«I chose statistics because I thought it was just more interesting, had more life to it than pure mathematics, even though I was trained mathematically». Alan Gelfand explains his passion for statistics, to which he devoted his life, thusly. In the nineties he popularised, together with Adrian Smith, the Markov Chain Monte Carlo method (MCMC), a contribution that considerably improved the subset of Bayesian statistics by sampling from probability distributions. «It is one of those things where you can feel very fortunate because there are lots of really smart people out there who work really hard for their careers and don’t happen to find something. You discover something and it happens to be a breakthrough. The only thing I can say for me is at least I took advantage of it», he remembers.
During the last years, Alan Gelfand’s research revolved around the area of space-time statistics, a booming field with a lot of possibilities, as he explains humbly yet passionately: «There is an expression we use in English, “low-hanging fruit”, which you can reach up and grab without working so hard because it hasn’t been picked yet and in spatial-temporal analysis there was so much low-hanging fruit that you could play, enjoy all the possibilities that are out there. Many other areas have been developed so much, have been pushed so much that you have to reach much higher to find some fruit. I have been very lucky». In fact, he published four books and more than 250 scientific papers regarding these questions, and received a number of awards. The latest one was the Distinguished Research medal from the ASA Section on Statistics and the Environment.
Alan Gelfand is currently a Professor of Statistics at Duke University (Durham, USA) and Fellow at the American Statistical Association, the Institute of Mathematical Statistics and the International Statistical Institute. Applications of his work can be found predominantly in areas such as environmental exposure, space-time ecological processes and the development of climate models.
What is your vision on the applicability of statistics?
What exactly is Bayesian thinking?
What did the Markov Chain Monte Carlo (MCMC) mean for Bayesian statistics at the moment?
You also have been working in Spacial Statistics. What was your experience in that?
In what direction do you think statistics will grow? Do you think it will be more theoretical or more applied?
Viktor Mayer-Schönberger and Kenneth Cukier start their book about Big Data1 telling how Google searches did predict the spread of the H1N1 flu outbreak in 2009. This example serves to authors to cite the article by Chris Anderson, «The End of Theory: The Data Deluge Makes the Scientific Method Obsolete»2, published in 2008 in Wired magazine, who provocatively proclaimed «Petabytes allow us to say: “Correlation is enough”. We can stop looking for models». What do you think about it? Is the future of Statistic to become just simple descriptive data analysis?
«But Big Data is not the same as statistics. We try to understand complex processes, explain, predict, capture uncertainty»
|«If statisticians are not visible enough, people would just assume that we don’t have that much to contribute»
We would like to talk about the visibility of statistics. What should statisticians do to make them more visible?
Which is your opinion about the future of financial funding in research?
Misuse of statistics in the media can lead, on purpose or unintentionally, to manipulating the figures. Do you think this is a common issue?
Do you consider the population is prepared to correctly understand statistical analysis in the media?
1. Mayer-Schönberger, V. i K. Cukier, 2013. Big Data: A Revolution That Will Transform How We Live, Work and Think. Eamon Dolan/Houghton Mifflin Harcourt. Boston. (Go back)
|«The future of statistics is going to be in working in these complicated interdisciplinary projects, working on challenging processes, challenging systems»
«There’s a real challenge with misuse of the data and, in the public mind, it creates a real skepticism, a real doubt about the validity of statistical analysis»
© Mètode 2014 - 83. Online only. The Digits of Science - Autumn 2014