Can students self-generate appropriately targeted feedback on their own solutions in a problem-solving context?

It was observed that in a lecture-format problem-solving course, the amount of homework completed by students only had a weak effect upon exams scores. Three changes were made to the approach used to teach the course in an effort to increase the effectiveness of homework assignments. First the class was flipped so that in-class problem-solving activities could be used to provide practice and feedback to students prior to assessing their learning. Next, homework grading was changed from a correctness basis to an effort basis. Finally, the identification of problem-type was explicitly taught, allowing the use of homework wrappers. The homework wrappers asked students to reflect upon their thought processes at the time they solved each homework assignment, comparing it to the general approach they had been taught for that problem type. After the changes were implemented, completion of homework was observed to have a stronger effect upon exams scores. The slope of the trendline in a plot of exams scores as a function of percent homework submitted increased from 0.171 to 0.435 after the changes in the teaching approach, and the difference in exams scores for those who submitted more vs. less than the average amount of homework increased from 1.43 to 12.45.