This is a long story and you may want to read some previous instalments (see here and here). However, the short version goes something like this.
Jo Boaler, professor of mathematics education at Stanford University in the US, has long claimed that timed maths tests cause a debilitating condition known as ‘maths anxiety’ that affects students’ performance in maths and motivates them to avoid maths. Boaler’s claims range from stating that, “Evidence strongly suggests that timed tests cause the early onset of math anxiety for students across the achievement range,” to, “For about one-third of students, the onset of timed testing is the beginning of math anxiety.”
I have been following the citation trails for these claims for a number of years and cannot find anything to substantiate them. For example, the ‘one-third’ claim made in American Educator points to a page on Boaler’s YouCubed website. In turn, the YouCubed page references an article titled, “Research suggests timed tests cause maths anxiety”, (although it did not always do so and the current link to it is broken). The article does not substantiate the ‘one-third’ claim and when challenged on this by Stephanie Lee of The Chronicle of Higher Education, Boaler then cited a range of different articles which, again, do not substantiate this claim. Instead, Boaler stated that the ‘one-third’ figure, “…is clearly an estimate, that draws from different scientific papers and decades of work in schools.” I strongly disagree that this is ‘clearly’ an estimate. Boaler made this statement and offered a citation for it with no indication that it was an estimate based on ‘decades of work in schools’. This is not accepted academic practice and it is the sort of thing a PhD student would be strongly criticised for by their supervisor.
I then examined the range of articles Boaler offered in response to Lee and did not find compelling evidence for any causal link between timed testing and maths anxiety.
As I noted, if we wind back a little to the, “Research suggests timed tests cause maths anxiety,” article, there is nothing to substantiate the ‘one-third’ claim. Boaler refers to some research she conducted that involved students writing about how tests made them feel. This design cannot demonstrate that timed tests cause maths anxiety. Boaler also — now infamously — notes that, “researchers now know that students experience stress on timed tests that they do not experience even when working on the same math questions in untimed conditions (Engle 2002).” The Engle 2002 reference has nothing to do with timed tests or maths anxiety. I assumed this was a citation error and have been keen for Boaler to correct it.
Even if it could be shown that students experience stress on timed tests that they do not experience when working on the same questions in untimed conditions, this would not prove that timed tests cause maths anxiety. Maths anxiety is not meant to be a temporary state, but a chronic and debilitating one that causes students to avoid mathematics electives. Temporary stress can even be a positive in some contexts, sharpening performance. Nevertheless, it is important to understand what Boaler intended to reference here.
Recently, an anonymous complaint was lodged with Stanford about Boaler’s approach to citations and the ‘Engle 2002’ issue was part of it. In response, Boaler has finally cleared this up. She first posted an extraordinary statement to YouCubed in which she accused Stephanie Lee of being ‘shoddy and indolent’ and the Washington Free Beacon, a conservative outlet that broke the story of the complaint, of being ‘far-right’. She also stated that, “All the studies were accurately reported, with one misplaced citation, which has now been replaced with the correct one.”
Boaler did not link to this correction, but with a little informed detective work, I found it on this YouCubed page. In reference to the, “Research suggests timed tests cause maths anxiety,” article, the page states, “Note: The article includes a misplaced citation. On page 469, instead of Engle (2002) the correct citation is Ashcraft (2002).”
So, we can now see that the article Boaler intended to reference is this 2002 review by Mark Ashcraft.
Does it support Boaler’s claim that timed tests cause anxiety when the same questions in untimed conditions do not? Yes and no. Yes, in the sense that I see this as an acceptable citation that is open to interpretation and, as such, could reasonably be cited by Boaler in this context. No, in the sense that my interpretation is that this is not very strong evidence at all.
In a long article that addresses many aspects of maths anxiety, Ashcraft writes the following relevant to Boaler’s claim:
“….we (Faust, Ashcraft, & Fleck, 1996) found no anxiety effects on whole-number arithmetic problems when participants were tested using a pencil-and-paper format. But when participants were tested on-line (i.e., when they were timed as they solved the problems mentally under time pressure in the lab), there were substantial anxiety effects on the same problems.”
More that one thing varies at a time in this experiment. It is not just the presence of timing. The format — pencil and paper versus online — changes. And there would undoubtedly be some effect of being ‘in the lab’.
Significantly, the abstract of the paper does not support the contention of Boaler’s that sent us down this rabbit hole in the first place. It states, “Although the causes of math anxiety are undetermined, some teaching styles are implicated as risk factors.” Ashcraft expands on this in the body of the paper:
“Math anxiety is a bona fide anxiety reaction, a phobia (Faust, 1992), with both immediate cognitive and long-term educational implications. Unfortunately, there has been no thorough empirical work on the origins or causes of math anxiety, although there are some strong hints. For instance, Turner et al. (2002) documented the patterns of student avoidance (e.g., not being involved or seeking help) that result from teachers who con-vey a high demand for correctness but provide little cognitive or motivational support during lessons(e.g., the teacher “typically did not respond to mistakes and misunderstandings with explanations,” p.101; “he often showed annoyance when students gave wrong answers . . . . He held them responsible for their lack of understanding,” p. 102). Turner et al. speculated that students with such teachers may feel “vulnerable to public displays of incompetence” (p. 101), a hypothesis consistent with our participants’ anecdotal reports that public embarrassment in math class contributed to their math anxiety. Thus, it is entirely plausible, but as yet undocumented, that such classroom methods are risk factors for math anxiety.” [My emphasis]
So, there is no strong evidence on the causes of maths anxiety but, when speculating, Ashcraft does not mention timed tests.
Ashcraft also notes that anxiety is most likely to show up when performing multi-step calculations or work with fractions that imposes a strain on working memory. It is less likely to show up for simple recall of maths facts.
Why is this final point important? That goes to the reason why I have spent so much time down this rabbit hole in the first place. Jo Boaler is extraordinarily influential in the world of maths education. She even managed to visit my hometown of Ballarat last year to spread her agenda. Many teachers are convinced by Boaler that evidence has conclusively demonstrated that timed tests cause maths anxiety and have adjusted their teacher methods accordingly.
In Boaler’s perhaps most influential article on timed tests, she repeatedly criticises the use of timed tests of multiplication facts.
As Ashcraft states:
“In our view, routine arithmetic processes like retrieval of simple facts require little in the way of working memory processing, and therefore show only minimal effects of math anxiety.”
And yet the benefits for fluency of swift recall of maths facts, as facilitated through timed retrieval, are likely to be significant. Students may well be missing out on this beneficial retrieval practice because their teachers have been convinced by Boaler that it will cause harm.
That is not OK and so, despite the hyperbolic denunciations, I will continue to prosecute my case.
I think this is an important topic, do you think there are interested maths researchers to investigate or organise a grant for it?
Seems ridiculous that in 2024 we haven't got recent and more insightful studies interrogating this highly critical topic?