven their transgressions more easily and get more opportunities. Unsurprisingly they tend to be more confident in their abilities, and confidence often begets its own rewards.

As most of us know, the right sort of encouragement can work wonders in getting us to work harder, while being continually discouraged and ignored is likely to create people who don’t put as much effort in. So, strange as it may be, it’s at least possible that more attractive people are better at intellectual stuff. The problem for researchers is how do you separate these effects – how do you tease apart the possibility that beautiful people are genuinely better at stuff from the idea that they’re given preferential treatment due to an accident of genetics?

The Beauty Premium

The so-called beauty premium was originally demonstrated by Hamermesh and Biddle, who showed that physically attractive people earn upwards of 10% more than their plug-ugly cousins. When Mobius and Rosenblatt in Why Beauty Matters decomposed these results in an experimental situation they both replicated the original findings and came up with a rather precise formula for explaining this wage premium: 20% of it was down to confidence, 40% due to visual perception of supervisors and 40% due to the attractive people having superior oral skills, which is generally regarded as an outcome of greater confidence. However, they also demonstrated that the additional wages earned by the beautiful are actually unjustified – simply looking better doesn’t actually make you better at doing your job. Now there’s a surprise.

One of the problems with these types of experiments is that they’re laboratory based. As we saw in Be A Sceptical Economist unreal experimental scenarios can often trick people into behaving in an abnormal fashion, as their brains try to figure out what the experimenters want them to do rather than just behaving “naturally” – whatever that is.

Although the beauty effect and premium are interesting findings they’re also exactly the kind of daft idea that gets generated from commonsense expectations and exactly the kind of result that you might artificially generate from an unrealistic experiment. To confirm this idea what we really need is a real-world situation that we can mine for data. In particular we need a way of showing that attractive people get an economic reward for looking different and not for being smarter.

Trust, Not Beauty

In Trust and Credit Jefferson Duarte, Stephan Segal and Lance Young found just such a natural setting in which to investigate the beauty effect but went on to discover something slightly different: that us humans are rather good at judging whether or not to trust people based on their looks. What they latched onto were the opportunities for behavioural economic researchers in the development of peer-to-peer lending sites where individuals lend money to other individuals without the overhead of the banking system. They hypothesised that there might be a correlation between the way people appeared in their photographs and the likelihood of them getting access to lending.

The idea that a lender would put their money at risk based on a perception of trust from nothing more than a photograph is, of course, a bizarre and stupid one. Naturally, then, that’s exactly what the researchers found as the people perceived as being more trustworthy received more loan offers at lower interest rates than those perceived otherwise. Of course, this may just be another example of people being stupidly and unconsciously biased but they found otherwise: they found that this perception of trust is genuinely successful at predicting loan defaults even after controlling for stuff like credit scores. As they remark:
“This indicates that borrowers’ photographs offer relevant information about trustworthiness that is not embedded in the standard model used for credit scoring.”
Which suggests, at the very least, that the financial industry’s determination to reduce costs and automate lending decisions comes at a cost: the loss of individual human judgement about the trustworthiness of individuals. If this finding is correct it indicates that there is more to assessing trustworthiness than a credit report.

Trust and Beauty

Clearly, assessing trustworthiness from a photo is not exactly the same as assessing attractiveness. However, there’s obviously some relationship between the two. In fact the researchers also controlled for attractiveness as well as trustworthiness and found that although the two are related attractiveness doesn’t appear to increase the probability of a loan being funded – in fact, if anything, more attractive borrowers seem to have less chance of getting a loan funded than everyone else.

So this field experiment seems to suggest that the beauty premium may be less of a factor than laboratory based experimenters may think, if it's really a factor at all, and that that researchers are getting confused with perceptions of trustworthiness. It also raises some intriguing questions about the nature of trust and our propensity to engage in risky transactions based on our perceptions of it. The peer-to-peer lending data appears to be showing not just that lenders will take risks with their money based purely on their judgement of trustworthiness from photographs alone but also that this judgement is remarkably accurate.

All of which shows that we shouldn't accept every believable theory we hear even when, or perhaps especially when, it seems intuitively correct. Of course, we already knew that removing human judgement from credit scoring was a bad thing, but it's also nice to know that appropriately incentivised individuals can do better than sophisticated computer systems: when it comes to judging people, people are still the best.

Related Articles: The Psychology of Scams, The Halo Effect: What's In A Company Name?, Investors, Embrace Your Feminine Side


  1. Interesting post, as always.

    It makes sense from an evolutionary perspective that trustworthiness = attractiveness, at least at some level.

  2. Which suggests, at the very least, that the financial industry’s determination to reduce costs and automate lending decisions comes at a cost: the loss of individual human judgement about the trustworthiness of individuals.

    I think this is a big deal.

    As we have seen advances in technology, more and more decisions that used to be made by the exercise of human judgment have been taken over by machines. Google doesn't decide what is the best article for a given search term by having a human assess the value of the article -- it uses an algo that can be gamed.

    The machine does a not-bad job. And at far less cost than what it would cost to hire humans to do the same job. But machines make mistakes that even the dumbest humans would never make. Sooner or later, that sort of thing catches up with you.

    We have never tried putting the machines in control before. This is something new!


  3. Fascinating post. The role of intuition is evident and clearly lost in the computer world in which we live.

    Thanks for the info.

    KBK Wealth Connection