Learning how to use a new statistical GUI? The question is in the header, but I would extend the context a bit.
Next semester I am due to be a teaching assistant (TA) in a course in statistics, where I would need to help sociology students learn to use SPSS. I don't know SPSS, yet, and would like to learn how to use it.
I was thinking of taking a simple dataset, and start reviewing it with methods I know, thus starting to map out where are methods I know of.  And once finished, to try and explore more options.
Can someone propose other/better strategies to master a new statistical graphical user interface (GUI)? (in my case SPSS, but it could apply to many other GUI's).
 A: Since you are pretty well versed in R, get a copy of Muenchen's "R for SAS and SPSS Users" (Springer, 2009) and work backwards.
A: As someone who made the shift the other way from SPSS to R, I'd say that SPSS is relatively simple and intuitive relative to R. The menus and dialog boxes guide you through the process. Of course this means that it is also fairly easy to run analyses that don't make sense. And the GUI  leads to less flexible analyses and tedious button pressing especially for repetitive analyses.
Thus, your approach of taking a dataset and just playing around might be sufficient.
There's plenty of how-to books out there, such as:


*

*Discovering Statistics Using SPSS

*SPSS Survival Manual


There's also plenty of websites offering tutorials: 


*

*Andy Field

*My old notes

*UCLA on SPSS
I'd also recommend that if you are teaching students about SPSS, you encourage them to use syntax. Using SPSS syntax is not as good as using technologies like R and Sweave.
However, using syntax is much better than just pressing menus and buttons in an ad hoc way and then wondering later what you've actually done.
I wrote a post listing tips for using SPSS syntax in order to approximate reproducible research with SPSS.
A: In youtube there are a lot of videos related with SPSS, I usually learn a gui with them. I recommend you record videos to teach how to apply methods and encourage them to use the "paste" button to learn a little of syntax
A: Your method will certainly work for most introductory techniques you'd need to do. I am quite familiar with R, but I was required to learn Minitab to teach to non-statistics people. In one afternoon, I had a basic enough understanding of Minitab to explain how to use it in the limited framework of an introductory stats class. I figure SPSS would behave similarly.
I would suggest reviewing the course syllabus and trying to complete all techniques outlined in the syllabus. A large majority of your students' questions will pertain to doing tasks the teacher has asked them to do, and a smaller percentage of questions will be related to doing tasks in their other statistics endeavors. 
You will be able to answer a large majority of student questions by knowing how to use SPSS to solve a class-related problem, so just playing around within SPSS will benefit you.
