Questioning everything

If you know me, you’ll know I love a good rant. One of my many bugbears is the perception of statistics, mathematics, and just numbers in general, in the wider world. I’ve worked with data and numbers for most of my life, in many guises – from a data scientist, to a Masters student, to a tutor for six-year-olds.

The most common problem, which is so common it is a well-known joke, is: “I was never any good with numbers, there’s just no point trying with me!” So many people I come across instantly shy away from anything vaguely representing numbers, claiming they’re too hard to understand. It’s a real problem, and it needs fixing.

Enter The Tiger That Isn’t: Seeing Through a World of Numbers by Michael Blastland and Andrew Dilnot. This book has been recommended to me time and time again, and I can agree it’s worth the hype. Its premise is that ‘we all know more than we think we do’, and that with a little common sense, numbers aren’t scary if you follow some simple rules to help understand them.

I’ve taken a few of my favourite rules and outlined these below, but more detail, finesse and humour can be found in Blastand’s and Dilnot’s read.

Define your peas

Counting seems simple when you are counting things that are simple to count. But, in real-world examples, we aren’t counting simple, whole, discrete things, and are in fact counting mushy peas rather than your spherical garden variety. More often than not, we’re taking things that are difficult to define, and are squishing them into neat rectangular boxes so they can be counted.

Definitions can be difficult. Thatcher’s government, for example, changed the definition of unemployment 23 times. At first glance, you might think defining unemployment would be easy – do you have a job? Yes or no. But what constitutes a job? A full work week? How long is a full work week? What if you volunteer? What if you’re on an unpaid internship? What if you’re on furlough, or have just been made redundant, or work in live music and don’t have a contract?

If it has been counted, it has been defined. Check your definitions and know what you’re working with.

Size matters

Quite often, especially in the public domain and in the media world, numbers with a lot of zeros are thrown about without much of an explanation as to what those zeros actually mean. One of the first questions for anyone when confronted with a number should always be ‘and is that a big number?’ To answer that question, we need to know more about that number.

Numbers relating to public spend, government figures or populations will likely always look big. But are they actually big? Divide the spend over the number of weeks or days, and the number of people. It might not seem so big now. Six may seem like a relatively small number, unless we’re counting the number of times someone has been diagnosed with cancer – in which case, it suddenly seems quite terrifying.

Blastland and Dilnot sum it up nicely: “Our default position should be that no number, not a single one of them, is big or small until we know more about it.”

Averages are not typical

The definition of an average, according to Google, is: ‘a number expressing the central or typical value in a set of data.’ While this may technically be correct because of the ‘or’ in the middle, it doesn’t help the widespread notion that an average gives a good indication of the wider dataset. Most of the time, it’s quite the opposite.

My nana, a proud, patriotic Welsh woman, frequently jokes: “On average, Wales is bigger than England.” Whether or not this is true, I don’t know. But the idea is an important one. Taking the average of a group of numbers hides the ups and downs (in this case the glorious Welsh mountain ranges) and makes out like everything is flat. Life is not flat. Wales, I can guarantee you, is not flat.

On average, the colour of a rainbow is white. However, white does not portray the ‘typical’ colour seen in a rainbow. In fact, especially when dealing with areas such as wealth and populations amongst others, there is so much variation and so many people at each end of the scale, that the average ‘typical’ value is anything but typical.

When presented with an average, think about what the data is showing you. More importantly, think about what the data is not showing you.

While I don’t have the time here to go through each of the notions presented in the book, I do recommend you look it up and see for yourself. However, these rules all have one idea in common – never take anything at face value. Question everything. Relate numbers back to something more understandable to see the entire picture. Think about where the number has come from, who created it, what it’s based on, what it’s showing you, and what it isn’t.

Too often, the media, the government, and even Bob-from-down-the-road throw numbers around, never expecting anyone to dig any deeper into them. Develop your digging skills and trust your instincts. It’s amazing what you can uncover when you dig.

Bethan Blakeley is analytics director at Boxclever