Sources of Grade Variability in Multi-Session Lab Courses: Should We Seek Legitimate Correction Factors to Mitigate Detectable Variation in Student Grades?

 

Saewan Koh & Mark Wolansky

Tested Studies in Laboratory Teaching, 2014, Volume 35

Abstract

Grades amongst groups in large multiple section lab courses can be highly variable. The spread amongst lab sections in our own courses were at times +/- 4.5%. This spread is of concern since we strive for fairness in how students are evaluated. More importantly students themselves expect this to be true. We examined grades from two large multiple section lab courses (819 students and 358 students) to identify the sources of variation in student’s final lab grades. TA, lab section, room assignment, day of lab, time of lab and home program of individual students were used in a linear mixed effects model to determine which variables were important in determining a student’s final lab grade. Some of the sources of variation we detected such as room assignment can be controlled for administratively while best practices in TA training can reduce TA bias and thus a portion of the overall differences in scores amongst sections. Surprisingly, we discovered other significant sources of variation that were self-selected by the students and so out of reach of any mitigation process that we would be able to perform. In this mini workshop we will present the results from our analyses and then open the floor to discuss potential strategies to reduce grade variation. For each source of variation we will also discuss whether or not mathematical corrections should be applied to mitigate its effect on overall grades if it were possible. We ask when if ever would these corrections be appropriate.

Keywords:  grading, GTA training, multi-lab sections, variations in grading

University of Calgary (2013)