What’s wrong with the South African Institute of Race Relations?
Surveys on race by South Africa’s Institute of Race Relations (IRR) are deeply flawed and cynically used. Its influence on mainstream politics is significant and dangerous.
Media headlines across South Africa have repeatedly trumpeted findings by the South African Institute of Race Relations (IRR) that claim “more White people have experienced racism than Black people since 1994.” The website, TimesLive, for example, claimed the IRR found that Black people experience less racism than Whites. It reported an IRR poll opinion found that “… 54% of white people questioned said they experienced racism, while only 36% of black people said they experienced racism.” In December 2021, a headline on the website of radio station Jacaranda FM read “80% of people haven’t experienced racism in the past 5 years.”
These headlines, like the attitudinal survey results they are based on, show that Whites reported experiencing racism more than any other group in IRR polls. What they don’t tell us is if White people actually experience more racism.
At first glance you may think that the problem here is just that the media reports research in sensationalist ways. But when you take a closer look at IRR surveys, how it uses them in public relations campaigns and lobbying efforts, and how it analyses results in its reports, a disturbing picture emerges.
I have recently argued that IRR research is systematically misleading and riddled with methodological errors that seem designed to produce results favorable to the organization’s lobbying endeavors, many of which deny that racism is a serious problem in South Africa.
🚨 New paper alert 🚨 https://t.co/SRUvLrgyYf
The future of SA hangs in the balance. @IRR_SouthAfrica argues for a classically liberal vision for SA’s future & argues that our focus on racism is disastrous. But they cut corners to make their case: a thread 🧵 1/10
— Phila M. Msimang (@Phila_033) October 3, 2022
The IRR’s claims are neither supported by the data it collects nor by how it analyzes it. Some of the conclusions that the IRR draws from its surveys cannot possibly be supported by the kind of survey evidence that it draws on.
For instance, there’s a big difference between establishing what people believe about an issue and finding out what is factual about that same issue. An example of this is the difference between establishing people’s opinions about what policies they believe would work for economic advancement and finding out what policies would actually work for economic advancement. This distinction between opinion and fact is not one the IRR makes in its research. From asking people’s opinions on what economic policies they think would be best, the IRR makes the false inference that the opinions found to be most popular—beyond purportedly showing us what policies people support—are the ones that show us which economic policies are actually best.
The IRR also fails to make a distinction between popular opinion and fact when enquiring about how serious of a problem racism is in South Africa. The IRR claims that we can know how bad racism is simply by asking people to recall racist incidents in their experience that were directed against them. The IRR’s understanding of racism focuses on incidents of interpersonal conflict on the basis of race. Although this seems like a reasonable view, as a research institution the IRR should know that this is not the only way in which racism manifests.
Take the example of job discrimination by race group in hiring practices. For the last few decades, empirical studies of racial discrimination in hiring practices have shown that how likely a person is to be recruited for a job depends significantly on the race group that they belong to, or that they are assumed to belong to. Numerous media outlets have shared the findings of a recent report by the World Bank that found that race is still the leading indicator for inequality in South Africa. Other studies have found that South Africa has racially inequitable employment outcomes even when we have controlled for relevant factors such as age, education, experience, and skill level discrimination. In audit studies in different regions, racial discrimination is tested through the likelihood that someone would be called back given the name written on their CV. In the unfortunate case that you are at the receiving end of job discrimination by race during the recruitment process, you might not know it. You just won’t be called back, or you simply won’t be recruited for the job.
Racial discrimination can happen even when you are unaware of it, like when the outcome (e.g., not being called back or recruited) obscures the racist causes (e.g., employers tend to prefer a White candidate over a Black candidate). It is this kind of empirical research that is often used as a rationale for affirmative action as opposed to views such as the IRR’s against affirmative action.
The IRR’s characterization of racism as the racist incidents people report that they have experienced is far too narrow. It cannot even capture all attitudinal aspects of racism (like in the aspects of job discrimination described above), or the institutional, structural, and systemic aspects of racism that do not require a racist person or racial animus to drive further racial injustice. Examples of these different kinds of racism can be readily found in the relationship between racism and health as international studies show. All of these aspects of racism are not accounted for or captured in the IRR’s research or reflected in how it understands racism in its work.
When research so poorly captures the problem it purports to study, its data tends not to be scientifically useful and does not produce reliable evidence. This is what I argue is clear from the research of the IRR on racism. Notwithstanding, the IRR has received tremendous mileage from the research and use it as a political weapon.
The IRR’s research is used to lobby against legislative and social interventions targeting racism and its socio-economic legacies. It is used to shut down conversations about racism unless they pertain to the purported victimization of White people, such as in the presently ongoing Dischem affirmative action controversy or in consideration of the possible racial motivations behind some farm attacks. The IRR’s research is referred to and cited in discussions by a range of liberal and conservative commentators in South Africa and abroad on social issues. The IRR is cited in discussions on farm attacks to bolster the view that Whites are being specially targeted on the basis of race. This is related to the myth of White genocide, a myth that the IRR keeps an official distance from in discussing farm attacks. More junior IRR staff like the IRR’s policy analyst and contributor to the Free Market Foundation Mpiyakhe Dhlamini draw a closer connection in conservative media between the White genocide conspiracy theory and the position of the IRR. The research of the IRR also features in certain court cases, such as in the recent hate speech case brought to the courts by AfriForum, a controversial conservative Afrikaner pressure group, in which Gabriel Crouse of the IRR was called as an expert witness.
The IRR’s research has been used by the likes of John Steenhuisen, the leader of the opposition party Democratic Alliance, and Gareth Cliff, a former radio DJ but now a controversial podcaster, to downplay the racial tensions and the role of racism in the vigilante murders that occurred during the July riots of 2021 in Phoenix. The conversation between Mudzuli Rakhivhane, Cliff, and Steenhuisen is exemplary of how the IRR’s data is used to shut down conversations about anti-Black racism and its connection to socio-economic and other systemic injustices. In this way, the IRR’s work is used to play up the victimization of White people while downplaying ongoing racism against Black people.
Although the IRR’s research is of questionable scientific value, its influence on mainstream politics is significant and dangerous. This is the conversation we should be having next.