For some students, there is pressure to be at the top of their classes. This stress can come from parents, peers, or personal expectations. Final grades seem to follow a certain pattern that measure various students’ performances. What if grade distributions are not a common trend though, but instead they are a common belief that has not been adequately tested?
U of T researchers Elizabeth Patitsas, Jesse Berlin, Michelle Craig, and Steve Easterbrook have disproved the ongoing myth that computer science grades are bimodal. They also found through a psychology test done on Computer Science (CS) instructors that the cause of this myth is the ongoing belief of the ‘geek gene hypothesis.’
In the computer science community, there is a belief that final CS grades have a bimodal distribution. A bimodal distribution means that more than one set of grades occur the most. Hence, a bimodal graph would show two ‘modes’ indicating the most common range of grades — for example, students who got an ‘A’ and those who got a ‘C.’
The geek gene hypothesis describes those students who have the ‘natural talent’ to program, explaining the mode centred around higher grades. However, Elizabeth Patitsas notes that “[The study’s] findings support existing literature that CS students cannot be simply categorized into those who ‘get it’ and those who don’t. Instead, most students are somewhere in the middle: they get some things but not all things.”
Patitsas’ research on bimodality broke down into two studies: the first study tested for multimodality in University of British Columbia (UBC) CS grades. The second study was a psychology test done on CS instructors to see if the geek gene hypothesis influenced their ideas on bimodal grades.
In the first study, Patitsas and associates did a statistical analysis of 778 CS lecture’s final grades from UBC collected over 18 years. They found that only 5.8 per cent of the distributions were multimodal. The rest of the distributions were normal with some skew.
Now that the researchers proved that bimodality was not a common distribution, they wanted to find the source of this belief. Patitsas and associates did a psychology test on 60 CS instructors to observe if the geek gene hypothesis affected their ideas on bimodal grades.
They separated the instructors into two groups. Treatment 0 was not primed to think about bimodality in grades. They had to categorize 6 ambiguous distributions, including normal, skewed, bimodal, etc. Treatment 1 was primed to think about the geek gene hypothesis. The researchers asked the instructors about the geek gene hypothesis before the experiment and then the instructors categorized the distributions.
The results showed that for Treatment 0, there was a significant relationship between instructors interpreting the ambiguous graphs as bimodal and the participant’s responses to geek gene hypothesis questions. There was also a stronger relationship for Treatment 1 over Treatment 0 for academic success and bimodality.
According to Patitsas, it was “found that confirmation bias and the geek gene hypothesis” plays a role in the bimodality belief, because the instructors whose beliefs consisted with the geek gene hypothesis were more likely to label the ambiguous histograms as bimodal.
This research provides insight for what needs to change in the CS community. The article notes that the geek gene hypothesis is popular because it is used as a social defense. As an instructor lives out their career, their faith in their teaching abilities fluctuates. According to the article, a teacher with high self-efficacy who sees poor performance believes the problems lay within the students themselves, rather than their teaching abilities. They may subconsciously use the geek gene theory to rationalize the performance and therefore use that as the reason for bimodal grade distribution.
The bimodal distribution has not only been disproved but is an ineffective belief if CS instructors wish to improve students’ marks. There should be an emphasis on improving curricula.
“Disproving the geek gene is important for CS education. It means that education can make a difference to our students. It means instructors can’t simply write off the students they think ‘don’t get it,’” states Patitsas.
“If CS educators think of their job to help the average student — rather than simply the students who magically ‘get it’ — they will have more motivation to switch to effective instructional methods like peer instruction,” notes Patitsas. This can be done through smaller tutorials and labs for students to get more one-on-one help.
Overall, Patitsas and associates have not only disproved the bimodality belief but have also proved that the geek gene hypothesis should not be a factor for how instructors judge their final grades. The CS program as a whole must take steps to suit the curriculum to the students needs so the ‘have it’ versus ‘don’t have it’ mentality can die out.
