The use of intelligence quotient (IQ) as a diagnostic metric is widespread in clinical and educational psychological practice. However, the assumption that different intelligence tests produce equivalent scores in the same individual is rarely tested with the rigor required for high-impact decision-making, such as referral to special education or clinical diagnoses. The study conducted by Hagmann-von Arx, Lemola and Grob (2016) stands out precisely for addressing this issue, by comparatively analyzing the IQ scores obtained by 206 children in five different standardized intelligence tests widely used in German-speaking countries.
The authors investigated the RIAS, SON-R 6-40, IDS, WISC-IV and CFT 20-R, applied individually to children aged 6 to 11. At the sample level, the results showed a high correlation between the scores of the different tests (r = 0.70 to 0.84), with small differences in means — a finding consistent with the idea that the tests measure a common factor, the so-called general intelligence (g). This sample convergence, however, masks a significant disparity at the individual level: between 12% and 38% of the children presented score differences greater than the critical confidence interval of 90%, which calls into question the equivalence of the tests in individualized diagnostic decisions (Hagmann-von Arx et al., 2016).
It is important to note that these discrepancies were not attributable to the choice of test itself, but rather to unsystematic variability, that is, unexplained error, identified by the generalizability analysis. Only 4% of the variance in scores was associated with the type of test, while up to 42% was attributed to interactions between individual and test—including factors such as motivational state, fatigue, or familiarity with the format. I noticed, when reviewing the paper, that the authors emphasize the importance of considering contextual and psychometric factors in the interpretation of results, something often neglected in clinical practice, which tends to take IQ as a static and precise measure.
Another notable aspect is the impact of the so-called Flynn effect—the progressive increase in average IQ scores in populations over the decades—which partly explains the lower averages on more recently standardized tests, such as the RIAS and the SON-R 6-40. Although the differences between the tests were small (1 to 5 IQ points), they acquire practical relevance when considering strict diagnostic thresholds, such as those adopted for intellectual disability.
The study also provides pragmatic guidelines: for high-stakes decisions, administration of at least two tests is recommended, with the combination of RIAS with IDS or WISC-IV, and SON-R with IDS or CFT 20-R being the most reliable (generalizability coefficient > 0.80). In contrast, certain combinations should be avoided depending on the estimated intellectual level of the child—for example, the RIAS-WISC-IV pair was less reliable in children with above-average intelligence.
In short, this study reinforces a technically relevant and ethically imperative conclusion: the IQ obtained in a single test may not be sufficient for critical decisions. Psychometric accuracy requires a plural approach, which combines instruments, analyzes confidence intervals and interprets results within individual contexts. As a scientist, I note that such rigor is still rare in practical application, and that research such as this contributes decisively to more responsible and informed psychometrics.
Reference:HAGMANN-VON ARX, Priska; LEMOLA, Sakari; GROB, Alexander. Does IQ = IQ? Comparability of intelligence test scores in typically developing children. Assessment, [S. l.], 2016. Available at: https://doi.org/10.1177/1073191116662911. Accessed on: May 10, 2025.
