A reader points me to Steve Sailer’s breakdown.
I ritually point out that each race within the U.S. (see the red bars in my graph) did pretty darn good compared to the rest of the world.
PISA scores are measures of educational attainment or 15 year olds. Sailer points out that American Asians do pretty well compared with Asians in other countries, American whites do pretty well compared with whites in other countries, and so on. If you are trying to explain differences, race appears to be a more powerful variable than country.
Of course, this is consistent with the Null Hypothesis, which is that differences in education systems do not matter. Of course, as Sailer points out, most folks prefer to disregard the Null Hypothesis and instead condemn the “poor performance” of our education system.
Maybe I am misunderstanding this, but doesn’t “doing pretty well compared to” mean doing better than most others? In which case it would not be a piece of evidence confirming the null hypothesis.
I don’t think Steve and most others would argue schools have no effects, especially on tests somewhat geared to detect acquired skills, but that they’re very modest and some of these differences are likely transient (noise from influence economic cycles or some other external factor).
https://twitter.com/RCAFDM/status/1205241914297126912
Clearly, educational inputs explain little of the variance within countries and between countries. Steve’s point is that ancestry is a rather strong predictor, especially once selective immigration is accounted for.
Also, diminishing marginal returns. To the extent the US does as well or better, it comes at a high price.
Apples and oranges. Why break race out only for the USA? I would be interested to see how US blacks compare against Canadian, French, and UK blacks.
Most other developed countries haven’t been tracking and reporting data by race/ethnicity historically, save perhaps for certain indigenous groups (Canada, Australia, and New Zealand). The US does this routinely, so it’s part of their (our) PISA questionnaire and tabulated in data furnished to OECD and other groups. While it’s true that 1st and 2nd generation immigrants have performed poorly in most other OECD countries, we can provide more of an “apples-to-apples” comparison by comparing US whites to 3rd+ generation (~= native born Europeans).
https://randomcriticalanalysis.com/2016/05/09/my-response-to-the-nytimes-article-on-school-districts-test-scores-and-income/#groups_internationally
While it’s a little hard to draw strong inferences about recent immigrants from distant low-income countries because of the role of selective immigration, the patterns in the available data are mostly pretty comparable to what we observe in the US.
Thank you. Much appreciated.
I think that France has been multi-racial since WW2 with currently about 3 million non-white, non-European citizens. However, they are truly a color-blind country and by law it is illegal for their government to collect race data.
Nevertheless, OECD does collect economic status data and if we assume that non-whites are typically “less-advantaged” than whites, we can take a look at the OECD country by country write ups (for France see: http://www.oecd.org/pisa/publications/PISA2018_CN_FRA.pdf ) to get the sense that the null hypothesis might possibly be rejected.
In the USA, the OECD states ” In reading, the performance gap between socio-economically advantaged and disadvantaged students was 99 score points (OECD
average: 89 score points). ”
While in France, OECD writes “Advantaged students in France outperformed disadvantaged students in reading by 107 score points in PISA 2018 – a difference significantly larger than the OECD average difference of 88 score points.”
OECD for Canada: “In Canada, socio-economically advantaged students outperformed disadvantaged students in reading
by 68 score points in PISA 2018. This is smaller than the average difference between the two groups
(89 score points) across OECD countries. ”
And for the UK: “In the United Kingdom, advantaged students outperformed disadvantaged students in reading by 80 score points in PISA 2018. This is not significantly different than the average difference between the two groups (89 score points) across OECD countries. In PISA 2009, the performance gap related to socio-economic status was 92 score points in the United Kingdom (and 87 score points on average
across OECD countries).”
Possibly some educational systems have features that disadvantage or advantage the different socio-economic designations differently. At any rate, I am still unwilling to conced that the USA education system has achieved perfection and that there is nothing that can be done to improve.
Socioeconomic variables are ultimately proxies for the variables of interest (genetics, culture, etc). Although the socioeconomic gaps are clearly a large and persistent feature across high-income countries, they can’t necessarily be compared directly to each other to say one country is clearly doing better than another. For example, cognitive ability is clearly a significantly stronger predictor of income in the United States than in most Scandinavian countries (which likely relates to larger public sector, union density, and other factors that make these countries somewhat less meritocratic)
https://twitter.com/RCAFDM/status/835287572200964096
Because intelligence is also known to be highly heritable, this implies we should expect there to be larger links between parental income and child test scores. In other words, the weaker predictive value of parental income vis-a-vis child test scores in some welfare states likely has much more to do with the role of markets on parental income levels (not to mention incentive effects!) than the education system. There are also likely to be differences in the underlying ability distribution, which further confounds ready interpretation of such “gaps”.
https://twitter.com/NoamJStein/status/1193611551749672960
Educational attainment is likely somewhat more comparable, but even here it can’t be compared apples-to-apples because of different rates of educational attainment, types of degrees awarded, and so on. Long story short, though the existence of large and persistent gaps is likely significant, I wouldn’t bother trying to parse these differences too finely to say one country is clearly doing a better job than another (especially not without doing more to try to normalize these…. somehow)
The null hypothesis doesn’t say “that the USA education system has achieved perfection” any more than the efficient market hypothesis says the stock market perfectly prices stocks. It says that there is no scalable change that will improve student performance on a consistent long-term basis.
Thus, a school run by a charismatic headmaster and teachers who work 60 hours a week might “improve student performance on a consistent long-term basis” but would not be scalable. Head Start is very scalable and improves performance in the first years of elementary school, but the improvement washes out by the third grade.
It seems the only way to get conservatives to side with school systems is well to point out racial IQ differences.
1) The IQ racial differences are shrinking and even Charles Murray admitted this. It may never completely disappear but it is shrinking.
2) I really do wonder with assimilation and increased English at home if Hispanic-Americans will catch up at some point. (I suspect the lowering of H-A birth rate since 2008 will help here as well.)
3) I do believe the white assortative locations and mating will turn the worst underclass into the Rust Belt and Appalachian mountain areas. (The white assortative locations is the smartest WWC move to MA.)
“The IQ racial differences are shrinking and even Charles Murray admitted this.”
Link please.
https://www.aei.org/economics/bell-curve-20-years-later-qa-charles-murray/
Of course Charles hates to admit they are shrinking slightly and the evidence is not completely proven. (I still hold Murray wrote the Bell Curve to convince good white people to stop supporting welfare for African-American communities.)
When in doubt I think how Trump got elected is US economy has turned the Bell Curve realities against the white working classes in semi-rural United States. The first area of lost manufacturing happened in the Urban areas (New York in 1950s had a garment district!) and which led the complete collapse of the urban minorities areas. (Yes there are other issues here as well.) Most charts show the the African-American vs. white unemployment hit its highpoint in 1982 and has slowly been shrinking to historical lowpoints.
What happened?:
1) most new jobs have been in the urban service sector.
2) The loss of manufacturing jobs hit the WWC harest in the early Bush years.
https://www.blackenterprise.com/black-unemployment-rate-falls/
“The IQ racial differences are shrinking and even Charles Murray admitted this….. Of course Charles hates to admit they are shrinking slightly and the evidence is not completely proven.”
And you know that he hates to admit this how?
I suspect you of inserting thoughts and motivations into CM’s head that possibly aren’t there.
Your link contradicts your statement.
These are Murray’s exact words (from collin’s link):
Here’s what Dick and I said: There is a mean difference in black and white scores on mental tests, historically about one standard deviation in magnitude on IQ tests (IQ tests are normed so that the mean is 100 points and the standard deviation is 15). This difference is not the result of test bias, but reflects differences in cognitive functioning. The predictive validity of IQ scores for educational and socioeconomic outcomes is about the same for blacks and whites.
Those were our confidently stated conclusions about the black-white difference in IQ, and none of them was scientifically controversial. See the report of the task force on intelligence that the American Psychological Association formed in the wake of the furor over “The Bell Curve.”
What’s happened in the 20 years since then? Not much. The National Assessment of Educational Progress shows a small narrowing of the gap between 1994 and 2012 on its reading test for 9-year-olds and 13-year-olds (each by the equivalent of about 3 IQ points), but hardly any change for 17-year-olds (about 1 IQ-point-equivalent). For the math test, the gap remained effectively unchanged for all three age groups.
On the SAT, the black-white difference increased slightly from 1994 to 2014 on both the verbal and math tests. On the reading test, it rose from .91 to .96 standard deviations. On the math test, it rose from .95 to 1.03 standard deviations.
If you want to say that the NAEP and SAT results show an academic achievement gap instead of an IQ gap, that’s fine with me, but it doesn’t change anything. The mean group difference for white and African American young people as they complete high school and head to college or the labor force is effectively unchanged since 1994. Whatever the implications were in 1994, they are about the same in 2014.
I couldn’t immediately find data for IQ, but the black/white math SAT gap isn’t shrinking:
https://i1.wp.com/www.brookings.edu/wp-content/uploads/2017/01/ccf_20170201_reeves_3.png?w=768&crop=0%2C0px%2C100%2C9999px&ssl=1
Aren’t a lot more African-Americans taking the SATs?
Possibly, but I don’t find that rationalization mentioned here, and it wouldn’t explain why there are “almost no blacks among the top scorers on the SAT”:
The Widening Racial Scoring Gap on the SAT College Admissions Test (Journal of Blacks in Higher Education) http://www.jbhe.com/features/49_college_admissions-test.html
Why are you assuming that the observation is strictly about IQ, if mostly, or if at all? Race — for the purposes of explaining the efficacy of public education — in the real world embeds also socio-economic and cultural characteristics which correlate with it.
Well said. If you meet enough families, it is quite clear that culture is correlated to intelligence. That obviously doesn’t prove it’s causal, but it’s just common sense that how you raise children, what values you promote, etc will drive how a child’s brain develops.
What is interesting to me is whether gaps in raw IQ now are just the result of culture at an evolutionary level. If you take a thought experiment of two cultures over time (one valuing education, the other not) that you’d see genetic differences eventually.
If you meet enough (large) families (to be fair, an increasingly scarce sample set), you will observe that to the extent ‘culture’ or ‘nurture’ is reflected in the same shared environments, there is still a great deal of variance among siblings, both in terms of cognitive talent and other important and highly market-valued aspects of personality, which is fairly attributable in large part to even the relatively small genetic differences between them.
Interesting. I would still say from the data that there is an indication that US *Math* education lags meaningfully.
– US Asians are on par with Singapore in Reading/Science but 30 points behind in math, similar for other)
– It’s the lowest score for every single US subgroup by 20-30 points.
That’s the interesting one. It could be cultural (cross-cultural in the US) or a true indication of subpar education given this paradigm that teaching matters relatively little.
I still the say of the issues laid out by The Bell Curve is the true solution for any society is the Singapore Solution which is make it hard to any family want lots of children. All the top scoring nations have low birth rates so that means families long term enforce education attainment on society. (And note the lowest birth rates in the US are Asian-Americans.)
😐
I love Sailer’s chart and the slew of interesting questions that arise. I think we should guard against reading the numbers like Olympic sport results and focus mostly on how “our” country ranks. The United States is a fascinating outlier since it leads the world in so many cognitive pursuits.
For me, there are a couple of natural experiments unfolding since the fall of the Berlin Wall and with China/India/South-East-Asia fully embracing market economies. Estonia is a fascinating outlier compared to its Slavic neighbours and its Finnish ethnolinguistic brethren. Slovenia is also an interesting outlier compared to the former Yugolavian nation states. The former Yugoslavia even has a non-Indoeuropean ethnolinguistic counterpart to Estonians with Albanian communities.
I wish ethnic and IQ data collection was not taboo. As Edgar points out, French policies prevent useful data from being collected. The French colony of Reunion seems like a treasure trove of comparative data as do many former colonies with large Indian and Chinese diasporas. For instance, if South Asians have a naturally high PISA score then why isn’t Guyana ahead of its Latin American neighbors?
I think that a thoughtful analysis would show that institutions do matter and account for the differential success of Estonia and Slovenia. We just don’t know to build these institutions from scratch (yet).
I would hasten to point out that Sailer’s chart does not address causality. “Race” and “Ethnicity” correlate with many other possible explanatory variables, including economic conditions, geography, and language.
However, clearly, if one wanted to invest toward improving PISA scores in the US, the plan would (directly or indirectly) disproportionately invest in improving the scores black and hispanic students.
Problems of this sort are clearly not unique to the United States. The US is mostly distinguished in the relative magnitude and types of racial/ethnic diversity. Also, I believe the United States is investing relatively more in black and hispanic students than whites.
https://randomcriticalanalysis.com/2017/06/22/no-us-school-funding-is-actually-somewhat-progressive/
Unless the gains per dollar spent are larger for these minority groups, which seems unlikely, I’m not sure I see the economic rationale to deprive taxpayers or to short change other students.
I find Sailer’s chart incredibly depressing because the naive “Man from Mars” implication is that it wouldn’t do much good. “The gap” is not unique to America but worldwide.
Some support for that depressing view comes from American history. In the last 50 years, a tremendous amount of money and effort in the education business has tried to do just that, but “the gap” has gotten only slightly smaller. One hastens to blame “white privilege” and “structural racism” but east Asians in America (including the descendents of Japanese-Americans put in concentration camps during WW II) do better than American whites.
The utility of this test is only worth a second place, so that makes whitey smartest.
Aren’t we all missing the big news. Steve Sailer and Optimistic View of our Education System actually crossed paths for a brief moment!
Various gaps will probably close as people get more mixed race and we have more selective immigration, but that doesn’t really mean we made minorities smarter via education.
The big issue noted in the Bell Curve is that we do a poor job of educating top candidates pre-college. Yet there is no political will to get the most out of our native high IQ, and in fact gap closing is in danger of destroying what we have (Stuyvesant, my high school, etc)
I would like to see a more rigorous definition of what exactly the Null Hypothesis means when we discuss comparisons between countries, or even within a country if we think it can be applied to any old country (there are about 200–are we talking about Sierra Leone? Burma? Papua New Guinea? Tunisia? only PISA participants? ).
I started writing a comment about this yesterday and then gave up. I’ll try again right now.
It seems that there are tacit assumptions about what schools all have in common already, and what students have in common already, for the Null Hypothesis to actually be confirmed when testing new policy interventions at the margin.
What do we assume? That …
1. schools meet regularly.
2. they have desks and chairs and books, and the teachers are paid salaries on time.
3. the students are not acutely malnourished or chronically underfed, and probably not chronically ill with malaria or serious infections.
4. most students are being educated at least in part in their mother tongue or a language they are adequately proficient in, with appropriate curricular materials.
5. The teachers have a decent education. To choose an extreme counterexample, a Peace Corps veteran told me that at times the qualification to teach in a bush school in upcountry Sierra Leone was to have passed the grade one was teaching. Teachers could not teach 3d grade unless they had passed 3d grade themselves. This example is extreme and irrelevant here, because Sierra Leone is not participating in PISA. But for me it focuses the mind. How qualified does the teacher have to be?
I also like that example (Sierra Leone bush school) because a good teacher might be able to just teach 3d grade. Imagine doing it 20 years in a row.
= – = – END LIST = – = – = – =
My comment may not be especially relevant here, but perhaps it is appropriate for arguing that the Null Hypothesis can be expected to adhere within a restricted range of all educational observations.
Perhaps the Null Hypothesis is most useful
1. within a particular country,
2. If that country is already middle income
3. If that country is moderately well governed and administered.
4. Many of the easiest gains from formal education have already been exhausted, and we are largely tinkering at the margin
I can’t speak for Arnold, but I am a fairly strong believer in the null hypothesis for America and very much NOT a believer in the null hypothesis for much of the rest of the world (basically the global “south”). The five things you enumerate are pretty much true everywhere in the USA. They are very much NOT true all over the world.
Lant Pritchett’s The Rebirth of Education: Schooling Ain’t Learning has a lot of depressing statistics about what many schools are like and how little is learned.
A depressing anecdote from education in Sub-Saharan Africa at the secondary level is that many male teachers view sleeping with one’s female students in high school as perk, a fringe benefit. This would become more motivating if the salaries come late.
Those male teachers with rectitude restrain themselves from sleeping with junior high school girls.
This was reported in the London _Economist_, with two teachers “taking it outside” to engage in fisticuffs because they both wanted the same girl. At the same time, they declared students in junior high to be off-limits, and a violation of norms. “High School is where we play.”
I don’t recall the country. It was a big Anglophone country.
Is this an accurate portrayal of reality? Beats me. The _Economist_ style guide once instructed writers to “simplify, then exaggerate.”
To suggest more reading, there is the book _The Beautiful Tree_, by Tooley, published by Cato. It’s for a popular audience. I do recognize Pritchett’s name from the scholarly literature, either at the World Bank or often published by him.