To graph or not to graph, that is the question...

I have generated most of my statistical graphs in Systat, the statistics program I use predominantly (learning R was too hard (read: time consuming), though I realize its MANY great advantages). The problem with these graphs is that they are not easily edited and often appear jumbled together.

I am preparing my thesis in manuscript form, which means ready to submit to journals for publication. I've been sending my supervisor bits and pieces, and the terrible Systat graphs were one of the things he pointed out as needing to change. Just regraph them in Excel, he says, no big problem. And it wouldn't be, if I could find my spreadsheet with all the data. Apparently I generated it in Systat, and then accidently only saved the graph, not the data sheet.

So after spending two hours searching through every data sheet I had on my computer (in Systat, Statistica AND Excel form), I come to the conclusion that I definitley do not have it. So I'll have to start from scratch. Redo it. The most dreaded of all stats problems (at least for me).

And then I did it... took all of 30 minutes! Some quick calculations and a pivot table and I'm done like dinner. I could have saved myself two hours if I had just done that to begin with. Sometimes I think it's subconcious procrastination. My mind knows it will be faster, so it convinces me that it would be more work... and then I go about it the roundabout way (usually with distractions).

And now I'm writing, writing and writing some more

I haven't written anything since I got back from CSEE '09 because I've been incredibly busy doing and redoing modelling selection for my squirrel nests. Logistic regression + AIC by hand (and by hand I mean excel spreadsheet) is time consuming. And even more so when you input values as words and not as 0s and 1s, because Systat doesn't tell you which it labels as 0 or 1. Of course, I don't realize this until I'm sitting in my supervisor's office proudly showing him my results (which I hadn't looked at in-depth yet). He asks me, which is 0 and which is 1? I stare dumbfounded at the results.

So back to the drawing board, I recode my data and run 3 series of logistic regressions for 3 different questions... plus all the AIC model selection all over again. Needless to say, all my data will be coded from now on. Such a simple thing, but I assumed that it would tell me which one it coded as 0, and which one as 1. And you know what assuming does... Seems so reasonable, but I have to remind myself that it's a statistics program, it is made to be overly complicated and tricky. It can't have its user's analyses go smoothly :S

And so, finally, with the modelling done and redone, I'm actually at that point in time. ACADEMIC WRITING. The bane of my life, the culmination of two years of hard work, sweat and tears.. I'm not looking forward to this.

On the bright side, it means I'm ALMOST done. Almost being a vague word meaning anytime with the next 2-6 months, depending on my work ethic. Too bad I live in Ontario, land of tempting camping trips and sunny afternoons too nice to sit in an office.