“Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write.” H.G.Wells

Today! I swung by the penultimate session of the VIIIth *Summer School in Statistics for Astronomers*, just to catch a couple of old friends who were at the school, chat a bit at the coffee break, and get a quick refresher in clustering…

The school, being the 8th in a series, is offered by the Penn State Center for Astrostatistics, an NSF funded interdisciplinary center in, well, astronomy and statistics.

The school is very popular, and is aimed at graduate students in astronomy, and postdocs and researchers. Typically a little over a couple of dozen students are accepted, who get a week immersion course in modern statistical techniques and a crash course in using the ever popular R software package.

The department of astronomy, incidentally, also offers a graduate class in astrostatistics, covering a similar range of material.

So, when I was a teenager, I was, for obscure reasons, in a school in England and had just taken my mathematics “Ordinary”-level in the GCE exams. I was in the accelerated math class and raring to go on to the Advanced level, looking to get it done two years early.

Our new math-teacher-from-hell decided the advanced class would deviate from this track and spend a year doing the O-level Statistics class instead.

So, I got a year of intro Probability Theory and Statistics.

ZOMG was I bored, mad and jaded.

Didn’t handle it well, but had the sense to take the easy A and then got back into the whole math thing and did the Advanced Math, Advanced Pure Math, Advanced Applied Math GCEs.

For undergrad I did a Math.Phys. degree so got the mandatory dose of stats from the math side – Probability Theory, Statistical Techniques and Numerical Analysis; as well as baby stats for baby experimental physics slotted in the physics class curriculum.

Still not impressed.

Though I noted a lot of my pure Math friends were peeling off into the Stats, and most, as it happened, did very very well for themselves in various Chartered and Certified Occupations.

Grad school came upon me, and the obligatory year long Math Methods class was taught by an Experimental Particle Physicist – not so much trendy topology and rather a lot of monte-carlo-the-crap-out-of-this-particle-detector (~~baby~~ toddler version).

Shock immersion in fortran coding, and a lot of the stuff I had soaked up earlier was now actually useful, and I think I date my ~~corruption~~ conversion to computational techniques to that term, and it wasn’t that bad actually…

I fondly remember one of the problems as actually being challenging in an interesting sort of way.

So, “my fine young friends/now that I am Full Professor”, what do I really think of stats?

I’ve gradually come to think that comprehension of statistical techniques is at least as necessary for our profession as calculus, and arguably more important in a range of subfields.

It is also one of the technical prerequisites students are least likely to have taken in any depth, and are most resistant to get into seriously.

It is also, unarguably, the single most useful skill we can have our students learn, both for being usable as researchers, and for alternate career options.

Serious demonstrated statistical skills, at the level of user, not necessarily developer, is the skill most likely to keep you comfortably fed, sheltered and provided with the latest iThingies if you decide the whole academia thing is not for you, or find external circumstances leaving no choices about such.

Er, so, come take our course.

Next year.

It is full.