Thursday, July 02, 2009

Lies, Damn Lies, and Statistics

Finished up statistics training this week. I swear, it's like the summer of lost productivity. Between meetings and training, I'm surprised anything is getting done! At any rate, we went through this statistics training. Blergh. When the instructor started getting into SAS, my mind turned to goop. Sure, I can do most of my stuff on Excel, but I'm afraid that when I write up in my M&M that I used Excel, I'll get laughed out of review. So, SAS it is (besides, it's provided to us for free). At any rate, if anyone knows of a good SAS book, drop me the tip in the comments, I'd be most appreciative.

7 comments:

soil mama said...

I haven't found any good SAS books that are of any use. I've used a combo of SAS and S-Plus programs and although I like the pull down menu's of S-plus, I have been using SAS for most of my "real" data. I've found several great websites that walk you through the code and tell you how to decode the output (which is half the battle). I've thought of switching to using R for stats, but it would require learning a whole new code. The appeal of R (for me) is that it is free and I don't have to worry about which stats program my next employer will provide.

... just went to check some of my SAS links and several were no longer working, But this one was still good and has MANY SAS resources and codes. http://www.ats.ucla.edu/stat/sas/

Do you have a statistician with your agency that can help you, or do you ever hire a consultant? The people I work with at the forest service usually pay for the stats to be ran, or at least double checked.

Tom said...

Do you have a statistician with your agency ...

Area decided that rather than pay for full-time statisticians, they'd purchase SAS licenses and make them available to the research scientists. That helps me not at all I am afraid.

I've never hired a statistician, though I suppose I could. It sort of makes my SAS license a bit obsolete though. I just need to find the right people to collaborate with I suppose. :)

soil mama said...

you could potentially hire a statistician to "consult" for you and help you work out some SAS codes. once you get the basic codes down, you can do quite a bit by just tweeking them a little. The stats guy my FS friends work with is a biologist and uses PCord too! In just a quick conversation with him, I've gotten a new outlook on the PCord process and how to test for community differences.

Tom said...

PCord ... that's another program that I have that I really need to sit down and figure out completely. I have the "Analysis of Ecological Communities" book they also sell, but I'm thinking of taking one of their workshops.

soil mama said...

PCord: I'm lucky since the author is at my school, I was able to take his community analysis course. Bruce is amazingly brilliant and totally normal at the same time. I have no idea how helpful their workshops would be, I suppose part of it depends on if other microbial ecologists are in there too. Many examples are geared more towards macroecology (plants & animals) so some of the examples and methods don't really apply to those of us working with sequence or TRFLP data... at least w/o tweeking a bit.
I usually stick with NMS since it works with so many data structures. relativising still confuses me a bit, but the book covers it pretty well.

I think I already told you about it, but I'll be attending the FESIN work shop in a few weeks (in Snowbird, UT) "ECOLOGICAL APPROACHES TO ANALYZING COMPLEX COMMUNITY DATASETS" it's more fungal focused, but I'm sure much of the info can be applied to bacterial communities as well.
you can get more info here: http://www.bio.utk.edu/fesin/MSA2009/workshop.htm

Anonymous said...

I've found both "Applied Statistics and the SAS Programming Language" (Cody and Smith) and "SAS for linear models" (Littel, Stroup, and Freund) to be quite useful. They both contain lots of examples, which is the easiest way for me to learn new code and what the output means.

Tom said...

Thanks Anon. I'll look into them!