9 minute read
The beginning of 2020 has been an eventful time for Professor Alan Welsh at the ANU Research School of Finance, Actuarial Studies & Statistics (RSFAS): He was awarded the 2019 Hannan Medal, a top distinction in the field of statistics from the Australian Academy of Science and, more recently, was appointed as Co-Editor of The International Biometric Society’s (IBS’s) Biometrics journal.
We caught up with Alan to discuss his new appointment, the University’s legendary statistician Edward James (Ted) Hannan, the importance of restraint in covering high-profile topics and the pros and cons of discovering something very few people know about.
Congratulations on your appointment as the newest Co-Editor of Biometrics. What direction do you envision for the journal under your editorship, and for global biometric trends in general?
There is a very steep learning curve with the editorial position. Partly because of the volume and partly because of the high standard of the papers that get sent in. There are 650 to 700 papers a year, of which three co-editors handle a third each.
Overall editors have limited conscious control over the direction of the journal, since most articles are self-selected by those who submit them. On the other hand, if people know that an editor has a particular interest, perhaps they’re more likely to submit articles in those areas. However, with three different co-editors that unconscious impact on the journal is diluted a little bit. The journal is getting a lot of submissions in biostatistics, which is medical applications of statistics, and within that area there are a lot of papers in casual inference, trying to tease out factors causing particular effects from observational studies rather than from randomised experiments. This is also applicable in economics and other social experiments.
Ted Hannan, after whom the Hannan Medal is named, is a significant part of the University’s history. How did it feel to win an honour bearing his name?
It was great, and it felt really appropriate: Ted was the first Professor of Statistics in the University’s old faculty, the teaching part. He was also one of the first two PhD students in statistics at the ANU. I was lucky enough to know him and interact with him a bit because, when he retired from the Institute of Advanced Studies, he came over to ANU and spent his last few years at what was one of the precursor departments to RSFAS.
Do you have any stories from your time with Ted you’d be willing to share?
Ted loved reading a lot. He would come in and knock on the door in the morning and share what he was reading. He would tell you again at morning tea and over lunch and sometimes again at afternoon tea. One book he was reading was a biography of James Joyce, an Irish novelist, and he was very keen on him. He was so enthused by this book he would talk about it to everyone he bumped into. He had a great sense of humour and a short temper, but he would apologise pretty quickly afterwards.
You’ve previously mentioned that you often have to develop new methods for analysing data where no methods have been previously available. Is there a particular field you’re excited about exploring further by creating new methodology?
I’ve always liked to have a very broad view. I never felt like I wanted to become a great expert in a very narrow area. I’ve been lucky to maintain my curiosity about lots of things. A lot of these projects are a bit serendipitous — somebody suggests it, it catches my attention and then we are up and running. I hope to continue in that style, and I’m really open to interesting social scientific questions that require interesting statistical methodology.
Which biometrical applications do you see as most significant for the years ahead, especially following the latest environmental and bio-diversity upheavals in Australia caused by the unprecedented bushfire season of 2019-2020?
In terms of thinking now what data should be collected to make things possible in the future, that’s a worthwhile thing, but it’s too early trying to do analysis because we don’t have the data and long-term effects take a while to appear.
Is there a particular area in your own research you’re looking forward to? Any recent highlights?
I’ve got an ARC (Australian Research Council) grant on model selection and dimension reduction, two areas with overlapping goals but running separately. The goal is to bring them together and get into the interface between them. I’m working with a postdoctoral student who is doing very good work in the area, and it would be really nice to see it develop.
With a PhD student last year, we did some very nice theoretical work on asymptotic approximations in simple linear mixed models, for estimators in simple linear mixed models. I’m very pleased with the results that are under submission. I think they are quite exciting and will give us some new insights. A lot of the attraction is the little window when you know something almost nobody else knows, and we’re still in that little window now. Once it’s out there, you hopefully get some credit for it. The pleasure of discovering new things for that brief moment, knowing stuff that no one else does, is unbeatable.
Thinking back, can you name one of those eureka moments that still has relevance today?
It’s challenging because sometimes you close things off by discovering them. You don’t mean to, but that’s what happens when you solve the problem that’s been unsolved, you remove it from other people’s interest.
One thing that has come up again recently is the sports science work. Sports scientists developed a method for analysing their data, which they were promoting quite widely. A young colleague who went to work at AIS (Australian Institute of Sport) got involved. We looked into it from a statistical point of view and managed to understand how it was working and achieving apparent miracles. Of course, that unfortunately invalidated it as a method for analysing data. The Sydney Morning Herald quoted the research in an article. Six, seven years it’s still bubbling along.