the perils of curving: centering the grade average
“you are just inflating the test scores to make it look like someone has a higher understanding than they actually do.” (talking about leaving revisited, p 203)
“the test score becomes ‘how did i do compared with other students?’ the average for our last test ranged from 40-60, so i don’t think anyone knows what they are doing, but they are content because they did well compared to others.” (talking about leaving revisited, p 203)
it can be hard to write an exam that targets precisely the level of knowledge and skill your students should have; sometimes questions are harder than we intend, sometimes students misread them, sometimes the exam is too long for the allotted time, all of which can result in low exam scores. it’s therefore common practice to curve–or more accurately, to adjust the meaning of the exam score, typically centering the exam average at a b or a b-. the intentions behind the adjustments are all positive: no one wants their exam to have a d average, even if the numerical average is in the 60s.
unfortunately, however, curving decreases student trust in the meaning of grades as well as student motivation–and the bigger the curve, the greater the impact (wolfe and powell, 2015).
some of the most compelling evidence comes from students’ own words. in talking about leaving revisited, elaine seymour and anne-barrie hunter report their findings from interviews with 346 undergraduates; the interviewees were a mix of students who left intended stem majors and students who persisted in a stem major. about a quarter of them (29 and 24%, respectively) cited curving as problematic, saying that it decreased their trust in the grading process, reduced their motivation to learn, and promoted a competitive environment.
if curving has these impacts, we would be wise to avoid it. if students have already taken an exam and the scores are low, one approach is to allow students to do exam corrections for partial credit, an idea described more fully in this part of the guide. we also, of course, want to examine the questions on the exam to determine whether there are some invalid or otherwise problematic questions, a process called item analysis. but ultimately, we also want to try to avoid the problem. the rest of the guide describes methods that can help us write more valid exams and as well as course structures that reduce students’ test anxiety. both can help us achieve the goal of writing exams that measure our students’ learning in meaningful ways.
other sections:
- steps to creating more valid (and less stressful) exams
- class structures that minimize anxiety about exams
<<back to the beginning
.
.
.
.
.
.