How do you pitch R to new users?

[rdpeng]

I thought it would be nice to have a discussion of

1. Who are the types of people who are new to R (SAS users, non-statisticians, programmers, etc.)
2. How do we craft a pitch for R to these different audiences?

[adam-gruer]

Great idea. We could tease out a lot of interesting ideas in a discussion like that.

[timchurches]

This year I've been teaching (or attempting to teach) R as part of an informal fortnightly seminar/tutorial series on computing for health research to researchers and researchers-in-training at the institute/clinical school where I work. It's been a partial success, and I'm sure it will be better next year... (the swirl tutorials by @rdpeng and colleagues were a godsend in the first few months, btw, and we plan to develop a few swirl packages of own own).

I pitched the seminar/tutorial series at research students: undergrad medical student doing their mandatory research year, Honours students, and Masters and doctoral level research candidates.

The response was good and the sessions were over-subscribed, but quite a few research students told me that they had asked their supervisors whether they should attend the seminars/tutorials, and they had been told "no, we use Stata/SAS/SPSS/Excel for analysis, it's a waste of time learning anything else", or words to that effect. Now, from the supervisors' points of view, that's understandable - they don't really want their research students to be writing code in R or some other language that they can't make heads nor tails of (although many research supervisors don't seem to use anything other than Excel). However, this just perpetuates the reliance on and entrenchment of these legacy products - if you are forced to use SAS or SPSS for your doctoral thesis, you are very likely to want to keep using it. The same supervisors sometimes then complain at research committee meetings about the cost of licenses for the software products they and their team rely upon.

How to address this problem? That is, introducing trainee researchers to R (and/or python) in the face of resistance from their supervisors?

It is closely related to the problem of introducing formal or semi-formal training in computational thinking (in which learning to programme plays a major part) into research training in general. We have encountered some push-back, or just incredulity or lack of comprehension, from more senior faculty members to any suggestion that doctors-in-training should learn how to think computationally, despite the fact that it is now pretty clear that their professional careers will be dominated by statistical machine learning models helping them decide what to do (or increasingly, telling them what to do). Which is somewhat ironic, because the teaching of evidence-based medicine is now completely pervasive, and most of that evidence is based on statistical (and increasingly ML) models of what's best.

[rdpeng]

The state of medical education today is probably something that deserves its own conference! I'll leave that aside for now.

The question for me is which facets of R do you highlight to which audience? For example, earlier on, I used to teach R to people who were more advanced (older) and had already established themselves with a particular system (S-PLUS or SAS, for example). So my pitch there was that the programming language of R was very flexible and expressive and allowed you to do new things that the multi-billion dollar corporations hadn't bothered to code yet.

But as I started teaching R to younger and less experienced people, the relevant comparison was not to S-PLUS or SAS (they likely have never heard of them) but rather to Excel or even just doing some things by hand. Here, I tend to emphasize things like knitr and tidyverse tools and automation.

For people who are C coders (or the like), I tend to emphasize things like rapid prototyping, access to packages/libraries for machine learning and statistics.

I haven't yet developed a pitch for people who are using python yet....

My point is that if R is an elephant, then for some people I show then the tail, some people get the leg, and some people get the trunk. I wonder how others do it?

[Lingtax]

I regularly teach R to Psychological researchers (and increasingly to other health disciplines). Given that they typically come in competent (loosely defined) in SPSS, I tend to try to sell them on direct relative advantages e.g. easy to do everything they already do in SPSS, plus the things they need to use other software for (e.g. path analysis, meta-analysis, etc.). Then I start to sell based on things they didn't even know was possible, e.g. shiny, markdown etc. and the advantages of learning a non-proprietary system. I find the conversion rate is higher from SPSS to R than it is from STATA to R.

When they start to get over the initial aversion, the fact that R doesn't take 3-6 minutes to load is also appealing.

[stefaniebutland]

rOpenSci Community Manager here. I love this idea.

If you @rdpeng or a group writes something up, I'd love to feature it on the rOpenSci blog. (I know you have plenty of outlets for your stuff Roger so no obligation.) I'm hoping a couple of ozunconf projects will get written up as blog posts that we can share on the rOpenSci blog to highlight the Aussie community. @njtierney has agreed to write a general ozunconf post to go up next week.

This reminds me of something that came out of our unconf17: Hey! You there! You are welcome here by Shannon Ellis from this issue: https://github.com/ropensci/unconf17/issues/63

[rdpeng]

I'm happy to write something up if there's interest at the unconf this week.

[anikobtoth]

Plenty of researchers/academics from the natural sciences are also just starting to use R! I have been trying to encourage the use of R in my labs and working groups, but it's often very difficult to get people on board. I am interested in how to get more people to use R for the sake of producing good science that is easily reproducible.

[duttashi]

statistics and mathematics if taught incorrectly, leaves an indelible mark of "fear" in young minds. Well, this is what happened to me when I was taught mathematics at school. To overcome this, I present R thinking in a story-based format to beginners or novices on my blog. I will be teaching R programming this semester at the undergraduate and master's level. The audience largely have never heard of R. To have the audience captivated, I propose an arbitrary question to them like, "A large income is the best recipe for happiness I ever heard of” quotes the famous English novelist Jane Austen" and then build my R-powered story around it. My point here is. R is a powerful language and its power can best be articulated iff wielded correctly.

[stefaniebutland]

@rdpeng's blog post on How do you convince others to use R?: https://simplystatistics.org/2017/10/30/how-do-you-convince-others-to-use-r/