Cognitive performance trends among European older adults: exploring variations across cohorts, gender, and educational levels (2007-2017)

BACKGROUND

This study explores recent cohort trends in cognitive performance among older Europeans from 2007 to 2017, addressing three key questions: (1) Did cognitive performance improve universally and across the performance distribution during this period? (2) Did these improvements occur across educational levels and for both men and women? (3) Can established risk factors explain these performance gains?

METHODS

Using data from the Survey of Health, Ageing and Retirement in Europe (SHARE) across 12 European countries, we assessed immediate recall, delayed recall, and verbal fluency in individuals aged 60 to 94 in both 2007 and 2017 (n = 32 773). Differences between the two time points were estimated with linear mixed effects regression models and quantile regression.

RESULTS

Cognitive performance improved in all age groups, across educational levels, and for both men and women between 2007 and 2017. Notably, improvements were more pronounced at the upper end of the performance distribution for delayed recall and verbal fluency. Education explained approximately 20% of the observed improvements. Risk factors did not explain the observed improvements.

CONCLUSIONS

European cohorts of both younger-old and older adults continue to exhibit improvements in cognitive performance. Variation in the size of the cohort improvements across the performance distributions in delayed recall and in verbal fluency may contribute to growing inequalities in cognitive outcomes. Future research should further investigate the potential heterogeneity in cognitive performance gains.

TRIAL REGISTRATION

Not applicable.

Neurodiversity and Intelligence: Evaluating the Flynn Effect in Children with Autism Spectrum Disorder

The Flynn Effect (FE) among child and adolescent populations indicates that intelligence scores improve by about three points per decade. Using nine years of data from the National Database for Autism Research, this study examined whether general intelligence changed significantly for nine cohorts with autism spectrum disorder (ASD; N = 671). Analyses demonstrated a downward trend such that Cohen's d from 1998 to 2006 was - 0.27. The mean IQ is 92.74 for years 1-3, 91.54 for years 4-6, and 87.34 for years 7-9, indicating a reverse FE of 5.4 points per decade. A linear regression revealed a significant negative FE comparable to the positive effect of age on IQ among those with ASD. Implications for research, practice, and law are discussed.

Flaws in Flynn Effect Research With the Wechsler Scales

Many Flynn effect (FE) studies compare scores across different editions of Wechsler’s IQ tests. When construct changes are introduced by the test developers in the new edition, however, the presumed generational effects are difficult to untangle from changes due to test content. To remove this confound, we use the same edition of Wechsler Intelligence Scale for Children–Fourth Edition (WISC-IV) across an 11-year period. Whereas previous research has reported the FE to be less than half the theoretical rate when comparing WISC-IV with Wechsler Intelligence Scale for Children–Fifth Edition (WISC-V), we find the rate of gain to be nearly identical to Flynn’s prediction when comparing only WISC-IV scores over the same time period. The FE is shown to vary significantly across the domains of cognitive ability, and thus changes to the construct coverage of the WISC-V Full Scale IQ (FSIQ) composite between editions significantly affect FE findings. Implications for future FE research are discussed.

Using Score Equating and Measurement Invariance to Examine the Flynn Effect in the Wechsler Adult Intelligence Scale

The Flynn effect (FE; i.e., increase in mean IQ scores over time) is commonly viewed as reflecting population shifts in intelligence, despite the fact that most FE studies have not investigated the assumption of score comparability. Consequently, the extent to which these mean differences in IQ scores reflect population shifts in cognitive abilities versus changes in the instruments used to measure these abilities is unclear. In this study, we used modern psychometric tools to examine the FE. First, we equated raw scores for each common subtest to be on the same scale across instruments. This enabled the combination of scores from all three instruments into one of 13 age groups before converting raw scores into Z scores. Second, using age-based standardized scores for standardization samples, we examined measurement invariance across the second (revised), third, and fourth editions of the Wechsler Adult Intelligence Scale. Results indicate that while scores were equivalent across the third and fourth editions, they were not equivalent across the second and third editions. Results suggest that there is some evidence for an increase in intelligence, but also call into question many published FE findings as presuming the instruments' scores are invariant when this assumption is not warranted.

Cognitive Tests in Early Childhood

Cognitive assessment of young children contributes to high-stakes decisions because results are often used to determine eligibility for early intervention and special education. Previous reviews of cognitive measures for young children highlighted concerns regarding adequacy of standardization samples, steep item gradients, and insufficient floors for young children functioning at lower levels. The present report extends previous reviews by including measures recently published or revised, nonverbal cognitive assessment tools, and issues specific to assessing bilingual or non-English-speaking children. Sixteen tests were reviewed, including all available measures of cognitive functioning for 2- to 4-year-old children normed in the United States. Test characteristics evaluated included (a) representativeness and recency of standardization data, (b) item bias analysis, (c) psychometric characteristics, and (d) appropriateness for assessing young children with developmental delays and non-English-speaking children. Implications are discussed for clinicians, researchers, and test developers.

IQ Scores Should Not Be Adjusted for the Flynn Effect in Capital Punishment Cases

Atkins v. Virginia (2002) dramatically raised the stakes for mental retardation in capital punishment cases, but neither defined this condition nor imposed uniform standards for its assessment. The basic premise that mean IQ scores shift over time enjoys wide recognition, but its application— including the appropriateness of characterizing it in terms of an allegedly predictable “Flynn effect”— is frequently debated in the course of death penalty litigation. The scientifically and ethically sound approach to this issue is to report IQ scores as obtained and be prepared to address those factors that might affect their reliability. Altering the IQ scores themselves is insufficiently supported by professional literature, legal authority, or prevailing standards of practice.

Failure to Apply the Flynn Correction in Death Penalty Litigation: Standard Practice of Today Maybe, but Certainly Malpractice of Tomorrow

The Flynn Effect is a well documented phenomenon demonstrating score increases on IQ measures over time that average about 0.3 points per year. Normative adjustments to scores derived from IQ measures normed more than a year or so prior to the time of testing an individual have become controversial in several settings but especially so in matters of death penalty litigation. Here we make the argument that if the Flynn Effect is real, then a Flynn Correction should be applied to obtained IQs in order to obtain the most accurate estimate of IQ possible. To fail to provide the most accurate estimate possible in matters that are truly life and death decisions seems wholly indefensible.

Problems With IQ Gains: The Huge Vocabulary Gap

Despite Kaufman, Raven’s Progressive Matrices and the Wechsler subtest Similarities are tests whose gains call for special explanation. The spread of “scientific spectacles” is the key, but its explanatory potential has been exhausted. Three trends force us to look elsewhere: (a) gains on Wechsler subtests such as Picture Arrangement, (b) gains in developed nations persisting into the 21st century, (c) the growing gap between the active vocabularies of parents and their children.

IQ Scores Should Be Corrected for the Flynn Effect in High-Stakes Decisions

IQ test scores should be corrected for high stakes decisions that employ these assessments, including capital offense cases. If scores are not corrected, then diagnostic standards must change with each generation. Arguments against corrections, based on standards of practice, information present and absent in test manuals, and related issues, ignore expert consensus about the assessment of intellectual disabilities and the acceptance of the Flynn effect in the field. Most psychometric concerns about correction are based on validity studies with small subgroups and do not reflect sufficient effort to estimate the precision of the Flynn estimate. We computed a confidence interval for the Wechsler PIQ across four validity studies that shows a SEM of about 1 around a mean of about 3 points per decade. A meta-analytic weighted mean of the 14 studies in Flynn (2009) is 2.80 (2.50, 3.09), close to Flynn’s (2009) unweighted average (2.99). More psychometric research would be helpful, but this level of precision supports the Flynn adjustment of 3 points per decade.

The Flynn Effect and Its Critics: Rusty Linchpins and “Lookin’ for g and Gf in Some of the Wrong Places”

The consensus of most intelligence scholars is that the Flynn effect (FE) is real, IQ test batteries are now routinely restandardized on a regular basis. A cornerstone in Flynn’s explanation of the FE is his analysis of select Wechsler subtest scores across time. The featured articles by Kaufman and Zhou, Zhu, and Weiss question whether Flynn’s arguments are grounded in the unproven assumption that similarly named Wechsler scores measure the same constructs across editions. Kaufman raises the issue by means of a detailed task analysis of changes in test administration and scoring directions for similarly named tests across different Wechsler editions. The author applauds Zhou et al. for bringing methodological rigor to the comparison of similarly named Wechsler Performance composite scores across time. Unfortunately, both Kaufman and Zhou et al. inadvertently perpetuate some of Flynn’s incorrect interpretations of select Wechsler measures (Similarities and Performance tests) as measures of the novel abstract problem solving that characterizes fluid intelligence (Gf). The author presents empirical Wechsler subtest g-loadings based on seven Wechsler joint- or cross-battery factor analyses (with other cognitive batteries). The results suggest that the extant Wechsler FE data and its system of interpretations, hypotheses, and resultant theory are held together by multiple anchors, a number that, in the words of Kaufman, are “seriously coated in rust.” The author briefly discusses the theory, tools, and technologies that currently exist to place a more reasonable degree of order in the house built by Flynn.

Considerations on the Flynn Effect

Flynn has proposed a grand integrative theory, which he calls “scientific spectacles,” to explain the phenomenon of rising IQ scores across multiple decades known as the Flynn effect (FE). In his theory, he purports that modern society has placed increasing value and emphasis on the application and education of scientific principles—which include abstract fluid reasoning—and this has mirrored larger score increases in the abstract fluid reasoning tasks included as part of most intelligence tests. He highlights huge gains in the Wechsler Similarities subtest over time as one linchpin supporting his theory that fluid reasoning is rising faster than other aspects of intelligence, but further points to large increases in Wechsler performance tasks (i.e., perceptual organization and reasoning) relative to crystallized tasks, and to the many studies of Raven’s Progressive Matrices, which is widely held to be a fairly pure measure of fluid reasoning. As with any new theory of this magnitude, Flynn’s grand proposal has invited numerous criticisms that must now be addressed—and this is as it should be in the spirit of scientific progress. Each of these issues is multilayered and has many legitimate perspectives. The present article attempts to carefully consider each issue and perspective in sequence.

“In What Way Are Apples and Oranges Alike?” A Critique of Flynn’s Interpretation of the Flynn Effect

Flynn wrote a book devoted to the Flynn effect, featuring his theoretical explanation of why the intelligence of worldwide populations has apparently increased from generation to generation. The essence of his theorizing is that because of the societal impact of scientific technology, people of today are much more guided by abstract, rather than concrete, approaches to problem solving. He bases his theory in large part on gains on specific tasks, most notably Raven’s matrices, Wechsler Intelligence Scale for Children (WISC) Performance subtests, and, most important, on WISC Similarities. The gains on these separate tasks over more than half a century (1947-2002) are striking. However, Flynn failed to take into account the sweeping changes in test content, administration procedures, and scoring guidelines when the 1974 WISC-R was developed from the 1949 WISC. These substantial changes challenge the meaningfulness of comparing children’s performance in 1947 with their performance in 2002 on Similarities and other WISC subtests— and therefore challenge Flynn’s explanation of the effect that bears his name.

Looking Through Flynn’s Rose-Colored Scientific Spectacles

In the first article of this special issue of the Journal of Psychoeducational Assessment, I critiqued Flynn’s theoretical explanation of the Flynn effect because he depended too heavily on an apparently huge generational gain on the WISC Similarities subtest; I claimed he was comparing apples with oranges because that subtest changed too much when the WISC was first revised. Four sets of esteemed researchers were invited to respond to my article and also to an article by Zhou, Zhu, and Weiss—Flynn, Sternberg, McGrew, and Ceci and Kanaya. Flynn disagrees strongly with my critique of his theory, a theory that posits striking generational shifts from concrete (utilitarian) thinking to a kind of fluid reasoning that Flynn nicknames “scientific spectacles.” In this final article of the special issue, I respond to Flynn’s claims, and also to the points made by the other invited respondents. In addition, I respond to the range of opinions expressed by the scholars who were invited to write an essay on whether or not IQs should be adjusted for the Flynn effect in capital punishment cases (Fletcher et al., Hagan et al., and Reynolds et al.). Ultimately, I disagree with Flynn’s explanation of the Flynn effect, but I agree with his position that IQs should be adjusted for the effect in death penalty cases.

“Apples and Oranges Are Both Round”: Furthering the Discussion on the Flynn Effect

While the magnitude of the Flynn effect is well established (approximately 3 points a decade on the Wechsler scales), the causes behind it are still unknown and hotly debated. Kaufman argues that, because of the administrative and scoring changes that occurred with the introduction of the Wechsler Intelligence Scale for Children—Revised, Flynn’s interpretation of the effect is not appropriate. Although agreeing that these changes account for some aspects of rising IQ, this study questions the impact of these administrative/scoring changes to account for most of the impact, given the heavy documentation of the Flynn effect on multiple IQ tests and norms over time and around the world. The authors also add to the discussion led by Zhou, Zhu, and Weiss by stressing the importance of examining the role of individual differences within the Flynn effect to understand fully the exact nuances and cause(s) of it.

The Flynn Effect: So What?

The Flynn effect probably has multiple causes, and the target essays in this issue have expanded the number of possible causes behind it. This essay deals primarily with a different question: How important is IQ in the current world and should it perhaps be understood also in conjunction with a consideration of some kind of ethical intelligence?