From an educational philosophy perspective, learning to program is well aligned with what we strive to teach our students. Becoming proficient in statistical programming requires the ability to think critically about complex problems, to develop scientific research questions, and to apply rigorous analytical methods to answer your questions. I believe these skills are key elements to a strong liberal arts education. We want our students not only to be able to think critically about the world around them, but to have the skills to critically engage with the world.
Using R will require our students to conduct their work using a more scientific process. R will require students to plan ahead and to think more holistically about their research process. Students can learn to conduct their research in a reproducible manner using R, rmarkdown and knitr. The entire analysis process can be conducted in R (data manipulation, exploration, analysis, visualization, etc.). Students can even combine their text and R code into a completely reproducible analysis. Applying a reproducible research methodology to student work can greatly contribute to the learning process at a small liberal arts college. Students’ successes will be validated, reproduced and built upon, while their mistakes and misunderstandings will be more clearly identifiable and correctable.
From a more practical perspective, R is quickly becoming the default programming language of data science. If we want our students to be prepared to compete for opportunities in quantitative fields it is in their best interest to learn R. Not only is R used at leading tech companies like Facebook and Google, but it is also used throughout academia. Additionally, R is open-source and free. Unlike many of the software tools students use at Reed, students will always have access to R. There is something beautiful about the best tool being free and continually developed by its users.
A typical argument against learning R in an academic environment is that it takes longer to learn than other statistical applications often used in teaching (i.e., Stata and SPSS). I no longer believe that this is a valid or even a reasonable argument. Many people are investing tremendous effort into making R easier to learn (e.g., http://swirlstats.com/ and http://tryr.codeschool.com/ ). There are clear online (free) resources for learning R and a supportive and expanding user community. Even if R did take longer to learn (which I don’t believe), the pedagogical and practical advantages to learning R almost certainly outweigh any additional cognitive costs to learning it.
There is a lot more that can and should be said on this topic. Please feel free to contact me at rich.majerus@gmail.com if you have comments, critiques, or concerns about this post. To learn more about using R at Reed College please see the resources that we have assembled here: http://reed.edu/data-at-reed/resources/