Jarrett Byrnes assistant professor, UMass Boston

R
#load relevant libraries
library(plyr)
 
#read in data
change_data_all_t2<-read.csv("../analysis/change_data_all_t2.csv")
 
#fit a series of models for comparison
nullMod<-lm(percent.cover.change.t2~cover.t0, 
         data
=change_data_all_t2)
a<-lm(percent.cover.change.t2~ urchins+cover.t0,
         data
=change_data_all_t2)
b<-lm(percent.cover.change.t2~ richness.t0+cover.t0,
         data
=change_data_all_t2)
d<-lm(percent.cover.change.t2~richness.t0+urchins+cover.t0,
         data
=change_data_all_t2)
e<-lm(percent.cover.change.t2~richness.t0*urchins+cover.t0,
         data
=change_data_all_t2)   #compare models using AICc aicW.mmi(list(nullMod, a,b,d,e))
A little code from some recent analyses

I grew up, like so many ecologists, using point-and-click stats programs. I designed experiments for ANOVA, considered non-normality a sin, and was a slave to excel's bar graphs. Maybe sometimes I'd go wild and think on a logit scale. Eventually I got blessed into the brotherhood of SAS, loading and reloading giant sets of CDs every year and hoping that the University would keep our license updated.

Then one day that license expired. And I heard about this thing called R. A free stats program with a growing supportive community of users. I loaded it up, played around with a few simple functions, and began asking for help as I was forced to confront the details of the statistics behind my analyses like an adult.

It really blew my mind. As I trudged up the learning curve, the world of statistics opened up. The support, tips, and tricks from the vast community of users were signposts and shouts of encouragement. My data was no longer a static chunk of stuff to plug-and-chug through formulae I could casually ignore. It became a living thing. I could learn a technique one day, and then find or begin to write code for it the next.

This kind of empowerment is necessary for 21st century science. We are not statisticians. But to derive Truth from Nature, we need to be comfortable in their house. For we are all data scientists. Like it or not, the scale and complexity of questions we ask require increasing sophistication. And we need to become comfortable with that.

R is a brilliant way into this world. I am a user. I am a developer. I am an advocate. And I am a teacher. If you have not used it before, it will change your relationship to your data. I hope you enjoy it, and would be happy to help you find the way.



As a final note to those experimentalists who are coming to R from the world of jmp/sas/spss/etc, pretty much the first question I get once someone runs their first analysis is "What's up with my ANOVA giving me different results from my old software?" The answer has to do with different methods for calculating sums of squares. There are a lot of issues here, and they are worth reading up on and being aware of their meaning. As a good starting point, see this blog post and the links to mailing lists posts and journal papers in the comments.