Wednesday, November 10, 2010


In a few weeks I'm going to have access to more language testing data that you can shake a whole copse full of sticks at, so I'm sharpening the number crunching skills with Bachman (2004) and its companion workbook.  The latter relies on SPSS, so I went in search and downloaded a 21 day version of PASW v18.  When the 21 days ran out, I thought, well, it's a fair cop, and prepared to put my hand in my pocket.  FFS.  I spent an entire morning in a vicious IBM loop, credit card in hand, looking for an licence authorization code.  Forget about it.  It was like trying to find a shop located 1000 feet underground, with no access from the surface.

Fortunately, an old chum had put me in the direction of comparable open-source software.  This morning I've downloaded and installed (in about 2 minutes) PSPP .  I haven't worked with it yet, but it looks like it does everything that PASW v18 does - everything that a language tester wants of it, anyhow.

Or to put it another way, SPSS seems to have become a pile of shite since IBM got its hands on it.  Go open source, kids, it's the way ahead for non-numpties.

Bachman, LF (2004). Statistical analyses for language assessment. Cambridge, MA: Cambridge
University Press.