The Puget News Logo Banner

Examine the face to learn whether somebody is credit-worthy

by Eric Franklin on April 22, 2009

I’ve been interviewing applicants for a house rental for the past couple of weeks and have believed in most cases that I could spot whether somebody was trustworthy or not with some reliability, and within moments of meeting them. Of course, I then walk away and second-guess myself the rest of the time. So far though, I’d say my snap decisions about whether somebody is going to turn in the application like they said they would, or about what I would find on somebody’s credit report, have turned out to be fairly inaccurate.

According to a study by Jefferson Duarte of Rice University (and written up in The Economist), facial aesthetics and cues really can (and do) provide a window into whether somebody is deserving of a loan. Using the Mechanical Turk program at Amazon.com – a feature which allows anyone to “hire” workers for small tasks (such as rating photos for credit-worthiness):

The researchers looked at 6,821 loan applications, 733 of which were successful. Their first finding was that the assessments of trustworthiness, and of likelihood to repay a loan, that were made by Mechanical Turk workers did indeed correlate with potential borrowers’ credit ratings based on their credit history. That continued to be so when the other variables, from beauty to race to obesity, were controlled for statistically. Shifty physiognomy, it seems, is independent of these things.

Obviously, personal biases can weigh on what one person views as shifty and another views as normal, but the numbers don’t lie, there’s something primal about our visual cues and we rely on them because they tend to work. Accuracy flies out the window on an individual basis when I’m looking at people for my house or if you’re the former president who said:

“I looked the man in the eye. I was able to get a sense of his soul.” – George W. Bush upon meeting Vladimir Putin for the first time

Leave a Comment

Previous post:

Next post: