Citizen Beta #23: statistics, fact checking and data journalism

It’s taken me a couple of days to recover from the brilliant Citizen Beta we’ve just had. So much to process!

We were joined by Andy Dudfield from the Office for National Statistcs, Simon Rogers from Google, John Walton from the BBC, and Lev and I showed off some recent Full Fact work.

The audience was made up of journalists, technologists, civil servants and all round amazing folk.

Citizen Beta happens once a month. After a year long hiatus it’s back with two new organisers: Ira Bolychevsky and Imeh Akpan. Check out the latest events.

This month was all about statistics, fact checking and data journalism.

There are some numbers which mean more than others. GDP rising, prices falling, the average cost of a house. Numbers tell us stories about the lives people are leading, whether they have jobs, how much they are earning and how that’s changed year on year.

Access to these numbers governs how we understand and tell those stories, and who we hold accountable for the choices that have lead us there.

The Office for National Statistics has done a spectacular job in recent years of moving us to a position where we can do more.

Andy Dudfield, Chief Publishing Officer, walked us through some of those changes—a move away from spreadsheets that give you the right table, to an API that can give you the right number. This lets you ask more granular and complex questions.

Lev Konstantinovskiy, Full Fact’s NLP engineer, and I decided to premiere a part of our automated fact checking work that we call “robochecking”. It takes a spoken sentence like “unemployment has risen by 5% since 2016” and generates a graph in real-time, using ONS data.

Imagine a world where a journalist in a press conference can push back in real time when bad stats are used. We know that cutting down the time it takes to respond to misinformation makes a real difference. That’s the world we’re trying to create, but we can only do that when the data is accessible in real-time.

The ONS numbers also support the downstream task of data journalism. John, data editor at the BBC, talked us through some of their projects and their newly found love of R—a statistical programming language.

We were also joined by Simon Rogers, Data editor at Google, who called in from San Francisco. He spoke through the new dataset search which is run off schema.org

Then we topped off the night with a tribute to the heroes who don’t always get the awards, or the start-up bucks, but who continue to work incredibly difficult jobs because they know it will affect the lives of millions.

Taking an institution that is hard to manoeuvre and making incremental improvements from the the inside is no easy feat.

We owe a lot to Laura Dewis, Matt Jukes, Rob Chambers and Andy Dudfield, who have been fighting our corner at the ONS for years and years.

I wanted them to know that we see the work they do and appreciate it. So what better way to thank them than with an impromptu Mexican wave?

Speakers

Andrew Dudfield, Chief Publishing Officer, Office for National Statistics
Andy is responsible for the running and ongoing development of all Office for National Statistics websites. He has played a significant role in organisational wide service design and has been instrumental in creating the culture for a radically open team.

Mevan Babakar, Head of Automated Fact checking, Full Fact
Mevan has worked at the UK’s leading independent fact checking charity, Full Fact, since 2014. There she leads a tech team that aims to build tools to scale the work of factcheckers around the world. She’ll be sharing how they use ONS data to check statistics against primary data, in real-time.

Lev Konstantinovskiy, NLP Engineer, Full Fact
Lev sits in the automating fact checking team at Full Fact, where he focuses on Natural Language Processing. Formerly he was a maintainer of Gensim, a Python open-source machine learning package.

Simon Rogers, Data Editor, Google
Simon is a data journalist, writer and speaker. He is the author of ‘Facts are Sacred’, published by Faber & Faber. He is the data editor at Google in California and formerly at Twitter, San Francisco. He is director of the Data Journalism Awards 2015.

John Walton, Data Journalism Editor, BBC
John Walton is the data journalism editor in the BBC’s Visual & Data Journalism team. John leads a team of award-winning data journalists, whose recent projects have covered the health service, crime, house prices and life expectancy . In 2017 John contributed a chapter to the book, Data Journalism: Past, Present and Future.

Next Citizen Beta

Citizen Beta happens once a month at Newspeak House, often at the end of the month. Sign up to the mailing list to be the first to hear about the next one.