Over 600 statisticians, data scientists and analysts gathered in Belfast to share findings, exchange ideas, and build friendships.
This was my second time at the conference, and it was another swift learning experience.
(As I recall, this was my first time being used in promotional material.)
- The Mirror and the Lamp: Statistics are a mirror on society, as well as a lamp for illumination.
- Rapid-fire: I gave a five-minute talk on the purported curse of Aaron Ramsey.
- Getting involved: The conference gave many opportunities for getting involved with the wider statistical community.
The Mirror and the Lamp
After rushing through an airport, I managed to arrive on time in the ICC Belfast. The first keynote presentation was by Siobhan Carey, the Chief Executive of Northern Ireland Statistics and Research Agency. Called ‘The Mirror and the Lamp’, the talk went through the usage and value of statistics in Northern Ireland, including for drawing the Irish border.
We were reminded that statisticians may have an image problem, with word clouds from selected audiences.
On Tuesday morning, the Young Statisticians Section gave their guide to the conference from some seasoned survivors. The need to eat and sleep was restated.
The first parallel talk I attended was on quality and value in official statistics. Methods were discussed on how to estimate the value of EU trade by different ports of entry. Afterwards, there were two talks on hospital standardised mortality ratios from NHS National Services Scotland, including using R Markdown to turn seven days of report processing into 50 minutes.
Next, I went to presentations on the usage of statistics in schools, including by Darren Macey (Cambridge Maths; RSS Statistical Ambassador). There were some shocking examples of poor questions in stats examinations, and Darren sought to develop a central resource of large data sets for educational purposes.
Charlotte Watts (Chief Scientific Adviser in the Department for International Development) then gave her keynote talk on how statistics is helping DFID with its work.
After a quick lunch and reception, I then went to a talk on the Social Metrics Commission’s new measure of poverty in the UK. This new measure utilises estimates of liquid assets and inescapable costs, such as childcare, housing and extra costs associated with disabilities.
On Tuesday afternoon, there was the rapid-fire talks, which last five minutes per speaker. My choice about which set of rapid-fire talks to attend was determined by the fact I was giving one.
My five minutes were spent on the purported curse of Aaron Ramsey: the suggestion that goals by a Welsh and former Arsenal footballer somehow caused the death of famous celebrities. The evidence for this ‘curse’ is not strong, and the subjective choice over who is a celebrity affects any calculated results. My calculations were simplistic and did not account for the day distribution of celebrity deaths, which may be driving the slightly-elevated rate after Ramsey goals.
Despite my nerves, the talk was well-received, particularly by the statistician who sparked my interest in the topic.
The day ended with a discussion about fake news and falsehoods with James Ball, Peter Cunliffe-Jones and Timandra Harkness, followed by an awards ceremony and poster reception.
Analysis and Engagement
After arising for the data ethics breakfast briefing, I went to the Young Statisticians Section talk on the usage of Twitter. I go on Twitter myself, with active use of the mute button. The session was helpful, as I hope to be more engaging in all means of communication.
Marian Scott from Glasgow University gave the next keynote, on the rise of mass digital data, forming data lakes from the deluge. Data visualisation is an important of both of my paid and voluntary roles, so I attend the talks on well-chosen graphs. Jennifer Rogers (Phastar; RSS Statistical Ambassador) gave a talk on how uncertainty is perceived in graphs. Next, Lucy Teece (University of Leicester; RSS Statistical Ambassador) led a discussion on statistic writing and books.
I then went to presentations from the Statistical Excellence Awards for early career writing, which covered: repeating a famous analysis on V-1 bombs on London, a simple algorithm, and polarisation of popular votes in Switzerland.
The discussion meeting on replicating statistical findings was followed by a very fun pub quiz.
The final day started with another set of talks from prize winners, including on multi-state models in hospital epidemiology. I then went to the keynote on Bayesian categorical matrix factorisation via double feature allocation, which has uses for analysing patients with latent diseases.
I then went to a fascinating set of presentations by the Office for National Statistics, which is currently transforming the Labour Force Survey into a push-to-web Labour Market Survey. I am awaiting the promised October publication into its progress and key findings.
After learning more about to get involved with the Royal Statistical Society, with Rob Mastrodomenico (Global Sports Statistics; RSS Statistical Ambassador) talking more about the programme and media engagement.
The final parallel talk was on the much-awaited developments in regional economic statistics. It was a fitting talk to attend, given the publication of regional quarterly economic growth estimates that morning.
The final keynote presentation was by Anthony Reuben, who heads the BBC Reality Check team and wrote a book called Statistical. The talk showed many examples of confusing statistical claims, such as from Buzzfeed: ‘As many as 300 managers at BBC News earn up to £77,000 or more’.
The talk included a simple question to start assessing a statistic or other assertion (which I will now call the Reuben question):
Is it reasonably likely to be true?
The conference ended with a wonderful meal at the Titanic Belfast.
The next conference is in Bournemouth, which is easier for some (me) than others (people better than me). Hopefully, the Ambassadors will assemble in 2020.