An entertaining reckoner on how to look at data and how it cannot be taken for granted
The months since the pandemic began have been a puzzle when interpreting various data on Covid, be it tests, infections, deaths, vaccine efficacy and side effects and so on. Media reports spin a new story everyday, which often evokes the fear element. That is the problem with data and interpretation. Mood swings can be driven by how we see such data that generally works on the negativity instinct. Therefore, you need to “step back and enjoy the view”. This is one of the 10 rules put forward by Tim Harford in another entertaining book on how to look at data or the caution to be exercised when interpreting it.
Presentation of data is not just tricky but also dangerous in some countries, as Harford shows us. In Tanzania criticising official data is a criminal offence. Graciela Bevacqua of Argentina was asked to show low inflation in 2007 and non-compliance meant getting the sack. In 2010 Andreas Georgiou of Greece was forced to show a lower fiscal deficit and when he came up with a number of 15.4%, he was removed from the statistics department. Another principle of Harford here is that “don’t take statistical bedrock for granted”. We in India will feel closer to this rule as all the data on employment, revision in methodology of calculating GDP and so on has ignited similar debate.
The problem with statistics is that it can prove anything. Early on it was found that countries that had high population also had large number of storks and hence the two were linked with causality. Then this was dismissed as being nonsense. The same negative argument was used to say that merely because a large number of smokers were detected with cancer, it did not prove anything. The tobacco lobby now had a strong argument on their side!
It is with these stories and examples that Harford takes us through the pitfalls of drawing conclusions based on data. At times we feel emotionally strong toward something and are plagued by the ‘ostrich effect’. Can one think of the most common example? Yes, it is the stock market. When markets boom, we refuse to believe that there is anything amiss in Covid times and find some rationalisation for the new highs, as experts tell us things will get even better. It is hence quite weird that an entire market can be interpreted in this manner as we are seeing today, to the extent that everyone believes these movements and it becomes self-fulfilling.
Often, we have surveys or experiments that are carried out and then generalised, and this holds for several trials of medical or even electronic products. Or for that matter even unemployment surveys, which are sample-based and then blown up to explain the universe. As evidently the samples cannot be more than half per cent of the population, we need to ask the question “what if someone is left out from this exercise”. This, according to the author, is important as often it is the entire female population that could be left out from the survey or experiment, which is normally the case with any such trials. The same can hold for children or the differently-abled and responses tend to be skewed. While we need not reject the findings, they may not be taken at face value.
Another rule that Harford puts across is: misinformation can be beautiful. This we can see when we look at company or government presentations when charts are shown in a selective manner and worded carefully to show facts without disclosing the truth. Using the right scales and pictures and focusing on a few achievements, the message conveyed can be very different from the true state of the country or company.
Even computers and algorithms can be misleading and there have been cases in the USA where they have supposedly scored over clinical research in terms of predicting the spread of influenza. By using Google to track the number of hits which search for pharmacies or information on the flu, programmes have been able to predict the spread of disease in specific locations. Fascinating as it may sound, these algos have limitation in the sense of not being able to read the mind through such tracking. Hence there have also been instances of false reporting, as a search may be out of curiosity rather than being afflicted by the disease, as has been the case with the Covid pandemic.
At a personal level he warns us to keep an open mind and not be dogmatic on our views on data movements. Here he gives the example of the genius economist, Irving Fisher, who made a lot of money predicting future events but refused to change his view as the Depression set in and lost his entire wealth and became a big debtor! This was the result of being dogmatic in views.
It is true that we live in a world where there is a plethora of data that is being thrown at us to show that things are really great. The language and data point to show that the Indian economy is recovering from the pandemic is not very different from what we saw in the USA before the presidential elections. This is where Harford’s 10 rules can help us sieve the cloud of noise and see the true picture.
That’s why he ends the book by saying that we have to constantly remain curious because often “we think we know which we don’t” and this frailty is used by the purveyor of information to drive home the messages that may not be right.
Madan Sabnavis is chief economist, CARE Ratings
How to Make the World Add Up: Ten Rules for Thinking Differently About Numbers
Pp 338, Rs 699