Making sense of statistics by studying Covid-19 data

MUCH as many people make a habit of checking the weather forecast every day, some have become used to looking at daily updates of Covid-19 statistics. The Worldometers website is probably the most visited by those who are looking for information on a truly global scale.

One thing that has struck me in visiting this site is its potential value in assisting people of all ages to gain a better understanding of how to make sense of statistics.

It can illustrate the value or otherwise of using percentages to show trends. An increase of 20 per cent, whether it is in production of goods or the incidence of infection, sounds big, but its significance depends on the base from which the increase took place: a relatively small change from a low base figure may mean very little in reality, while appearing impressive or ominous in percentage terms, whereas it would require a huge numerical change to produce a 20 per cent change from a far larger base. An extreme illustration of this point occurred in the novel coronavirus statistics one day in September, when Singapore experienced a rise of 25 per cent in its number of new infections in a single day - from four to five.

Figures such as these clearly show why percentages should rarely, if ever, be used to describe changes taking place on a base of less than 100 where people are concerned (in polls and most academic fields, the figure would be much higher) and why the base should be much bigger when the production of goods is concerned.

Another good learning point is about the difference between absolute figures and those that are put into a context that allows their meaning to be understood in terms of relative impact. This can be seen by considering the infection and death rate figures for the individual US states.

For example, California has the highest infection numbers of any US state, with 1,617,823 cases on Dec 14, but then, it is also the most populous state, with 39,512,000 people. Its rate of infection per million people puts that figure in perspective: California ranks 40th among the states by that measure.

USING STATISTICS

Other figures can offer a basis for discussing how states compile and share information: at the beginning of the outbreak, some that had a heavy reliance on tourism reported suspiciously few cases at a time when tourists returning to their home countries were reported to have caught the infection there.

Other statistics may also stimulate discussion: for example, when countries are reported to have hundreds of new infections daily and dozens or more deaths, but are said to have few or no "serious, critical cases", what does it mean? Presumably, many of those who died were in a "serious" or "critical" condition for at least a day or two before their demise: do national statistics simply not reflect this or might the countries concerned lack facilities to provide care for many seriously ill people, and they choose not to record them?

Statistics are open to both misunderstanding and abuse. They can be used selectively or in a misleading way to score a political point, and to play on people's anxieties, including to sell them products they don't need or to dissuade them from having medical treatment that they really do need.

An understanding of how to make sense of them does not come naturally, but, as it can be an asset in many spheres of human life, it pays to learn.

At a time when political leaders and the media in various countries are given to making factually unsupported claims to which all too many people are receptive, knowing how to make sense of statistics can be of particular value.