Synopsis: Two stats geeks methodically unpick common statistical misrepresentations while giving readers a tool kit to allow them to test statistical claims that they come across themselves.
My Take: “The Tiger That Isn’t” really should be compulsory reading for anyone involved in public policy – advisers, politicians, activists and in particular, journalists. In their capacity as hosts of the BBC Radio 4 series “More or Less” Blastland and Dilnot have applied themselves to debunking statistical misrepresentations in the public debate for many years. This book is a lightweight and accessible distillation of the most common lessons of the show and includes a deceptively useful section on estimation techniques/ball park evaluations that I now use quite frequently (the value of questions as simple as “Is that a big number?” or “What exactly is being counted?” can be quite surprising).
Highlight: From a personal perspective, one of the greatest services that this book carries out is the debunking of the Average Wage myth. Anyone with a rudimentary understanding of statistics know that averages, while a useful summary statistic, can be appallingly misleading if you are trying to measure some form of central tendency in a group. All it takes is a handful of outliers to render the statistic basically useless for finding the ‘middle’ of any population.
One area in which this basic short-coming of the use of averages is routinely, flagrantly disregarded is in the analysis of national wage data. Lazy or ignorant journalists and deceptive politicians frequently cite the ‘Average wage’ or ‘Average Household Income’ as though this is a useful figure in some way. However, as Blastland and Dilnot vigorously point out, the presence of a handful of extreme outliers at the top end of the income scale inevitably skews average wage data dramatically above that of a ‘middle’ earner. As the old joke says, Bill Gates walks into a bar and everyone became a millionaire – on average. As such, as Blastland and Dilnot point out, a much better measure of central tendency for most policy discussions is the population’s median ie the middle value of an ordered set of values (or the average of the middle two in a set with an even number).
In Australia, while the average wage is frequently cited as being around $55,000 for an individual, as Andrew Leigh has pointed out, the median wage is around half this:
As you can see from Andrew’s invaluable chart, someone earning $55,000 would actually be at around the 80th percentile of income earners ie earning more than 80% of workers. These figures come as a shock to most people when they first hear them – so ingrained is this statistical misconception in the public debate.
This misleading use of average wage data distorts public debate, by leaving punters under the impression that the ‘average Australian’ is doing substantially better than an economically ‘middle class’ Australian actually is. Similarly, it buttresses the provision of (the misleadingly named) middle class welfare by government by creating the impression that the quite well off are amongst the bulk of Australians. It really is a malignant feature of the public debate.
This misrepresentation continues to be perpetuated for a number of reasons. Some journalists are ignorant of basic statistics. Other journalists resort to using average wage data when they discover how difficult it is to locate up to date median income data. Finally, the political elite have little incentive to question the data as inflated average wage figures bring their relatively incomes closer towards the fabled ‘middle class’ to which most Australians aspire to attribute themselves. In this context, ooks like “The Tiger That Isn’t” that swim against the stream and highlight this popular misconception are performing an important service to the public debate.