Statistical Engineering @ Brighton - Slides available

Thanks to those of you that came to my talk - if you would like a copy of the slides, they can be downloaded here.


Thanks


Tim Davis

The Poster Competition

This year I have been allowed to judge a bit of the poster competition. Judging is fun. Specially when you are allowed to do it with a glass of wine in your hand.

I and the proper judges wandered round the posters on Tuesday evening, chatting as we went to the posters’ creators. My colleagues judged no doubt according to quality of statistics, innovative thought, relevance of methodology and analysis and suchlike. I had a much easier brief. It was: Could I understand it?

This year, the poster competition has an additional prize: The Significance prize for the poster which best communicates its message. That’s the bit that I was judging. Judges in past years have complained that some entries contained excellent and advanced statistics that were completely incomprehensible. I was looking for the reverse: I was looking for a poster which might translate into the sort of article that could go into Significance – in other words, a poster that even a non-statistician would find interesting.

Going round, you notice some points straight away. Print size is one. Does it contain acres of small print that you have to read with your nose up against the poster, blocking everyone else’s view for the 20 minutes it takes to slog through it? Or is there nice big print, not too many words, easily readable by the middle-aged like me from a reasonable distance? Layout helps. I was slightly baffled by one otherwise excellent poster, laid out in three columns, til I realised that I was supposed to start reading at the top of the middle column; I’d started, as with a newspaper, on the left.

Say ‘Poster’ to most people and what they think of is pictures not words. A picture is worth a thousand words; and a thousand words is far too many to fit on a poster. But you can fit three or four pictures on, – so, presumably, three or four times the amount of information.

So I was tempted by posters like P31, Andrea Roalfe’s ‘Working as an Applied Statistician in Primary Care and General Practice’; each type of illness she worked on had a picture to illustrate it. Much better than a dry list. Even more than pictures, people like pretty pictures. P9, ‘Supporting statistics in schools,’ scored highly there with its fluffy koala – but came from Australia, where they are entitled to use fluffy koalas as relevant to almost everything. How do you get an attractive picture into, for example, P14, ‘Assessing the Effect of Informative Censoring in Piecewise Parametric Survival Models?’ I don’t know. Probably you cannot. Still, those ones can still be in contention for the proper prize. P6, Tom Gerlach’s ‘Census rehearsal 2009’, with its image of a red-curtained theatre stage, looked marvellous – though it did perhaps lack content. Still, if statistics ever fails him, he should have a great future in design or advertising.

The real key, though, is clarity. Call it the quality of explaining things. Can you understand it at a glance? At first reading? It obviously helps if the message itself is fairly simple. As I have said, I was not looking for cutting-edge statistics. A useful application of standard statistical techniques is just as good for telling the great public why we all need statistics and statisticians.

The winner I chose uses simple statistics and clear graphs and explains its purpose clearly. I could understand what it was saying almost at first glance; three minutes reading it enhanced that understanding. It has a clear conclusion as well, and that conclusion is an important and an interesting and a useful one, and one which should influence real-life decisions and what people actually do. All of which makes it ideal for the basis of an article in Significance magazine – and, I suppose, a near-ideal poster as well. It will be announced at the conference dinner on Thursday. Before then, see if you can guess which one I have chosen.

Julian Champkin,Editor, Significance

breadth of statistics profession evident at RSS annual meeting

The conference was launched with an interesting and compelling talk by Peter Donnelly, Professor of Statistics at the Wellcome Trust Centre for Human Genetics, which is affiliated with Oxford. His talk, “Statistics, Genomics, and Human Diseases,” focused on recent developments in identifying parts of the genome related to disease and traits in humans. These discoveries may provide new ways to understand the biology of disease development. At one time, Donnelly said, statisticians in genomics could be described humorously as people who “model genes.” “Now,” he said, “we are more explorers than modelers.”

David Hand, RSS President, brought words of welcome to the conferees. He noted the upcoming World Statistics Day on 20.10.2010, and emphasized the significance of the United Nations’ recognition of the important contributions of statistics and statisticians. Unfortunately, Hand noted, the level of public recognition is still rather small.

President Hand then presented an Honorary Fellowship to Janet Derbyshire for her important work in clinical trials, especially for her consistent recognition of the role of statistics in drug development. According to the RSS website, “Honorary fellowships are awarded for the Society to recognise the contribution of individuals of great eminence working in fields related to statistics who are not members of the statistical profession.”

In the concurrent sessions, there were a rich variety of themes addressed. Kudos are due to Vern Farewell and the rest of the programme committee for a job very well done.

Genomics, climate science, statistics and the law, and adaptive clinical trials are just some of examples of today’s presentations, reflective of the breadth of our profession. I attended a particularly interesting session in which three sources of data for climate science (plankton, surface temperatures, and ice cores) were discussed. Methods of measurement and modeling of uncertainty were shown. The speakers were clear, and did a nice job of talking in understandable terms about the science of measurement and the statistical science involved as well. However, as happens sometimes in discussions of climates, some of the questions after the talks reflected the biases of the questioners rather than the content presented.

A very nice set of posters were on display during the day and featured in the late afternoon. The topics and the locales represented were diverse, though a majority of them focused on some aspect of statistical work in the health sciences. The quality of the work was quite impressive.

The day was capped by the conference social, held out on the Brighton Pier, which was virtually empty (except for RSS conferees) due to the high winds lashing the area. But inside the pub I found cold drinks, warm food and even warmer fellowship. Some of the planned activities had to be cut back because of the weather, but it was a delightful evening nonetheless.

I am already regretting that I have to return home on Thursday, as I realize how many excellent sessions I will miss. In the future I will ask my friends to look at the RSS calendar before setting their wedding dates.

Fishy Business

Our opening social event had us getting lost in the labyrinth of fish tanks in the Brighton Sea Life centre. The catfish, parrotfish, angelfish, sea anaemonies and companions looked on as we enjoyed a glass of wine and canapes in the Victorian splendour, meet up with old friends and welcomed first-timers to conference. Once you ventured through to the hall of mirrors (aagh!) to the coral reef, you found a 'variance' of statisticians in an orderly queue waiting for a trip on the new glass bottom boat. Looking down on the sharks and fish living on the coral reef, you could imagine being transported to Australia on the Great Barrier Reef. Certainly the walk through the glass canopy under the tank, gave you the perception of diving in the reef. It was feeding time, so the fish were very active, apart from a sleeping basking shark. The finale was the feeding of the turtles with their daily ration of lettuce leaves.

One hot topic for canapes chatter was who are the top 10 statisticians in the last century. Leading names were Cox, Fisher, Lindley, Box, Nelder ...now thrown open to the conference to add another 5 and agree the ranking!


Mary Sweetland

lovely reception, slightly less lovely weather

The RSS conference social programme got underway this evening with a lovely reception at the Sea Life Centre in Brighton. The centre is a large aquarium with a series of interesting displays of aquatic life, and the venue was open this evening only for participants of the conference. So we had the run of the place, enjoying juice, wine, and hors d'oeuvres, while marveling at the undersea world and reveling in the joy of renewing old acquaintances and making new ones.

Registration began this afternoon, and in typical sharp RSS fashion, it was well organized. I was in and out of registration in under a minute, but spent several very pleasant minutes afterward meeting and greeting RSS staff.

The weather got quite blustery and then rather wet this afternoon. Medium-sized waves crashed loudly against the beach and the Brighton Pier. It is a bit chilly as well, or so it seems to someone coming from the lingering summer weather in the US. Nonetheless, the coast and the town are spectacular, a great setting for the conference.

The scientific programme begins tomorrow with opening remarks from David Hand, RSS President and a plenary talk by Peter Donnelly of Oxford. We’ll all be there Brighton-early (sorry, I have a rather nasty addiction to puns).

Statistical Engineering and a theory that dates back to 1914

One of the best methods that I have come across which exemplifies the inductive-deductive iterative nature of statistical investigations (see my first post) dates back to 1914 – the so-called “Pi” theorem of E Buckingham; I will illustrate the use of the “Pi” theorem using the well known paper-helicopter experiment, which many people who have taught statistical methods to engineers will be familiar with. If we adopt a completely empirical approach, we might decide to run a response surface experiment to model the flight time of the helicopter as a function of various design parameters; three design parameters might require about 15 runs in the experiment to develop the transfer function. However, if we think for a minute about the physics, we know that the flight time will be a function of the mass of the helicopter, and the area swept out by the rotors, together with the force due to gravity, and the density of air – and all of these quantities are known. The application of the “Pi” theorem, which reduces the dimensionality of the problem, and does not require linearity to ensure dimensional consistency, reveals that the number of experimental runs can be reduced to about three. It is a mystery as to why the “Pi” theorem isn’t referenced in any of the classic texts on response surface methodology and design of experiments; is it because not enough statisticians are interested in engineering?

getstats zone at RSS 2010

The getstats team are looking forward to RSS 2010 in Brighton next week. All of the delegates are encouraged to visit the getstats zone which will have information about the campaign and the ways you can get involved and answer any questions. We are also going to be doing some filming of vox pops to get everyone’s views on the importance of statistics and the campaign. Please drop by the zone to find out more! We look forward to seeing you there.