Skill Progression in MIT App Inventor — Benjamin Xie

App Inventor can be summarized as Scratch for grown ups 🙂

Like Scratch, App Inventor is open ended, so how do we know people learn programming? Benji wants to measure what and how people learn. He examined a dataset of 10k users that created at least 20 projects, because he wants to study skill progression across projects. He is not just counting the number of blocks, but the block types, discounting blocks that are not connected (dead)

Benji will investigate six computational concepts: {procedure, variable, logic, loop, conditional, list}. Of the 1,333 different types of blocks found in projects we analyzed, 39 of them are computational concept blocks (CC blocks) that relate to computational concepts, stemming from a computational framework commonly used for Scratch [1].


He will measure breadth (new blocks used) and depth (number of unique blocks).  The results show that over time, breadth decreases, but depth increases. So, users initially try out new blocks, and then use many of them in projects.

2016-09-07 10.58.47


All stuff related to this talk on Benji’s website (commendably well organized!)

[1]  K. Brennan and M. Resnick, “New frameworks for studying and assessing the development of computational thinking,” in 2012 annual meeting of the American Educational Research Association, 2012.