The so-called blood transfusion scandal illustrates precisely why we need technocrats and why it can be dangerous to air technocratic arguments in public—there should not be a simple opposition to technocracy, as argued in a recent Mail & Guardian supplement.
In many instances, there is little room for misinterpretation of statistics, as the process of interpretation is itself just a conversion of the statistics into a different form. The calculation of “risk” is one such case. Assume that we want to find a way of determining the risk of someone in South Africa being knocked down by a car.
A statistical analysis will begin by speculating about what factors might make a person more likely to suffer such a fate.
Mode of transport is an obvious one; people who walk to get about are more likely to be knocked down than someone who uses a car. Nationality is another less obvious factor. Foreigners may be at greater risk because they do not know that speed limits are not as strictly enforced as in their home country, or because they drive on the other side of the road.
The beauty of statistical analysis is that it can tell us whether the evidence supports such guesses. But imagine that some overenthusiastic policymakers then decided to ban all foreigners from crossing roads. Would this be xenophobic? I would argue not. It might be stupid policy, because there are less extreme ways of dealing with the problem, but it is clearly motivated by concern for the foreigners’ well-being.
So now we get to the recent uproar about racial profiling. Here are the facts. HIV tests are not 100% foolproof, so some blood tests negative even though it is actually positive. To reduce the chance of using such HIV-positive blood, the Blood Bank classifies blood into different “risk groups”. The process does not eliminate accidental HIV transfer, but seeks to reduce it as much as possible.
There are many factors associated with a higher risk of HIV that will be used, among them having multiple sexual partners and/or unprotected sex and previous infection with sexually transmitted diseases. Many studies have indicated a higher HIV prevalence rate in South Africa’s black population than among other racial groups for reasons that have not been wholly explained by statistical factors.
If it is true that, even for two people who are otherwise identical by the other factors, a black person is more likely to have HIV than a white person then, based on the reasoning explained earlier, it is a statistical fact that black people should be considered higher risk. This is a big “if”. It might be that poverty is a risk factor and because more black than white people are poor the analysis has incorrectly made race, instead of wealth, the risk factor. The problem is that, to my knowledge, no combination of risk factors has been found that totally eliminates the influence of the race factor.
A similar problem, referred to as the “African dummy”, plagued cross-country analyses of economic growth. (This is not an insult; a “dummy variable” is something that is used in many statistical analyses). These found that African countries otherwise identical in all the social, political and economic factors to countries in another continent, still had lower growth rates than those countries.
This is troubling to economists, who are reluctant to believe that merely being on a different continent has a negative effect on economic growth. Recently a few studies have come up with other socio-economic factors that might explain this phenomenon, but there is no reason to believe that will be possible in every such case.
In the current situation, it would be best for all parties if the South African National Blood Service could find a factor, other than race, that explains the difference in prevalence between the population groups. Assume for a moment though that this is not possible. Virtually all of the public comment on this subject has condemned outright the use of race as a risk factor, regardless of any statistical explanation.
While many have found little comfort in the Department of Health’s recent statement that racial profiling will not be used in future because it doesn’t change what has already happened, I am concerned for a different reason. The department’s statement means that, unless the Bank has found a “proxy” or substitute variable for race, we are sacrificing not “lies and statistics” but lives—black and white—and statistics on the altar of political correctness.
Statistical profiling that uses race where there is no alternative measure is no more racist than saying “foreigners are more at risk of being run over by cars” is xenophobic.
Sean Muller, an economics graduate, is a Public Policy Partnership Fellow. He writes in his personal capacity