Quote of the day

June 27, 2011 § Leave a comment

Bob Ward, policy and communications director at the London School of Economics Grantham Research Institute on Climate Change and the Environment:

“Scientists are not very good at framing the unknowns in ways that are meaningful to the decision-making process. Too often when you are trying to bring your research into the policymaking domain, uncertainty can be used as [an] excuse for inaction. Some have downplayed uncertainties not because they are trying to pull the wool over your eyes but because they are afraid they would be amplified by those who have vested interests in resisting potential policy responses. We need to get better at explaining uncertainties.”

Sceptics would argue that capability to deal with uncertainty varies a lot from science to science. I personally float in and out of scepticism, depending on how close I am to getting a job.


Wait, the Royal Society gave this book a prize?

June 21, 2011 § Leave a comment

Nick Lane’s Life Ascending: The Ten Great Inventions of Evolution, winner of the Royal Society Prize, has me rolling me eyes on the first page of chapter with the witticism “The earth was named badly. ‘Sea’ would have been better.” On the next page:

“If the probability of life in the universe is one in a million billion, then in a million billion planets there is a chance of exactly 1 that life should emerge somewhere.”

If I ever have a job again, I’ll ask my upper division students to find (i) the correct number assuming independence and (ii) bounds without assuming independence.

(There’s a much less questionable use of probability in chapter 2 when he says that though we share 50% of our genome sequence with bananas, we’re not 50% banana, because we’d expect 1 in 4 bases to match up by chance. But then aren’t we one-third banana?)

Et tu, Economist

June 18, 2011 § Leave a comment

AMERICANS are getting fatter: obesity rates have risen 74% in the past 15 years to nearly 28% of the adult population. And they are driving more: the number of miles driven by each licensed driver (VMT/LD), excluding commercial vehicles, increased by an average of 0.6% a year between 1988 and 2008. Academics at the University of Illinois have found a striking correlation between these two variables—but with a large time lag. They noted that previous research had found that changes in diet had an effect on body weight only after some six years. Therefore VMT/LD in 2004 is correlated with obesity in 2010… This near-perfect correlation (99.6%) permits predictions about obesity rates. Since VMT/LD fell in 2007 and 2008, America’s obesity rate could fall to as low as 24% in 2014…

Teaching Glivenko-Cantelli

June 13, 2011 § Leave a comment

That, when data are iid, the empirical distribution function converges (uniformly) to the true distribution function is one of the fundamental theorems underlying statistics. I don’t know if we (meaning I) emphasise this enough in classes.

What do undergraduates need to know? They certainly don’t need to hear the words “uniform convergence” or “almost surely”. But I don’t think a few explicit or implicit words about laws of large numbers is enough to convince students that statistics works.

“Statistics works because if you gather enough data, you get the right answer.” We know this isn’t quite true, but how much more do we need to tell our undergraduates to avoid misleading them?

Where Am I?

You are currently viewing the archives for June, 2011 at "But it's under .05!".