Finding ‘good’ numbers: my secondment to the BBC 

When you start a new job, you expect to have plenty of time to settle in. That was not the case when I started my secondment at the BBC. Parliament was just back from recess with the announcement that schools across the country were closed because of safety risks around reinforced autoclaved aerated concrete (RAAC). Within ten minutes of getting my laptop, I was asked to help find information on Government spending on school maintenance over the last 25 years. Quite an ask for day one – especially when Robert Cuffe, the Head of Statistics at the BBC, is presenting the figures on TV that afternoon. 

I probably should have taken this as a sign of what was to come over the next 6 months. The breadth of topics I covered was mind-blowing. In one morning, you could be asked questions on 8 or more different topics ranging from potholes to prisons, and productivity to pest control. 

Finding ‘good’ numbers 

I learnt quite quickly that there’s a number that could be included in every story. The issue is which of these numbers are ‘good’ ones. The role of the secondees is to make sure those numbers are appropriate and correctly represented. How do you do this? Next time you are given a number try thinking about these questions: 

• What is included in the numbers? 
• What is not included in the numbers?

When someone asks you about a statistic, it’s always good to know the definition. It’s a useful starting point. But the next best question to ask is what you’re not seeing too. 

Did you know a prisoner only ‘escapes’ if they’re out of prison for more than 15 minutes? Or that the total crime estimate from the Crime Survey for England and Wales excludes the theft of milk bottle from doorways? Or, in a more potentially detrimental way, that the total crime estimate does not include some sexual offences?

When you use using statistics to explain what’s going on, you need to know which aspect of an issue you are highlighting. You should then consider the effect of this. 

Understanding the context

When there’s a pattern in the data, it’s important to try to understand the context behind the change. Have there been policy changes? Is there an external factor at play? 

Small boat migration was, understandably, often a topic of discussion in the BBC Verify Newsroom. In February 2024, the BBC got some freedom of information (FOI) data on the weather in the English Channel over the last few years. I analysed this data to explore if the weather in 2022 was one of the reasons for fewer small boat crossings. The piece, rightly, received a lot of scrutiny. There were requests for the analysis to go further and show the proof that the weather had affected the number of crossings. In this scenario, it was important to recognise when there is enough information to show a connection, but that you cannot reasonably take the analysis further. 

In probably my proudest moment, this made it into an article on the front page and a TV piece. It was fascinating to see this process start to finish and definitely challenged me. 

Knowing what you can and cannot say

It’s a skill to know what you can say from highlighting the source and describing the trend. 

There was one debate about whether we could state the number of reported ticket scams had ‘surged’. This led to an interesting discussion about trying to define a surge. Although we concluded the increase had not been large enough to warrant the use of the word ‘surge’, it really made you think about the translation of these more generic descriptions, especially for the more general audience. 

The other skill though, as I have mentioned, is knowing when you cannot say things. Sometimes the numbers alone are not enough. Sometimes you need a hard-hitting case study on homelessness to demonstrate the scale of the numbers, or a well-designed chart to explain the intricate financial forecasts in the Autumn Statement. 

Summing it up in one sentence

So, you have a good idea of the number and how ‘good’ it is. Now you need to, get everything into one sentence: the top line to tell the whole story.  

There will always be more you can say, but you must decide what is actually needed. Alongside this there’s probably: 

• a journalist who wants to make your sentence shorter

• the consideration of whether the content is going on TV, radio, or whether it will become a written argument

After I had attempted to write a live page post post I was once told “you write like a statistician”. To be honest, I didn’t think that was a bad thing, but it helped illustrate demonstrate the point perfectly. As statisticians, we do tend to waffle a bit! And when you only have 120 words for a live page post, every single one matters. 

So, in the interest of being concise, I will sum up my reflections on my secondment in one sentence: I’m never taking a number on face value again and I’m absolutely fine with that.