By Althea Need Kaminske

Does test anxiety cause poorer performance on exams? Meta-analyses show that students with higher test anxiety tend to perform worse on exams (1). We also know that anxiety can affect cognitive processes through working memory capacity (2). Therefore, the general consensus is that test anxiety interferes with our working memory, which in turn leads to poorer exam performance. According to this interference hypothesis, anxiety interferes with our already limited working memory capacity, making it harder to focus and process information in anxiety-filled situations like exams. However, a recent study with German medical students found that test anxiety did not predict exam performance when prior knowledge was controlled for, claiming strong evidence against the interference hypothesis (3). 

This finding seems to run counter to the perceived wisdom that otherwise capable students are underperforming on exams because of anxiety. Educators are cautioned against using high-stakes assessments that might exacerbate student’s test anxiety (I think there are other reasons to avoid high-stakes assessments, namely that they don’t actually reinforce learning in any long-term, meaningful way as they don’t encourage spaced retrieval practice). Moreover, any increase in anxiety is especially concerning as student mental health appears to be in crisis (4). Teachers aren’t faring any better - one study found that teachers were more likely to report increased anxiety during the pandemic than health care workers (5). With all the stress placed on educators, the finding that test anxiety may not be the root cause of poor exam performance might be somewhat of a relief. While I have no love of high-stakes exams, in many scenarios they are unavoidable. How much do educators need to worry about anxiety influencing exam scores?

What did the study look like?

Theobald and colleagues (2022) examined German medical students preparing for their second state exam using a digital learning platform that provided mock exams similar to the state exam (3). Using log files from the digital learning platform, the researchers were able to monitor students’ knowledge of the material during the study period leading up to the exam. In addition to measuring study performances and exam performance, researchers measured both state and trait test anxiety in students as well as their working memory capacity. Trait test anxiety refers to an enduring characteristic - someone with high trait test anxiety tends to be anxious about tests regardless of the situation. State test anxiety refers to anxiety that stems more from the situation. Regardless of someone’s trait test anxiety, they might still find a particular test to be anxiety provoking. 

I want to highlight some features of the study design because it has important implications for what types of conclusions we can draw from it. While different levels of test anxiety, working memory, study performance, and exam performance were measured, none were directly manipulated. Therefore, this was a correlational study and not an experiment. Students were not randomly assigned to have lower or higher test anxiety or higher or lower working memory capacity. This means that it is difficult to disentangle all of the factors that might influence both test anxiety and exam performance. This is one of the challenges in studying people - they come preloaded with lots of complicated interconnected variables.

While it is difficult to disentangle all of the factors that might influence both student characteristics (like test-anxiety and working memory capacity) and our variable of interest, exam performance, we can use some very cool statistical models to help us out. The researchers had a good idea of what factors might be important to measure (in this case, working memory capacity and study performance) so they put those into the model to help them tease out how all these things influence each other.

What did they find?

If the researchers only looked at test anxiety and exam performance then there was a negative relationship - students with higher test anxiety performed worse on the exam. However, when study performance was included, test anxiety no longer predicted exam performance. Further, working memory capacity did not seem to make any difference. “...we did not find evidence for the hypothesis that test-anxious students with low working memory capacity performed more poorly on the final exam than those with high working memory capacity” (Theobald et al., 2022, pg. 2077). 

The researchers theorize that students with higher test anxiety do poorly on exams not because of constraints on working memory capacity at the time of the exam, but because they had poorer exam preparation leading up to the exam - perhaps as a result of the anxiety leading up to the exam. In other words, the time to address test anxiety is not during the test itself, but during the studying leading up to the test. 

Does test anxiety cause poorer performance on exams?

So, does test anxiety cause poorer performance on exams? I would say, yes, actually. It just may not happen during the test itself. I think this study provides evidence that test anxiety affects learning which, in turn, affects exam performance.

I also want to go back to the researchers’ assumption that test anxiety would more disproportionately affect students with lower working memory. On the surface this follows from the idea that anxiety ties up resources in working memory. Students with lower working memory capacity would have fewer resources from the beginning, so anything that affects working memory should affect them more. Somewhat paradoxically, research on the effects of pressure on working memory finds the opposite. Students with higher working memory capacity are disproportionately affected by stress (6). Beilock & DeCaro (2007) examined stress, working memory capacity, and strategy selection during mathematical problem solving (6). They found that students with lower working memory capacity solved problems using heuristics: strategies that did not require much working memory capacity and worked most - but not all - of the time. On the other hand, students with higher working memory capacity tended to solve problems using algorithms: strategies that were rule-based and required more working memory capacity but were guaranteed to produce a correct outcome. Students tended to apply these strategies regardless of conditions of the test. Thus, students with high working memory capacity tended to perform worse under pressure compared to students with low working memory capacity. They persisted in using the more working memory intensive algorithmic approaches even when they were under pressure and had less capacity to work with.

Importantly, this research on pressure and performance was experimental - the researchers manipulated the amount of pressure during the task (6). This was not the case in the Theobald et al. study (3). All of the students took the same exam under the same conditions. However, this is not to say that the Theobald et al (2022) study  was somehow flawed. What is most impressive about this study is its ecological validity - it examines real students taking a real test. And I certainly don’t think it would be ethical to have some medical students take their state exam under higher pressure conditions while others take it under lower pressure conditions. Rather, this highlights the complexity of studying learning (something I’ve touched on before when talking about prior knowledge). 

Bottom Line

The relationship between test anxiety, working memory, and exam performance is complicated! One thing that struck me about this study was that the students who were anxious about the exam had every reason to be anxious. They were not adequately prepared and their level of anxiety reflected that. A bad faith reading of this might imply that these students deserved to be anxious. Instead, I think this means that we might reframe test anxiety as a sort of early warning system for students. It’s a feature rather than a flaw. If students should are better equipped with effective study habits and support, they can both preform better on exams and manage their test-anxiety. Theobald et al. (2022) stress this as well: 

“Thus, test-anxious students became more aware of their knowledge deficits, which might have led to higher anxiety. Hence, test-anxious students need to acquire strategies that help them effectively acquire new knowledge and avoid repeated failure… interventions aiming to reduce test anxiety shortly before or during evaluative situations may not have the intended impact because they cannot offset knowledge deficits.'' (pg. 2081-2082). 

 If students have high test-anxiety then the best time to address that is before the test. 

References

1. Richardson, M., Abraham, C., & Bond, R. (2012). Psychological correlates of university students’ academic performance: A systematic review and meta-analysis. Psychological Bulletin, 138(2), 353-387. https://doi.org/10.1037/a0026838. 

2. Moran, T. P. (2016). Anxiety and working memory capacity: A meta-analysis and narrative review. Psychological bulletin, 142(8), 831. https://doi.org/10.1037/bul0000051 

3. Theobald, M., Breitwieser, J., & Brod, G. (2022). Test anxiety does not predict exam performance when knowledge is controlled for: Strong evidence against the interference hypothesis of test anxiety. Psychological Science, 33(12), 2073-2083. https://doi.org/10.1177/09567976221119391 

4. Abrams, Z. (2022). Student mental health is in crisis. Campuses are rethinking their approach. Monitor on Psychology, 53(7), 60. https://www.apa.org/monitor/2022/10/mental-health-campus-care

5. Kush, J. M., Goicoechea, E. B.,. Musci, R. J., Stuart, E. A. (2022). Teacher mental health during the COVID-19 pandemic. Educational Researcher, 51(9). https://doi.org/10.3102/0013189X221134281 

6. Beilock, S. L., & DeCaro, M. S. (2007). From poor performance to success under stress: Working memory, strategy selection, and mathematical problem solving under pressure. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(6), 983-998. https:/doi.org/10.1037/0278-7393.33.6.983