Statistics...

Should be reading, instead, refreshing my familiarity with R.

For anyone that doesn't know, R is essentially an open-source (read: "free") software package using a command line interface (read: "You type commands, generally line-by-line") used to do statistics and some other data-handling tasks quickly.  The help manuals are written in language somewhere between a math text, a computer manual, and a guide to statistics where--if you don't know them all--the particular tests are described using the names for the test you didn't learn the first time.

Still, once you get a set of scripts you know works, it's mostly a matter of picking the right tests and changing the specifics in the script.  Which isn't all that hard.  Compared to SPSS, things run faster and you're in and out of the program quicker until you want to make your output pretty.  (My copy of SPSS is older, so I sort of just play with the settings until I get something good.)

The "problem" with R is you really have to be sure what you're doing procedure wise because it will merrily calculate almost anything if the math doesn't kick out an error or you type it in wrong.  Which--in spite of several statistics and analysis courses I've taken--the first thing I do when I have to use stats is actually dig out one of the several copies of the PowerPoint presentations (PPT's) and other, assorted information my Biometry Instructor piled on us in class.

I make a point to download every possible file from the course websites at the end of each semester and my Biometry course's folder is 286MB, many of them huge PPT's.  What Dr. Sabo did on every PPT was embed the MS Excel spreadsheets he used to run up the data he discussed.  What this means for me is not only do I get explanations in the presentations (and the textbook I kept) but also examples of how to run the procedure in Excel and (courtesy of the lab materials) in R or SPSS.

Really cool, huh?

Totally worth the money.  Oh, and you have to total respect an instructor who requires critiques of published research designs and analyses as a regular part of the course and even includes a paper he co-authored.

Makes me wish my Qualitative Analysis for Anthropology (QA4A) instructor had done something similar.  Maybe even actually used PPT's for later reference.  Oddly, beyond the briefest of forays into descriptive statistics (very brief in Biometry), the only real overlap between both statistics courses was actually non-repeated measures ANOVA and Chi-square. 

Where Biometry pulled apart a lot of the parametric and non-parametric ways of approaching hypothesis testing, crashed through things like bootstrapping, jackknifing, and multimodel inference (my favorite), QA4A really focused on probability and non-parametric tests of correlation and frequencies with almost everything plotted out by hand on paper.

Still, I learned a lot that way as well.  I just hope I can find and read my notes when I need to...