Conversations with Data: #53
Do you want to receive Conversations with Data? Subscribe
Analysing and communicating uncertainty in data has never been more important for journalists. But what is the best way to interrogate data and deem it trustworthy?
To help us better understand this, we spoke with Professor Denise Lievesley from the University of Oxford in our latest Conversations with Data podcast. As an experienced statistician in government and academia, she discusses what data journalists can learn from statisticians and the parallels between the two professions.
What we asked
You've had an impressive career across academia and government. Tell us about your background.
I would describe myself as an applied social statistician. I've worked in a variety of jobs as a statistician, including as the director of statistics for UNESCO. My role as an applied statistician involved collecting data for the Millennium Development Goals and monitoring progress across the world. I was also the chief executive of NHS Digital (formerly the NHS Health and Social Care Information Centre). My last two jobs have been in academia. I was the dean of social sciences at King's College London. Currently, I'm principal of Green Templeton College, a specialist graduate college of the University of Oxford where the students study medicine or applied social science.
Let's start with the basics. How do you define statistics?
Statistics are described as the science of uncertainty. The difficulty is that statistics has two meanings. It relates to the numbers themselves but also to the science, which is about the understanding of the numbers. One way of thinking about statistics is that it's about understanding why patterns occur and whether they happen by chance or by some other external factor. Statisticians help you understand that uncertainty and draw conclusions from the data.
Tell us about your time at UNESCO and what it was like to set up their Institute for Statistics.
I joined UNESCO in 1999 when the UN agency had a division of statistics. It was viewed as a service to the organisation rather than a profession in its own right, providing external services to the world. I was recruited to dissolve the division of statistics and to set up a new institute. And that is indeed what I did. The UNESCO Institute for Statistics still exists today. Based in Montreal, it is still a jewel in the crown of UNESCO. It collects data and sets the standards for collecting data and works on statistical capacity building concerning education, science, technology, culture, and communications.
How can journalists decide if they can trust data?
There are two ways. The first is by looking at the science that has underpinned the data. How were they collected? Are they likely to be representative? How up to date are they? What is known about the error in the data? These are questions journalists are used to asking when validating if a story is true.
The second way to determine its trustworthiness is to understand why the data was collected in the first place. How was the agenda set for the collection of the data? What are the incentives to report in a particular way? What happens to the government statistician in a country if they produce data that is unpopular? Understanding the job security and the support systems for statisticians is essential, too. Like journalists, statisticians may face pressure to report good news.
In the wake of the death of George Floyd, how can statistics help bring a voice to society's most vulnerable?
One of the reasons why I'm a statistician is because I think it gives a voice to the marginalised in our societies. The challenges in doing this is that very often those marginalised people are missing from the data. Often they don't trust the government well enough to be prepared to participate and respond in our studies. The participation rate is an important aspect of representativeness in data collection. It is critical that our data reflects inequalities within our societies. We haven't always found it easy to collect high-quality data on really sensitive issues.
One of the things that I take away from the current anguish experienced by a part of the population in the United States is that they haven't had visibility or a voice. Statistics are an important part of giving them that voice. But we have to learn ways in which we can help them. I don't think we've got enough black and ethnic minority staff in our statistical offices to build that understanding.
In what situations would timely data be more useful than data released later on?
As with COVID-19, there are occasions where timely data is absolutely critical in a changing situation. In such scenarios, it may be better to have less than perfect data fast. There are other cases where the data from last week or from last month makes very little difference and it would be better to have data that has been through some greater checks. That could mean more time for reflection or good data analysis. For COVID-19, we need data fast, but we'll also want better data later in order to do a post-mortem. So I think the message for a journalist is that the latest data is not always better.
What can journalists do to encourage governments to collect better quality data?
There are two things they can do. First of all, I would like to see more journalists get involved in the consultations that take place about what data ought to be collected and what is collected in the first place. One of the problems we've got in many countries of the world is that the agenda about what is collected is only settled by governments. The other thing that journalists can do is push hard for data to be published. Almost all countries have signed up to the UN declaration on statistical ethics. However, not all countries abide by it.
Latest on DataJournalism.com
Journalist Sherry Ricchiardi explores how reporters can tell stories about the impact COVID-19 is having on society's most marginalised and vulnerable groups. From The New York Times to the Associated Press and The Guardian, our latest long read article cites useful resources and examples of data-led storytelling.
Our next conversation
In the next episode of our Conversations with Data podcast, we'll be speaking with Leonardo Milano, the predictive analytics team lead at the United Nations OCHA Centre for Humanitarian Data. The discussion will focus on how predictive data supports the humanitarian community and their response to future crises.
As always, don’t forget to let us know what you’d like us to feature in our future editions. You can also read all of our past editions here.
Tara from the EJC Data team,
P.S. Are you interested in supporting this newsletter as well? Get in touch to discuss sponsorship opportunities.
If you experience any other problems, feel free to contact us at email@example.com