Distribution across Age of Raw and Scaled Intelligence Scores - the DARSIS model

The mean raw score on intelligence tests rises steeply in childhood before stabilizing for adolescents and adults, Little is known however about how the different percentiles, let alone the entire raw score distribution, change from early childhood to adulthood. This study will show that there is a regular, mathematical relationship between age, raw scores and scaled scores on intelligence test subtests with a high g-loading. Using the norm tables from 60 subtests from 19 different intelligence tests normed in 5 different countries between 1984-2020, a relatively simple model with three coefficients and a variable ’difficulty’ parameter is shown to explain almost all of the variance in the norm tables. Smaller errors are found for the mean of norm tables, and for higher performance, but not for greater age. The year in which a test was normed also did not appear to influence the fit of the model. Possible applications of the DARSIS model, such as the creation of norms for above-level testing, creating extended norms, a reduced parameter space when norming new intelligence tests and calculating reference ages are discussed.