Introduction

Maybe you have heard of the trick. Before a big exam, you take ten minutes and write down everything you are afraid of. The bad grade. The parent who will be disappointed. The blank feeling that hits when you turn the first page. You get it all onto paper, and then you walk into the room lighter.

It sounds like folk wisdom, but it came from one of the most respected journals in science. In January 2011 two psychologists at the University of Chicago reported in Science that this exact exercise protected math scores under pressure in a laboratory, then lifted anxious students' grades in real high school biology finals [1].

The idea spread fast. Teachers added it to exam days, study guides still recommend it, and our own article on the psychology of high-stakes exams describes it with real enthusiasm.

Here is the part most of those pages leave out. In 2018 a large team of researchers repeated the core experiment with a sample more than six times bigger, and the effect was gone, with a correlation of r = -0.098 where the original study's effect had been r = .793 [2].

That does not make test anxiety a myth. It is one of the most studied problems in educational psychology, and it is real, common and measurable. It just means the famous fix deserves a second look, and so does the question underneath it. What is test anxiety actually doing to you, and what helps?

This article works through both questions. You will see what test anxiety is, who has it, what it costs and why psychologists came to believe that worry steals working memory. Then we follow the writing study from its clever beginning to its replication, and finish with the treatments that have held up.

Worry and the Racing Heart

Test anxiety has a long paper trail. In 1952 George Mandler and Seymour Sarason built a questionnaire that picked out students whose anxiety was tied specifically to being tested, and six years later Sarason and colleagues published a version for children [3] [4]. The most useful idea came in 1967, when Robert Liebert and Larry Morris split test anxiety in two [5].

Worry is the thinking part: the doubts, the what-ifs, the vivid picture of failing. Emotionality is the body part: the pounding heart, the damp hands, the knot in the stomach. In their college sample it was worry, and not emotionality, that tracked how badly students expected to do.

Most people can tell which half they know better. One is a voice narrating disaster. The other is a body ringing like an alarm. You may well have both.

You can hear the split in the way students talk. Before one biology exam, a student wrote that they felt mentally prepared but always felt the anxiety physically, which is emotionality with very little worry [6]. The bodily side is not a sideshow, either: a 2024 study of 42 adolescents found that it tracked attention problems on a reaction-time task and in resting brain activity more closely than worry did [7]. If the bodily side is the part you notice most, our piece on stress, cortisol and memory explains what stress hormones do to recall.

Worry became a story about attention. In 1971 Jeri Wine reviewed the evidence and argued that highly anxious people spend part of every test watching themselves, splitting attention between the problem and their own self-evaluation [8]. In 1984 Irwin Sarason sharpened the point across three studies, concluding that test anxiety is largely a problem of intrusive thoughts that crowd out task-focused thinking, and showing that a simple instruction to concentrate on the task reduced them [9].

Here is what that account means for you in an exam room. The problem, on this view, is not that you feel nervous. It is that part of your mind is busy grading you while the rest of it tries to answer the question.

Psychologists also built better rulers. The Cognitive Test Anxiety scale, published in 2002, measures the thinking side only, and a 2022 analysis found three strands inside it (general worry, freezing up and fear of failure) that behave like a stable trait rather than a passing mood [10] [11]. None of this makes test anxiety a medical diagnosis, and it has no entry of its own in the DSM-5. It does keep company with diagnoses, though. In a 1988 study of elementary schoolchildren whose test anxiety was judged clinically significant by self-report and a clinical interview, 60 percent met the criteria for an anxiety disorder [12].

How Many Students, and Which Ones

Ask how common test anxiety is and you get a range so wide it is almost useless. That is not because nobody has counted. It is because nobody agrees where high anxiety begins.

A 2025 systematic review of university research found that most studies treat test anxiety as a continuous score and never define levels at all, and that studies which do draw a line mostly use weak methods, such as splitting their own sample down the middle [13]. With that warning in mind, the clearest school figure comes from England. In 2014 a survey of 2,435 students in 11 secondary schools found that 16.4 percent described themselves as highly test anxious, with girls at 22.5 percent and boys at 10.3 percent [14].

Students reporting high test anxiety, English secondary schoolsAll studentsGirlsBoys302826242220181614121086420Percent of students

That gap between girls and boys turns up almost everywhere. In a 2005 study of 4,000 undergraduates and 1,414 graduate students, women reported more test anxiety than men, and yet women also earned higher grade point averages [15]. A 1990 study of 361 college applicants in Israel found the same gender gap, mostly on the bodily side, but test anxiety did not distort their aptitude scores by sex or ethnic group [16].

In both studies the women were more anxious without being less able. Feeling anxious and performing badly are not the same thing, and that will matter later.

Medicine is where the numbers balloon. A 2026 meta-analysis pooled 35 studies covering 13,489 students in medical school and estimated that 49.8 percent had test anxiety, though the studies disagreed so sharply that the pooled figure means little on its own [17]. A second 2026 review of 13 studies put the share anywhere from 13.5 to 72 percent [18]. In a German anatomy course with 625 first-year students, about half showed pronounced anxiety on at least one dimension, yet only around 10 percent crossed the line on the overall scale [19].

Same students, different ruler, very different headline. When you see a single percentage quoted for test anxiety, ask which ruler produced it.

What pushes someone toward the high end? A 1989 review traced test anxiety back to how parents react to a young child's first achievements, and described it growing as school evaluation becomes stricter and more comparative [20]. In a 2013 study of 335 children in grades 3 to 5, students reported more anxiety around No Child Left Behind state testing than around ordinary classroom tests [21]. And in a 2010 study of 132 British secondary students, math teachers' warnings about failure went with rising worry and tension, but only when students experienced those warnings as threats [22].

Most of those pressures come from outside the student. Test anxiety is partly something you carry into the room and partly something the room does to you.

Personality, goals and attention all feed in. In 134 university students, cognitive test anxiety rose with maladaptive perfectionism and with goals built around avoiding failure, and those factors plus gender accounted for nearly 50 percent of the differences in anxiety [23]. A 1999 study found that worry was the bridge between performance-avoidance goals, the urge not to look incompetent, and lower exam scores [24]. Our articles on perfectionism and on mastery goals versus performance goals go deeper into both ideas. Students with ADHD carry an extra load: in a 2014 study, college students with ADHD reported far more test anxiety than their peers, and nearly half scored in the range the researchers classed as clinically significant for worry [25]. Test anxiety is not a symptom of ADHD, but the two travel together.

High achievers are not immune, though they report less. In a 2023 Korean comparison, 102 high achievers showed lower test anxiety than 120 comparison students [26], and a 2026 study of about 800 high-achieving adolescents found that anxiety about a specific subject tracked grades in that subject while general anxiety explained very little [27]. In that group, at least, the anxiety that mattered was the kind attached to a particular subject.

What It Costs on the Scoreboard

So test anxiety is common. Does it really lower your score? It does, but by less than most people assume.

The landmark is Ray Hembree's 1988 meta-analysis of 562 studies, which concluded that test anxiety causes poorer performance, goes with low self-esteem and fear of negative evaluation, and that treatments which reduce it are followed by better test scores and grades [28]. A 1991 meta-analysis of 126 studies and 36,626 participants put the average correlation between anxiety and academic performance at r = -.21, and found the link tighter when the measure focused on worry and on testing in particular [29].

What does a correlation of that size mean for you? It means a real cost, but a modest one, the kind that shows up as a notch down in grade rather than a collapse.

The thirty-year update arrived in 2018 and pooled 238 studies, finding small to moderate negative links with standardized tests, university entrance exams and grade point average, strongest in the middle grades [30]. In the 2005 college sample, low-anxiety undergraduates averaged a B+ and high-anxiety undergraduates a B [15]. A 2024 survey of 201 US undergraduates found the same pattern as a staircase, with every step up in test anxiety matched by a lower grade point average [31].

Math deserves its own mention, because it comes with a brand of dread you may know by name.

A 2021 meta-analysis of 747 effect sizes put the link between math anxiety and math achievement at r = -.28 [32]. Another 2021 meta-analysis, pooling 177 studies and 906,311 participants, found that math anxiety and test anxiety both hurt math performance, and that working memory carried only a weak share of that link [33]. Keep that detail in mind, because the remedy at the center of this article leans on working memory.

Time matters too. In a one-year study of 290 students in eleventh grade, worry predicted falling grades and falling life satisfaction over the following twelve months [34]. Worry was not just a snapshot. It was a slope.

Where the Worry Goes

Why would anxious thoughts lower a score at all? The most popular answer involves working memory, the small mental workspace you use to hold a problem in mind while you solve it.

The idea grew out of research on choking. In 1984 Roy Baumeister defined choking under pressure as doing worse exactly when doing well matters most, and across six experiments showed that pressure pushes people to watch their own performance [35]. In 1992 Michael Eysenck and Manuel Calvo proposed that worry occupies storage and processing space in working memory while also driving extra effort, so anxiety damages efficiency more than the final result [36]. Their 2007 update, attentional control theory, added that anxiety weakens the ability to block out distractions and shift focus [37].

Think of working memory as the desk you do the problem on. Worry, in this account, piles its own papers on that desk, so there is less room left for yours.

The numbers back a link. A 2016 meta-analysis of 177 samples and 22,061 people found anxiety tied to lower working memory capacity by about 0.334 standard deviations [38]. Then came a twist. In 2005 Sian Beilock and Thomas Carr found that performance pressure hurt the students with the most working memory, and only on the hardest math problems [39]. A 2012 study of 96 adolescents complicated things further: anxiety went with worse test scores in teenagers with low working memory and with better scores in those with high working memory [40].

So anxiety does not act the same way on everyone, and these two studies do not even agree on who is most at risk. That is a good reason to distrust any one-line explanation, including the one this article is about to test.

The working memory story also reaches beyond test anxiety. In a 2003 series of three experiments, the threat of confirming a negative stereotype lowered working memory capacity in women and in Latino students, and that drop explained the women's lower math scores [41]. Our article on stereotype threat follows that research and its replication troubles.

Back in the late 2000s, though, the model looked clear enough to suggest a remedy. If worry fills working memory, then emptying the worry out before the test should free that space for you.

Why Writing Was Supposed to Work

The remedy came from a separate line of research. In 1986 James Pennebaker and Sandra Beall asked students to write about a traumatic experience over several days, and those who wrote about both what happened and how they felt made fewer visits to the campus health center in the months that followed [42]. A 1998 synthesis found that expressive writing improved several kinds of health outcome in healthy people, even though it made them more upset in the short term [43], and a far larger 2006 meta-analysis of 146 randomized studies confirmed a positive but small effect, an average correlation of r = .075 [44].

Pause on that last number. An effect of that size is real, but you would never notice it in your own life without a large study pointing it out.

Two studies made the jump to exams look plausible. In 1997 Stephen Lepore had students preparing for graduate entrance exams write their deepest thoughts and feelings about the test, and their depressive symptoms fell over the month before the exam; their intrusive thoughts did not become less frequent, but those thoughts stopped predicting how low they felt [45]. In 2001 Kitty Klein and Adriel Boals reported that 35 first-year students who wrote about coming to college showed larger working memory gains seven weeks later than 36 students who wrote about trivial topics, and in a second experiment writing about a negative experience also reduced intrusive thinking [46].

Notice the fine print, though. Those programs ran over days or weeks. Nobody had yet shown that a single session, a few minutes before a test, could do the same job.

The Ten Minutes

That is the experiment Gerardo Ramirez and Sian Beilock set out to run, and if you have ever sat in an exam hall feeling your mind fill up, its logic will feel right.

In their first laboratory study, 20 college students took two short math tests [1]. The second came with pressure: the promise of money for a good score and the news that a partner's reward depended on their performance. Just before it, half the students sat quietly for 10 minutes and half spent 10 minutes writing about their thoughts and feelings about the test. By the original data, the quiet group's accuracy fell from 82 to 70 percent while the writing group's rose from 86 to 91 percent [2].

Math accuracy before and after pressure, original 2011 lab studyQuiet beforeQuiet afterWriting beforeWriting after1009080706050403020100Percent correct

Remember how small that study was. Ten students per group can produce a striking chart from a handful of unusual scores.

A second laboratory study tackled the obvious objection that any writing might help. According to the authors, writing about an unrelated topic brought no benefit, so the content had to be the test worries themselves.

Then the team went into classrooms. In two cohorts of ninth graders, each about 50 strong, students did the writing or a control task right before the first final exam of their high school careers, and among the most test-anxious students, those who wrote averaged a B+ while those who did not averaged a B- [1]. Students low in test anxiety did about the same either way.

On paper, ten minutes of writing had leveled the playing field. The study was short, cheap to copy and easy to like, exactly the kind of finding that travels.

An early challenge came within months. A pair of researchers named Lang had found in 2010 that a different writing exercise, describing a person who is good at solving problems, raised the scores of cognitively test-anxious students in two school studies of 219 and 232 students but lowered the scores of calmer ones [47]. In a 2011 letter to Science, they argued that the writing study showed the same hidden cost for low-anxiety students [48]. Ramirez and Beilock replied that low-anxiety students in their classroom studies scored just as well whether they wrote or not [49].

That exchange was about the original data. The bigger test, whether anyone else could get the same result, was still to come.

From a 1952 Questionnaire to a 2025 Meta-Analysis

The writing study sat on top of decades of work on worry, attention and expressive writing, and it was later tested by one of the largest replication efforts in psychology.

1952
Mandler and Sarason publish an early test anxiety questionnaire
1958
Sarason and colleagues release a test anxiety scale for children
1967
Liebert and Morris separate worry from emotionality
1984
Irwin Sarason links test anxiety to intrusive thoughts
1986
Pennebaker and Beall launch the expressive writing method
1988
Hembree links test anxiety to lower scores across 562 studies
1992
Eysenck and Calvo argue that worry uses up working memory
2001
Klein and Boals report working memory gains after weeks of writing
2002
The Cognitive Test Anxiety scale is published
2007
Attentional control theory updates the working memory account
2011
Ramirez and Beilock report the ten-minute writing study in Science
2018
A preregistered replication of the lab study finds no effect
2019
A meta-analysis of 44 trials supports structured treatment
2022
A study of 309 medical students questions the interference idea
2025
A meta-analysis of 21 single-session studies finds no reliable benefit

Read it top to bottom and you will recognize a familiar story in psychology: a plausible mechanism, a striking small study, quick adoption, then the slower work of checking.

The Replication

Psychology spent the 2010s re-running its famous experiments. In 2018 the Social Sciences Replication Project re-ran 21 experiments published in Nature and Science between 2010 and 2015, with samples about five times larger than the originals on average and every plan preregistered and sent to the original authors [2].

Ramirez and Beilock's first laboratory study was one of the 21, rebuilt in an online survey tool from the authors' materials because the original software was not available.

In the first round, with 26 people, the effect was r = -0.075, nowhere near significant. The protocol called for a second, larger round, and with 131 people pooled the effect was r = -0.098, with p = .394, against an original r = .793 from 20 people. That made it one of 8 studies out of 21 that did not replicate, and across all 21 the replication effects averaged about half the size of the originals [2].

That is a striking reversal. A sample more than six times larger did not just find a smaller effect. It found essentially none.

The original authors did not accept the verdict. They argued that the replication never created real pressure, because students in the high-pressure and low-pressure conditions scored about the same, so it could not test whether writing protects people under pressure.

The replication team disagreed and pointed to their preregistered checks, which showed that the pressure manipulation had raised anxiety, and both positions were posted publicly along with the correspondence [2].

You can weigh that dispute for yourself. Either way it changed the question. After 2018 the claim could no longer rest on the original paper. It needed new evidence.

What the Classrooms Found Next

New evidence did arrive, and at first it pointed in more than one direction.

Some of it was encouraging. In a 2011 trial, students preparing for graduate admissions exams such as the GRE, MCAT and LSAT were randomly assigned to write about the upcoming exam or about something neutral, and the expressive writers scored higher, but only among those taking the MCAT or LSAT [50]. In 2014 Daeun Park, working with Ramirez and Beilock, tested the exercise on 80 students who varied in math anxiety, and writing shrank the gap between high and low math-anxious students on math problems [51]. Our guide on how to study for a math test deals with that kind of anxiety in more depth.

Then the real classrooms reported in. In a large introductory biology course, Jennifer Doherty and Mary Pat Wenderoth printed a writing prompt on the cover page of every exam, allowed five minutes, and then asked students to crumple the page and throw it into the aisle. Most wrote for only about two minutes, and pep talks were the most common thing on the pages across all four exams. One student wrote: "Feeling axious so anxious I didn't even spell it right the 1st time. Ha!" The crumpling, the authors noted, seemed to break the tension, complete with laughter and the odd paper airplane, though their report measured how the exercise worked in practice rather than what it did to scores [6].

Hold on to that picture, because it shows what the exercise feels like in a real room. It is quick, a little silly and often comforting. Whether it changes your grade is a separate question.

The controlled trials mostly said no. In a 2021 study at Iowa State University, 195 students in a cognitive psychology course alternated between expressive writing and a control writing task across four real exams, so each student served as their own comparison [52].

The writing did not reduce anxiety or raise scores. Among the more anxious half of the class, 87 students, the average exam score was 72 percent with and without it, and a second experiment with an instruction-based exercise borrowed from stereotype threat research did no better.

The same year another team randomly assigned 300 students to one of four five-minute exercises, including expressive writing and two versions of reappraisal, or to no exercise before a hard math test. None of them changed anxiety or performance, and the writing group actually reported more anxiety in the moment [53]. A 2026 trial in China tried a stronger dose, with 197 students in their final years of high school writing about past bad test experiences for 10 minutes a day on three days before a semester exam, and it changed neither anxiety nor scores [54].

StudyWho took partWhat they didWhat happened
Ramirez and Beilock 2011, lab20 college students10 minutes of writing before a pressured math testWriters rose from 86 to 91 percent correct, the quiet group fell from 82 to 70
Ramirez and Beilock 2011, classroomTwo ninth grade cohorts of about 50Writing before the first high school finalHighly anxious writers averaged B+, highly anxious non-writers B-
Frattaroli and colleagues 2011Graduate admissions test takersWriting about the exam versus neutral writingHigher scores for MCAT and LSAT takers only
Park and colleagues 201480 college studentsWriting before math and word problemsSmaller gap between high and low math-anxious students
Camerer and colleagues 2018131 peoplePreregistered replication of the 2011 lab studyNo effect, r = -0.098
Rozek and colleagues 20191,175 ninth gradersWriting, reappraisal or both before two semester finalsLower-income failure rate fell from 39 to 18 percent
Myers and colleagues 2021195 psychology studentsWriting versus control writing across four real examsNo change in anxiety or scores
Ganley and colleagues 2021300 college studentsFour five-minute exercises including writingNo benefit, and more anxiety after writing
O'Meara and Lovett 202521 studies, 1,457 participantsMeta-analysis of single-session writingNegligible effects on anxiety and on scores
Liu and colleagues 2026197 high school students10 minutes a day for three days before examsNo change in anxiety or scores

You can see the pattern in the table. Most of the striking results come from small samples or from the original research group. Apart from a partial result on graduate admissions tests, the independent trials of writing on its own come back flat.

Putting the Numbers Together

In 2025 Patti O'Meara and Benjamin Lovett gathered every study they could find that tested a single session of expressive writing before a test, 21 studies with 1,457 participants in all [55].

The effect on anxiety was negligible, at r = -.05 (p = .57). The effect on test scores first looked significant at r = .09, but it shrank to r = .06 (p = .06) once one extreme study was set aside. There was little sign of unpublished null results hiding in drawers, and the authors concluded that the research, taken together, gives no support to single-session writing as a treatment for test anxiety.

Put plainly, if you ran this exercise across a whole school, the average student's anxiety and score would barely move, and you could not be sure they had moved at all.

There is a direct test of the mechanism, too. In a 2016 study of 150 participants, an expressive writing session changed neither mind wandering nor working memory performance [56]. So is writing useless? Not quite. In a 2022 trial with 141 first-year college students, three days of expressive writing reduced general anxiety at a one-month follow-up compared with students still waiting their turn [57].

A 2018 trial in China found that 75 highly test-anxious students who wrote about positive emotions for 20 minutes a day for 30 days reported less test anxiety afterward, although it did not measure exam scores [58]. And a 2025 review of 51 studies of positive writing found its most consistent benefits on well-being and positive mood rather than on negative feelings [59].

So writing has real uses. They are slower and gentler than the exam-day promise, and they show up in how people feel over weeks rather than in the next test score, with one notable exception.

There is one striking positive result, and it came from the original research group. In a 2019 field experiment with 1,175 students in ninth grade biology, Christopher Rozek and colleagues assigned expressive writing, a reappraisal exercise that framed stress as useful, both, or a control task right before the two semester finals. Among lower-income students, 39 percent failed the course in the control group, compared with 18 percent after any of the exercises, which worked about equally well [60].

Lower-income students who failed ninth grade biologyControl groupAny exercise50454035302520151050Percent failing

That is a big effect, and it deserves attention. But look at what it measured. It showed up for lower-income students rather than for everyone, it counted passing the course rather than an average score, and it sits alongside the independent trials above, which found little or nothing.

Do Anxious Students Blank, or Learn Less?

There is a second argument hiding under the first one, and it may matter more to you.

The writing fix assumes that anxious students know the material and lose access to it in the exam room. That is the interference idea. But it has always had a rival. Maybe anxious students study less effectively in the first place, and part of their anxiety is the accurate sense of being underprepared.

In 1985 Sigmund Tobias argued that the two accounts were complementary rather than competing [61]. Two years later Moshe Naveh-Benjamin and colleagues described two kinds of test-anxious students, one with weak study skills that never learned the material well and another that learned well but struggled to retrieve it under test conditions [62].

Recent work has sharpened the challenge. In 2022 Maria Theobald and colleagues followed 309 students in medical school as they prepared for a high-stakes exam on a digital platform that tracked their knowledge through practice sessions and mock exams [63].

Once knowledge was accounted for, test anxiety did not predict exam scores at all. What it did predict was learning, since students high in trait anxiety gained less knowledge during their preparation.

Read that again, because it flips the usual picture. Those anxious students were not falling apart on exam day. They were falling behind in the weeks before it.

Other studies pull the other way. A 2021 study of 730 university students found that interference from irrelevant thoughts and a lack of confidence predicted test performance, while worry and bodily arousal did not [64]. And in the 2010 competence study, priming competence raised anxious students' scores, which suggests they had more ability than they usually showed [47].

No single study settles this, and the honest answer is that both things happen. Some students do go blank. Others, if the medical school study holds up more widely, learn less in the weeks before. For you, that shifts where help should go. If anxiety is quietly undermining your preparation, a ritual at the exam door cannot fix that on its own.

What Has Evidence Behind It

Now the encouraging part. Test anxiety is treatable, and the evidence for that is far stronger than the evidence for the ten-minute trick.

Hembree's 1988 review already found that a range of treatments reduced test anxiety and were followed by better grades [28]. An earlier 1980 review noticed something that still matters: treatments aimed at worry did more for grades than treatments aimed at bodily arousal, though self-reported anxiety fell after almost any treatment, including fake ones [65].

That last detail is a useful warning about trusting how calm people say they feel.

A 2003 meta-analysis of 56 studies with 2,482 participants found an average effect size of 0.65 for test anxiety programs, with the best results from programs that combined study skills with behavioral or cognitive methods [66]. In 2019 Christopher Huntley and colleagues pooled 44 randomized trials with 2,209 university students and found that treatment cut test anxiety with an effect size of 0.76 and raised academic performance with an effect size of 0.37, with the most support for behavior therapy [67].

They also found signs of publication bias, so the true effects are probably somewhat smaller.

Translated, those numbers look like this. A change of 0.76 in reported anxiety is large by the standards of psychology, though the warning about fake treatments applies to it. The 0.37 gain in grades is the harder evidence, a meaningful gain but nothing like a miracle.

The same picture holds across ages, with one twist. A 2021 umbrella review of 31 meta-analyses on college mental health listed test anxiety among the problems with effective treatments, while warning that most underlying studies were of low quality [68]. In a 2020 trial, a six-session cognitive behavioral program for students aged 14 to 16 preparing for high-stakes exams produced a large drop in test anxiety, compared with a moderate drop in a waiting-list group [69]. And a 2025 meta-analysis of 19 studies with 2,377 secondary students found that academic interventions raised achievement with an effect size of 0.66 but had no significant effect on anxiety [70].

That result fits the medical school study from another angle. Better studying lifts scores without calming nerves, while treating the anxiety tends to lift scores too, so each needs its own attention.

One study habit seems to do double duty. In 2014 Pooja Agarwal and colleagues surveyed 1,408 students in middle and high school whose classes used regular low-stakes quizzing, and 72 percent said the practice made them less nervous for tests [71]. That is self-report rather than a trial, but it fits what is known about retrieval practice, which our article on the testing effect explains. Reappraisal, learning to read a racing heart as fuel rather than danger, has a modest record too: in 2010 students preparing for the GRE who were told that arousal can improve performance scored higher on the math section in the lab and on the real exam one to three months later [72].

The wider reappraisal evidence is consistent but modest. In a 2018 study, a single paragraph in an instructor's email the night before an exam improved first-year students' exam and course performance, though it did nothing for older students [73]. A 2024 meta-analysis of 44 effect sizes put the average benefit of stress reappraisal and similar mindset interventions at d = 0.23, and at d = 0.20 for written thinking tasks like exams [74]. A longer 2022 program combining growth and stress mindsets improved stress responses across six experiments with US high school and college students, and in one of them improved academic success [75].

Remember the 300-student trial, though, where two reappraisal exercises squeezed into five minutes did nothing. A brief version of a good idea can still fail, and you should be wary of any technique sold as a quick fix.

The rest of the evidence is thinner. Online mindfulness programs reduced test anxiety in a 2022 trial of 71 students, but no more than a sham version did [76]. A five-day, eight-minute mindfulness routine in a 2025 study of 61 high school students lowered anxiety and raised exam scores, but the comparison group simply waited [77]. Working memory training with emotional material cut test anxiety more than neutral training in a 2021 trial of 60 students [78].

A small biofeedback breathing study of 34 students found benefits only for the group using the device [79], and a single online counseling session reduced test anxiety among 57 students in 2025, apparently by raising their self-compassion [80]. Each of these is a lead. None of them is a verdict.

ApproachBest evidenceWhat it found
Behavior therapy and cognitive behavioral programsHuntley and colleagues 2019, 44 trialsAnxiety down by an effect size of 0.76, grades up by 0.37
Study skills combined with anxiety treatmentErgene 2003, 56 studiesAverage effect size of 0.65, best for combined programs
Cognitive behavioral program for teenagersPutwain and von der Embse 2020Large drop in test anxiety versus a moderate drop on a waiting list
Academic help aloneFishstrom and colleagues 2025, 19 studiesAchievement up by 0.66, no significant change in anxiety
Stress reappraisalBosshard and Gomez 2024, 44 effect sizesSmall average benefit, d = 0.23
Practice testingAgarwal and colleagues 2014, 1,408 students72 percent said it made them less nervous
Single-session expressive writingO'Meara and Lovett 2025, 21 studiesNo reliable effect on anxiety or scores

The pattern is fairly consistent. Structured treatment over several sessions has the strongest record, practice and better study habits help the preparation side, and brief messages about stress have a smaller, patchier one.

If You Still Want to Try the Ten Minutes

None of this means you should feel silly for writing before an exam. It costs almost nothing, and plenty of students find it settling. The real question is what to expect from it.

The fair reading is that it is a comfort ritual, not a score booster. If it helps you feel steadier, keep it. If it stirs up more worry, as it seemed to for some students in the 2021 math trial, let it go.

If you do try it, the original studies offer a few details worth copying. The writing was about the upcoming test specifically, not about life in general. It happened right before the exam and lasted about ten minutes. And in the original classroom studies, the students who seemed to benefit were the highly anxious ones.

For teachers, the crumpled-page version shows the exercise can fit inside a real exam session in a couple of minutes. Just do not expect it to close an achievement gap on its own. The one large success came from the original research group, and its benefit showed up among lower-income students.

If test anxiety regularly keeps you from showing what you know, the stronger options are slower and less glamorous. Practice tests under realistic conditions. Study methods that build genuine recall. Structured help from someone trained in cognitive behavioral approaches.

And if anxiety leads someone to avoid exams altogether, or spills over into everyday life, that is a reason to talk to a school counselor or a clinician rather than to search for a trick.

What Is Settled and What Is Not

Some things are solid. Test anxiety is common, it is reported more often by girls and women, and it rises with the stakes. It is linked to lower scores by a modest but reliable margin. Anxiety goes with lower working memory capacity on average. And structured treatment, especially behavior therapy and programs that combine study skills with anxiety work, reduces test anxiety and modestly improves performance.

Other things are still open. Researchers disagree about how much test anxiety blocks retrieval on the day and how much it undermines learning beforehand. They disagree about whether worry or bodily arousal does more of the damage. And the authors of the original writing study still dispute whether the 2018 replication tested their idea fairly.

The ten-minute claim itself sits in an awkward spot. It has one striking original paper, one large positive field study from the same group, a failed high-powered replication, several null classroom trials and a meta-analysis that finds no reliable benefit. That is not a proven technique. At best it is a promising idea that has not held up as a stand-alone fix.

Conclusion

The ten-minute writing study was a good idea, tested cleverly and published in a top journal, and then repeated by people who wanted to know whether it was true. That is science working, even when the answer disappoints.

What survives is more useful than the trick. Test anxiety is real, it costs points, and it can start doing damage weeks before the exam, when it gets in the way of studying. The tools with the best record work on that whole stretch: practice that builds real recall, help with study habits, and proper treatment when the anxiety is severe.

So if you want to write down your worries before your next exam, go ahead. Just do the slower things too. They are the ones with the evidence behind them.

Frequently Asked Questions

What causes test anxiety?

Test anxiety grows from a mix of temperament, experience and circumstance. Research links it to fear of negative evaluation, low self-esteem, perfectionism focused on avoiding mistakes and goals built around not failing. High stakes make it worse, and warnings about failure from teachers can raise worry when students experience them as threats. Weak preparation feeds it too, because students who sense they have not learned the material well have a real reason to worry.

Is test anxiety a symptom of ADHD?

No. Test anxiety is not a symptom of ADHD, and it is not a diagnosis of its own in the DSM-5. The two often occur together, though. A 2014 study of college students found that those with ADHD reported much more test anxiety than their peers, and nearly half scored in the range the researchers classed as clinically significant for worry. Students with ADHD who struggle with exams may need support for both.

Does writing about your worries before a test really raise your score?

Probably not on its own. A famous 2011 study reported large benefits, but a high-powered 2018 replication of its laboratory experiment found no effect, several later classroom trials found none, and a 2025 meta-analysis of 21 studies found no reliable effect on anxiety or on scores. A large 2019 field study by the original research group found that lower-income students failed less often after writing or reappraisal exercises, and many people find the exercise calming, but it is not a proven way to raise a grade.

How do you get rid of test anxiety?

The approaches with the strongest evidence are structured ones. Meta-analyses of randomized trials find that behavior therapy and cognitive behavioral programs reduce test anxiety substantially and improve grades modestly, especially when they are combined with study skills training. Regular practice testing builds recall, and many students say it makes exams less nerve-racking. Learning to see stress as helpful has a small average benefit. Anxiety that disrupts exams or daily life is a reason to see a school counselor or a clinician.

How common is test anxiety?

It depends heavily on where the line is drawn. A large survey of English secondary schools found 16.4 percent of students highly test anxious, with girls at 22.5 percent and boys at 10.3 percent. Studies of medical students report anywhere from about 13 to 72 percent, and a 2026 meta-analysis pooled them at around 50 percent while noting that the studies disagree enormously. There is no agreed cut-off for high test anxiety, so any single figure should be read with care.