Patterns in student assessment of problem-solving software

We investigated whether there were patterns in student assessment of (the usability, ability to learn from, and usefulness of) online problem-solving software tutors. We found that the correlation between the learning of students (how much they needed the software tutor, how long they worked with it, how much they learned from it) and their assessment of the tutor was weak. Further, we found that four-year college students assess the software tutor more favorably than two-year students, non-Caucasian students assess the tutor more favorably than Caucasian students, and female students assess the tutor more favorably than male students.