THE WORLD of arcane statistical modelling would appear to be far removed from the tortured decisions that many couples have to make when they undergo IVF.
Many are presented with a dilemma about whether to implant one embryo or two.
If you implant two embryos you have a better chance of implantation than if you implant one, but you also have a better chance of losing both.
Solving the dilemma of which is the better option is just one of the issues facing the best minds in statistics, many of whom are gathered in the Conference Centre Dublin this week for the World Statistics Congress.
The problem cannot be solved by a controlled experiment because no couples would consent to it and it would be regarded as unethical. The objective data that has been culled from previous studies of IVF has proved to be inconclusive.
“You don’t care about how many embryos are alive at one month after conception,” said Harvard professor of statistics Donald Rubin.
“You care about having live children that are healthy. Which is better for having live children that are healthy? We don’t know the answer.” Prof Rubin was one of four of the world’s top statisticians dubbed the “dream team” for the day who took part in a generalised discussion about statistics and the world we live in.
The other three were Prof Stephen Stigler, son of the Nobel Prize-winning economist George Stigler, Prof Peter Huber, formerly of MIT and Harvard, and Sir David Cox.
Now 87, Sir David is known as one of the doyens of the profession, and received a knighthood for his services to statistics.
Outside the world of statistics he is not well known, but his research on proportional hazards model has proved critical in helping doctors assess the potential benefits of different medical treatments.
Sir David acknowledged that the profession would appear to have an image problem. He recalled the “last time but two” he visited the United States, he told an immigration official he was a statistician.
“What rubbish, every day they are telling us the opposite of what they told us the day before,” the official replied.
“The fact that people see it that way undermines the very careful way in which we can get secure answers,” Sir David said.
A quite bewildering array of subjects are being discussed over the length of the congress, from the downright obscure “semiparametric propensity score weighting method to adjust for nonresponse using multivariate auxillary information” to the practical: forecasting avalanches in advance, or deciding on a global standard for judging house prices.
One session had the provocative title “where are the young people?” and it was chaired by Claire Gormley, a lecturer in mathematics in UCD. Statistics may be terminally unfashionable, but statisticians have never been so much in demand, she says.
