Data disruptor: look beyond the numbers at the probabilities
3 March 2020: vast information streams engulf just about every business sector but, armed with algorithmic tools, professional statistician Nigel Marriott is akin to a modern-day fortune teller. He tells Penny Sukhraj what statistics can bring to business.
Marriott’s unique skill set allows him to delve into hot topics, driving the agenda for politicians and companies alike. His work as a professional statistician encompasses a vast variety of subjects, from gender pay to weather, from sport to politics.
He also finds time to champion access for those with disabilities as a former Trustee of SENSE, the world’s largest deafblind charity. Most recently, Marriott also predicted the election result with more accuracy than the exit polls.
Marriott always knew he wanted to work with numbers, but instead of becoming an aeronautical engineer – his initial desire – he found himself drawn to the world of statistics. He worked for two multinational organisations, ED&F Man Ltd and Mars Inc, in statistics-led roles, before breaking off to set up Marriott Statistical Consulting Ltd in 2006.
Ahead of presenting an ICAEW Academy course on statistical data analysis, Marriott spoke with ICAEW Insights about his work in a number of different fields, and why in business it’s important to be able to estimate the probabilities of things going wrong.
Disconnect in climate change management
Climate change has just about every sector in a spin over how to calculate, prevent or get ahead of it. But for Marriott, it’s about considering the “disconnect between what you hear in the news and what policymakers can actually do.”
The real question is whether we can manage. “You’ve got to start looking at whether you would cope in different scenarios. What climate change is about, is the probability of certain scenarios being more likely than others.”
And thinking locally is key too. Bath, for example, wouldn’t be considered a high risk city for flooding. “But what would be catastrophic would be a landslide because the city is built on the side of a valley, and if the climate got a lot wetter, that could raise the risk of a landslide.”
What’s really needed to close the gender pay gap?
Regarding gender pay, Marriott says that many have made “incorrect claims” about what a gender pay gap indicates. While aware that objectors to the current system can come across as “a pedant or a negative voice”, he insists the problem is that the official measure – the difference between the median man and the median woman – is just a single number. “That single number does not capture the full nuance of what is going on,” Marriott says.
People in HR are not statisticians, he points out. “The government required HR professionals to be statisticians. This had the potential to be a big problem.” For Marriott, it presented an opportunity. “It meant working with HR, doing analysis for their teams and companies and analysing data to understand why they have a pay gap, and what’s needed to close it.”
Case of trial and error
Marriott recently shot to prominence when his predictions for last year’s general election ended up being more accurate than the BBC’s own exit polls. However, he reveals this was as a result of trial and error.
“Two years ago, I was so badly out – but I didn’t give up. I looked at why my model got it so wrong and figured that out, and I tried to learn from it.”
He also points out that bad luck happens too in the world of statistics – take the Rugby World Cup. “I got it all right until the final. South Africa won and their victory was larger than expected from the data. But that happens sometimes. It’s not a bad point, but it’s important … to know what to do to cope when things go wrong.
“Sometimes businesses don’t realise they’re at risk of things going wrong so it’s important to be able to estimate the probabilities of things going wrong.”
Nigel Marriott is presenting a one-day ICAEW Academy course. It introduces delegates to statistics and analyses and takes them through the six key concepts that underlie all statistical thinking and analysis. For more information click here or email firstname.lastname@example.org.