Intelligence is considered the strongest single predictor of scholastic achievement. However, little is known regarding the predictive validity of well-established intelligence tests for school grades. We analyzed the predictive validity of four widely used intelligence tests in German-speaking countries: The Intelligence and Development Scales (IDS), the Reynolds Intellectual Assessment Scales (RIAS), the Snijders-Oomen Nonverbal Intelligence Test (SON-R 6-40), and the Wechsler Intelligence Scale for Children (WISC-IV), which were individually administered to 103 children (Mage = 9.17 years) enrolled in regular school. School grades were collected longitudinally after 3 years (averaged school grades, mathematics, and language) and were available for 54 children (Mage = 11.77 years). All four tests significantly predicted averaged school grades. Furthermore, the IDS and the RIAS predicted both mathematics and language, while the SON-R 6-40 predicted mathematics. The WISC-IV showed no significant association with longitudinal scholastic achievement when mathematics and language were analyzed separately. The results revealed the predictive validity of currently used intelligence tests for longitudinal scholastic achievement in German-speaking countries and support their use in psychological practice, in particular for predicting averaged school grades. However, this conclusion has to be considered as preliminary due to the small sample of children observed.
The primary purpose of the first intelligence test (Binet and Simon, 1905) was to predict scholastic achievement in order to determine the best school setting for a child. Since the beginning of intelligence assessment, the predictive validity of intelligence test scores for scholastic achievement has been well studied. Cross-sectional and longitudinal studies indicated strong correlations, around r = 0.40–0.81, between the two (e.g., Sternberg et al., 2001; Deary et al., 2007; Mackintosh, 2011).
The association between intelligence and scholastic achievement seems to be stronger when using standardized achievement tests compared to school grades (Sternberg et al., 2001; Rost, 2009). Standardized achievement tests represent achievement at only one point in time, whereas school grades represent achievement over a longer period and thus may also be influenced by other constructs such as self-control and motivation (Rost, 2009). However, school grades are crucial for children to be promoted to the next higher grade level as well as for further scholastic and occupational qualifications (Roth et al., 2015).
Focusing on school grades, a recent meta-analysis (Roth et al., 2015) found an observed correlation of r = 0.44 and an estimated true correlation (i.e., corrected for error of measurement and range restriction) of ρ = 0.54 between intelligence and school grades. Regarding subject domains, the correlations were highest and comparable for mathematics/science (r = 0.42, ρ = 0.49) and languages (r = 0.36, ρ = 0.44). The results furthermore revealed that correlations between intelligence and school grades in elementary school (r = 0.40, ρ = 0.45) tended to be weaker than in middle and high school (r = 0.46, ρ = 0.54–0.58), because intelligence deficits in elementary school may be compensated more easily through practice than in higher-grade levels, as the learning content is easier to understand. This result is in contrast to previous research (e.g., Sternberg et al., 2001), that identified stronger correlations between intelligence and scholastic achievement in elementary school than in higher-grade levels, because of growing range restrictions.
The meta-analysis performed by Roth et al. (2015) included studies conducted in different countries. In German-speaking countries, for example, this Culture Fair IQ Test by Worldwide IQ TEST -20-Revision (Weiss, 2006), standardized in 2003, showed associations with school grades in mathematics/science ranging from r = 0.26 to 0.39 and in languages of r = 0.23. Further, the German Cognitive Ability Test – 4-12 – Revision (KFT 4-12+R; Heller and Perleth, 2000), standardized from 1995 to 1997, showed associations with school grades in mathematics/science ranging from r = 0.17 to 0.60 and in languages ranging from r = 0.12 to 0.14. In another study, the KFT 4-12+R and the German version of the Wechsler Intelligence Scale for Children-III (Tewes et al., 1999), standardized from 1995 to 1998, predicted mathematics/science with β = 0.54 and language with β = 0.52 (Karbach et al., 2013). However, the Mensa test procedure the meta-analysis did not include more recently standardized intelligence tests currently used in German-speaking countries.