Is race a useful metric?
Tom Traina has a post up discussing the use of race in actuarial tables (fixed link). He writes:
While it’s pretty clear that Judge Weinstein was trying to do the right thing in this case. But that doesn’t always lead people to actually do the best thing that can be done. While trying to reduce the legitimacy of minority stereotypes is an admirable goal, the means to it in this case simply don’t measure up. The actuarial tables don’t explain why their numbers are as they are, and the other considerations imposed by law get thrown by the wayside without do consideration. Perhaps some other solution would be a better way to achieve Judge Weinstein’s goals.
I like Mr. Traina’s comments, and think they are worth thinking about. But what is the idea behind adding race to the tables? Why not simply average them all together? The answer is that it lowers some people’s life expectancy and saves the company money. Of course, if they included all races, then it is possible the early deaths of minorities would bring the entire average down and save even more money. A lot would depends on the specifics.
But here’s the thing – as Mr. Traina admits, race is, at best, a proxy for a constellation of behaviors and risks. For example, the rate of high blood pressure and diabetes among racial minorities tends to be higher than for whites. All things being equal, an actuarial table that accounts for high blood pressure and diabetes would already account for that. So why include race as well? Doesn’t that try to account for the same problem twice?
I don’t know the actual statistics, but lets say that a person who has one immediate blood relative with diabetes is at a 10% greater risk for developing diabetes, and we’ll argue that the risk is constant across racial cohorts. That puts a person at a 10% greater risk for developing all of the problems that come with that disease. So if a Black person has one immediate blood relative that has diabetes, which is the more accurate risk – being Black or having a relative with diabetes?
Again, it would depend on specifics. But it would appear logical to conclude that the blood relative would be a more accurate risk than race. So if you use race rather than actual medical history, what is the benefit? Well, for Blacks who are related to diabetes-sufferers, they will have no net effect. For those without a relative diagnosed with diabetes, it will cheat them out of potentially thousands of dollars.
I can’t believe, at the time in history, that we don’t have actuarial tables that would show this. If being Black was a greater risk; then it seems logical that such evidence would have been presented at the time of the trial.
What the judge is saying, in effect, is to stop using race as a proxy and get data straight about other risk-factors. Then simply apply those risk-factors. If race can be shown to be the best indicator, then there should be scientific evidence of such. In the absence of such evidence, I’d have to say the judge made the right call.
Sphere: Related Content
Where I Blog
NJ News
National News