Letter to the Editorial Board of Aging, Neuropsychology and Cognition
2 May 2019 - 11:30
Headshot of Dr Barbara Boswell
We write to strongly object to the publication (March 2019) of the article ‘Age- and education-related effects on cognitive functioning in Colored South African women’ by Nieuwoudt, Dickie, Coetsee, Engelbrecht and Terblanche. We ask that you retract it because of its racist ideological underpinnings, flawed methodology, and its reproduction of harmful stereotypes of ‘Coloured’ women.
The authors ignore a large body of postcolonial and critical race theory which shows (1) that the idea of ‘race’ is a set of articulated political relations and (2) that racial categories are highly unstable, fluid, and provisional. Instead, they uncritically use the apartheid racial designation ‘Coloured’ (p1-2). Their definition of communities so classified does not problematise the idea of ‘mixed race’; incorrectly suggests that these communities are a homogeneous class; conflates ‘race’ and ethnicity; and suggests what can only be read as percentages of biological inheritance by ‘race’ and ‘clan’. The latter is akin to eugenics.
Equally important, with respect to ‘social class’ the authors ignore the latest data from Statistics South Africa about economic activity, labour force, poverty, and indigency, instead electing to cite incorrect, outdated sources in order to support their argument .
This article is scientifically flawed. Its title, abstract and introduction infer that the results are applicable to all ‘Coloured South African women’. However, the authors acknowledge that they draw on a small sample size; that the 60 participants were from only one geographic community; and admit that their methodology produced a result that ‘is likely not fully representative of the larger Colored population of SA’ (14).
One instrument used to measure cognitive function, the Montreal Cognitive Assessment (MoCA) has been shown, in the South African context, to yield results that are fundamentally flawed. Robbins et. al (2013) found that HIV- Xhosa-speaking South Africans tested with this instrument performed similarly on several tasks to North Americans aged 70 and over who had Alzheimers disease. They concluded that:
“… the normative data from the MoCA samples is likely wholly inappropriate for this [South African] population. Using [this data]… may lead to misclassification of healthy individuals as impaired in populations similar to our sample. Further research is needed to establish locally appropriate normative data and to determine the most sensitive and specific cut-off scores. Most participants in this study, regardless of HIV status, would be classified as impaired when compared to the MoCA normative sample.” (Robbins et.al, 2013: 450)
From the authors’ own data (Table 1) the group of women aged 40-49 are within their ‘normal’ range, attaining an average MoCA score of 26.1. They fail to explain this as it contradicts not only their claim of age-related decline but also their spurious link to educational level. This group had the second lowest average education level, namely 10.3 years of formal schooling. Shockingly, the authors do not discuss this discrepancy. Koen et al (2015) in their examination of the strengths and limitations of MoCA, posit that it is a measure of mild cognitive impairment, it is not robust, it requires a longitudinal study, it cannot be viewed as a substitute for in-depth neuropsychological assessment, and it is a more accurate and reliable indicator when a MoCA upper limit of mild cognitive impairment of 22 is used and 17 for Alzheimer’s Disease.
Nieuwoudt et al did not use a longitudinal study. All the average MoCA scores in their study were above 22 for all age groups.
The authors use a cognitive measurement instrument which has been demonstrated to be deficient and inapplicable to South African contexts. Moreover, the authors acknowledge that their normative sample data is dated and drawn from a population of white, adult Americans aged 7 to 90 years (see p10). The value of comparing their sample of 60 ‘Coloured’ women to this normative sample is highly questionable.
Their argument is circuitous and biased: the authors start off with the premise that ‘Coloured’ women are at ‘increased risk for low cognitive functioning’ (1) and work from this assumption to confirm that these women are ‘at high risk...for low cognitive functioning’ (10). They attribute their findings to ‘a combination of low education level, poor quality of education and socio-demographic factors such as ethnicity, employment, marital status, income and health status…’ (10). In other words, without a comparison or matching group, the authors conclude that low cognitive scores are attributable ethnicity (i.e., what they define as being coloured).
On these grounds we find the article fundamentally flawed. Their own data does not support their assertions. There is no new finding here; just a repackaged Verwoerdian paradigm. We thus ask that you retract this article.
Why is this important?
The issue is important because the article is published as scientific research but draws on colonial stereotypes of African women, and 'Coloured' South African women specifically, as intellectually deficient. The article relies on flawed methodology and science, perpetuating harmful, racist stereotypes.