31 lines
7.9 KiB
Plaintext
31 lines
7.9 KiB
Plaintext
Objective evidence against racism
|
|
<p>A theme I have touched on several times in my blogging is that the best way to defeat racism and other forms of invidious discrimination is to develop and apply objective psychometric tests.</p>
|
|
<p>Usually I make this argument with respect to IQ. But: one of my commenters, an obnoxious racist who I refrain from banning only on free-speech principle, recently argued that drug use should be (as, in fact, it now often illegally is) treated differently by police depending on the subject’s race. </p>
|
|
<p>His argument (if you want to call it that) is that blacks, due to a low baseline level of self-regulation, are significantly more prone to criminality and violence than whites when intoxication further impairs that ability. Thus, the law should treat these cases differently as a matter of public safety.</p>
|
|
<p>As presented, this prescription is racist, repugnant, and wrong. Because even if you believe that blacks as a group have less ability on average to self-regulate, this belief tells you nothing about any individual black person. Acting on it would infringe the foundational right of individuals to be treated equally by the law.</p>
|
|
<p>But now let’s perform a thought experiment. Actually, a couple of related ones.</p>
|
|
<p><span id="more-4994"></span></p>
|
|
<p>Let us suppose that scientists were able to develop a behavioral test that reliably and repeatably measures an individual’s ability to self-regulate as a number, which we’ll call RQ. In fact there have been <a href="http://en.wikipedia.org/wiki/Stanford_marshmallow_experiment">attempts at this</a>.</p>
|
|
<p>For purposes of this thought experiment, suppose that (a) we have such a test that is highly repeatable, (b) the distribution is like most psychometric scores Gaussian, and (c) there is a range of low scores on the left tail of the Gaussian that is known to correlate strongly with criminal and violent behavior. </p>
|
|
<p>If you don’t know much about psychometry, you might think these premises are implausible. Follow the logic anyway, because I’m really chasing a point about how we should perform ethical reasoning. Besides, they’re actually quite plausible – there are known psychometrics with these properties.</p>
|
|
<p>Our test divides the population into low-RQ and high-RQ contingents naturally, because if you plot both Gaussians the place where they intersect defines equal likelihood that a person with that RQ score is or is not part of the criminal population. </p>
|
|
<p>Now suppose that we consider drug-fueled crime a sufficiently present danger that all citizens are required to have their RQ measured and registered with the police. (I’m not advocating this, but it’s necessary for the thought experiment. If you like, assume instead that it appears on driver’s licenses along with height and weight.)</p>
|
|
<p>What would the ethics be of a law treating as a crime or disorderly conduct public intoxication of any person who cannot exhbit an RQ score above a specified minimum?</p>
|
|
<p>(Note that we can, in principle, deal with measurement uncertainty by contracting the low-RQ range. That is, if we believe with 95% confidence that an individual’s score on RQ tests won’t vary by more than N points over a series of tries, we simply reduce the defining bound of the low-RQ cohort by N below wherever the RQ distributions for our defined criminal and noncriminal populations intersect.)</p>
|
|
<p>Would this violate the principle of equal treatment under law? No, not any more than (for example) forbidding people with epilepsy from driving cars. In both cases, any reasonable person (including the offender himself) would know that the combination of his condition and behavior made him a danger to those around him. Legal discrimination against epileptics is therefore justified.</p>
|
|
<p>The epileptic-in-the-driver’s-seat case enables us to dispose of another objection. Should our attitude about the laws disqualifying epileptics from driving change if we learn that epilepsy is not evenly distributed across racial groups, however we define race? </p>
|
|
<p>Actually this <em>isn’t</em> a hypothetical; racial minorities in the U.S. do in fact have higher incidence of epilepsy than whites. But no civil-rights lawsuits on a disparate-impact theory have been filed, because that would be too insane for even the most extreme demagogues in our racial-grievance industry. The intent and effect of keeping epileptics off the roads is to avoid preventable injury and death due to a medical condition that can be unambiguously diagnosed. </p>
|
|
<p>By the same reasoning, temporarily jailing low-RQ people for disorderly conduct while only monitoring high-RQ people for actual crime would not be racially discriminatory even if different races turn out to have different mean RQs.</p>
|
|
<p>The key point in the conditions of our thought experiment is that an RQ is not assigned to a person on the basis or race or other ascriptive grouping; it is an individual measure of an individual used to make rationally justified inferences about that individual’s behavioral risks.</p>
|
|
<p>Having got this far, we now reach the part of the hypothetical that is likely to truly upset the tender-minded. Suppose RQ turns out to be highly heritable? Does that change the soundness of the law at all, or make it racially discriminatory?</p>
|
|
<p>Well, suppose epilepsy turned out to be highly heritable? Would that render our prohibition on epileptics driving instantly unsound and racist? Clearly not; it does not matter to the intent or effect of the law <em>why</em> they occasionally zone out or have convulsions, just that you do not want that happening behind the wheel of a car on a public road.</p>
|
|
<p>By the same reasoning, legal discrimination against low-RQ persons based on RQ would not become invidious if we knew RQ were heritable.</p>
|
|
<p>Now we come to the worst case: RQ is highly heritable, and the means differences among racial groups are large. Summing up the argument, I have shown that legal discrimination on the basis of RQ should <em>even so</em> not be considered racist or wrong.</p>
|
|
<p>Now we come back to where I began this post and consider again the difference between racist and non-racist thinking. The least prejudiced and nasty version of my commenter’s argument would read like this:</p>
|
|
<p>“People with poor ability to self-regulate should be arrested and jailed when they get drunk, because they’re dangerous to themselves and others. Black people have low mean ability at self-regulation. Therefore, blacks should be presumptively arrested and jailed when drunk in public.”</p>
|
|
<p>The point I’ve been working up to this whole essay is this. Many people think the racist part of this argument is the assertion “Black people have low mean ability at self-regulation.” In fact, I expect many of my readers type that as racism so strongly that they would consider any attempt to measure RQ automatically suspect.</p>
|
|
<p>But this is silly. That claim is no more racist than the following: “Black people have a higher incidence of epilepsy than white people.”</p>
|
|
<p>These claims may be true or false: we know the latter is true, we don’t know if the former is. But neither claim is ‘racist'; or, to put it another way if either claim is racist then both are and we have emptied the predicate “racist” of any meaning a serious person should care about.</p>
|
|
<p>No. The racism in the argument happens after the “therefore”, at the place where it leaps from a claim about the statistical distribution of RQ to a pre-judgment about all blacks and about whatever individual blacks we might encounter.</p>
|
|
<p>I have said it before, and I expect I’ll have to say it again. The individual is not the mass. The point is not the distribution. Racism is not merely hatred, it is a fundamental failure of reasoning. It is not just vicious, it is <em>stupid</em>.</p>
|