Back in January, we published a paper showing that the obvious explanation of variation is in GRE scores by nationality/origin fits the data well.

Mean scores on English-based graduate admissions exams vary strongly country by country. Here we test four hypotheses for this variance. The first is that samples of would-be graduate students are not representative of the countries from which they came. Second, English Language familiarity might account for cross-national variance on these exams. Third, national differences in well-being explain national differences in exam scores. Finally, variance in national IQ (NIQ) captures most of the variance in scores on admissions exams. We thus coded two extant measures of NIQ, and national scores on the following three admissions exams: The Graduate Management Admission Test (GMAT; n = 200 countries), Graduate Record Exam (GRE; n = 149), and Test of English as a Foreign Language (TOEFL; n = 175). Partial support was found for both the unrepresentative samples and language familiarity hypotheses. The well-being hypothesis was not supported because despite large correlations between it and exam scores, these effects were wiped out when NIQ also appeared in the regression equations. The NIQ hypothesis received the most support in that it strongly predicted admissions exam total scores (GMAT, r = 0.68 (N = 161); GRE, r = 0.71 (N = 141); TOEFL, r = 0.61 (N = 153). Moreover, these relationships (excluding those with the TOEFL) were robust to the controls of English familiarity and well-being. We end by discussing how our data also validate the existence of Rindermann’s (2007) “Big G” nexus.

It appears someone else had the same idea, just more limited approach:

This study develops a straightforward transformation of Graduate Management Admissions Test (GMAT) scores into an equivalent IQ score using countries for which both measures are available; and, validates the same for countries for which direct measurements of National IQ are not available. The GMAT is an aptitude test used for admission into many MBA and other graduate programs in business. National average GMAT scores are found to be positively correlated with well-established measures of cognitive ability provided the GMAT scores are corrected for participation rate. Even though GMAT scores are indicated to be prone to measurement error, they drive ethnic-based proxies of National IQ to insignificance in regression analysis of GDP per capita.

All of this really began many years ago, in 2014, on the HumanVarieties blog: Quick Post: L&V’s National IQs predict GMAT scores across 173 nations.


Leave a Reply