Q: What statistics software do pirates use?
A: ‘We use R, me hearties!’
Turn away now, it doesn’t get any better….
I’ve spent the last two days in a workshop on R, run by the lovely people at the New Zealand Social Statistics Network. It was billed as ‘Intro to R’, and ended up being partly intro, partly intermediate, and also an intro to statistical packages generally, a refresher in statistics and lecture on good data visualisation. Something for everyone.
I went along because I’ve taught myself R from various online sources, and cobbled together analysis from bits of code pillaged from the Web and other people’s projects (’cause I’m down with OPP). I needed to see what I’d missed through my autodidacticism and pick up new tips and tricks. Keep the tools sharp, as it were. The course absolutely delivered — there are a few things I should do differently and lots I can do better.
The open-source approach — giving it away for free — creates different incentives. Several people asked me why I use R. It’s simple — R is free, and it gets the job done. But it isn’t the price itself that’s the issue. The issue is that I’ve moved around from job to job over the last 20 years, and everybody uses something different. I’ve used SPSS, SAS, Limdep, GAMS, Eviews, and probably others. I got quite good at SPSS, for example, and then went somewhere that didn’t have a licence. When I started my current job, I had the choice of retraining in Stata or improving my Eviews, or venturing out into the wilds of R. I chose R, because I figure that jobs will keep coming and going, and there’s no guarantee that the next place will use Stata or whatever.
In a world where data is largely accessible and files are in the cloud and people change jobs regularly, there is a lot less call for investing in site-specific resources. That includes training in software that ties you to a particular employer. Portable tools make sense, and open source is portable. It does raise the question of who should pay for training, the employer or the employee. Training will increase the employee’s potential contribution to the current workplace, but also raises her value on the job market. That suggests that open-source software might have more value in economies with greater agglomeration, where a number of employers essentially train a pooled workforce from which they all draw. It also may have more value where there is greater job entry and exit — if people don’t tend to change jobs, then having portable skills isn’t as valuable.
Now I’ll leave you with my earworm. On the first day of the course, in the first presentation, the presenter said about our data files, ‘They should be comma-separated.’ Since then, I’ve had that line going around my head, but sung by The Offspring: