This study investigated the systematic rise in cognitive ability scores over generations, known as the Flynn Effect, across middle childhood and early adolescence (7–15 years; 291 monozygotic pairs, 298 dizygotic pairs; 89% White).
Leveraging the unique structure of the Louisville Twin Study (longitudinal data collected continuously from 1957–42199925ya using the Wechsler Intelligence Scale for Children [WISC], WISC-R, and WISC-III editions), multilevel analyses revealed between-subjects Flynn Effects—as both decrease in mean scores upon test re-standardization and increase in mean scores across cohorts—as well as within-child Flynn Effects on cognitive growth across age. Overall gains equaled ~3 IQ points per decade.
Novel genetically informed analyses suggested that individual sensitivity to the Flynn Effect was moderated by an interplay of genetic and environmental factors.
…Within-level and between-level FEs: The FE has usually been documented as a between-subjects phenomenon, either as mean increases in cognitive ability scores over generations or mean decreases between test versions. There is reason to suspect that environmental influences associated with between-subjects FEs also influence cognitive ability at the individual level, boosting intellectual growth across development above and beyond typical age-related gains. 2001, for example, proposed that environmental enrichment makes it easier for children to select into environments that match their cognitive ability. Cognitively beneficial environments, in turn, boost cognitive growth, which facilitates further self-selection into more positive environments, in turn boosting cognitive ability, and so on, creating a reciprocal cycle that makes individual children exhibit greater intellectual growth across development [eg. mutualism]. The individual gains brought about by ‘individual multipliers’, as 2001 called them, can be thought of as within-person FEs. If enough children in a given population show such within-person FEs, the group mean will also rise, eventually resulting in a between-subjects FE across cohorts. A reciprocal process of social multipliers, which is the between-subjects analog of the individual multipliers process, can compound mean cognitive ability gains over time.
Although the possible connection between within-level and between-level FEs has been discussed for 2 decades, few studies have examined this empirically. Effective investigation of within-person FEs requires longitudinal data to model individual cognitive growth across age and isolate within-person FEs from age-related gains, as well as multilevel data across cohorts to distinguish within-level FEs from between-level FEs. Datasets meeting those criteria are rare and, perhaps because of this rarity, nearly all previous FE studies have performed solely crosssectional, between-subjects analyses. Only 2 previous reports have investigated within-person FEs using a multilevel approach. In 2 distinct multilevel analyses of math scores collected longitudinally across childhood 1986–26201212ya, 2017 observed statistically-significant within-person FEs, along with between-subjects gains. Later, 2020 performed a followup analysis of the same data. Results highlighted the utility of examining within-person FEs using longitudinal, multilevel approaches, to arrive at a more nuanced understanding of the FE.
…Current study: In this study, we examined the FE in data from the Louisville Twin Study (LTS), an intensive longitudinal study of cognitive development (2015; 1983). Analyses focused on middle childhood and early adolescence (ages 7–15 years). Several features of the LTS data make them particularly well-suited for FE analyses and for addressing the gaps in the literature described above. First, initial data were collected continuously 1957–42199925ya, making it possible to test for rising IQ scores across generational cohorts of US boys and girls over a long time span. Second, 3 versions of the WISC were administered over the course of the study (WISC, WISC-R, and WISC, 3rd ed. [WISC-III]). This allowed us to examine whether test restandardization resulted in systematic drops in mean IQ scores. Third, children were followed longitudinally, with some taking multiple versions of the WISC over the course of their participation. This enabled us to test for FEs not only between subjects (ie. rises in mean scores), but also within children (ie. rate of within-person cognitive growth), all while taking advantage of the statistical benefits offered by longitudinal models (eg. distinguishing age effects from cohort effects, increased power). Finally, because the LTS is a twin study, we were able to partition the variance of the within-person FE into genetic and environmental components (also referred to as ‘biometric components’) and examine the relative influence of genetic and environmental factors on individual sensitivity to the FE. In doing so, we performed the first-ever genetically informed analyses of the FE.
Thus, the unique structure of the LTS data enabled us to examine the FE as both cohort effects and test version effects simultaneously in a single sample. This data structure also made it possible to analyze FEs both within children and between children, and to examine the relative influence of genetic and environmental factors on within-person FEs. By modeling all of these elements, we were able to isolate specific aspects of the FE while controlling for alternate effects (ie. cohort vs. test version effects, within-level vs. between-level effects, genetic vs. environmental components). At the between-subjects level, we hypothesized that we would observe evidence of the FE in 2 ways: (1) for a given age and test version, children who participated more recently in the LTS testing period would have higher cognitive ability scores on average than previous cohorts; (2) for a given age and cohort, children who took newer WISC versions would score lower on average than children who took older versions. Within individual children, we expected that within-person FEs would boost the rate at which children grew intellectually between ages 7 and 15 beyond expected age-related growth. Because this was the first genetically-informed study of the FE, we treated our biometric analyses as exploratory.
…Developmental changes in cognitive ability can be difficult to observe when cohort, test version, and age are all varying simultaneously. To our knowledge, this was the first study to document cohort and test version FEs together in a single sample. Studies that use either the cohort or test version approach face major limitations inherent in each method (limited representativeness of military conscripts in the former, vulnerability to changes in content between test versions in the latter). The fact that we documented both types of effects substantially increases our confidence that the FE is robust in the LTS sample. Furthermore, modeling both cohort and test version effects enabled us to document the full manifestation of the FE in the LTS, which otherwise might not have been apparent. Although mean IQ scores were ~100 at each age (Table 1), this apparent stability was the result of a complex process in which gains across cohorts and within individuals across age were balanced out by decreases in scores due to test restandardization. Analyzing one type of FE without controlling for the other would have revealed only half of the story.
…These results suggest that the FE boosted both individual cognitive growth between ages 7 and 15 relative to age-based norms and mean cognitive ability scores across generations. Our results provide novel evidence of within-person FEs not only on fluid intelligence, as documented previously (2017), but also crystallized and general cognitive ability (as measured by VIQ and FSIQ, respectively). Moreover, our within-person FE findings speak to the importance of modeling the FE at multiple levels of analysis, where possible. By capturing both within-level and between-level FEs, multi-level models offer a more nuanced understanding of how the FE operates across development. At least in our sample, the FE was not only a population-level phenomenon that drives broad gains in mean cognitive ability across generations. The FE also appeared to influence the cognitive development of individual children, boosting their intellectual growth beyond what would have occurred without positive environmental inputs. Had we only measured between-level FEs, as is traditionally done in FE research, we would have missed this important aspect of cognitive development.
…Unique to the LTS, a multilevel approach also set the stage for twin analyses. Theories about the relative role of genetic and environmental factors in the FE have been debated for decades, but no previous study has examined this empirically. By partitioning the variance in within-pair sensitivity to the FE, we were able to perform the first-ever genetically informed investigation of the FE. As 2001 hypothesized, individual differences in sensitivity to the FE were associated with variance in both genetic and environmental background, suggesting that the FE reflects a complex interplay of genetic sensitivity and environmental change that unfolds across cognitive development. Our finding that sensitivity to the FE on FSIQ, VIQ, and PIQ all showed substantial heritability (A) serves as a fascinating example of gene-environment interplay—the extent to which a child’s growth in cognitive ability received a boost from the environment was influenced by genetic factors. The environment also plays an important role, as variance in shared (C) and non-shared (E) environmental factors were both associated with individual-level sensitivity to the FE.
…Given the magnitudes of the observed standard errors, interpretations about possible biometric differences across cognitive domains should be made with some caution. That being said, results suggested that sensitivity to the FE on FSIQ may be more heritable than sensitivity to the FE on more specific cognitive domains (ie. crystallized intelligence and fluid intelligence as estimated by VIQ and PIQ, respectively). If robust, this variability speaks to the utility of analyzing the FE in multiple cognitive domains, as biometric results from FE analyses of general intelligence (eg. IQ) may not apply directly to more specific measures of cognitive ability.