Speed isn’t everything, and slowness may in fact be more beneficial (2017)

> Even then Borghans concedes that "Grades, scores on achievement tests, and IQ are strongly positively correlated", just not perfectly so.

Sure. There are also strongly positively correlations between how rich your parents are and grades, scores on achievement tests, IQ, and career success.

Before making my earlier HN comment I had found the 2007 paper you cited, and noticed the comment "but there is little evidence of any historical trend in the relationship."

Thing is, I had also found "Investigating Core Assumptions of the “American Dream”: Historical Changes in How Adolescents’ Socioeconomic Status, IQ, and GPA Are Related to Key Life Outcomes in Adulthood" from 2019, at https://web.s.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=0&... , which did find a difference - IQ was becoming a less relevant predictor:

] In a nutshell, the U.S. has faced dramatic socioeconomic changes during the last decades. The present findings suggest that for two birth cohorts—born in the early 1960s and in the early 1980s inequalities in educational, occupational, and health-related outcomes, which can be assumed to increase in later adulthood and old age, can be predicted by individual differences in adolescents’ IQ, GPA, and parental SES. The patterns of associations between life-outcomes and parental SES did not vary much, and that of IQ remained largely the same or varied inconsistently between cohorts. However, the observed increased predictive utility of GPA suggests that combined effects of social, economic, and labor-market changes may have increased stratification processes due to individual differences in GPA for adolescents belonging to the 1980s cohort.

This is for the US. They argue that the transition to a service/post-industrial economy combined with the increasing need for a college degree, increases the measure of GPA as a predictive value over intelligence.

It also comments the earlier "studies could not answer questions about whether predictive relations could generalize across cohorts who grew up in different socioeconomic environments."

FWIW, this study was pre-registered, at https://osf.io/hm5pt/ , so there's less chance of p-hacking.

Because of that 2019 paper, I restricted my search to only include papers published in the last few years.

Statistics of course can't work if there is no deviation, but I can't think of a country which is ethnically homogeneous.

Countries with a high degree of homogeneity, like South Korea and Japan, still have their own forms of racism. For example, https://en.wikipedia.org/wiki/Racism_in_South_Korea says many students "have difficulties maintaining relationships with students who do not share the same nationality background ... because of their classmates' different skin colors".