Numbers Matter

We live in a dark age.

It is an age where truth doesn't matter and where learning is dismissed as elitism, a world where people latch onto the words of demagogues who promise them authenticity in a world which has none. It is an age - yet another age - where we persecute groups for political power and prestige.

We are on the edge of something terrible and unspoken.

I am an optimist. I believe we can pull ourselves back from this edge, if we have radical political overhaul and evidence-based policy. For the latter part at least, we will need numbers.

In the age where people have had enough of experts, it's going to be difficult to get people to trust numbers. In many ways, mathematical manipulation represents everything popularly presented as elitism: mathematics is abstruse and frequently inscrutable to those who haven't spent several years studying its ways. Arguments made from mathematics are dry and emotionally unavailable. Economics (and crashing the economy) is based on incredibly fancy mathematical models with somewhat questionable applicability. It's no wonder that evidence-based policy, with its reams and reams of data, is getting utterly trashed by populist appeals to our psychology.

Traditionally, the response to this has been utterly ineffective. The political class have utterly ignored mathematics in favour of ideology. Scientists of all stripes do attempt to combat this, but they have their own research to do too. Besides, outreach is difficult and scientists are as fallible as the rest of us once they venture outside their specialty. Some people have even been outright dismissive, because pissing off a large group of people and driving them into the arms of demagogues is definitely the best thing to do.

I could attempt to defend the value of numbers with more numbers. It probably wouldn't convince people who are already disdainful of numbers. I could shout at people who dislike numbers. That would only convince them that numbers are for elitist snobs. So I'm going to do my best to defend numbers with a heartfelt appeal to emotion. Maybe others can do better - and I'd like to see them try - but I intend to lay the groundwork for such an approach.

I'm going to start by conceding a point: on their own, numbers don't necessarily matter to the average person on the street. What are they going to do with a 2.57 or a 4.6? But outside of certain branches of mathematics, numbers exist as a representation of what's going on, a way of expressing what's going on down the street in terms of what's going on in the wider world.

They are important because of what they represent.

I'm going to make a point now that's so straightforward I hope it doesn't come off as condescending. Numbers are important because they represent you, and me, and your family, and your friends, and your neighbours, and your communities, and your country, and the world. Each number is a representation of us. Each rise or fall in a number reflects real people being affected, not just some nebulous elite class.

Numbers also matter because of their scope. When you look at what's going on, you can see what's going on in an extremely limited area around you. You can see the people you interact with, and maybe the people those people interact with. Beyond that, it's difficult to see very far. Having numbers is like having a view from everywhere at once, because it allows you to see so much more. Think about how the world seems so big when you walk everywhere, and how much it shrinks when you drive or fly. That's what using statistics feels like. It feels like being able to see everything all at the same time...

...well, not quite everything. We use numbers as a representation of many things because numbers are easy to work with. There are problems with this.

Firstly, not everything can be quantified - how much you love your partner, how much you hate your job, how happy you are...sure, you can try to put it on a scale from 1 to 5, but there's always going to be some ambiguity in how different people interpret those scales. I love numbers, but I don't believe everything should be quantified.

Secondly, quantification is fallible. The numbers you get out depend on the numbers you put in - and it's easy to skew the numbers you put in. You can cherry-pick your sample to be small, or disproportionately from one demographic (over 60, or from Lydd, for example...). Those are just two examples. You can also report relative risk (relative jump in a percentage) as opposed to absolute risk. You can play around with your results until you demonstrate something of statistical significance for one particular group. Though numbers are a powerful representation, they're often distorted. I would consider myself as bad as any demagogue if I didn't openly acknowledge this.

So why care about numbers and quantification in the first place if it's so easy to distort data?

It's an important question. I'd be doing you a disservice to gloss it over.

My answer is idealistic: because just as it's possible to distort data, it's possible to be aware of possible biases and distortions and design metrics accordingly. It's possible to critically evaluate statistics and decide whether they're reliable or not. It's possible to judge what is a fair representation and what is not without politicians and the media getting in the way. It really is, with mathematical training, but also to a certain extent with experience. I particularly recommend Ben Goldacre for all things to do with bad and heavily manipulated data, and Respectful Insolence. For hard mathematical skills S1 and S2 statistics (and integration) are needed, but speaking from experience, most of understanding statistics is about experience and getting a "feel" for the data. That probably accounts for why numbers are seen as so abstruse and remote - most people will never touch an integral. If you're most people in the UK, you probably dropped maths after GCSE.

I appreciate that most people simply will not have the time to learn about calculus and data handling - but a good way to get a feel for how statistics work and why numbers are important is to read Goldacre or Respectful Insolence, or many of the other science advocacy and skeptic blogs out there. Because numbers are important and they do matter, both as a powerful representation of our world and as a tool which is often misused.