Many college level exams are poorly constructed. Doing good in these exams in no way represents that students actually learnt what they needed to from the course. This is mostly because many of these exams are "hack-able", where students have learnt many strategies on how to do well without having learnt much in the course.
Three reasons for such a poor state of college exams:
- Conflict of interest: For most public Universities, graduation and academic progression rates are key factors which determine how much funding they receive from the government, and to attract new students. The University administrators in turn transfer these incentives to the faculty members who are rewarded for passing more students. This inherently creates a conflict of interest as in almost all cases, the faculty teaching the course also prepares the exams and grades them. This conflict of interest in exacerbated by the fact that students provide good ratings in their course evaluations to easy graders.
- GPA used as a screening factor: Many companies use student's GPA as an initial screening factor during interviews. This incentivises students to emphasize grades over learning, and to reward faculty members who give good grades.
- Lack of will: For most faculty members, their success in career depends almost entirely on their research accomplishments, and their teaching skills play a very insignificant role in promotions and market value. Creating good exams takes a lot of time and effort, and when you do, you get push-back from students who are used to a particular type of testing. And as noted in the first point, the incentives from the administration are also most often not geared towards adequate testing (it takes a lot of effort for the administration to verify this too).
There are solutions to all of these. The COI can be easily resolved by having someone other the course instructor prepare and grade the exams.
Western Governors University for example has implemented such practice. If more industries adopt a wholesome screening approach instead of just relying on the GPA, students may be less inclined to overemphasize their grades. I think machine learning has a role to play here in making this process more efficient. For the third point, I think we should start seeing bifurcated faculty roles in Universities where some faculty members have a heavier teaching load (and evaluated accordingly) and others have a heavier research load. We also need innovative tools which help students make informed choices about Universities in terms of teaching quality, likelihood of success and return on investment. If these tools are easily available, I think students will start holding Universities accountable for their teaching practices.
