Some similarities with Felleisen's work teaching programming to early high schoolers,
- http://www.programbydesign.org/
Those folks have tons of data on what works, what doesn't.
I get a bit of a NIH feeling from this effort, that I hope is unfounded.