The first person in line gets to pick. The first passenger in the car sits shotgun. And the first student graded gets … a better grade? 

 Sequential grading bias is the unintentional difference between grades based on the order in which students are graded. 

 “Sequential grading biases may be at play if grades and/or feedback are related to the order in which assignments were graded or the perceived quality of preceding submissions,” Kim Sherman, licensed psychologist and senior professor of practice of psychology, said. 

 A University of Michigan study showed that submissions graded later alphabetically received up to a 0.6-point difference compared with their earlier-graded counterparts. Students with surnames in the middle of the alphabet also tended to receive more negative or impolite feedback than classmates graded first or last.  

 “Biases of all kinds may be more likely to shape our responses when we make snap judgments,” Sherman said. 

Interestingly, students with last names U-Z receive higher grades in both random and alphabetical orders, while last names P-T receive worse grades when alphabetically ordered. One possible explanation for this phenomenon is fatigue.  

 Fatigue: 

Tulane University has approximately 2,000 faculty as of 2024, with an 8:1 student-to-faculty ratio. While the average class size is 21 students, some lecture courses have up to 120 enrolled students in multiple classes. 

 “We kind of suspect that fatigue is one of the major factors that is driving this effect, because when you’re working on something for a long period of time, you get tired and then you start to lose your attention and your cognitive abilities are dropping,” Jiaxin Pei, a doctoral student at the University of Michigan, said in the study. 

 To counter this, professors can grade all answers to one question at a time before moving on to the next. 

 Multitasking: 

Switching between answers leads to unnecessary strain on graders. As the brain is only capable of performing one task at a time, multitasking leads to lower task quality and more frequent burnout.  

 Sherman suggests dividing assignments into chunks instead of trying to grade students all at once. In doing so, personal feelings or tiredness are less likely to affect grade quality. 

 Furthermore, the online grading system itself may be a problem. 

 Randomized grading: 

Learning management systems often automatically order students alphabetically by their last name. Canvas is the fastest growing LMS — implemented at 32% of U.S. and Canadian higher education institutions — yet its default setting poses a disadvantage for students sandwiched in the middle of the alphabet.  

 Compared to students graded randomly, submissions graded alphabetically tend to have a greater disparity between grades — a difference of up to 0.004 points in their cumulative GPA, which is statistically significant in large sample sizes.   

 “To address potential sequential biases, instructors can grade submissions in a random order … by manually shuffling the sequence,” Sherman said.  

 But even with random ordering, students graded first still receive about 3.5 out of 100 points more than those graded last.  

Decision biases: 

Sequential decision biases, or perceived shifts in quality between submissions, can also affect how students are graded. If the student previously graded received an extraordinary grade, the student graded afterwards — though still scoring well — appears to have done worse than they did. 

 A study from the University of California, Berkeley shows that after giving three low scores, there is a 12% increase in grades while there is a 6% grade reduction after grading three high scores. This representative heuristic results in inconsistent grading, especially for subjects like the social sciences and humanities, which are more subjective in their grading criteria than STEM subjects. 

 “Rubrics or other clear grading guidelines can help promote consistent decision making and focused feedback,” Sherman said. “When there is more than one grader, everyone should be oriented to the same rubric or grading guideline.” 

 While grading biases may be discouraging for students, acknowledgment is the first step towards making amends. Simply randomizing students on Canvas and taking frequent breaks during grading could prevent students from receiving the short end of the stick. 

 “It is … important for instructors and teaching assistants to slow down our thinking and ensure our responses center student learning and self-efficacy,” Sherman said.