In the spring of 2016, KTH changed its grading system for degree projects to simple Pass/Fail scale. The reason behind this move is probably familiar: heterogeneity in the application of the A-F scale between different departments. However, given that the study plans which the students had registered on included this scale, they were offered a one-off chance to actually choose their grading system in the degree project.
The author used this opportunity not only to record the choices of the students (a very useful instance of revealed preferences) but also to enquire about the reasons of their choices. A questionnaire was sent to them and their answers were then matched with their actual pick. After that, a logit econometric model was implemented in order to validate the students’ stated reasons and to establish the actual drivers of their decision.
As it turned out, a perceived reduction in stress and an increased motivation to do a good job seem to be the major explanation for those going for the simple scale whereas a perception of a greater confidence in earning an A were invoked in the case of the multistep scale. Interestingly enough, the GPA grade before the project of those who went for this latter scale was significantly lower which is compatible with them making this choice in order to raise their grade. This interpretation receives larger support knowing that a post-analysis of the results showed that their grade in their project was also higher than their previous GPA.
Comment. This study has several interesting aspects. First, it is a very good example of how one can extract reliable information from self-reported data. Matching the answers to a questionnaire with actual revealed preferences is a procedure that validates what someone says which should always be taken with caution as guidance to explain what people do (an important lesson to be learned by many of those in the business of interview-based research).
Second, the paper has also an important takeaway for our university administrators. At a time when their pockets seem to be shrinking considerably, a simple Pass/Fail scale could save resources in as much as it reduces the task of examiners. Note that in a population of high achievers with an average GPA score of 4.219, only 45% of them pick the multistep scale in this sample. Also, many of these students do not seem to particularly believe that a good grade on this finer scale will influence their wage at first employment (or throughout their career for that matter).
It would have been interesting to compare the choices of the students with their expected performance in a more direct way. Indeed, the examiners could have been instructed to grade on an A-F scale anyway, to be simplified to a Pass/Fail one for those pupils who requested it. A more interesting comparison could have been made then. Actually, a post survey analysis could have communicated the students their grade in the finer scale and ask them if they would have changed their choice. This could have illuminated the issue even further.
Finally, there seems to be a gender aspect to the results given that the proportion of women that chooses the simpler scale is larger than the one that chooses the more sophisticated one. This also raises the suspicion that issues like risk-aversion and overconfidence could also be playing a role. Indeed, a multistep scale gives opportunities to shine if the grade is A but also to raise the eyebrows of future employers if the grade is E. The latter can be avoided at the expense of the former. A preference by female students for a simple scale is compatible with what we know from research in psychology: men tend to be more overconfident and risk-loving than women.
Text: Iñaki Rodríguez Longarela, Stockholm Business School
