Some people in the US may have been wrongly executed because of inaccuracies in the IQ tests used to assess them. But the inaccuracies may also have seen some escape execution if they scored lower on the tests than their real IQ.
, a psychologist at the University of Huddersfield, UK, searched the literature to identify sources of error in two of the most commonly used IQ tests: the third edition of the Wechsler Adult Intelligence Scale (WAIS-III), which can be used on people aged between 16 and 93, and the fourth edition of the Wechsler Intelligence Scale for Children (WISC-IV), suitable for children aged between 6 and 16.
In both cases, the test manuals state that you can be 95 per cent confident that a person鈥檚 true IQ lies within 4 points of their test score.
Advertisement
However, Whitaker found that for people with extremely poor WAIS-III scores, their actual IQ could be up to 16 points higher or 26 points lower than the score achieved. In the WISC-IV test, actual IQ may be up to 25 points higher or 16 points lower than the score achieved.
In 2002, the US Supreme Court ruled that the execution of any mentally retarded person against 鈥渃ruel and unusual punishments鈥�. An IQ of 70 is generally considered the threshold 鈥� making Teresa Lewis (IQ 72), who was executed in September last year, a borderline case.
Too simplistic
Whitaker says that setting a cut-off point at 70 is too simplistic, given that his results suggest IQ tests may provide inaccurate results at this end of the IQ scale. He suggests looking at how an individual is functioning in the real world 鈥� 鈥渨hether they are failing at everyday tasks that require intellectual ability鈥� 鈥� to establish whether or not they are mentally retarded.
He offers a simple explanation for the wide error margins in IQ readings at the low end of the scale. The statistics used to arrive at the 95 per cent confidence level for the tests are based on the IQs of a representative sample of the population. 鈥淏ut by definition, most people in the population have average IQs,鈥� says Whitaker. 鈥淭his causes problems when statistics based on the performance of people with average IQs are assumed to apply to people with low IQs.鈥�
What鈥檚 more, the manuals usually take into account only one type of possible error 鈥� how well the items in the test measure the psychological factor that they are supposed to assess. In performing his calculation, Whitaker considered other sources of error such as the differences that may arise between scores an individual achieves if they take the IQ test several times.
of the University of New Mexico, Albuquerque, who represented a convicted murderer in the case that resulted in the Supreme Court鈥檚 ruling in 2002, says: 鈥淐ourts are becoming more sophisticated in dealing with the details of individual evaluations, particularly in states with a substantial number of death penalty cases.鈥�
He points out that last month the Tennessee Supreme Court ruled that in cases where a defendant鈥檚 IQ test score was slightly above 70, the lower courts should nonetheless not reject mental retardation as a defence.
High stakes
, a psychologist at Goldsmiths, University of London, says Whitaker鈥檚 research makes an important point. 鈥淎nxiety, low confidence or even overconfidence can distort a person鈥檚 IQ test score,鈥� he says.
Sitting a test to determine whether you are sentenced to death would be classed as a 鈥渉igh-stakes setting鈥� by psychologists. 鈥淯nder pressure like that, most people would not do their best,鈥� he says, and their score would underestimate their true IQ 鈥� again suggesting problems with using IQ alone to decide whether or not a convicted person should be spared the death penalty.
Whitaker reported his findings at the in Glasgow, UK, last week.
