Simon asks for a sense of proportion about the allegations of Hillary Clinton’s misuse of emails:
The media obsesses over whether Clinton might have sent an email containing confidential information from her personal account while secretary of state, and also wonders about whether Trump tells lies, pays any taxes, bribes officials and assaults women. Anyone who reads these stories can see that there is no equivalence here. But anyone who just reads the headlines would be tempted to think otherwise.
There might be a more widespread cognitive bias at work here. We can call it the counting heuristic – a tendency to judge things by frequency rather than impact. Here are three examples:
- A lovely paper by Michael Ungehauer and Martin Weber shows that people evaluate correlations between asset returns simply by looking at the frequency of co-movement. This can lead them to under-estimate correlations by putting too little weight upon a few large moves. But in risk management it is these moves that matter: a stock that falls only slightly when markets slump is worth having. Neglecting this causes defensive stocks to be under-priced.
- In the mid-00s, a long run of stable returns led banks to presume that mortgage derivatives were relatively safe. They therefore under-estimated tail risk – the small chance of disaster. Investors and corporate managers can under-estimate the importance of low-frequency, high-impact events in non-financial firms too.
- The peak-end rule tells us that we look back more fondly upon a nasty experience if it is followed by discomfort than if it suddenly ends. For example, colonoscopy patients who had the scope left in for a few minutes at the end of the procedure evaluated their experience as less unpleasant than those who had it immediately removed. In this way, the duration of an experience can mislead us as to its impact.
These experiences all imply that a steady drip-drip of minor allegations such as those against Clinton can weigh more heavily in people’s minds than more serious allegations – just as the drip-drip of small co-movements distracts people from important large co-movements, or frequent small returns distract people from the risk of big losses, or a frequent discomfort can cause them to misremember real pain.
One simple question should help us combat the bias generated by the counting heuristic. Just as the first question you must ask of any statistic is: is that a lot or a little? So we must always ask of any news story: so what?
Size matters. For example, the countless stories about Clinton’s misuse of emails (which might be exaggerated by the error of correlation neglect) pose the question: is a neglect of bureaucratic protocols really more serious than tax-dodging and sexual assault?
This over-weighting of frequency and under-weighting of impact has also infected UK politics. Here are four examples:
- In the 2015 election campaign there were numerous stories about the “cost” of Labour’s policies. Such stories generally failed to point out that these costs were much smaller than the cost of Tory austerity.
- Tory attacks on the Bank of England pose (but do not answer) the question: even if we grant that there have been monetary policy errors, how big are these relative to the cost of fiscal policy errors or Brexit?
- The tabloids’ fuss about some migrants falsely claiming to be children raises the question: so what if we let into the country a few 20-something refugees?
My point here should be a trivial one – that quantification matters. However, the interaction of a vicious media with cognitive biases means that even trivial points get neglected in politics.