CBA? More like LOL.

I was a bit disappointed that the van der Kleij et al. article didn't find more substantial results. Computer-based assessment is something that figures fairly heavily into the classes I teach: much of the homework my students compete is via an online platform that has automated versions of some of the textbook activities. While there are some redeeming aspects of this platform--for example, there are animated grammar tutorials that students can watch (and that I can assign to make sure that they watch), and the videos are usually fairly well done--one of my biggest critiques is the quality of feedback students are given. When they get an answer wrong, they're told that they got it wrong, but they aren't told why. After they've exhausted their two allowed attempts, they're shown an answer key with the correct answers, but once again, they aren't offered any sort of explanation as to why their answer was incorrect, why the correct answer is the correct answer, and what they can do to improve moving forward. My "solution" to this has been to try to give them much more elaborated feedback on the homework they had in during class, and to give them credit for this homework based on completion rather than accuracy. This approach seems to have at least partially mollified my students, but honestly, my preferred way of fixing this would be to ax the online platform and use GauchoSpace to design my own activities. :D In a perfect world (read: in a world where I actually had the power to throw out part of the current course requirements and put in something that [I think] would work better), I'd do it in a heartbeat, but things like power and, of course, available time are currently not in my favor. :-/

What have your experiences been with CBA? Do you find the type of CBA offered to your students to be effective? If so, why? If not, what would you like to change about it?