Putting test anxiety to the test
Test anxiety may be a bigger problem for some of your students than you realize. What can we do to help?
Most of us, probably all of us in fact, have experienced the increase in stress that comes before taking an exam, and especially if that exam was particularly important, or a “high stakes” exam. Of course it is also likely that not all of us experienced the same degree of stress, and that some of us were able to interpret and deal with that stress in a more positive way than others. When this stress is focused on the negative consequences of failing a test, this is called test anxiety. While there are differences in how anxious people tend to be on average, called trait anxiety, test anxiety focuses specifically on test-taking situations and the social evaluation that takes place. Unsurprisingly, research has shown that test anxiety has negative effects on exam performance, but also that some students seem to experience more test anxiety than other students. This has in turn inspired the development of short interventions intended to help students with test anxiety, and these have shown positive effects in small-scale studies. A new study in CBE—Life Sciences Education expands this research to include a wider range of student backgrounds and types of institution by examining patterns of test anxiety in students in STEM (science, technology, engineering, and math) courses. They found the same relationships between test anxiety and exam performance in a larger sample. However, they did not find the same positive effects for the intervention at this scale. While this was not what the authors predicted, this null result is still an important finding for the lessons it provides.
Better understanding test anxiety is important for most areas of study, but the authors point out that it may be particularly relevant for students in STEM disciplines. There is a problem in higher education with the retention of STEM students, and the authors note the importance of addressing this problem because of the future needs of an international STEM workforce. One factor in this retention problem may be that introductory STEM courses tend to be dominated by high-stakes exams. In addition, students who leave the field are disproportionately marginalized groups of students, such as those facing stereotypes or discrimination based on gender or ethnicity. In fact, students from these groups also show higher levels of test anxiety in STEM courses, so for example test anxiety tends to be higher in women, first-generation university students, and in students from marginalized ethnic backgrounds. Thus the problem is clearly framed – the STEM retention problem affects certain groups of students more than others, and one potential pathway is higher levels of test anxiety in these groups.
This pattern has led several research groups to explore ways to address this problem, and one promising approach that is the focus of this study is interventions that can potentially reduce test anxiety. One particularly promising intervention is based on research in psychology that explores how people can reinterpret, or reappraise, feelings in certain situations. For example, the arousal that can happen in a high-stakes testing environment, such as increased heart rate and “knots” in the stomach, can be interpreted by some students as something bad or threatening. This can then both increase stress and impair performance. Importantly, it is not the arousal itself that causes problems, but rather the interpretation of that arousal – i.e., increased anxiety can take up important resources that are then not available to meet the demands of the test. Reappraisal-based interventions address this by helping students interpret this arousal in a more positive way, for example with materials that discuss why some arousal in contexts like testing situations is beneficial and adaptive, for example by helping to focus attention.
The authors addressed this problem with an ambitious research design in which they implemented and evaluated a reappraisal-based intervention across 12 courses at seven higher education institutions in the US. The sample of institutions was intentionally diverse, including both predominantly white institutions as well as several Historically Black Colleges and Universities, which the authors note have been underrepresented in studies of STEM students. Students were asked to complete several questionnaires before and after the semester, and course grades were also examined. Students were randomly assigned to receive the intervention or a placebo intervention that discussed stress but not reappraisal specifically. The reappraisal-based intervention included instruction about how stress is natural and can benefit performance, and both interventions included multiple-choice questions to assess engagement.
Before evaluating the intervention, the authors examined the results from the questionnaires at the beginning of the semester. They replicated previous findings on test anxiety in this more diverse sample. Students with higher test anxiety, both as a general trait as well as immediately before the exam, performed worse on the exam as well as in the course.
Results from the intervention evaluation did not show the expected positive effects, neither for test anxiety right before the exam nor for performance on the exam. One potential issue was that there was a relatively low response rate for the questionnaires, and the authors also note the possibility that students with higher levels of anxiety might have been less likely to participate. Supporting this possibility, participants in the study scored higher on average in the course than non-participants.
Independent of potential sample bias, the authors discuss other important lessons from the null result. Most relevant to other researchers and educators interested in helping students with test anxiety, they speculate that the intervention might not have been intensive enough because it was a “low effort” approach focused on a brief reappraisal of stress. They note that a more comprehensive intervention might be better suited to students experiencing high levels of test anxiety, because reframing anxiety in a brief intervention may not give students the type of longer-term coping mechanisms to manage persistent anxiety. More intensive approaches that have also shown positive effects include giving students numerous opportunities to practice and internalize reappraisal strategies and/or including an expressive writing component that encourages deeper reflection.
Comment: Despite the overall null result, this study makes several important contributions that may be particularly relevant for instructors in STEM courses, but that I think are also relevant for any university instructor. As instructors, we want our method of examination to reflect a student’s “true” performance, to the degree possible. The research on test anxiety suggests that this is one factor that might make this more difficult, at least for some students in some contexts. Because these effects are larger for higher stakes exams, this suggests that it may be unfair to have most or all of a student’s grade depend on the results of a single exam. This also suggests that one does not need an intervention to help mitigate the effects of test anxiety – for courses that rely more on higher stakes exams, one approach would be to grade students based on their performance on a wider range of lower stakes evaluations. This can include, for example, relying more on formative assessment, which will be the topic of an upcoming special issue of AHF. CeUL has online resources for those interested in learning more about formative assessment.
It is also important to remember the importance of context, and that some students may struggle more with test anxiety by virtue of their personal characteristics or background, relative to the university setting. The idea that students who belong to marginalized groups may be more likely to experience test anxiety is consistent with a large body of evidence in psychology showing how factors like context and an individual’s background can influence how we perceive the same situation. This is also broadly reminiscent of stereotype threat, where something as simple as including a question about gender before a math test can change performance for some female students by providing a reminder of gender stereotypes about math. For students from backgrounds that are still underrepresented in university education generally, as well as STEM education specifically, we may underestimate both how common and how strong these effects may be in their everyday university experience.
This study also illustrates how difficult it can be to expand a successful intervention developed at a smaller scale. This is consistent with research from intervention science, where it is all too common that interventions show promising results in small-scale evaluations but then show weaker, or no, effects when delivered at larger scale. The authors discuss several factors related to this, such as the difficulty of getting large numbers of students to participate (and to follow through on all aspects of the study if they do agree to participate), the difficulty of making sure that participants did in fact engage with the intervention materials, as well as possible variability in the way that multiple instructors across several institutions implemented the intervention. The detailed discussion of these factors is a strength of the study.
Another important contribution that should not be overlooked is the fact that the authors published the study despite the null result, which is unfortunately still too rare in many disciplines. This is also consistent with intervention science, where there are calls for researchers and journals to make more of an effort to publish such studies, along with a rigorous interpretation of null results that can inform future research. The authors of this study echo that call in their conclusion. By doing just that in this study, the authors have raised many important issues for future research that will continue to help students who experience test anxiety.
Eric Pakulak, Department of Psychology
Keywords: test anxiety, stress, student performance
Senast uppdaterad: 2026-09-29
Sidansvarig: Centrum för universitetslärarutbildning