An earlier study using sequences of online learning modules found that a significant fraction of undergraduate students were unable to solve similar new problems after learning from an online problem solving tutorial. The current study examines the effectiveness of two methods to improve students' subsequent problem solving performance. First, an "on-ramp" module designed to help students develop proficiency in relevant basic skills was added prior to the tutorial. We found that students' performance on subsequent modules improved significantly over the previous year, and in one of the two sequences we found evidence to support that the improvement was due to the addition of the on-ramp module rather than other irrelevant factors. Second, a new "transfer" module was added after the tutorial and before the final quiz module in which half of the students were given a compare-contrast task and the other half were asked to answer tutorial-style scaffolding questions. On the subsequent quiz module, we did not find significant performance differences between the two conditions, nor did students' performance significantly improve over the previous year. The study demonstrated that mastery-style online homework can serve as an efficient and flexible method for evaluating the effectiveness of new instructional designs.