Thursday, February 26, 2015

How do we love thee, R? Let us count the ways....


We have been talking about R (well, actually pushing R) as a good replacement for proprietary statistical software for some time, but we thought we’d take this opportunity to share with you WHY we think this is.

By the way, we have developed this thinking more clearly recently because of conversations we’ve had with our esteemed editor, Dana Bliss, at Oxford University Press, and our colleagues, Dr. John Orme and Matthew Cuellar, down in the relative warmth of Knoxville, Tennessee.

1)  As educators, we should be teaching our students to use software that they have access to once they go out into the field.  From our own little anecdotal survey, very few social service agencies have SPSS, which is what most schools teach, so what good is it to teach students skills that they can’t apply later?

Can we convince agencies to use SPSS?  Wouldn’t that be easier?  The answer to that is most likely not.  SPSS is expensive.  Really, really expensive.  SPSS is sold in different versions and with different modules.  If you’d like to purchase the high-end networked version, it will set you back tens of thousands of dollars.  Per year.  You’ve got it – licenses are only good for one year.

2)  R is cutting edge.  We have only talked about using R for statistical analysis, but it can be used for much, much more. 

Think social service agency again.  You may be producing similar reports regularly.  You might, for instance, have to report to your board how many clients you served in a quarter, how many terminated, etc.  You may have to show graphs.  Or you may need to do more sophisticated analysis, such as determining how successful your programs are.

You can actually create dynamic documents in R with some user-contributed packages that will run your analysis and insert your findings, tables and graphics AUTOMATICALLY in your documents. Way cool.

3)  R has statistical capabilities that that really expensive software can’t do.  For all that money you just spent, you still can’t do it all! How disheartening is that?

4)  We are pretty certain that the steep learning curve that people write about is not quite as steep as you may think.  Sure, if you want to run diagnostics on your logistic regression, you may have some learning to do, but you can easily solve that problem by purchasing a REALLY GOOD BOOK THAT IS GOING TO BE AVAILABLE FROM OXFORD UNIVERSITY PRESS IN JUNE FOR A GREAT PRICE.  Or you can spend thousands and thousand per year for SPSS.  But then you’ll still have some learning to do.

Bottom line – we have been teaching R to our students for a couple of years now (yes, we converted from SPSS) and we have seen NO difference in our students’ abilities to learn statistical methods.  If anything, our students are learning more now because they install R on their home computers and can practice skills we teach in class at home.

5)  All the cool kids are doing it!  R is growing in popularity across all sectors.  This includes academia, science, and business.  Microsoft, for example, just purchased Revolution Analytics, a firm that focuses on using R for analyzing Big Data.  R is powerful and everyone knows it, including the “big boys.”

6)  Resources abound.  You can learn about R, in large part, for FREE!  A good starting point is http://cran.r-project.org/, the home of all things R.  Here, you can find, downloadable manuals, a refereed journal, and a list of recommended books.  Or go to YouTube and enter “R + statistics.”  The last time we did that, we got 381,000 results.  And in the spirit of things being freely available, top-notch universities including Harvard, UC Berkley, and Johns Hopkins are offering classes on R via massive open online courses such as Coursera and edX.

R is just good for the planet.

And did we mention R and all of its packages are FREE?

Yup.  We love R.


No comments:

Post a Comment