Michael Rönnlund and colleagues have a very nice paper out in Intelligence. They show that the individual differences in general intelligence that exist at age 18 are almost perfectly preserved to age 60, after which this stability starts to slowly break down.
There are many longitudinal studies of the temporal stability of cognitive ability, almost all showing that it is moderately to highly stable across the lifespan. There are two limitations in this literature that make it difficult to say what the stability of intelligence is.
Firstly, most studies compare test scores from childhood to ones obtained in adulthood or old age. As children mature at different rates, their IQs have substantial volatility, which may lead to underestimation of actual stability. For example, Deary and colleagues found that the correlation between scores on the Moray House Test between ages 11 and 77–80 in Scotland was, after correction for range restriction, 0.73. Because cognitive maturation is still underway at age 11 (and cognitive degeneration has kicked in by age 77, although only healthy individuals were retained in the sample), this estimate is perhaps not the best value for the “true” stability of intelligence.
The second problem is that almost all stability studies, including that Scottish one, use observed test scores. This, too, leads to underestimation of stability because observed IQ scores, while mostly reflecting g variance, are also influenced by other sources of variance, especially random measurement error. An exception to this is Schalke and colleagues’ study of more than 300 Luxembourgers who were tested at ages 12 and 52. This study used latent variable methods to model g and other cognitive abilities as latent factors, enabling the estimation of the stability of “true score” variables that are more or less uncontaminated by random measurement error and other irrelevant sources of variance. Schalke and colleagues found that the stability of g between ages 12 and 52 was 0.85. Interestingly, they found that the stability of most “broad” cognitive abilities, such as Crystallized intelligence and Visual processing, was similar to that of g, indicating that intelligence in general and not just g is highly stable. A limitation of Schalke and colleagues’ study is that there are still maturation-related cognitive differences between individuals at age 12.
Rönnlund and colleagues’ new study suffers from neither of these shortcomings. They had a sample of 262 Swedish men (with some attrition at older ages) whose intelligence was first measured in a military IQ test at age 18 (when cognitive maturation is more or less complete), and then remeasured at ages 50, 55, 60, and 65 by the researchers. They used latent variable methods to estimate stability coefficients between g factors at these different ages, circumventing the problem of measurement error.
They found that the stability coefficient for g was 0.95 between ages 18 and 50, and 0.94 between ages 18 and 55 and 60. Only at age 65 did some signs of deteriorating stability emerge: the coefficient was 0.86. These results mean that g differences between individuals at age 18 explain about 90 percent of g differences as late as at age 60. Even at 65, some 74 percent of g differences are explainable by g differences measured in late adolescence. These estimates indicate that the cognitive rank order of individuals stays almost fixed throughout the years of active adult life.
A limitation in the study was that only three (age 18) or four (ages 50-65) tests were used to define g, suggesting that a substantial portion of the non-stable variance may have been due to psychometric sampling error. Therefore, the stability estimates, high as they are, may nevertheless be somewhat downwardly biased. Conversely, a potentially upwardly biasing influence on the stability coefficients was that Rönnlund and colleagues’ models didn’t include paths from the residual, test-specific variances at age 18 to corresponding variances at ages 50-65, which may have resulted in some non-g stability, if any existed, getting attributed to g.
Deary, I. J., et al. (2004). The impact of childhood intelligence on later life: Following up the Scottish mental surveys of 1932 and 1947. Journal of Personality and Social Psychology, 86, 130–147.
Rönnlund, M., et al. (2015). Interindividual differences in general cognitive ability from age 18 to age 65 years are extremely stable and strongly associated with working memory capacity. Intelligence, 53, 59–64.
Schalke, D., et al. (2013). Stability and change in intelligence from age 12 to age 52: Results from the Luxemburg MAGRIP study. Developmental Psychology, 49, 1529–1543.