Full opinion text
MEMORANDUM OPINION C. LYNWOOD SMITH, JR., District Judge. Petitioner, Kenneth Glenn Thomas, is an inmate in the custody of the Alabama Department of Corrections. He was sentenced to death by the Circuit Court of Limestone County, Alabama, for the intentional murder of Mrs. Flossie McLemore during the course of a burglary. See Ala. Code § 13A-5-40(a)(4) (1975). Following exhaustion of his direct appeal rights and post-conviction remedies in the state court system, Thomas filed a petition in this court, seeking relief in the nature of habeas corpus. See 28 U.S.C. § 2254. All but one of his claims for relief were dismissed in an order and memorandum opinion entered on March 6, 2007; the sole claim that survived the motion for summary judgment filed by the respondent Commissioner of the Alabama Department of Corrections was Thomas’s contention that he is mentally retarded. If Thomas is retarded, then the Supreme Court’s decision in Atkins v. Virginia, 536 U.S. 304, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002) — holding under the Eighth Amendment that “death is not a suitable punishment for a mentally retarded criminal,” id. at 321, 122 S.Ct. 2242—will require his death sentence to be vacated. This court originally ordered that Thomas’s Atkins contention be remanded to the state court system, with instructions to reevaluate his claim in accordance with standards established by binding authorities. Upon joint motion of the parties, however, this court was persuaded to withdraw that remedy in favor of an order amending the judgment to reflect that the claim would be litigated on the merits in this court. An evidentiary hearing commenced on May 19, 2008, and concluded on May 20, 2008. Thereafter, the parties filed post-hearing briefs. Respondent’s brief concedes that the evidence presented during the May hearing establishes that “Thomas’s IQ now falls in the mild mental retardation range,” but goes on to argue that Atkins provides him no relief because he failed to prove that “he suffered from significantly sub-average general intellectual functioning or substantial deficits in adaptive behavior in the developmental period.” Respondent thus reduced his argument against a finding of mental retardation to this question: Has petitioner proven, by a preponderance of the evidence, that he met the criteria for mental retardation prior to the age of eighteen years (the so-called “developmental period”)? The remainder of this opinion addresses that question, as well as other issues. Table of Contents I. The Legal Criteria Defining Mental Retardation............................1262 II. Diagnostic Criteria Defining Mental Retardation ...........................1263 A. Assessment of Intellectual Functioning................................1263 1. Standardized assessment instruments...............................1264 a. The Wechsler Adult Intelligence Scales...........................1265 b. The Stanford-Binet Intelligence Scales...........................1267 2. Measurement errors and cut scores..................................1268 a. The effect of “standard errors of measurement” on “true” IQ scores......................................................1269 b. The stipulated SEM and its effect upon determination of petitioner’s IQ..............................................1271 3. The “Flynn Effect, ” IQ gains over time, and cut scores.................1275 4. Conclusions ......................................................1279 B. Assessment of Adaptive Behavior......................................1281 1. Standardized assessment instruments...............................1282 2. The importance of clinical judgment.................................1283 3. The assessment instrument used in this case .........................1284 III. Expert Witnesses and “Best Practices”.....................................1286 A. Petitioner’s Experts..................................................1286 1. Dr. Karen Salekin.................................................1286 2. Dr. Daniel Marson................................................1287 B. Respondent’s Expert.................................................1288 C. “Best Practices”.....................................................1289 IV. Assessments of Petitioner’s Intellectual Functioning........................1293 A. Intelligence Assessments Performed Prior to Age Eighteen..............1293 1. October I, 1968 — age nine years and seven months....................1293 2. October 12,1972 — age thirteen years and seven months................1294 3. June 28,1978 — age fourteen years and two months....................1294 4. March 18,1975 — age sixteen........................................1296 5. Findings.........................................................1296 B. The Intelligence Assessment Performed on April 11, 1977................1297 C. Intelligence Assessments Performed Near the Date of the Offense........1300 1. March 21, 1985 — age twenty-six years ...............................1300 2. January 2j, 1986 — age twenty-six years and ten months...............1301 3. Findings.........................................................1301 D. Intelligence Assessment Performed in Preparation for Hearing..........1301 V. Assessments of Petitioner’s Adaptive Behavior..............................1303 A. Petitioner’s Deficiencies Prior to Age Eighteen.........................1303 1. The home environment — birth to age 12..............................1304 2. Adolescence and foster care — ages 12 through 18......................1305 3. Schooling ........................................................1307 4. Assessments by third-party informants..............................1309 5. Findings.........................................................1311 B. Petitioner’s Deficiencies at the Time of the Offense.....................1312 C. Petitioner’s Present Adaptive Functioning Abilities.....................1321 VI.Conclusions 1322 I. THE LEGAL CRITERIA DEFINING MENTAL RETARDATION The Supreme Court’s decision in Atkins v. Virginia did not dictate a national standard for determining whether a criminal defendant is mentally retarded and, for that reason, not subject to the ultimate sanction of the law. Instead, the Court left to the states “the task of developing appropriate ways to enforce the constitutional restriction” upon the execution of mentally retarded convicts. Atkins, 536 U.S. at 317, 122 S.Ct. 2242 (citation, internal quotation marks, and footnote omitted). The Court’s reticence to propound hard and fast rules undoubtedly was grounded in the fact that the statutory definitions of mental retardation adopted by Congress and those states that then prohibited the execution of mentally retarded persons were not identical. Even so, the Court observed that all of the existing statutes generally conformed to diagnostic criteria promulgated by the American Association on Mental Retardation and the American Psychiatric Association. The American Association on Mental Retardation (AAMR) defines mental retardation as follows: “Mental retardation refers to substantial limitations in present functioning. It is characterized by significantly subaverage intellectual functioning, existing concurrently with related limitations in two or more of the following applicable adaptive skill areas: communication, self-care, home living, social skills, community use, self-direction, health and safety, functional academics, leisure, and work. Mental retardation manifests before age 18.” Mental Retardation: Definition, Classification, and Systems of Supports 5 (9th ed.1992). The American Psychiatric Association’s definition is similar: “The essential feature of Mental Retardation is significantly subaverage general intellectual functioning (Criterion A) that is accompanied by significant limitations in adaptive functioning in at least two of the following skill areas: communication,' self-care, home living, social/interpersonal skills, use of community resources, self-direction, functional academic skills, work, leisure, health, and safety (Criterion B). The onset must occur before age 18 years (Criterion C). Mental Retardation has many different etiologies and may be seen as a final common pathway of various pathological processes that affect the functioning of the central nervous system.” Diagnostic and Statistical Manual of Mental Disorders 41 (4th ed.2000). “Mild” mental retardation is typically used to describe people with an IQ level of 50-55 to approximately 70. Id., at 42-43. Atkins, 536 U.S. at 309 n. 3, 122 S.Ct. 2242 (emphasis in original). The Atkins opinion thus pointed the states in the direction of clinical definitions that have three constituent parts: that is, in order to be diagnosed as “mentally retarded,” the person under evaluation must exhibit (i) before the age of eighteen years (ii) significantly sub-average intellectual functioning, accompanied by (iii) significant limitations in adaptive functioning. Similarly, the Alabama Supreme Court subsequently held that a defendant seeking the benefit of Atkins “must have significantly subaverage intellectual functioning (an IQ of 70 or below), and significant or substantial deficits in adaptive behavior. Additionally, these problems must have manifested themselves during the developmental period (i.e., before the defendant reached age 18).” Ex parte Perkins, 851 So.2d 453, 456 (Ala.2002). The Alabama Supreme Court layered a gloss on the Perkins definition in Smith v. State, No. 1060427, 2007 WL 1519869 (Ala. May 25, 2007), holding that a defendant must exhibit significantly subaverage intellectual functioning abilities and significant deficits in adaptive behavior during three periods of his life: before the age of eighteen; on the date of the capital offense; and currently. All three factors must be met in order for a person to be classified as mentally retarded for purposes of an Atkins claim. Implicit in the definition is that the subaverage intellectual functioning and the deficits in adaptive behavior must be present at the time the crime was committed as well as having manifested themselves before age 18. This conclusion finds support in examining the facts we found relevant in Ex parte Perkins and Ex parte Smith and finds further support in the Atkins decision itself, in which the United States Supreme Court noted: “The American Association on Mental Retardation (AAMR) defines mental retardation as follows: ‘Mental retardation refers to substantial limitations in present functioning.’ ” 536 U.S. at 308 n. 3, 122 S.Ct. 2242, 153 L.Ed.2d 335 (second emphasis added). Therefore, in order for an offender to be considered mentally retarded in the Atkins context, the offender must currently exhibit subaverage intellectual functioning, currently exhibit deficits in adaptive behavior, and these problems must have manifested themselves before the age of 18. Smith, 2007 WL 1519869, at *8 (emphasis supplied). See also Holladay v. Allen, 555 F.3d 1346, 1353 (11th Cir.2009) (same). II. DIAGNOSTIC CRITERIA DEFINING MENTAL RETARDATION A. Assessment of Intellectual Functioning “The assessment of intellectual functioning is essential to making a diagnosis of mental retardation, as virtually all definitions of mental retardation make reference to significantly subaverage intellectual functioning as one of the diagnostic criteria.” Ruth Luckasson et al., Mental Retardation: Definition, Classification, and Systems of Supports 51 (Washington, D.C.: American Association on Mental Retardation 10th ed.2002) (hereafter, “AAMR, Mental Retardation ”). [intelligence is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capacity for comprehending our surroundings — catching on, making sense of things, or figuring out what to do. Thus the concept of intelligence represents an attempt to clarify, organize, and explain the fact that individuals differ in their ability to understand complex ideas, to adapt effectively to their environments, to learn from experience, to engage in various forms of reasoning, to overcome obstacles by thinking and communicating. Id. at 40 (citation omitted). 1. Standardized assessment instruments “Although far from perfect, intellectual functioning is still best represented by IQ scores when obtained from appropriate assessment instruments.” AAMR, Mental Retardation at 14. The “Wechsler Adult Intelligence Scales — Third Edition” (WAIS-III) and the “Stanford-Binet Intelligence Scales — Fifth Edition” (SB5) are the two most widely used IQ tests, id. at 59, and they were utilized by the parties’ expert witnesses to assess petitioner’s current intellectual functioning. Both are “standardized” assessment instruments, meaning that: (a) during the design phase, each was administered to a large, representative sample of the population for which the test was intended to provide reliable, normative data; (6) the reliability and validity of each test has been established over time by cumulative empirical applications and analysis; and (c) each test must be administered, scored, and interpreted by trained examiners in strict accordance with instructions issued by the test developers. a. The Weehsler Adult Intelligence Scales The first IQ assessment instrument to be named the “Weehsler Adult Intelligence Scales” (WAIS) was published in 1955 as a revision of the “Wechsler-Bellevue Intelligence Scales” developed in 1939 by Dr. David Weehsler, a clinical psychologist, during his association with Bellevue Psychiatric Hospital in New York City. The theoretical basis for the device was Dr. Wechsler’s belief that intelligence is a multifaceted construct that enables an individual to comprehend and deal effectively with the environment in which he or she lives and works. After dividing intelligence into two major types of skill sets, verbal and performance, Weehsler used the statistical technique of factor analysis to determine specific skills within those two major domains. The most recent iteration of this assessment instrument, the so-called “Weehsler Adult Intelligence Scales — Third Edition” (WAIS-III), is a 1997 revision of the “Weehsler Adult Intelligence Scales — Revised Edition” (WAIS-R) published in 1981. It is an individually administered test designed to assess the intelligence of individuals ranging in age from 16 years to 89 years. The WAIS-III was standardized on 2,450 adults from the United States. Thirteen separate .standardization groups were created by age classification. Within each group, the number of males and females was roughly equal (except for the 65 to 89 age group, which contained more females), and there was Census-based stratification for race or ethnicity (White, African American, Hispanic), education, and geographic region based on Census reports. AAMR, Mental Retardation at 61. As shown in the following table, fourteen sub-tests are equally divided ■ between seven Verbal Scale subtests and seven Performance Seale subtests. The number preceding each subtest indicates the standardized order of administration: i.e., the “Picture Completion” subtest (on the Performance Scale) is administered first; the “Vocabulary” subtest (Verbal Scale) is given next, and so on in alternating order to assist an examiner in maintaining the' test-subject’s interest. WAIS-III SUBTESTS (Grouped According to Verbal and Performance Scales) Verbal Scale Performance Scale 2. Vocabulary A series of orally and visually presented words that the examinee orally defines. 1. Picture Completion A set of color pictures of common objects and settings, each of which is missing an important part that the examinee must identify. 4. Similarities A series of orally presented pairs of words for which the examinee explains the similarity of the common objects or concepts they represent._ 3. Digit Symbol — Coding A series of numbers, each of which is paired with its own corresponding hieroglyphic-like symbol. Using a key, the examinee writes the symbol corresponding to its number. 6. Arithmetic A series of arithmetic problems that the examinee solves mentally and responds to orally._ 5. Block Design A set of modeled or printed two-dimensional geometric patterns that the examinee replicates using two-color cubes._ 8. Digit Span A series of orally presented number sequences that the examinee repeats verbatim for Digits Forward and in reverse for Digits Backward._ 7. Matrix Reasoning A series of incomplete gridded patterns that the examinee completes by pointing to or saying the number of the correct response from five possible choices._ 9. Information A series or orally presented questions that tap the examinee’s knowledge of common events, objects, places, and people 10. Picture Arrangement A set of pictures presented in a mixed-up order that the examinee rearranges into a story sequence._ 11. Comprehension A series of orally presented questions that require the examinee to understand and articulate social rules and concepts or solutions to everyday problems. 12. Symbol Search A series of paired groups, each pair consisting of a target group and a search group, The examinee indicates, by marking the appropriate box, whether either target symbol appears in the search group. 13. Letter-Number Sequencing A series of orally presented sequences of letters and numbers that the examinee simultaneously tracks and orally repeats, with numbers in ascending order and the letters in alphabetical order. 14. Object Assembly A set of puzzles of common objects, each presented in a standardized configuration, that the examinee assembles to form a meaningful whole. Source: D. Wechsler, WAIS-III Administration and Scoring Manual (San Antonio: The Psychological Corp. 1997) As the Supreme Court observed, the WAIS-III is scored by adding together the number of points earned by a test subject on these different subtests, and then using a mathematical formula to convert this raw score into a scaled score. The test measures an intelligence range from 45 to 155. The mean score of the test is 100, which means that a person receiving a score of 100 is considered to have an average level of cognitive functioning. It is estimated that between 1 and 3 percent of the population has an IQ between 70 and 75 or lower, which is typically considered the cutoff IQ score for the intellectual function prong of the mental retardation definition. Atkins, 536 U.S. at 309 n. 5, 122 S.Ct. 2242 (emphasis supplied, citations omitted). b. The Stanfordr-Binet Intelligence Scales The Stanford-Binet battery of intelligence assessment instruments is far older than the Wechsler series. The original test was devised by psychologist Alfred Binet, who was charged by the French government with developing a method of identifying intellectually-deficient school-age children for placement in special education programs. Research conducted by Binet and physician Theophilus Simon between 1905 and 1908 at a school for mentally-retarded boys led to the development of the Binet-Simon test, which employed questions of increasing difficulty to measure such attributes as attention, memory, and verbal skills. In 1916, Lewis Terman, a psychologist at Stanford University in California, released the “Stanford Revision of the Binet-Simon Scale.” That test has been revised several times since its inception, and currently is in its fifth edition — a version that is generally referred to as either the “Stanford-Binet 5” or “SB5.” The SB5 consists of a battery of tests that assess a person’s intelligence across four areas of intellectual functioning: verbal reasoning; quantitative reasoning; abstract and visual reasoning; and short-term memory. These areas are covered by fifteen subtests, including vocabulary, comprehension, verbal absurdities, pattern analysis, matrices, paper folding and cutting, copying, quantitative, number series, equation building, memory for sentences, memory for digits, memory for objects, and bead memory. All test subjects take an initial vocabulary test that, together with the subject’s age, determines the number and level of sub-tests to be administered. Total testing time is 45 to 90 minutes, depending upon the subject’s age and the number of subtests administered. Raw scores are based on the number of items answered, and are converted into a “standard score” corresponding to the test subject’s age group, similar to an IQ measure. According to the website of the SB5’s publisher, Riverside Publishing Company, “[njormative data for the SB5 were gathered from 4,800 individuals between the ages of 2 and 85 + years. The normative sample closely matches the 2000 U.S. Census. Bias reviews were conducted on all items for the following variables: gender, ethnicity, culture, religion, region, and socioeconomic status.” The SB5 has a mean IQ score of 100, and a standard deviation of 16: the “standard deviation” indicates how far a particular individual’s score falls above or below the mean score for the test subject’s age group. For example, if an eight-year-old child achieved a score of 116 on the SB5, the child’s score would be one standard deviation above the mean performance score of all eight-year-olds in the representative sample. 2. Measurement errors and cut scores A key task for the ... analyst applying a scientific method to conduct a particular analysis, is to identify as many sources of error as possible, to control or to eliminate as many as possible, and, to estimate the magnitude of remaining errors so that the conclusions drawn from the study are valid. A critical question that must be addressed is: “How much confidence can this court place in the IQ scores produced by the tests administered to petitioner?” Even though most of the intelligence tests that will be discussed later in this opinion are generally considered to be reliable assessment instruments that produce valid IQ scores, there still exists an inherent potential for “measurement error.” Measurement errors can be either random or systematic. “Random errors” are caused by any factors that randomly affect measurement of test variables. Examples include factors peculiar to the individual test-subject (e.g., fatigue, poor health), the testing situation (e.g., environmental distractions inhibiting concentration), the manner in which the test was administered (e.g., the examiner’s failure to adequately explain each segment of the test, .or to strictly follow the test developer’s instructions), the examiner’s lack of training, or a multitude of other, unpredictable variables that artificially inflate or deflate a test subject’s performance. The important attribute of random errors is that they do not have consistent effects across the entire population of persons to whom the test instrument is administered. “Systematic errors,” on the other hand, are test-specific sources of error that are caused by any factors that systematically affect IQ measurements across the entire population of test subjects. Systematic errors also can be generated by many variables, but usually they can be traced to inadequacies in the assessment instrument itself. Unlike random errors, systematic errors tend to have consistently positive or negative effects upon the performance scores generated by each individual to whom the test is administered. To use a pedestrian example, suppose “you recorded the temperature every day in your backyard. If your thermometer was incorrectly calibrated, so that it was always 4 degrees too high, the faulty thermometer would produce a systematic error (an upward bias) in your measurement.” One such systematic inaccuracy in the intelligence assessment instruments administered to petitioner over the course of his life will be discussed in Part 11(A)(3) infra, addressing the so-called “Flynn effect.” a. The effect of “standard errors of measurement” on “true” IQ scores A “true” IQ score is the hypothetical score a test subject would obtain if no measurement error influenced his or her performance during the administration of an intelligence assessment instrument. No clinician, much less this court, can state a test subject’s “true” score with absolute certainty, because error always is present in any testing situation. For well-designed test instruments, however, the random errors of individual test subjects are randomly distributed. Accordingly, the more scores that are grouped together, the more likely it is that an individual test subject’s errors will be cancelled out by the results obtained by administration of the assessment instrument to a very large sample of the population during the “normative research and survey process” leading to the development of a “standardized test.” That is the reason reputable IQ assessment instruments like the Wechsler and Stanford-Binet batteries were standardized (“normed”) on large, representative samples of the general population. During the design process of developing normative standards, measurement errors are taken into account through use of a mathematical concept that statisticians and psychologists refer to as the “Standard Error of Measurement” (SEM), The SEM is an index of the variability of test scores produced by persons forming the normative sample. The SEM makes it possible to determine the reliability of a particular intelligence assessment instrument, and the level of confidence that can be placed in the scores produced by an administration of the test to an individual test subject. Every intelligence test has a SEM, which is used to calculate a range of scores lying along a continuum (think of a yardstick), and evenly arranged on each side of the IQ score obtained during an individual administration of the test. The test subject’s “true” IQ most likely lies within that range above and below his or her actual test score. For example, Kenneth Glenn Thomas was nearly 49 years of age on the date WAIS-III and SB5 IQ assessment instruments were administered to him by the parties’ expert witnesses. The SEM for a full-scale IQ score produced by the administration of a WAIS-III test to a person between the ages of 45 to 54 years is 2.23 points. The SEM for the full-scale IQ score produced by the administration of an SB5 assessment instrument to a person between the ages of 40 and 49 years is 2.12 points. Taking those factors into account in connection with the IQ scores obtained from petitioner, this court can be confident that Thomas’s true “intellectual functioning ability” fell within a band of scores bordered on the high-end by adding two SEMs to the full-scale IQ score obtained by administration of each of the foregoing test instruments, and bordered on the low-end by subtracting two SEMs from his full-scale IQ score: in other words, and in the case of the WAIS-III, by adding 4.46 points (2 x 2.23 SEMs = 4.46) to — and also subtracting 4.46 points from — the full-scale IQ score produced by administration of that IQ assessment instrument. The same process applies to the administration of an SB5 assessment instrument, except that the number of points added to and subtracted from petitioner’s obtained full-scale IQ score would be 4.24 (2 x 2.12 SEMs = 4.24 points). The 2002 AAMR manual explains this process as follows: The assessment of intellectual functioning through the primary reliance on intelligence tests is fraught with the potential for misuse if consideration is not given to possible errors in measurement. An obtained standard IQ score must always be considered in terms of the accuracy of its measurement. Because all measurement, and particularly psychological measurement, has some potential for error, obtained scores may actually represent a range of several points. This variation around a hypothetical “true score” may be hypothesized to be due to variations in test performance, examiner’s behavior, or other undetermined factors. Variance in scores may or may not represent changes in the individual’s actual or true level of functioning. Errors of measurement as well as true changes in outcome must be considered in the interpretations of test results. This process is facilitated by considering the concept of standard error of measurement (SEM), which has been estimated to be three to five points for well-standardized measures of general intellectual functioning. This means that if an individual is retested with the same instrument, the second obtained score would be within one SEM (i.e., ± 3 to 4 IQ points) of the first estimates about two thirds of the time. Thus an IQ standard score is best seen as bounded by a range that would be approximately three to four points above and below the obtained score.... Therefore, an IQ of 70 is most accurately understood not as a precise score, but as a range of confidence with parameters of at least one SEM (i.e., scores of about 66 to 74; 66% probability); or parameters of two SEMs (i.e., scores of 62 to 78; 95% probability). This is a critical consideration that must be part of any decision concerning a diagnosis of mental retardation. AAMR, Mental Retardation at 57-59 (all emphasis added, citation omitted). b. The stipulated SEM and its effect upon determination of petitioner’s IQ The attorneys for both parties and their expert witnesses stipulated that a standard error of measurement in the neighborhood of approximately ± 5 points is proper for full-scale IQ test scores produced by the intelligence assessment instruments discussed in this opinion. The American Psychiatric Association agrees: the most recent edition of its Diagnostic and Statistical Manual of Mental Disorders notes that “there is a measurement error of approximately 5 points in assessing IQ.” APA, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision 41-42 (2000) (emphasis added) (hereafter “DSM-IV-TR”). , The parties disagree, however, as to how the “band of confidence” produced by the process of adding and subtracting that value from petitioner’s obtained, full-scale IQ scores should be interpreted. Petitioner argues, in combination with the “Flynn effect” discussed in the following section, that his IQ scores should be adjusted downward, because a “compelling showing of lifelong deficits in adaptive behavior ... yields scores that are perfectly consistent across [his] life.” Petitioner points to several cases as support for this position, but none are particularly illustrative, much less precedential. Respondent, on the other hand, argues that this court should refuse to adjust petitioner’s IQ scores downward using the standard error of measurement. Respondent also contends that he has been unable to find any authority to support a downward adjustment of petitioner’s full-scale IQ scores, “depending on ... assessment of Thomas’s adaptive functioning.” Respondent asserts that, even though petitioner “argued during the evidentiary hearing [that] the range for mildly mentally retarded is from 70-75 ... under Alabama law, the cut-off for mild mental retardation is 70 or below.” The resolution of these conflicting positions lies in a careful examination of the diagnostic criteria contained in the authoritative treatises published by the AAMR and APA, and referenced in decisions of the Supreme Courts of the United States and the State of Alabama. It is clear that neither of the professional organizations dedicated to the diagnosis and treatment of mental deficiencies advocates a fixed, finite IQ “cut score” as an impregnable barrier, separating persons who are mentally retarded from those who are not. The AAMR explicitly states that “a fixed cutoff for diagnosing an individual as having mental retardation was not intended, and cannot be justified psycho-metrically.” Mental Retardation at 58 (emphasis supplied). That manual also states — not just once, but at least eight times — that the clinical standard for “significantly subaverage” intellectual functioning “is approximately two standard deviations below the mean, considering the standard, error of measurement for the specific assessment instruments used and the instruments’ strengths and limitations.” Id. at 13 (emphasis supplied); see also id. at 14, Table 1.2 (“Intelligence”) (same); 17 (same); 23, Table 2.1 (“IQ Cutoff’) (same); 27 (same); 37 (same); 58 (same); 198 (same). In effect, this expands the operational definition of mental retardation to 75, and that score of 75 may still contain measurement error. Any trained examiner is aware that all tests contain measurement error; many present scores as confidence bands rather than finite scores. Incorporating measurement error in the definition of mental retardation serves to remind test administrators (who should understand the concept) that an achieved Wechsler IQ score of 65 means that one can be about 95% confident that the true score is somewhere between 59 and 71. Id. at 59 (emphasis supplied). In like manner, the American Psychiatric Association recognizes that measurement error must be taken into account when interpreting a full-scale IQ score obtained by assessment with any of the standardized, individually administered, intelligence assessment instruments discussed in this opinion. Significantly subaverage intellectual functioning is defined as an IQ of about 70 or below (approximately two standard deviations below the mean). It should be noted that there is a measurement error of approximately 5 points in assessing IQ, although this may vary from instrument to instrument (e.g., a Wechsler IQ of 70 is considered to represent a range of 65-75). Thus, it is possible to diagnose Mental Retardation in individuals with IQs between 70 and 75 who exhibit significant deficits in adaptive behavior. Conversely, Mental Retardation would not be diagnosed in an individual with an IQ lower than 70 if there are no significant deficits or impairments in adaptive functioning.... APA, DSM-IV-TR at 41-42. The Fourth Circuit endorsed these generally-accepted clinical standards when instructing a district court to consider whether a state statute defining mental retardation permitted measurement error to be taken into account when determining whether a capital murder habeas petitioner’s raw IQ score of 72 was “ ‘two standard deviations below the mean’ as set forth under that statute.” Walker v. True, 399 F.3d 315, 323 (4th Cir.2005). See also In re: Bowling, 422 F.3d 434, 442 (6th Cir.2005) (observing that “there appears to be considerable evidence that irrebuttable IQ ceilings are inconsistent with current generally-accepted clinical definitions of mental retardation and that any IQ thresholds that are used should take into account factors, such as a test’s margin of error, that impact the accuracy of a particular test score”) (Moore, J., concurring in part and dissenting in part) (footnote omitted). In addition, many state courts have recognized that standard errors of measurement must be taken into account when interpreting IQ scores. See, e.g., State v. Burke, 2005 Ohio 7020, 2005 WL 3557641, at *13 (10th Dist. Dec. 30, 2005) (“In accord with the AAMR’s standard, measurement error must be considered in determining an individual’s IQ score.”); In re Hawthorne, 35 Cal.4th 40, 24 Cal.Rptr.3d 189, 105 P.3d 552, 557-58 (2005) (observing that IQ test scores are not precise due to measurement error, and holding that mental retardation should not be determined according to a fixed IQ cut score, but upon an assessment of the defendant’s overall capacity based on a consideration of all relevant evidence); State v. Williams, 831 So.2d 835, 853 & n. 26 (La.2002) (observing that any IQ test must account for standard margins of error). The California Supreme Court explained the point this way: With respect to the intellectual prong of [California’s mental retardation statute], respondent Attorney General urges the court to adopt an IQ of 70 as the upper limit for making a prima facie showing. We decline to do so for several reasons: First, unlike some states, the California Legislature has chosen not to include a numerical IQ score as part of the definition of mentally retarded. ... Moreover, statutes referencing a numerical IQ generally provide that a defendant is presumptively mentally retarded at or below that level, rather than — as respondent impliedly argues— that a defendant is presumptively not mentally retarded above it. Second, a fixed cutoff is inconsistent with established clinical definitions and fails to recognize that significantly subaverage intellectual functioning may be established by means other than IQ testing. Experts also agree that an IQ score below 70 may be anomalous as to an individual’s intellectual functioning and not indicative of mental impairment. Finally, IQ test scores are insufficiently precise to utilize a fixed cutoff in this context. In re Hawthorne, 24 Cal.Rptr.3d 189, 105 P.3d at 557 (emphasis added, citations and footnote omitted). See also State v. Lott, 97 Ohio St.3d 303, 779 N.E.2d 1011, 1014 (2002) (“While IQ tests are one of the many factors that need to be considered, they alone are not sufficient to make a final determination on this issue. We hold that there is a rebuttable presumption that a defendant is not mentally retarded if his or her IQ is above 70.”). One academician who surveyed state statutory developments following the Atkins decision has observed that many states have incorporated a specific IQ cutoff score in their definitions of mental retardation, most often using an IQ of seventy as the cutoff for this component of the mental retardation definition. However, most of these definitions do not acknowledge that each assessment instrument has a standard measurement error, usually between three and five points, and that the standard measurement error is not the same for all instruments. Recognizing the impact of the standard measurement error, in the previous AAMR definitions and the current APA definition, the IQ cutoff for mental retardation has been quantified between seventy and seventy-five, as noted by the Court in Atkins. To avoid mistaken reliance on and potential misuse of a particular IQ score, especially if it does not include consideration of standard measurement error, the AAMR stated its current IQ cutoff in terms of being at least two standard deviations below the mean of the specific instruments used, considering their particular standard measurement error, strengths, and limitations. The current APA definitional material also refers to the IQ cutoff as being approximately two standard deviations below the mean, with reference to measurement error of approximately five points. Thus, any state’s use of a fixed IQ cutoff score, without reference to standard measurement error and other factors concerning the specific instrument used, risks an inaccurate assessment of the intellectual functioning component of the mental retardation definition. Peggy M. Tobolowsky, Atkins Aftermath: Identifying Mentally Retarded Offenders and Excluding Them From Execution, 30 J. Legis. 77, 95-96 (2003) (emphasis added, footnote omitted); see also id. at 139 (“[SJtates that use a rigid IQ cutoff score of seventy for the intellectual functioning component may be excluding some individuals otherwise falling within the accepted clinical definition.”). 3. The “Flynn effect,” IQ gains over time, and cut scores The discussion in the preceding sections bears upon the following truism: “An IQ score is only as valid as the test the person takes, and the test is only as valid as the standardization sample on which it is normed.” James R. Flynn, What is Intelligence? Ill (Cambridge Univ. Press 2007) (emphasis supplied). The “Flynn effect” is the name given in recognition of the central role played by Professor James R. Flynn in discovering and, in a series of fifteen or more publications between 1984 and today, documenting the fact that IQ scores have been increasing from one generation to the next in all fourteen nations for which IQ data is available. Flynn recently explained the phenomenon this way: For the Wechsler (WISC and WAIS) and the Stanford-Binet IQ tests, the best rule of thumb is that Full Scale IQ gains have been proceeding at a rate of 0.30 points per year ever since 1947. This rate is based on comparisons of all of the Wechsler and Stanford-Binet tests used in recent years (see Box 11). It means that for every year that passes between when an IQ test was normed, that is, when its standardization sample was tested, obsolescence has inflated their IQs by 0.30 points. For example, if you took the WISC (normed in 1947-1948) in 1977-1978, you would get an unearned bonus of 9 IQ points (30 years x 0.30). Even though you might be dea,d average, you would be scored at 109 thanks to obsolete norms thirty years out of date.. After all, IQ gains over time mean that as we go back into the past, representative samples of Americans perform worse and worse. In this case, you are not being compared to your peers, the 14-year-olds of the late 1970s, but to a much lower-scoring group, the 14-year-olds of the late 1940s. Your score of 109 against the old norms makes you appear above average, but you are actually no better than average and deserve an IQ of 100. Flynn, supra at 112. In other words, as an intelligence test ages — or moves farther from the date on which it was standardized (“normed”) — the mean score of the population as a whole on that assessment instrument increases, thereby artificially inflating the IQ scores of individual test subjects. Stated somewhat differently, IQ scores have been increasing over time for reasons that are totally unrelated to the actual, “true” intelligence of test subjects. Even though the parties’ attorneys and expert witnesses uniformly agreed that the Flynn effect is an empirically proven statistical fact, they disagreed on the extent to which an individual test subject’s IQ score should be adjusted to take that phenomenon into account. For example, petitioner’s psychological expert, Dr. Karen Salekin, testified that she always applies Dr. Flynn’s recommendation to reduce a full-scale IQ score by 0.30 points for each year elapsed beyond the date on which the test instrument was standardized (“normed”). On the other hand, respondent’s expert, Dr. Harry McClaren, testified that the Flynn effect was something he would “take into consideration. But to slavishly say that this is definitely going to be right or a better estimate of this person’s true IQ, I don’t think that ... the profession [has come to a national consensus on] that point.” Dr. McClaren’s opinion about the lack of professional acceptance of the validity of the Flynn effect is refuted by the AAMR’s 2002 manual, which explicitly states that, “as others have shown (e.g., Flynn, 1987), it is critically important to use standardized tests with the most updated norms.” Mental Retardation at 56 (emphasis supplied); see also id. at 59 (noting “variances in scores between successive revisions of intelligence measures”). A manual published in 2007 under the present name of the organization formerly known as the “American Association on Mental Retardation” (AAMR) — i.e., the “American Association on Intellectual and Developmental Disabilities” (AAIDD) — is even more explicit. It recommends that clinicians take into account both the Flynn effect and the standard error of measurement when performing retrospective diagnoses in less than optimal circumstances (e.g., the legal and physical constraints of a maximum-security prison environment). Specifically, the User’s Guide: Mental Retardation Definition, Classification and Systems of Supports — 10th Edition (“User’s Guide”) directs diagnosticians to: Recognize the “Flynn Effect.” In his study of IQ tests across populations, Flynn (1984, 1987, 1989) discovered that IQ scores have been increasing from one generation to the next in all 14 nations for which IQ data existed. This .increase in IQ scores over time has been dubbed the Flynn Effect. Flynn reported a greater increase in the Wechsler Performance IQ, which is more heavily loaded on fluid abilities than on the Wechsler Verbal IQs. On average, the Full-Scale IQ increases by approximately 0.33 points for every year elapsed since the test was normed (Flynn, 1999). The main recommendation resulting from this work is that all intellectual assessments must use a reliablé and appropriate individually administered intelligence test. In cases of tests with multiple versions, the most most recent version with the most current norms should be used at all times. In cases where a test with aging norms is used, a correction for the age of the norms is warranted. For example, if the Wechsler Adult Intelligence Scale (WAIS-III, 1997) was used to assess an individual’s IQ in July 2005, the population mean on the WAIS-III was set at 100 when it was originally normed.in 1995 (published in 1997). However, based on Flynn’s data, the population mean on the Full-Scale IQ raises roughly 0.33 points per year; thus, the population mean on the WAIS-III Full-Scale IQ corrected for the Flynn Effect would be 103 in 2005 (9 years X 0.33 = 2.9). ■ Hence, using the “at least two standard deviations below the mean” (Luckasson et al., 2002), the approximate Full-Scale IQ cutoff would be approximately 73 (plus or minus the standard error of measurement). Thus the clinician needs to use the most current version of an individually administered test of intelligence and take into consideration the Flynn Effect as well as the standard error of measurement when estimating an individual’s true IQ score. AAIDD, User’s Guide at 20-21 (emphasis supplied). Respondent retorts that, with the exception of the AAMR/AAIDD, no other national organization or federal agency has officially endorsed the Flynn effect. That may be so, but it does not justify ignoring th'e phenomenon in the face of its unchallenged existence. The steady rise in IQ scores from year to year is a statistically proven fact. Dr. McClaren admitted that he had no knowledge of any-study arguing that IQ scores in the general population of the United States have not increased at the average rate of 0.30 points each year after a test instrument was standardized. Further, and even though “there is not a consensus among professionals as to why these gains -are occurring or what these gains actually mean (e.g., are we really getting smarter?), all are in agreement that the gains occur.” “Since 1998, when the American Psychological Association issued The Rising Curve: Long-Term Gains in IQ and Related Measures (Neisser, 1998), no s'cholar published in a first-line journal has ignored the relevant data on IQ gains over time.” It also is undisputed that Professor Flynn’s recommendation — i.e., “deduct 0.30 IQ points per year (3 points per decade) to cover the period between the year the test was normed and the year in which the subject took the test” — is a generally accepted adjustment. Outside the psychological and psychiatric communities, at least one Circuit Court of Appeals has held that the Flynn effect is relevant to the interpretation of IQ scores in capital cases. In Walker v. True, 399 F.3d 315 (4th Cir.2005), the habeas petitioner, who was raising a mental retardation claim under Atkins, argued that the district court committed reversible error by failing to -adjust his IQ scores to take into account both the Flynn effect and the test instrument’s standard error of measurement. The Fourth Circuit agreed, vacated the district court’s opinion dismissing the habeas petition, and remanded the case for consideration of “relevant evidence”: .The district court, without much explanation, did not consider the Flynn Effect or the measurement error, stating that such evidence “does not provide a legal basis for ignoring Walker’s WAIS test scores.” J.A. 266. But, as the Virginia statute makes clear, the relevant question is whether Walker scored two standard deviations below the mean, a question which is directly addressed by Walker’s expert opinion as to the Flynn Effect. Thus, not only did the district court resolve a factual dispute against Walker — contrary to the claims in his petition and where the facts remained materially disputed — it also refused to consider relevant evidence, namely the Flynn Effect evidence. Therefore, on remand the district court should consider the persuasiveness of Walker’s Flynn Effect evidence. And if the district court does credit that evidence, it should then consider whether the Virginia statute permits consideration of measurement error in order to determine whether Walker’s purported score of 72 is “two standard deviations below the mean” as set forth under that statute. Walker, 399 F.3d at 322-23 (emphasis added). See also Walton v. Johnson, 440 F.3d 160 (4th Cir.2006) (en banc); In re Hicks, 375 F.3d 1237, 1242 (11th Cir.2004) (Birch, J., dissenting from the denial of a stay of execution because the IQ scores generated by a 1985 administration of the Wechsler Adult Intelligence Scale to the habeas petitioner were “likely to have been artificially inflated by what has been labeled ‘The Flynn effect’ ”). State appellate courts have reached the same conclusion. The Ohio Court of Appeals, for example, has held that “a trial court must consider evidence presented on the Flynn effect, but, consistent with its prerogative to determine the persuasiveness of the evidence, the trial court is not bound to, but may, conclude the Flynn effect is a factor in a defendant’s IQ score.” State v. Burke, 2005 Ohio 7020, 2005 WL 3557641, at *13 (10th Dist. Dec. 30, 2005). 4. Conclusions Adherence to scientific principles is important for concrete reasons: they enable the reliable inference of knowledge from, uncertain information .... George Bernard Shaw famously said that England and America are two countries separated by a common language. In an analogous manner, the discussion in the previous Parts of this opinion demonstrates that attorneys, judges, psychologists, and psychiatrists share a common language, but our specialized vocabularies often separate one professional group from the other. Legal and mental health professionals differ in the acceptance of methods of analysis. In science in general and in statistics in particular, a method is evaluated according to well-established criteria. In the courts, theoretical justifications of a statistical method may be treated as if they are less important than the general acceptance of the method by statisticians and other scientists. The Evolving Role of Statistical Assessment as Evidence in the Courts 145 (Stephen E. Fienberg ed., 1989). For such reasons, when assessing the role that full-scale IQ scores play in determining mental retardation, courts must be careful to distinguish between the language {rules) of law and the language {diagnostic criteria) of psychologists and psychiatrists. Stated differently, it is important for courts to guard against resolving the factual question of mental retardation as a matter of law. In finding the facts of a particular case, courts and juries untrained in science are sometimes called upon to resolve contested scientific issues, but such factual findings do not establish generally applicable rules of law.... [A]n appellate court cannot convert a disputed factual assertion into a rule of law simply by labeling it a “legal standard” ... People v. Superior Court, 40 Cal.4th 999, 56 Cal.Rptr.3d 851, 155 P.3d 259, at 267 (2007). Contrary to respondent’s argument that there is no diagnostic or legal basis by which this court may properly adjust petitioner’s raw IQ scores in answering the question of whether he suffers from significantly subaverage intellectual functioning, the adjustments to raw IQ scores mandated by the “standard error of measurement” and the “Flynn effect” are well-supported by the accumulation of empirical data over many years. Both methodologies have been subjected to rigorous peer review and, while some psychologists may still ponder the precise cause(s) of the Flynn effect, no reputable member of the relevant professional communities denies that IQ scores have been increasing at the average rate of 0.30 points a year since the 1930s. General acceptance of both methodologies has come “as results and theories continue to hold, even under the scrutiny of peers, in an environment that encourages healthy skepticism.” Therefore, this court has taken both factors into account when evaluating the extent of petitioner’s intellectual functioning abilities. Stated differently, even though the legal cut-off score for a finding of “significantly subaverage intellectual functioning” is stated in opinions of the Alabama Supreme Court as “an IQ of 70 or below,” a court should not look at a raw IQ score as a precise measurement of intellectual functioning. A court must also consider the Flynn effect and the standard error of measurement in determining whether a petitioner’s IQ score falls within a range containing scores that are less than 70. B. Assessment of Adaptive Behavior Under Atkins and its progeny, a finding of “mental retardation” can only be made upon the basis of a conclusion that the petitioner’s significantly sub-average intellectual functioning is accompanied by substantial limitations in adaptive behavior. See Atkins, 536 U.S. at 309 n. 3, 122 S.Ct. 2242; Ex parte Perkins, 851 So.2d at 456. The term “adaptive behavior” is defined by the AAMR as “the collection of conceptual, social, and practical skills that have been learned by people in order to function in their everyday lives.” Mental Retardation at 41 (emphasis supplied). “Conceptual skills” include language, reading and writing, money concepts, and self-direction; in other words, a determination whether the test subject possesses a basic level of literacy and numeracy (so he can shop and make change), and remembers to do things on time. “Social skills” include interpersonal relationships, personal responsibility, self-esteem, gullibility and naiveté, following rules, obeying laws, and avoiding victimization. “Practical skills” include daily activities such as eating, personal hygiene, dressing, meal preparation, housekeeping, transportation, taking medication, money management, and telephone use, as well as occupational skills and maintaining a safe environment. Dr. Salekin explained the concept as follows: [Ajdaptive behaviors are everyday skills, such as walking, talking, grooming, cooking, cleaning, and participating in school or work. These abilities are learned over time in the context of one’s home and community, and they represent skills that are necessary to function within that context. It is important to emphasize that adaptive behaviors develop over the course of time and with experience, and thus individuals are evaluated against their same age peers. To measure adaptive skills, adaptive behavior scales have been developed and normed on individuals with and without intellectual disabilities. These scales require that an informant, typically a parent, teacher, or other individual who is very familiar with the individual’s daily level of functioning, rate the person of interest on a variety of skills. For instance, the informant may rate the extent to which the individual follows directions or balances a checkbook along a continuum ranging from “Never Does or Can’t Do” to “Always Does or Can Do Without Assistance.” There exist many misconceptions about how to conceptualize adaptive behavior for diagnostic purposes. First, some believe that adaptive behavior is measured by estimates of abilities or potential, but it is the individual’s actual performance that is important. Second, adaptive behavior is typical behavior that reflects an individual’s ability to function on a day-to-day basis. It is not measured by isolated successes or failures. Third, adaptive behavior is performance in one’s community, not in restricted settings, such as prison or therapeutic treatment programs. Fourth, the definition of mental retardation does not require that a cause of impairment be identified; diagnosis is determined by the presence of significant deficits in intellectual functioning and concomitant deficits in adaptive behavior that are evident before the age of 18 years. Doc. no. 110-2 (Report of Karen Salekin, Ph.D.), at 15,16 (emphasis supplied). 1. Standardized assessment instruments In the Tenth (2002) edition of the AAMR’s manual, diagnosticians are, for the first time, instructed that significant limitations in adaptive behavior should be established through the use of standardized measures normed on the general population, including people with disabilities and people without disabilities. On these standardized measures, significant limitations in adaptive behavior are operationally defined as performance that is at least two standard deviations below the mean of either (a) one of the following three types of adaptive behavior: conceptual, social, or practical or (b) an overall score on a standardized measure of conceptual, social, and practical skills. Mental Retardation at 76. There are many scales for measuring adaptive behavior, but it is important to recognize that “no one instrument can measure all of the relevant domains of adaptive behavior.” Id. at 84 (citation and internal quotation marks omitted). See also, e.g., doc. no. 110-2 (Report of Karen Salekin, Ph.D.) at 16 (“There exist numerous scales of adaptive behavior that can be used for the purposes of diagnosis, classification, and planning for supports; [but] no single measure is best for all three.”) (alteration and emphasis added). For that reason, review of school records, medical histories, public records, employment records, and personal observations of the test subject in face-to-face interviews, as well as consideration of other, collateral sources of information — such as interviews of “third-party informants” (e.g., family members, friends, school teachers, and other persons who know the subject well) — are all used by clinicians to complement, but not replace, standardized assessment measures. See AAMR, Mental Retardation at 84. Those who use most current adaptive behavior scales to gather information about typical behavior rely primarily on the recording of information obtained from a third person who is familiar with the individual being assessed. Thus assessment typically takes the form of an interview process, with the respondent being a parent, teacher, or direct service provider [e.g., prison guards] rather than from direct observation of adaptive behavior or from self-report of typical behavior. It is critical that the interviewer and informant or rater fully understand the meaning of each question and response category in order to provide valid and reliable information to the clinician. It is also essential that people interviewed about someone’s adaptive behavior be well-acquainted with the typical behavior of the person over an extended period of time, preferably in multiple settings. In some cases it may be necessary to obtain information from more than one informant. The consequences of scores to the rater, informant, or individual being rated should also be taken into consideration, as well as the positive or negative nature of the relationship between the rater or informant and the person being assessed. Observations made outside the context of community environments typical of the individual’s age peers and culture warrant severely reduced weight. Id. at 85 (citations omitted). 2. The importance of clinical judgment As a result of the fact that no currently available assessment instrument can measure all domains of adaptive behavior, the importance of “clinical judgment” increases. See id. at 84. Clinical judgment is particularly important in cases such as this one, where the passage of many years limits the use and valid interpr