We’ve all had good teachers and we’ve all had bad teachers, but the stories are largely anecdotal. But do some instructors have different grading patterns than others?
They are many different ways to quantify an instructor’s grading pattern. The simplest is just to calculate the average grade over all students and classes taught. This can be be biased because it mixes the advanced students (high grades) with the developmental (low grades) and required general education classes, and does not account for the number or variety of classes taught. Nevertheless, some interesting information appears even from this.
The average grade does not tell the whole story, however: two different instructors may have the same average by very different grading patterns, as the following data from Introductory Statistics shows.
Every class section of Introductory Statistics over a ten year period is represented in this heat map. Each horizontal bar represents a particular instructor, and the vertical thickness of the bar is proportional to the total number of students that instructor has taught (lifetime) in all sections of Math 140. The colors represent grades given. The bars are sorted by average grade. The instructor at the top has influenced a large number of students and given a lot of high grades, while the collection of instructors at the bottom tend to fail large numbers of students,
Tenure Track vs Lecturer: Does it matter?
In some cases, there is a significant difference in grading pattern between tenure track instructors and lecturers who are not hired on the tenure track. We compared all classes in the Math department over a ten year period that had over 100 enrollments and which were taught at least one tenure track instructor and one non-tenure-track instructor. We then compared the graded distributions using the Kolmogorov-Smirnov test. This test tells only if the grading patterns are different. It does not tell us if one class of instructor graded higher or lower or had a more tightly spread distribution. The lower the KS p-value in the following table, the more likely the classes had different distributions. The closer the KS p-value was to 1, the more likely the different types of instructors graded in the same manner.
|Number of Students in Sample||KS P Value||Difference|
|Developmental Math 1||12468||0.96||None|
|Developmental Math 2||24351||1||None|
|Algebra (Elementary and Middle School Teachers)||2300||0.96||None|
|Geometry I (Elem & Middle School Teachers)||3091||1||None|
|Geometry II (Elementary and Middle School Teachers)||187||1||None|
|Discrete Math (Math)||621||0.01||Significant|
|Discrete Math (ACM)||480||0.46||Significant|