More time on exams may not even help

I’m glad I don’t have to take exams. Not that I had a particularly strong aversion to them as a student but they were stressful, of course. Nothing like pitting your entire future on the choice between B, C, or D, all of the above.

Exams have come into the spotlight as more and more students apply for accommodations for extra time. Rose Horowitch wrote about this in The Atlantic last December. Educators across the country — myself included — have been thinking about it for even longer.

There's a nationwide trend in increased disability accommodations, but it's hitting different campuses differently. The Atlantic reported rates of 20-34% of students registered with disabilities, but this seems higher than on my UC San Diego campus. In one of my 100-student classes, typically three students will have some form of accommodation, on the order of the 5-6% rates reported across the UC system – we're likely under serving students [1]. 

Horowitch's article spurred a lively Commons section in The Atlantic's May issue. One law student noted that a curved grading system (common in law schools) means that you feel like you have to get accommodations in order to compete. This is a fair point, and in most contexts, curved grading is an outdated, unfair practice that usually isn't statistically defensible [2].

But there’s also an implicit assumption in the law student’s comment, which resembles other sentiments I’ve heard: students with more time on exams do better. Students themselves will ask for more time with the hope that having more time will help them improve – but does it?

There's at least one study that suggests it may not help most students, even those with test anxiety.

A physiology midterm in one or two hours

The 2025 study was conducted by two teaching professor colleagues of mine, and led by then Masters student, Nicole Marcus. It asks whether test anxiety and/or performance on an exam improves when students have more time.

They compared scores on midterms from human physiology courses in two different years, with about 200 students in each offering. The midterms were almost identical. The only thing that differed was how much time students were given: one or two hours. Nicole also collected information about students’ test anxiety, using a validated scale, and post-exam information about how students felt about the exam itself (e.g. “I had test-related anxiety during the midterm.”)

Here's the big takeaway: even with double the test time, students didn't perform better on the exam.

The average score of the two-hour group was actually one percentage point less than the one-hour group (67% versus 68%), but overall scores were not significantly different. There also wasn't a difference in whether students reported feeling anxious, though the percentages look different (81% versus 87%). Even students with the highest test anxiety didn't perform better with more time.

There are, as always, caveats. First, this is one course, two quarters, one university. It’s one data point that may or may not translate to other places. A slightly older study of an introductory physics course showed that students with more time performed slightly better, by 2%. Even still, it didn’t close opportunity gaps in the class. Other work has looked at when students turn in their exams and in most cases, finds that there isn't a relationship between submission time and grade received.

It is also not a study specifically about students who are requesting accommodations. Rather, this line of work addresses an assumption that drives people to request accommodations and also, less charitably, drives some to complain about others requesting them.

Second, my colleague actually does something very cool at the start of all of his exams — he gives students an expressive writing prompt, something that has been shown to help students with the highest levels of test anxiety. By briefly writing about them, students may be able to set aside their distracting and demotivating feelings to better focus on the task at hand. This is a great practice but one that may actually reduce the anxiety of all students in the sample, perhaps limiting the researchers' ability to see decreases in or based on test anxiety.

Finally, this exam wasn't intended to be time sensitive. Even in the one hour exam, 87% of students completed the last question. Many professors (including myself) write exams that they want students to be able to finish in the allotted time. In theory, if students understand the material, more time shouldn't be what they need.

A call to rethink assessment

When senior author Jim presented this work at our faculty meeting, the room was stunned. We've all heard requests from our students for more time, and many of us have wondered how to help struggling students while also assessing them in rigorous and logistically-feasible ways. The answer may not be giving students more time, but rethinking our assessment strategies.

In response to Horowitch's original article, one educator argues against gatekeeping around accommodations, suggesting instead, “Perhaps we ought to consider a more radical approach: rethinking pedagogies to be more inclusive so that accommodations are rendered moot.” Indeed, we have to ask ourselves whether speed is really the thing we’re hoping to test. If I’m training emergency room doctors then yes, I do care about the ability for someone to quickly tell me where they'd stick a needle. But if I’m training neuroscientists or data scientists who I want to be thorough, rigorous, and critical, time may not be of the essence. I left the faculty meeting thinking about how I can incorporate expressive writing and a little bit more breathing room into my exams.

The findings here also help us coach our students into better practices and direct their attention to aspects of test preparation or taking that might actually matter. We can help students understand what they personally would benefit from, given how they approach exams. Even if a professor doesn't offer an expressive writing time, students could take the time before an exam to intentionally offload some negative vibes. If more time means more second-guessing, students might try to reduce their urge to cross out answers. Extended time may indeed help some students, but not all.

A little bit of metacognition – about both our test taking and our test design – can go a long way in raising the bar for all students.

[1] If you talk to students they will tell you that it is very difficult to get accommodations — they need to re-apply each year, and our office (like many other college disability offices) is overburdened and slow to approve.

[2] Forcing your grades into a Gaussian distribution assumes that each time you teach your course (to let's say, 100 students), grades will be normally-distributed. Any less than 100 data points isn't really enough to accurately model a distribution, and there is no guarantee your students will always look the same. The pandemic had an unequal impact on students, which is also unequal across years.