Process
The Georgia Institute of Technology’s Student Government Association created a website that maintains a database of information on a breakdown of the grades of every section taught in the school. This information is publicly available for students and faculty. It helps students pick which class they want to take with which professor and it gives faculty feedback on the classes they have taught. While the website is useful for looking up a particular professor or course, our team was interested in what kinds of larger correlations and patterns we could gather from the data available from the course critique. In particular of interest was to find any correlation between the lower level and upper level courses and their respective grade point averages. The size of the database (over 34 thousand entries) makes it impractical to look at every course in the entire school. Thus, we decided to limit our queries to introductory courses in CS & Physics that all students must take as well as a few broader queries about the College of Computing, as this was the area we were most familiar in. Here are the specific questions we focused on:
Is there any correlation between the GPA of 1000 to 8000 level courses?
Are there any differences in terms of grades in the required intro CS courses that all Tech students must take?
Within the introductory Physics courses, what kinds of differences in GPA do we see across professors?
Are there differences in the GPA for a given course when you compare what term the course was taught (summer/fall/spring)?
SGA’s Course Critique in the form of an Excel spreadsheet provided the database that we used. Some of the data was calculated using Excel while most of the visualizations was done in “Many Eyes,” a web visualization program by IBM. It is important to note that some of the data was removed from the database for several reasons. Many of the classes contained 0 students, producing a 0 GPA which would have skewed the data. Also, we left out any “Undergraduate Research” or “Independent Study” courses, as these were one-sided as well.
Results
The average GPA of CS course, out of 1729 sections was a 3.06 [1]. Doing a calculation on the different levels of CS courses here are the results we found:
1000 level – 2.68 GPA
2000 level - 2.61 GPA
3000 level – 3.15 GPA
4000 level – 3.40 GPA
7000 level – 3.56 GPA
8000 level - 3.76 GPA
[1]
We can see a clear correlation between the level of courses and the GPA. Also interesting was the relationships between the percentages of As and the percentages of Fs with the lower level and upper level courses. [2] [3]
[3]
Intro Level Computer Science Classes
[4]
As we can see from the bubble chart, the classes with the most students are the intro level courses. Here are the classes with the largest sizes in order of greatest to least:
1371
1322
1321
4400
1315
1050
This is understandable considering that an entry level CS course is required for most majors at GA Tech. Another question that can be raised is what the grades of these classes look like and if there is a difference between the classes. Together these 723 sections of intro level CS courses + CS 4400 represents 25,198 students out 53,783 total. Without CS 4400, the numbers drop to 22,116 students.
[5]
When we plot these 5 intro CS courses we find something unexpected. While CS courses were required for non-CS majors, CS 1315 was created for a more media related and less intensive course that was generally seen as easy. However this data shows that there is no substantial difference between the intro courses [5]. 1050 has an average of 2.7, 1315 a 3.0, 1321 a 2.5, 1322 a 2.5 and 1371 a 2.7. While there is a slight bump in 1315, the results are negligible. However when we see the percentages of Fs we see a big difference between 1315 and any of the other intro courses [6].
[6]
Introductory Courses in Physics
In addition to computer science, every student at Georgia Tech also must take an introductory series in Physics and Calculus. As with the CS course, these freshman requirements are dreaded by many students and have a bad reputation around campus. By looking at the grade data for these introductory courses in several different ways, we found some interesting insights into which semester and professor might give students the best chance for success.
[7] GPA and Percent A for Physics I.
In figure 7 we see a plot for the average GPA for several different professors teaching the introductory Physics I class. The x-axis shows the percentage of students who received an A in the course and the y-axis shows the average GPA while the bubble size represents the number of students that a particular professor has taught. While the smaller circles, representing a professor who has only taught the course one or two times, are scattered about the space, we have highlighted the 4 professors who teach this course most frequently (each has taught over 1200 students). From this data we can clearly see that taking the course with Eric Murray might not be the best option for students who have a choice in the matter. While the GPA for students in Murray's offerings is only 2.0 and on average, a mere 9% of his students receive an A. In comparison, Schatz and Davidovic give out approximately 16% A's and the average GPA of their courses is a 2.4.
[8] GPA by term for Physics I - WDF rate and [9] GPA by term for Physics I - Withdraw rate
In figure 8, we explored the variations in grades for Physics I across different semesters from Spring 2002 (Term 0) through Fall 2006 (Term 14). In this scatter plot we have time along the x-axis and course GPA on the y-axis. Instead of representing the number of students in the course (a fairly consistent number in fall and spring with fewer in the summer), I chose to use the circle size to represent the percent of students who dropped, withdrew or failed the course in figure 8 and just those who withdrew in figure 9. Figure 8 is fairly straightforward, in the semesters with more D's and F's the overall GPA was much lower. What's interesting to note here, is that in several of the semesters when the overall course GPA was low, we also see the largest withdraw rate. That is, enough students realized early in the term that they would be better off dropping the course, and even after these students (presumably those with the most severe low grades) left the course, the final GPA remained quite low, suggesting that even when the course is filtered to contain the only the more capable students by drop day, grades remain quite low.
It is also interesting to note that while grades in Physics were quite low several semesters in a row (from Summer 2003-Fall 2004) the grades were higher both before and after this period. Looking back at figure 7, suggests taking a look at when the different professors taught the course. As one might have predicted, Schatz primarily taught Physics I in the first 3 terms, then Murray taught several sections of the course in terms 4-8 (with lower GPAs than other instructors during those terms), and we found several sections taught by Davidovic in terms 11, 12 and 14. However, Murray also taught some sections of Physics in the semesters with slightly higher average GPAs and the grades by Davidovic vary greatly from term to term with a low GPA of 2.04 in Fall 2004 and a high of 2.82 in Fall 2005. This suggests other factors are at play. This prompted a search of the campus newspaper for information about Physics courses, and sure enough, in Spring of 2005, the semester in which grades showed a sudden increase, the Physics department had just implemented a new curriculum. The Physics department had previously adopted a new textbook (in Fall 2004), and then in the spring they altered the curriculum in favor of teaching fewer topics, but covering concepts in more depth and detail. Furthermore, in the spring of 2005, course attendance also became mandatory and the outside of lecture laboratories were augmented with recitation-style meetings to help students with problems in a smaller instructional setting than the 200-person auditorium. It's encouraging that the grades suggest that the new policies and curriculum for the course is an improvement as both GPAs seem to be higher and there is a lower W and WDF rate.