I should have gotten around to posting on this before, but here, via Steve Hsu, is the 2009 paper The impact of smart fractions, cognitive ability of politicians and average competences of peoples on social development by Heiner Rindermann, Michael Sailer (no know relation), and James Thompson.
To test La Griffe du Lion's "smart fraction" theory, they first synthesize a large number of results from international school achievement tests (PISA, TIMSS, and PIRLS), then convert them to an IQ-like scale where Britain has a mean of 100 and the standard deviation is 15. (They don't use Lynn and Vanhanen's database of IQ tests, but Rindermann previously showed strong correlations between L & V's numbers and performance on international school achievement tests. Just looking through this data from international school achievement tests, it's remarkable how few surprises there are if you are familiar with Lynn and Vanhanen's data from IQ tests.
The most notable difference that jumps out at me is that the Irish do fine on school achievement tests (100), where they've done mediocre on some IQ tests, which I never quite believed. It might be worth investigating this discordance. For some reason, I'm reminded of the story of the English traveler in County Kerry who asks the Irish stationmaster why the clock at the north end of the railway station platform says 12:00 and the clock at the south end of the platform says 12:10: "And what would we be needing two clocks for if they both told the same time?" says the Irishman.
There are big concerns about school achievement tests, such as clarity of translations. Or, what does it mean to test fourth graders? For example, Finland doesn't start kids at regular school until seven. And how do we know the tests are nationally representative? And how do we know how hard the kids worked on the tests? Rindermann's aware of these problems (see his 2007 paper) and he's given it a pretty good shot at working out adjustments. But, the point is that we shouldn't put too much weight on any single number. For example, the Kazakhstan score is based on a single test of a single grade for a single year. (Rindermann should try to come up with a way to summarize how many datapoints he has for each country.)
The most notable difference that jumps out at me is that the Irish do fine on school achievement tests (100), where they've done mediocre on some IQ tests, which I never quite believed. It might be worth investigating this discordance. For some reason, I'm reminded of the story of the English traveler in County Kerry who asks the Irish stationmaster why the clock at the north end of the railway station platform says 12:00 and the clock at the south end of the platform says 12:10: "And what would we be needing two clocks for if they both told the same time?" says the Irishman.
There are big concerns about school achievement tests, such as clarity of translations. Or, what does it mean to test fourth graders? For example, Finland doesn't start kids at regular school until seven. And how do we know the tests are nationally representative? And how do we know how hard the kids worked on the tests? Rindermann's aware of these problems (see his 2007 paper) and he's given it a pretty good shot at working out adjustments. But, the point is that we shouldn't put too much weight on any single number. For example, the Kazakhstan score is based on a single test of a single grade for a single year. (Rindermann should try to come up with a way to summarize how many datapoints he has for each country.)
Since these tests report performance at the 5th and 95th percentiles, we can see not only the means but also the performance at the top and bottom by countries. Here, for example, is the last page of their results, Singapore to Yugoslavia/Serbia.
Singapore, a high-income high-cost, well-administered city-state, has the highest school achievement test scores in the world at their 95th percentile. It would be interesting to compare a metropolitan area, such as Silicon Valley, to Singapore. Judging by National Merit Scholar awards, the southern half of Silicon Valley would blow away even Singapore at the 95th percentile.
They write:
The highest values for the smart fractions are found in East Asia (1. Singapore IQ 127, 2. South Korea IQ 125, 3. Japan IQ 124, 5. Taiwan IQ 123, 9. Hong Kong IQ 122). A similar result was found in psychometric (average) intelligence or in student assessment studies (see Rindermann, 2007a). Different from the SAS, Scandinavia reaches in the cognitive elite not such a good rank (11. Finland IQ 121, 12. Estonia IQ 121 [the Baltics are added here], 16. Sweden IQ 120, 25. Denmark IQ 118, 34. Latvia IQ 117, 38. Lithuania IQ 116, 39. Iceland IQ 116, 41. Norway IQ 116). Maybe a homogenizing educational policy furthering weaker but disadvantaging high ability pupils leads to a smaller standard deviation and lower values for a gifted subgroup. Better are the traditional Commonwealth countries (5. New Zealand IQ 123, 7. Australia IQ 122 and 8. United Kingdom with IQ 122). They are followed by Western and Eastern European and North American countries, by South European countries, Arab or Muslim and Latin American countries and finally by sub-Saharan countries.
Most of the Gulf Arab countries do awful, but United Arab Emirates does quite well (mean 92).
The countries with the lowest results [at the 95th percentile] are 84. Botswana (IQ 96), 85. Saudi-Arabia (IQ 95), 86. Morocco (IQ 95), 87. Kyrgyzstan (IQ 94), 88. Belize (IQ 90), 89. Ghana (IQ 89) and 90. Yemen (IQ 84). Presumably many not participating countries would have lower values.Some astonishing results are observable like the high level of Kazakhstan (6., IQ 122) and the comparatively low for Israel (31., IQ 118, mean 93).
You mean, we were lied to by the movie Borat?
For Kazakhstan we have only results from TIMSS 2007 (4th grade); Mullis et al. (2008, p. 34) describe sample anomalies, a correction would be necessary. Israel has participated in several studies, compared to older studies and [for?] Jews in the Western World the results are deteriorating (e.g. Lynn & Longley, 2006). Most probably multiple reasons are responsible and not only the 20% fraction of Arabs (a thorough analysis would be necessary).
Israel's score at the 95th percentile is ahead of Norway's, so it's not that bad, but Long Island would probably do better. Israel is a country where Zionist intellectuals designed a populist, non-intellectual culture, so smart kids don't get as much cultural backing in Israel as in other parts of the Jewish diaspora.
There are also characteristic differences between mean, upper and lower levels. For instance between Canada and USA there is no difference in the upper level (IQ 120 and 120), but in the lower level (IQ 80 and 75). The past history of slavery and a different immigration policy (or different success of migration policies and geographical distance to societies with lower mean abilities) may be reflected into this difference. A similar pattern could be found for Finland and Germany: The difference in the upper level is only 1.20 IQ-points (IQ 121 and 120), but at the lower level 9.60 IQ-points (IQ 85 and 76). Most likely different immigration histories are reflected here, furthermore differences in educational policy (age of tracking, in Germany between age 10 and 12, in Finland at age 16). Early tracking increases ability variance.
I presume Thilo Sarrazin was thinking about results like this? It would be fun to see Jurgen Habermas respond to Rindermann.
Using regression analysis (as predictors mean and lower level) the largest residual (standing for difference between upper level and the rest) is found in South Africa (with its heterogeneous population of European, Asian and African descent), inverted the largest residual (standing for difference between lower level and the rest) is found in Belgium (probably a result of immigration and educational policy).
A few other notes: Armenia has a low bottom but a decent high end, close to Norway and Belgium. Armenia scores better than Georgia. Oddly, Azerbaijan beats Armenia at the bottom but has a very poor top.
Ireland, which has lagged in some IQ tests in the past, does fine (99.9), almost exactly the same as the U.K. (by definition, 100.0).
Mexico does crummy (65, 85, 105). If you want to complain about teacher's unions, start with Mexico, where teachers have a hereditary right to pass their jobs down to their offspring! Mexico ought to be able to bump these numbers up. Brazil is another country with a weak high end (105).
China and India aren't on the list.
Estonia and Finland, neither of which has many immigrants, have about the narrowest 5th to 95th percentile gaps among smart countries: 36 points. In contrast, Japan, which we like to think of as homogeneous, is 41 points, Taiwan 41, South Korea 39, Hong Kong 38, and Singapore 48.
South Korea (106) has the highest mean and highest 5th percentile (86).
They go on to evaluate La Griffe's Smart Fraction theory. Also, here's Rindermann's 2007 paper, with responses.
Ireland, which has lagged in some IQ tests in the past, does fine (99.9), almost exactly the same as the U.K. (by definition, 100.0).
Mexico does crummy (65, 85, 105). If you want to complain about teacher's unions, start with Mexico, where teachers have a hereditary right to pass their jobs down to their offspring! Mexico ought to be able to bump these numbers up. Brazil is another country with a weak high end (105).
China and India aren't on the list.
Estonia and Finland, neither of which has many immigrants, have about the narrowest 5th to 95th percentile gaps among smart countries: 36 points. In contrast, Japan, which we like to think of as homogeneous, is 41 points, Taiwan 41, South Korea 39, Hong Kong 38, and Singapore 48.
South Korea (106) has the highest mean and highest 5th percentile (86).
They go on to evaluate La Griffe's Smart Fraction theory. Also, here's Rindermann's 2007 paper, with responses.
Posts
