The ISA Gosset lecture 2017 will be delivered by the distinguished Professor David Hand OBE (Imperial College London) in the Royal Irish Academy, 19 Dawson St., Dublin on April 6^th at 6:00pm.

Please note tickets are required. These are available free of charge at https://www.eventbrite.ie/e/isa-gosset-lecture-2017-tickets-32088704178

The dangers of not seeing what isn’t there: selection bias in statistical modelling.

In general, in statistics and data science, more is better in the sense that larger data sets mean that smaller effects can be detected and more confidence can be placed in any statistical conclusions.  But this is only true if we have confidence that the data represent the underlying reality fairly. All too often the available data have undergone unobserved selection or distortion processes, which can mean they are potentially misleading. This applies in human interactions – where it has been suggested that the notion that ‘data=all’ can replace the need for careful theorising and statistical modelling – but also in the hard sciences and medicine.This talk gives examples of such cases, showing how ignorance of selection mechanisms has led to mistakes and even disasters. These mechanisms are described, and strategies for tackling the problem are outlined.This event is run in partnership with the Irish Statistical Association.  It will be followed by a reception partially sponsored by Diageo.

Read more

The inaugural ISA Gosset lecture 2014 will be delivered by Professor Adrian E. Raftery (University of Washington) in the Royal Irish Academy, 19 Dawson St., Dublin on May 29^th at 6:30pm.

Please note tickets are required. These are available free of charge at https://www.eventbrite.ie/e/population-projections-for-all-countries-tickets-11052998817

Population Projections for All Countries

Projections of countries’ future populations, broken down by age and sex, are widely used for planning and research. They are mostly done deterministically, but there is a widespread need for probabilistic projections. This lecture will describe a Bayesian statistical method for probabilistic population projections for all countries. These new methods have been used by the United Nations to produce their most recent population projections for all countries. The results suggest that world population will increase more than had recently been believed likely, reaching between 9 and 13 billion by the end of the century, with no end to population growth this century. The population of Africa, in particular, is likely to grow, from about 1 billion now to between 3 and 5 billion. The number of working age people per retired person will probably decline dramatically in most countries over the coming decades. The results also suggest that the current UN high and low variants underestimate uncertainty for high fertility countries, and overstate uncertainty for low fertility countries, mostly in Europe. Professor Raftery will comment on implications for carbon emissions this century.

Read more