Intelligence & Test Anxiety
Two meta-analyses have established a moderate negative correlation between intelligence and test anxiety (r=-0.33 & r=-0.23 respectively) [1, table 1; & 2].
To the layman, the explanation for such a finding is obvious: Test anxiety causes people to think about the consequences of failure and thereby prevents people from thinking about how to solve test questions. There is however another possibility: People can estimate their test performance somewhat well, and performing poorly on a test will cause people to be more worried about their results than those who perform well. In the literature assessing the relevance of these two possibilities, the explanation where low intelligence causes test anxiety is referred to as the deficit hypothesis whereas the interference hypothesis is the one which posits low test performance to be caused by test anxiety [3]. The causal effect of test anxiety on test performance is suspect to begin with as a lack of preparation is the most common reason college students report for experiencing test anxiety [8]; things like these are signs of poor performers merely having accurate perceptions of their test performance ability beforehand. If the people who have test anxiety for these sorts of reasons have their test performance further harmed by things like distraction effects, then test anxiety is demoted from being causal for test performance to having bidirectional causality with test performance right out of the gate. Multiple lines of evidence indicate however that the interference hypothesis cannot even be afforded this dignity, with causality going from intelligence to test anxiety rather than vice versa. These are:
-Evidence against Stereotype Threat effects
-The etiology of the IQ-Anxiety covariance
Stereotype Threat:
To begin with, it is worth thinking back to the stereotype threat literature. Stereotype threat occurs in a situation in which it is plausible that some members of a social group may exhibit behavior which is typical of a stereotype about their respective group. It is thought that belief in one’s groups’ stereotypes induces feelings of ‘threat’ (or test anxiety) that cause the stereotypes to become a self-fulfilling prophecy, and that stereotype threat effects partially contribute to long standing racial and gender gaps in academic performance, intelligence, etc. It is thought that these effects can be experimentally tested with so-called “primes” in tests. For example, let’s say two groups are given a test, and for one group the start of their test says that racial groups consistently perform equally on the test, while the control group gets no such prime, or perhaps the prime says that some group performs worse. If the prime group and the control group have different performances, this is supposed to be evidence for stereotype threat.
Or at least that’s the theory. The evidence? A bunch of small studies with various p-hacking issues and then some larger studies with null results (Reviewed here). Stereotype threat effects do not exist. In this wealth of literature covering an example of what may be considered experimentally-induced test anxiety, the anxiety does not result in reduced test performance.
Ability & Anxiety Source:
Next, various studies have assessed subjects’ anxiety before and after being administered their tests to compare how test anxiety relates to ability differently under the two conditions. In the most basic of such analyses, correlations between anxiety and performance have been repeatedly replicated to be larger when anxiety is measured post-assessment [4, 5, & 6]. To more rigorously make the point, variance in pre-test anxiety which does not co-vary with post-test anxiety does not correlate with ability, but variance in post-test anxiety which doesn’t co-vary with pre-test anxiety does correlate with ability [5].
To elucidate the relevance of this line of evidence, we can look at a similar approach which looks at how ability and test anxiety relate to general anxiety, and how general anxiety fits into the relationship between the two. Some people tend to be stable in a general trait of neuroticism throughout their lives such that they worry about everything, they tend to experience more negative emotions, they’re emotionally unstable in that their moods change frequently, and such that they have a tendency towards anxiety across the various domains of life (social anxiety, test anxiety, etc). Such a trait of general anxiety is a good candidate for being causal for test anxiety, and so any variance in test anxiety caused by general anxiety can be taken as probably not being influenced by ability; if this variance in test anxiety is disproportionately responsible for the correlation between test anxiety and ability, then this speaks to causality flowing from test anxiety to ability rather than flowing from ability to test anxiety.
One such paper on the relationship between math anxiety and math ability found the correlation to be mostly non-mediated by general anxiety [7]; the two types of anxiety were related to each other, and both were related to ability, but test anxiety could not have been an important mechanism by which general anxiety affects ability.
There may be a slight degree of collinearity between ability and the two measures of anxiety, but it could easily just be that generally intelligent people have less to worry about throughout life in general, with ability being causal for both general anxiety and test anxiety. If, instead, general anxiety were causal for ability by way of causing task-specific anxiety and distracting from task-specific performance, then the relationship between general anxiety and task performance should be disproportionately mediated by task-specific anxiety; the percent of variance in task anxiety explained by general anxiety should be smaller than the degree to which general anxiety mediates the relationship between ability and task-specific anxiety. Overall, this general line of evidence contradicts such a prediction; the contribution of general anxiety & pre-test anxiety to the relationship between ability and test anxiety / post-test anxiety is negligible in spite of a mean r = .56 correlation [2] between general anxiety and test anxiety.
SEM:
Finally, various studies assess the bias of test anxiety on test performance with a latent variable approach to see whether or not the correlation between test anxiety and test performance is mediated by the latent variable underlying test performance. This approach exploits the logic that test anxiety shouldn’t affect all test items equally; on less complex questions for example, test anxiety wouldn’t be expected to affect responses to items such as “2+2=? A) 22 B) 4 C) 5” This approach repeatedly shows that test anxiety relates to test performance via latent performance ability rather than other item specificities on the more anxiety-inducing items [3, 5, 9, 10, 11, & 12]. This is also true for high-stakes testing conditions [3, 10, 11, & 12] such as those involved in taking the ACT [3].
Sauce:
Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and interests: evidence for overlapping traits. Psychological bulletin, 121(2), 219. Retrieved from https://sci-hub.ru/https://doi.org/10.1037/0033-2909.121.2.219
Hembree, R. (1988). Correlates, causes, effects, and treatment of test anxiety. Review of educational research, 58(1), 47-77. Retrieved from https://sci-hub.ru/https://doi.org/10.3102/00346543058001047
Steedle, J. T. (2018). Keeping Your Cool: Does Test Anxiety Bias Performance on the ACT? Research Report 2018-3. ACT, Inc. Retrieved from https://files.eric.ed.gov/fulltext/ED593180.pdf
Zeidner, M. (1991). Test anxiety and aptitude test performance in an actual college admissions testing situation: Temporal considerations. Personality and Individual Differences, 12(2), 101-109. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/0191-8869(91)90092-P
Sommer, M., & Arendasy, M. E. (2014). Comparing different explanations of the effect of test anxiety on respondents' test scores. Intelligence, 42, 115-127. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.intell.2013.11.003
Seipp, B. (1991). Anxiety and academic performance: A meta-analysis of findings. Anxiety research, 4(1), 27-41. Retrieved from https://sci-hub.ru/10.1080/08917779108248762
Malanchini, M., Rimfeld, K., Wang, Z., Petrill, S. A., Tucker-Drob, E. M., Plomin, R., & Kovas, Y. (2020). Genetic factors underlie the association between anxiety, attitudes and performance in mathematics. Translational psychiatry, 10(1), 1-11. Retrieved from https://sci-hub.ru/https://doi.org/10.1038/s41398-020-0711-3
Bonaccio, S., & Reeve, C. L. (2010). The nature and relative importance of students' perceptions of the sources of test anxiety. Learning and Individual Differences, 20(6), 617-625. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.lindif.2010.09.007
Reeve, C. L., & Bonaccio, S. (2008). Does test anxiety induce measurement bias in cognitive ability tests?. Intelligence, 36(6), 526-538. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.intell.2007.11.003
Sommer, M., & Arendasy, M. E. (2015). Further evidence for the deficit account of the test anxiety–test performance relationship from a high-stakes admission testing setting. Intelligence, 53, 72-80. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.intell.2015.08.007
Sommer, M., & Arendasy, M. E. (2016). Does trait test anxiety compromise the measurement fairness of high-stakes scholastic achievement tests?. Learning and Individual Differences, 50, 1-10. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.lindif.2016.06.030
Sommer, M., Arendasy, M. E., Punter, J. F., Feldhammer-Kahr, M., & Rieder, A. (2019). Do individual differences in test-takers' appraisal of admission testing compromise measurement fairness?. Intelligence, 73, 16-29. Retrieved from https://sci-hub.ru/https://doi.org/10.1016/j.intell.2019.01.006