I think that covering installation/setup in greater depth has already been emphasized, but I wanted to further hit on some points.
Audience
In my biomededical science program, we were taught R/RStudio/Rcmdr for our completely online biostatistics course. So many of my fellow students never really wrapped their head around that Rmcdr, Rstudio were the same backend of R, just with different overlays. The concept of an IDE was completely foreign to our non-progammer class, although many people had been using SPSS, SYSTAT, Excel, Origin, etc for statistics and graphing for at least 1-2 years at this point. These were motivated, educated academics, but the way that we were introduced to R was poorly done, and lead to many people fighting back against it. I think Davis had a good approach (albeit probably a longer process and focused on in-person education).
"Why use a programming-based program when I can just point and click in SPSS/SYSTAT/etc." -- Many students in the course. They were hard to convince to switch from the "ease" of a GUI compared to a programming based statistical suite as almost all of the stats were on tiny (and already tidy) datasets.
I think it is so important to show how the proposed process is better and then build it up from there. The light bulb clicked for me when I saw the limitations/time costs of doing some things by hand in Excel versus programming/automation in R. Once you are convinced R is better, it is much easier to self-motivate to learn various things outside the scope of the course/community/task at hand.
The other big takeaway was the student's desire for video-based examples, as some people had difficulty following along purely text-based examples in the online-only course. Video fills in the gaps with some of the basic computer functions, and often WHY something is being done AS it is being done which allows highlighting the feature being taught. Gifs can also accomplish this, but sometimes the voice overlay and ability to pause and rewind on demand is more helpful.
Finally, I get the appeal of Rcmdr, but good lord did it set our group back in the actual use of programming-based R via RStudio and saving/writing scripts. 