ASA Joint Statistical Meeting 2012, San Diego

As I write this I'm sitting in a session at the JSM on "Advances in Bayesian Clinical Trial Design".

The JSM is the Joint Scientific Meeting of a selection of statistical associations

  • the American Statistical Association
  • the International Biometric Society
  • the Institute of Mathematical Statistics
  • the Statistical Society of Canada
  • the International Chinese Statistical Association
  • the International Indian Statistical Association

Each year these organisations hold a scientific meeting - rather like HTAi, but for statisticians. Despite these North American and Asian organisations hosting the conference there were a number of participants from the UK and Europe.

Late last year I was asked to join a 'topic contributed session' - a session where a set of presentations are presented with a common theme. I was invited because of a systematic review I published 3 years ago with Kath Barnard and Louise Dent on methods of predicting recruitment to clinical trials. I was asked to talk about the review in a session where statisticians would then talk about the implementation of the ideas therein. Neither Kath or Louise could come, so here I am in a meeting with 5,000 statisticians in San Diego, California.

This conference is huge. HTAi often looks large, with 6 parallel sessions going on at any one time. Here there are in excess of 25. The ASA is made up of 26 sections (eg biostatistics, marketing statistics, engineering statistics etc) and virtually every section and organisation as an activity going on in each session.

The JSM classifies sessions into at least 5 categories, but in reality there are only two types (especially to a non statistician) those which are understandable and those which aren't!

The understandable sessions tend to concentrate on the philosophy of an aspect of statistics or have a slightly humorous bent, for example a session on the statistics of elections or on what a statistician does day-to-day in particular jobs.

The more challenging sessions focus on statistical methodology, but aimed at practicing statisticians in the area of interest. Within a couple of minutes they usually get well beyond the ability of a humble physician to work out what's going on. Despite advice from the organising committee not to dwell on formulas, many presentations cover their slides densely with equations using notation I've never come across before. The statisticians in the audience seem to be very excited and engaged about what is said however, so I can only presume it's good stuff.

I'd hoped to pick up some hints on statistical aspects of clinical trials, and Bayesian statistics. Unfortunately the trial relevant sessions were mostly about phase 1 & 2 studies, and delivered by statisticians working for drug companies, with an agenda of profit generation rather than maximising the cost-effectiveness of healthcare. The Bayesian material was, as far as I could tell, cutting edge - and mostly fell into my non-understandable category. In recent years a further education course entitled 'Introduction to Bayesian Methods' has been offered, but unfortunately not this year.

The session I presented in was quite mixed - my presentation was on the philosophical end of the scale discussing a systematic review as was that of the discussant, with the others dwelling on the equations related to trial recruitment.

One possibly useful thing which came up in one session was - which provides among other things a free statistics text book and introduction to R (which is a free statistics tool, much like SPSS or SAS).

There are some good points about this conference though, which HTAi and others could learn from.

  1. The venue (the San Diego Convention Center) is good - with sufficient space on one floor to fit most of the sessions (HTAi could fit in here 3 or 4 times over) and sufficient space for meeting and socialising.

  2. On day 1 there was a briefing session for first time attendees. It was very useful to have the conference, how it is organised, and how to present to this audience put in context.

  3. There are facilities to plug your own equipment into the display equipment in each room. I've been stung at conferences in the past by conference-provided old laptops which were missing vital fonts or the ability to play movies. To avoid this I've taken to rendering all my slides as images, and not using movies - being able to plug in your own laptop (or iPad) is a most welcome solution.

In summary: if you have a hankering to attend a North American conference be aware that unless you're a statistician this one probably isn't for you. It's unlikely to be a good venue to present work arising from our RoR programme (the invitation I received notwithstanding). I can't imagine I'll be attending again - even if invited. I just don't know enough statistics to get enough out of this conference to justify a trip to North America. However, if you are a statistician there's enough content here to keep you occupied all week, and lots of opportunity to meet your professional colleagues from all over the world.