Schroders’ The Value Perspective blog carries this article about how one’s personal bias can lead to erroneous judgment.
In End of story, The Value Perspective outlined some of the issues that arise from the tendency of human beings to want to hear information presented in the form of a story. One snag about offering any sort of narrative in an investment case, we argued, is it only highlights one possible version of the future when of course there are so many others that could each play out instead.
This presents the risk of a fund manager or an analyst becoming psychologically attached to a single narrative prediction, which could then cloud their interpretation of new and potentially ‘inconvenient’ facts. All of which brings us to a draft academic paper with the not altogether promising title ofMotivated numeracy and enlightened self-government.
The paper, written by a team of US academics led by Yale Law School professor Dan Kahan, considers how people’s numerical skills can be affected when they are asked questions that conflict with firmly held beliefs. The US is a good place to research this since a number of ‘hot-button’ issues might be said broadly to mark Americans out as either liberal (Democrat) or conservative (Republican).
These “disputed empirical issues”, as Kahan calls them, occupy a conspicuous place in US political debate and would include such questions as whether or not fossil fuels contribute to global warming, whether nuclear waste can be disposed of safely and – the issue that appears to exercise the American psyche more than any other – the desirability of gun control.
Kahan and his team began, however, with the less controversial subject of skin cream. Scientists, the test subjects were told, divided a number of patients suffering from skin rashes into two groups – one that was administered the skin cream and another that was not – and then observed in how many the skin condition improved and in how many it grew worse after a two-week trial period.
In this scenario, of the patients who did use the skin cream, 223 got better and 75 got worse while, of those who did not use it, 107 got better and 21 got worse. Kahan’s test subjects were then asked whether they thought people who used the skin cream were more likely to get better or worse than those who did not use it. Before reading on, what would your answer be?
The correct answers stands on the maths that, of those people who did not use the skin cream, roughly five times as many got better than got worse whereas, of those who did use the skin cream, the ratio was roughly three to one. Human beings are, however, generally poor at judging this sort of question, with many inclined to focus on the biggest number – here, the 223 – rather than the ratios involved.
Thus, even though Kahan’s test subjects were Yale students and presumably very bright, 59% of them supplied the wrong answer, identifying the most important result as the one that was least supported by the information. That said, the more numerate a student was – as judged by an independent numeracy test – the more likely they were to find the right answer.
This remained the case even when Kahan and his team reversed the numbers in the question as a way of counteracting the possibility some students might be harbouring biases, one way or the other, about a cure for rashes. In either case, numeracy was the main driver of students’ responses when the question revolved around the effectiveness of skin cream, with more numerate students providing better answers – but what about the effectiveness of gun control?
Kahan and his team did the test again, with the same numbers, only now they related them to US cities that did or did not ban carrying concealed handguns in public and whether that led to an increase or decrease in crime. This time, the academics found, the students’ numeracy was clearly subordinate to their political leanings – be that liberal or conservative.
So, where the numbers contradicted their view on gun control, the most numerate students were every bit as bad at answering the question correctly as their less numerate fellow test subjects. On the other hand, where the numbers served to endorse their view on gun control – for example, the liberal students believing gun control reduced crime – the answers were almost 100% correct.
Kahan calls this “identity protective cognition thesis”, which is a self-sabotage of cognitive ability where it conflicts with a deeply-held belief. While that may not roll off the tongue, it should not prevent us from recognising the very real risk presented by a narrative. Say, for example, rather than having a committed belief in the right to bear arms, you have identified yourself as a raging bull on some US tech stock, the fact such a bias could lead you to make mistakes when analysing fresh data on that business does not bode well for the success of your portfolio.