The SEM must be considered in the interpretation of IQ scores

State of Florida has its psychometric head in the sand re: precision of IQ scores in Atkins death penatly cases

More on Dufour v Florida (in process) Atkins case. [Note. Dufour has been added to the Court Decisions section of the blog]

The other day I posted an AAIDD Amicus Brief supporting Dufour's appeal fo the Florida Supreme Court for a a diagnosis of mental retardation. It was clear from the brief that one of the main points of contention was the argument that the court needed to take into consideration the standard error of measurement (SEM) in IQ scores.

I've now obtained copies of the most recent Ninth Judicial Circuit Courts 2008 order denying Dufour's motion for determination of mental retardation (click here to view). Despite a number of total full scale IQ scores just above 70, and others below 70 (although questions were raised about their validity--e.g., malingering) it is my reading of this order that the state of Florida has drawn a line in the psychometric sand at the IQ score of 70, and that SEM is not to be taken into consideration in making a determination of mental retardation. The Florida Supreme Court has consistently invoked an absolute MR criterion of "IQ of 70 or below" as evidenced on page 3 of the order (citing Phillips v State, 2008; Cherry v State (2007), Jones v State (2007).

It appears that the state of Florida has its psychometric head in the sand regarding one of the most basic, fundamental and accepted concepts in psychological measurement--namely, the tests are fallable (contain some measurement error) and, more importantly, psychometrics allow for the quantification of this known measurement error in the form of the SEM (which allows for the establishment of a confidence band around an obtained IQ score, reflecting the empirically known amount of error present in each score). Pardon my psychometric bluntness...but the Florida Supreme Courts stance is simply wrong. Intelligence testing, although the best empirical technology to emerge from the field of psychology, is not as precise as some technologies from the hard sciences. You can not, and should not, establish single score cut-offs based on IQ tests. They have a quantifiable and known degree of measurement precision that must be taken into consideration when making decisons about people.

After the 2008 denial order, Dufour filed an ammended initial brief with the Florida Supreme Court (click here to view; apparently the first petition exceeded a maximum page length, thus resulting in the "ammended initial brief"). There is much to read and digest (e.g., early school performance suggestive of adaptive behavior and academic problems; issues of malingering during testing being raised; disagreements over scoring of items in subtests; etc --- I simply have not had time to read this brief in depth...I'm behind in reading all of the recent decisions I've located or people have sent me). The brief outlines three arguments, only one (and just one aspect of this argument) which I focus on in this post.

Tthe original SCOTUS Atkins v Virginia decision fall-out (produced by allowing each state to set its own definition and criteria for mental retardation) is part of the problem with Dufour and other MR death penalty cases in Florida. In the 2009 Dufour brief, it states that the Florida statute defines the subaverage intellectual component of MR as a score two standard deviations below the mean (which is IQ=70). The brief states that the relevant Florida statute does NOT specify a specific IQ cut-off score of 70. The argument is made that in a different decision (Foster v State, 2006), the court did allow for an IQ score as high as 75 (although the person was not considered MR due to the other "prongs" of the MR definition), which does reflect an acknowledgment of the SEM of IQ scores. However, the Cherry v State (2007) decision, which set a score of 70, appears to be pivitol in how the Florida court has viewed subsequent Atkins related cases, included Dufour.

Regardless of the Cherry court decision, and regardless of the particulars and nuances of the various Atkins-related decisions made in Florida, the Florida Supreme Court needs to recognize the universally accepted professional/expert fact that intelligence test scores contain a quantifiable degree of measurement error that is represented by the SEM. The SEM must be considered in the interpretation of IQ scores. The Florida Supreme Court would receive a failing grade in any introductory graduate level assessment course for failure to recognize the importance of the SEM of IQ scores....but, of course, I would not assign a failing grade until I had bracketed their performance with the appropriate SEM-based confidence band.