Perhaps the U.S would do even better if only Asian-Americans were allowed to participate.
American blacks make up nearly 13% of the population in the US. That's a pretty significant difference and if you include all minority groups, America's a significantly less homogeneous place than any country in Europe or Asia that I'm aware of.
I agree with your (unstated) premise that this isn't the point, though.
Non-ethnic germans make up 20% of the population, far greater than the 2.1%. See here: http://de.wikipedia.org/wiki/Deutschland#Demografie
Same goes with France, I just used 'arabs' for demonstration, but the actual non-ethnic French percentage is higher than the ethnic blacks in the U.S.
Or am I drawing erroneous conclusions from the data?
Also, there needs to be a critical mass of good researchers to create a rich ecosystem that in turn produces significant technology. The US lead in R&D is partially caused by its large number of active researchers and the collective amount of funding these scientists receive compared to other countries.
In some of these countries, school is mandatory for 15-year-olds (e.g. the Netherlands). In other countries, like the US, you can drop out by (or before) that time. I wonder if this skews the results in any way. In the NL, a random sample of 15-year-old students is essentially a sample of all 15-year-olds. In the US, it's a sample of all 15-year-olds who are still in school; in other words, you are not testing the results of kids who have dropped out. If you did, it seems to me that this would likely bring the overall score down.
Also, when judging how well a country will do in science you should not look at the average, but rather at the top.
What would be the point of such exclusions and comparisons?
I don't think that these surveys are merely meant to find out which countries are leading in certain subjects. Of course you can look at it that way, but I think a better way to look at it is to see it as a measure of how well each country's education system is doing at educating its students on average.
I think it's great if our top students are at the top worldwide in science, but I'm still really, really concerned that the US education system is failing to educate the average student.
Questions of political correctness aside, this seems like some fuzzy thinking to me. Why focus on skin pigment as a distinction? Presumably because these numbers are easily available, and the categories are relatively "neat".
But why not skip all that and just use the top quartile scores? Or the top percentile?
The relentless focus on skin pigment is damaging to our thought processes as a whole, globally. It's an easy target, and it is definitely a (relatively) reliable proxy indicator for certain attributes (e.g. income, test scores). But most interesting questions (and answers) are causative in nature, and with skin color only being a proxy indicator, it seems limited to correlative questions. Also, in our civilization's more primitive and less-informed times, skin pigment was a critical distinction in both daily life and foreign policy. However, this was simply due to the lack of better or easier-to-use distinctions.
I mean, does anyone truly believe that skin pigment causes lower test scores? I doubt it. Now, do inattentive parenting and poor schooling cause lower test scores? I think so.
It may be that skin pigment correlates with poor education at home and institutionally, and it may be the only reasonable "metric" for quality of upbringing. But the focus should be on the causative force, and shortcuts that we make for practicality's sake should be broadly disclaimed at the outset.
So let's drop our vestigial attachment to skin pigment as an explanatory or causative condition. Not for reasons of sensitivity or political correctness, but for reasons of epistemological correctness.
Also: most of the "issues" that people have with skin pigment are actually judgments of culture and not appearance. Again, appearance is merely a proxy for culture, here. Note that I see nothing wrong with condemning cultures that are judged to promote or result in "bad outcomes".
For 2003, the math scores by race are here: http://www.nces.ed.gov/surveys/PISA/PISA2003HighlightsFigure...
In 2003, students in the United States averaged 483 on the math literacy test and 477 on the problem solving test (24th place in the OECD in each). Non-Hispanic White students in the U.S. averaged 512 points on math literacy and 506 points on problem solving. That's equivalent to 15th place and 13th place, respectively.
Also, when judging how well a country will do in science you should not look at the average, but rather at the top. Only a small number of people will choose a career in the sciences, so as long as you have those people, the ones at the bottom don't matter for this purpose - they bring down the average, but don't prevent a country from being a leader in research and development.