So here’s what McGrayne was doing before she wrote the book about Bayes Theorem (the theorem that would not die). I stumbled across this book in the public library. It’s from 1993 and documents 14 women ranging from Emmeline Noether to Rosalind Franklin. What a range of histories, experiences and inspirational achievements.
Chris Lennard reported on his study leave through this talk on 10 April. He started with an introduction to methods of finger mark detection, then he reported on the International Fingerprint Research Group meeting in Israel last year. As a result he became involved in writing research guidelines. This wasn’t part of the original objectives of his study leave, but just like I discovered, undreamt-of opportunities can arise and it makes sense to grab them.
James Curran of Auckland University gave this talk to the Research School of Finance, Actuarial Studies and Applied Statistics on 4 April. Great turnout on a wet day just before semester break, for a great talk. James’ special topic is glass evidence and he was able to explode some of the big myths about glass (it’s not a liquid) and make reference to. Some of the classic forensics criminal cases of recent Australian and New Zealand interest, also of interest to the teaching of Forensic Statistics at the University of Canberra, James was able to direct me to some useful textbooks, and used some great examples of correct and incorrect interpretation of evidence in his talk. Well worth the bus ride in the rain to attend this talk.
This book has been mentioned on ANZSTAT a number of times, and I stumbled across a copy at the public library. Proofiness is defined as “the art of using bogus mathematics to prove something you know in your heart is true – even when it ‘s not.” Examples range from US presidential elections past and more recent; through famous criminal trials such as O.J. Simpson and Sally Clark; to various scientific studies reported in reputable journals. In the end parts of it became a bit repetitive and strident, but I can’t deny that three members of the household read it, so the message is percolating through!
The 24th meeting of CSSM took place in Goulburn on 26 March 2014. As always, four excellent and thoughtful talks on the interface between statistical methodology and statistical applications. James Chipperfield, ABS, spoke on “Using the bootstrap to make inference with estimating equations and probabilistic ally-linked data.” His talk was particularly relevant to the construction and use of linked data files. Emi Tanaka, University of Wollongong, spoke on “A short introduction to motifs and improving the comparison of motifs.” Her talk was my first chance to hear an explanation of sequence logos, and to learn about the additions that her team was able to have made to the software product TomTom that analyses motifs. Robert Clark, University of Wollongong, spoke on “Likelhood analysis of distance sampling data.” His talk was particularly relevant to anyone involved in distance sampling, and really seemed to go some way to reconciling the proponents and detractors of current practices in this method. Finally, Pavel Krivitsky, University of Wollongong, spoke on “Inference for exponential-family random graph models based on egocentrically-sampled data.” I was rather worried when his slide show showed “1 of 240″, but it turned out every highlighting of an equation generated another slide. So four hour-long talks made for a good day of networking at the cutting edge of statistical methods and applications.
That’s only the first half of the title of this talk! The subtitle was “The LOOK longitudinal study – background, statistics, and key findings.” The talk was presented to the Canberra Branch of the Statisticsl Society on 25 March 2014, after the AGM. The speakers were Professorial Felow of UC, Dick Telford and Adjunct Professor of ANU, Ross Cunningham. The talk described a tour de force in analysis of a study of 26 ACT primary schools that has been running since 2005. Some very rich mixed linear models fitted using REML revealed a range of results, some quite counter-intuitive. The presenters and their co-authors have over two dozen papers from the study, so there’s a heap more to it than could be covered in a one-hour talk. All in all it was a fantastic example of interdisciplinary collaboration between Sport and Statistics.
Vladimir Brusic is the Director of Bioinformatics at the Dana-Farber Cancer Institute in Boston, and presented this talk on 11 February. His talk was very approachable, going over many of the basic issues of data quality, asking the right question, and so on. I liked his five-step knowledge hierarchy model, from Data – Information – Understanding – Wisdom – Vision. His main application area is development of vaccines for dengue. Databases he used included FlaviDb, and FluKB.
In preparation for a holiday drive of seven hours through country New South Wales, I stocked up on podcasts: Better Homes and Gardens for the children, and the Royal Statistical Society for the adults. Problem with the RS podcasts, they were recorded at such a low level they were inaudible above the road noise! There are some specifically statistical podcasts at www.statsandstories.net, and I’m planning to listen to these podcasts soon. With no more long drives planned right now so they stand a good chance of being heard!