Friday, June 13, 2014

IQ and intellectual disability

One of the things about the death penalty is that the Supreme Court of the United States has held that no legitimate penological purpose is served by executing the intellectually disabled: Atkins v Virginia, 536 U.S. 304, 312 (2002).

So, what does “intellectually disabled” mean? Currently the medical community defines it as the existence of concurrent deficits in intellectual and adaptive functioning. The courts regard it as proper to consult the medical community’s opinions when it is necessary to come up with a definition.

In Hall v Florida, USSC No 12-10882, 27 May 2014 the issue was whether interpretation by the Florida Supreme Court of legislation, that could be read consistently with the USSC’s caselaw, was constitutional. The Florida court had held that intellectual disability requires an IQ of 70 or less. Mr Hall scored 71 (although as the USSC dissenters in Hall point out he had actually had several tests in which he scored higher). The case is not concerned with what “adaptive functioning” means, only with what “intellectually” means in the expression “intellectually disabled”.

The majority, Kennedy J, joined by Ginsburg, Breyer, Sotomayor, and Kagan JJ, held that the requirement for a score of 70 or under was unconstitutional because it ignored the margin of error inherent in testing. When a score of 70 is within the margin of error, the defendant should pass the first requirement and then have the opportunity to present evidence of lack of adaptive functioning. The decision of the Florida Supreme Court was reversed and the case remanded for determination consistent with the majority’s opinion.

The minority, Alito J, joined by Roberts CJ, Scalia and Thomas JJ, dissented on the grounds that the Florida court’s interpretation was reasonable because it allowed the defendant to present several IQ test scores, and it is known that accuracy increases when more than one test score is considered. The majority’s opinion introduced uncertainty, as to what scores could come within the range that passed the first requirement, and as to how courts should deal with changing medical opinion as to what score indicates intellectual disability. The law should apply society’s standards, not those of a professional body, and the proper focus is on whether the defendant is able to understand the possibility of execution as a penalty notwithstanding diminished ability to comprehend and process information, to learn from experience, to reason logically, or control impulses.

The minority criticised the majority’s method of counting the states when assessing what public opinion regards as appropriate, and pointed out that the majority did not address why Florida’s method of accounting for the risk of test error (by allowing more than one test to be considered) was ineffective.

There is plenty of interesting discussion of IQ testing in this case. The average IQ is 100, a standard deviation is about 15, and two standard deviations below average is taken to be 70, the upper limit for being intellectually disabled (more accurately: the upper limit for meeting the "intellectually" limb of the description "intellectually disabled"). But a test score comes with a range of error, and a person who scores 70 may really have an IQ between 65 and 75, but the range varies according to the range applicable for a given test and the extent to which one wishes to be sure of the result (that is, the degree of confidence). Some tests may carry a lower range of error, so that one could say with a high degree of confidence that a person who scored 70 really had an IQ between, for example, 68 and 72.

An interesting point, from a lawyer’s perspective, was made in the dissent: before this decision, a defendant had to satisfy the court on the balance of probabilities that his IQ was 70 or less. Now, if a defendant may have an IQ in a range that includes 70, he is entitled to go on to produce evidence of his lack of adaptive functioning. The standard of proof is, say the minority, altered.

I note (this is me now, not the Court) that in law we often receive scientific test results without reference to a range of error or a confidence interval. For example, regimes for testing in alcohol-related driving cases usually involve a test result which the defendant can challenge by independent analysis of a blood sample, where one has been taken. It is usual for the prosecutor's analysis laboratory to knock off a bit of the alcohol result just to be safe, so that even if the final reported level is just over the limit, an independent result would probably be higher. If a defendant facing a penalty for an alcohol-related driving offence has an opportunity to challenge the test result, so should a defendant facing the death penalty. Obviously everyone would agree with that. The analogy suggests that it is not necessary to express an IQ score as being within a range of error or as being within a given confidence interval.

Some test results received by courts, such as DNA results, are stated in probabilistic terms, and rightly so. If one were to require a probabilistic report of an IQ test score, one would look for some statistic such as a likelihood ratio: the probability of getting this score, given that that was the defendant’s IQ, compared to the probability of getting that score, given that that was not his IQ. But I suspect that the confidence interval says the same thing, and the question becomes, do courts really need that extra information?

If the reasoning of the majority in Hall were to be applied to evidence of the result of a blood test for alcohol, the effect would be to remove the legal limit. If, as I do, you instinctively side with the majority in cases like Hall, you still have to deal with the logic of the dissent. Instead of ignoring it, like the majority here, you could accept that sometimes in law logic is trumped by experience, and do as the Privy Council majority did in Ramdeen v The State (Trinidad and Tobago) noted here on 28 March 2014.