Saturday, August 04, 2012

IQ estimates from wordsum scores by ethnicity and community type

++Addition++Great comments. Using the GSS, Jason Malloy replicates what other studies have found regarding the urban/rural IQ gap in the US--that it has shrunk appreciably over the last three generations, but not due to rural areas catching up with urban ones. Instead, urban areas appear to have regressed while rural areas have treaded water:
I looked at urban/rural differences by birth cohort.

white's born 1900-1940
93 rural
100 suburb
102 city

white's born 1941-1969
95 rural
100 suburb
100 urban

white's born 1970-*
93 rural
96 suburb
96 city

The rural/urban difference has converged from 9 points to 5 points to 3 points across the three cohorts.
---

Continuing with the rural vs. urban IQ theme, let's consider the gaps among those of (relatively) shared ethnicity by the broadly based type of communities they were raised in. Ron Unz speculates that because urban dwellers outperform their bucolic counterparts on IQ tests and their proxies, groups that are skewed more heavily towards urban living and away from rural living are consequently more intelligent for it. The HBD chickadee poked some holes in Unz' initial assertions, but on net, the conclusion could plausibly follow from the contemporary data.

I wondered, though, if German and Dutch wordsum averages were lower because of a more rural skew than those of Italian and Irish descent, or if the differences were apparent among those of different ethnicities living in similar communities. The following table lists mean IQ values as estimated from wordsum scores by self-identified ethnicity among whites (with the exception of those of African descent) by community type. The last column shows the urban/rural gap. Due to concerns with sample sizes, the selection is somewhat limited. Unfortunately, Dutch numbers aren't high enough on the urban component for useful comparisons to be made:

AncestryRural IQSuburban IQUrban IQUrban advantage
Russian93.9103.1107.313.4
English/Welsh96.9103.3104.47.5
Scottish96.4102.1102.86.4
French96.699.2102.25.7
Polish94.599.7101.87.2
Irish93.7100.3100.97.2
Italian95.199.1100.65.5
German94.599.199.44.8
Native American89.293.696.67.5
African83.290.190.87.6
Mexican85.787.283.2(2.5)

With the exception of Mexicans, the trend is remarkably consistent. Rural rustics fare poorly, suburbanites are somewhere in the middle, and urban dwellers come out on top. That those of Mexican descent don't exhibit the same pattern that everyone else does, including blacks, makes Unz' argument that Hispanics will eventually get up to snuff with other Americans appear more than a little tenuous.

While the rural/urban gap is real, the presumption that the relative proportions of ethnics living in different community types accounts for the overall differences in intelligence isn't clear. Excluding Russians, about half of whom are Jewish with a heavy urban skew for Russian Jews and a non-urban skew for non-Jewish Russians, the overall differences between ethnic groups remain at each community level as well.

That is, the English and the Scottish outscore the Germans and the Irish in rural, suburban, and urban settings. It's not just that the Scottish are more urban than the Germans are, rural Scots outperform rural Germans and urban Scots outperform urban Germans. Since we're looking at the environments people grew up in as kids, it shouldn't take multiple generations for the urban premium to confer an advantage over rural living.

If Unz' theory was perfect, there would be no differences among varying ethnicity in rural environments, with the overall ethnic differences instead being accounted for by the fact that higher scoring ethnicities were skewed more heavily towards urban dwelling and less intelligent ethnicities skewed towards rural life. To the contrary, among white European ethnic groups, the correlation between rural mean wordsum scores and urban mean wordsum scores is a statistically significant, vigorous .78. Going from rural to urban helps everyone (except Mexicans!), but underlying differences remain irrespective of what community type is being considered.

Admittedly, this is stretching the suggestive value of wordsum scores as a predictor of intelligence to the maximum, and I'm not attempting to speak with authority about Unz' theory or the refutations of it here. I'm just offering up so more food for thought for those who've been following and thinking about it over the last week.

GSS variables used: ETHNIC, RACE(1)(2), RES16(1-2)(3-5)(6), BORN(1), WORDSUM

55 comments:

Ron Unz said...

I'd bet the problem with the Mexican numbers is they're massively distorted by recent immigration.

If you stratify by decade and almost by nativity, I bet you'll find something quite different.

Incidentally, I've now posted my reply to Lynn at www.ronunz.org.

Matt said...

I'd bet the problem with the Mexican numbers is they're massively distorted by recent immigration.

In what sense? That urban/suburban/rural regions are disproportionately effected or that urban/suburban/rural are all equally depressed?

Also, any opinion on this, while we're at it:

http://images.ocregister.com/newsimages/news/2007/02/27immigcrime_em.gif

Audacious Epigone said...

Ron,

It's a population in more flux than any other in the US, although the numbers here are restricted to those born in the US, so we're at least looking at the second generation that has putatively had their entire lives to learn English as well as anyone else.

Ron's response to Lynn and Nyborg is here.

Anonymous said...

I'd bet the problem with the Mexican numbers is they're massively distorted by recent immigration.


I'd bet that it is not possible for any evidence to exist which would cause Ron Unz to reconsider his deeply cherished beliefs. Attitudes which are not held rationally cannot be challenged by facts and reason.

Anonymous said...

Are German numbers a bit low because German Jews don't want to self identify as German?

Perhaps they choose Polish, etc instead?

Anonymous said...

Are German numbers a bit low because German Jews don't want to self identify as German?


Given the relative sizes of the German and Jewish populations in the US, I'm going to say "No".

Anonymous said...

In my experience, German Jews would not identify as Polish. Before WWII, German Jews both in Germany and the U.S. looked down on their brethren from farther east. That didn't entirely go away even after WWII. A lot of American Jews whose ancestors come from what is now Poland also will not identify as Polish. Most of their emigrant ancestors originated in the parts of Poland ruled by Russia or Austria as is the case for my family. Per older relatives, when they used to ask where their ancestors came from, they were often just told "Russia" or "Russia-Poland." None of those ancestors lived in Russia proper.

Dan said...

Wordsum isn't looking so hot, in my opinion.

For example, I was able to ace the SAT verbal in large part because my school district did intensive vocab exercises from a prepared series with 20 new words *every week for seven years*. That's many thousands of words, including almost all of the words that appear on the SAT (and probably WordSum as well).

I am fairly smart, but I got a big boost too. I am sure that urban districts, for HBD reasons, work very hard on vocab building for standard tests.

Also, it is no secret that urban people are better educated, and clearly Wordsum must correlate hard with education. You could have an uneducated genius completely bomb a test like Wordsum unless they read the kind of books that have lots of vocab words.

Anonymous said...

Is there a way to look at the rural/urban divide that is not loaded with vocab?

Audacious Epigone said...

Dan,

That's certainly a legitimate issue with an imperfect measure, but wordsum is the best we have to work with. Take it with qualifications in mind.

The GSS has asked a series of questions testing basic science literacy, so we could approach it from that direction, but I bet the results won't differ much from what wordsum scores tell us.

Anonymous said...

wordsum is the best we have to work with



I'm not so sure about that. We can look at things such as income and educational level attained for the different ethnic groups and see how well that lines up with wordsum.

Ron Unz said...

Anonymous: I'm not so sure about that. We can look at things such as income and educational level attained for the different ethnic groups and see how well that lines up with wordsum.

Actually, as I've mentioned in my various articles and columns, I also looked at the GSS measures for education and family income, and they followed a very similar pattern with regard to the relative ranking of ethnicities.

However, I didn't bother stratifying those by rural/urban, since they'd probably be far more obviously distorted than Wordsum, e.g. rural incomes and cost-of-living are much lower.

Anonymous said...


The GSS has asked a series of questions testing basic science literacy, so we could approach it from that direction, but I bet the results won't differ much from what wordsum scores tell us.


Oh, boy, would that be depressing. The fundies would probably out score them.

Anonymous said...

Ron, Epigone --

There is other data that is much better than the wordsum data and that is the NAEP data (National Assessment of Educational Progress)that you, Epigone, link to on your page.

http://www-958.ibm.com/software/data/cognos/manyeyes/visualizations/iq-by-state-us-2

Here you see a north-south divide but seemingly not much of a rural-urban divide. Fairly rural states like Montana, the Dakotas, Iowa, Nebraska, Idaho, New Hampshire and Vermont all do better than relatively urban California, Florida, Maryland, New York, Rhode Island and Connecticut.

The NAEP is far more comprehensive than the tiny Wordsum, I think.

Anonymous said...

The following table lists mean IQ values as estimated from wordsum scores by self-identified ethnicity among whites (with the exception of those of African descent) by community type.


Won't that tend to inflate the Mexican score by selecting only the whitest Mexicans? Presumably many of them don't identify as white at all.

Anonymous said...

Unz - "as I've mentioned in my various articles and columns, I also looked at the GSS measures for education and family income, and they followed a very similar pattern with regard to the relative ranking of ethnicities"


I'm afraid that based on your track record with data, I'm going to have to run those numbers myself. I'll post them here when I'm done.

Ron Unz said...

For some reason, my web browser doesn't probably serve that Java-rich NAEP page, but are those results stratified by race? Obviously, there's be a large demographic skew in all those states, with a heavy impact on average scores, which is why I was restricting RACE=WHITE during my GSS analyses, and further stratifying by white ethnicity in most runs.

Jason Malloy said...

The GSS dataset lumps together all the cohorts, which might explain why more recent immigrants would show less of a rural/urban difference. The word in the field is that US rural/urban differences mostly converged in the 1970s. For example, John Loehlin:

"A generation or two ago, average U.S. rural-urban IQ differences were substantial, averaging about 6 1/2 IQ points (Terman & Merrill, 1937). In more recent years, the difference declined to about 2 IQ points... The urban-rural difference is now so small that recent IQ test standardizations have not even bothered to stratify their samples by this variable." (p 181)

The APA's "Intelligence: Knowns and Unknowns":

Among other things, Kohn and Schooler's hypothesis may help us understand urban/rural differences. A generation ago these were substantial in the United States, averaging about 6 IQ points or 0.4 standard deviations (Terman & Merrill, 1937; Seashore, Wesman, & Doppelt, 1950). In recent years the difference has declined to about 2 points (Kaufman & Doppelt, 1976; Reynolds, Chastain, Kaufman, & McLean, 1987)."

Jason Malloy said...

The GSS dataset lumps together all the cohorts...

That is the cohorts will be lumped together unless you analyze cohorts separately, which is not always necessary.

Anonymous said...

Ron Unz wrote:
"Obviously, there's be a large demographic skew in all those states, with a heavy impact on average scores, which is why I was restricting RACE=WHITE during my GSS analyses, and further stratifying by white ethnicity in most runs."

Yes, those data at that link are for whites only. The rural areas seem to do admirably well.

The top German-American states (wikipedia) by percentage are:1.North Dakota 46.9
2.Wisconsin 43.9
3.South Dakota 44.5
4.Nebraska 42.7
5.Minnesota 38.4
6.Iowa 35.7
7.Montana 27.0
8.Ohio 26.5
9.Wyoming 25.9
10.Kansas 25.8
11.Pennsylvania 25.4
12.Missouri 23.5
13.Indiana 22.6
14.Colorado 22.0
15.Oregon 20.5

German American states do admirably well. Rural, German-American states do admirably well.

Severn said...

This is the order of several white ethnic groups in the US by wordsum estimated IQ.

English 102.4
Scottish 101.5
Swedish 101.3
Italian 100.4
French 100
Irish 99.7
Polish 99.7
German 98.5

There is only a four point difference between the highest and lowest group.

And here is the percentage of each group with at least a four year college degree

Scottish 42.2%
Swedish 38.5%
English 37.1%
Italian 33.6%
Polish 36%
German 32.7%
Irish 32.2%
French 29.2%

(White average is 29.3%)

The French seem to be statistical outliers along many demographic vectors. No doubt many are Louisiana Cajuns. Let's just ignore them.

Then the wordsum projected IQ matches up reasonably well with academic performance. The English under-perform their wordsum while the Poles over-perform theirs. Or, perhaps more accurately, the English wordsum slightly overstates their IQ while the Polish wordsum rather understates theirs.

In all cases, a sense of ethnic identity seems to be positively correlated with performance and success.

Russian Americans tend to be like Indian Americans and Chinese Americans - a small, select, high IQ group of people who came here recently on visas for tech jobs.

The proportion of Russians who are foreign born is quite large - 16.7% of Russian-Americans are foreign born compared to just 0.7% of Irish-Americans. In a few generations, when regression to the mean and intermarriage take effect, I'd expect them to more closely resemble other groups of whites.

Jason Malloy said...

Ok, using Epigone's variables I looked at urban/rural differences by birth cohort.

white's born 1900-1940
93 rural
100 suburb
102 city

white's born 1941-1969
95 rural
100 suburb
100 urban

white's born 1970-*
93 rural
96 suburb
96 city

The rural/urban difference has converged from 9 points to 5 points to 3 points across the three cohorts.

The convergence in the GSS is not due to a rise in rural scores (as is suggested in the above summaries), but is entirely because of a decline in city and suburban scores.

Mexican sample sizes are low for early cohorts when rural/urban differences were larger.

Olave d'Estienne said...

In answer to Matt's question about the OC Register crime gif:

It matches up pretty well to other things I've heard. Immigration has continued as violent crime has fallen from its peak in the early/mid-90s. (Previous to that, immigration continued as crime rose to its peak in the early/mid-90s.) Violent crime in the 90s was driven by wars to control the crack trade, which eventually died down a little as rival gangsters died off and were incarcerated. Superficial peace resulting from one gang controlling illegal drugs can reduce violence, but can do nothing to reduce drug mortality and morbidity, welfare dependency, illiteracy, or unemployment.

Recent immigration can temporarily suppress crime rates, since foreign-born Hispanics (mostly born in Mexico and Guatemala) are much less crime prone (and much older) than their sons.

It's not clear exactly why US-born Hispanics are so much more crime prone, but the cultural conservative answer fits the facts. The US justice system was designed by Anglo-Saxons for Anglo-Saxon culture. The Briton and the American were both honored by a system that treated them as innocent until proven guilty. They did regard this as the system showing weakness, or as something to be snickered at.

The HBD answer also fits the facts. One answer that does not fit the facts is that people are naturally drawn to environments where they will fit in, assimilate, and thrive.

The upshot is that stopping immigration and deporting foreign infiltrators may not be enough. "Non-Hispanic whites" (i.e., actual Whites) do not reproduce much in conditions of massive uncertainty and physical danger. Your whiterpeople, who think it's cool to live in Manhattan or L.A. or S.F., aren't actually going to reproduce there. They probably won't reproduce, and if they do, they'll be doing it in a low-crime area which, by necessity, will be an overwhelmingly White/ Chinese/ Japanese/ Korean area. (And no, S.F. doesn't count as a low-NAM area, because it's not an area. The Blacks will come over from Oakland to hold another Zebra killing spree exactly when they want to.)

As these areas dry up, White populations will continue to fall. This is a matter of habitat loss, nothing more.

Olave d'Estienne said...

They did regard this as the system showing weakness, or as something to be snickered at.

Ack, bad proofreading. They did not regard this as the system showing weakness....

Ron Unz said...

Jason Malloy: The convergence in the GSS is not due to a rise in rural scores (as is suggested in the above summaries), but is entirely because of a decline in city and suburban scores.

I wonder if this is actually correct...

One reason I prefer to look at the raw Wordsum scores rather than trying to convert them to IQs is that they're only somewhat correlated, but another is to avoid confusion over Flynn Effect adjustment issues.

For example, when I stratified the white Wordsum values by decade and also by rural/urban type factors, I noticed that the Suburban and Large City+Suburban mean Wordsums had barely changed at all over the 40 years, while there had been a dramatic rise in the Farm Wordsums. So offhand, it looked like there's been virtually no Flynn Effect for the urbanites, but a very strong one for the Farm rurals, which would make a lot of sense under various Flynn Effect hypotheses. Essentially, 1/3 of the Wordsum gap disappeared in 30 years, almost entirely due to a rise in the rural Wordsum.

However, if you continually renormalize your mean white IQ to 100, perhaps this would instead show up as a large drop in urban IQs.

My analysis only focused on the City/Suburb/Farm categories (leaving out e.g. small towns), since those seemed to provide the sharpest sign of some sort of surprising Rural/Urban Divide.

Anonymous said...

if you continually renormalize your mean white IQ to 100, perhaps this would instead show up as a large drop in urban IQs.


I don't follow you. If at some point in time the urban IQ is 103, the rural IQ is 95, and the meant white IQ is 101, and you then renormalize the mean white IQ to 100, how does that have a greater impact on urban IQ's than on rural ones?

I'm not saying that there can be no possible scenarios in which this does not happen. Perhaps it would happen for certain values of urban and rural IQ plus certain population sizes in urban vs rural areas. And, I suppose, it would depend on the precise formula used in re-normalizing the data. But that's all very speculative.

Audacious Epigone said...

Ron,

The conversion from wordsum to IQ doesn't change the distribution, so it looks like you and Jason are apparently finding somewhat contradictory trends over time.

Ron Unz said...

Audacious Epigone: it looks like you and Jason are apparently finding somewhat contradictory trends over time.

Well, it's perfectly possible there's an error in my spreadsheet calculation or something. Here's what I got for RACE=WHITE/WORDSUM stratified by RES16 and decade:

Farm: 1970s=5.30, 1980s=5.47, 1990s=5.64, 2000s=5.74

Large City: 1970s=6.70, 1980s=6.53, 1990s=6.62, 2000s=6.60

Suburb: 1970s=6.92, 1980s=6.89, 1990s=6.91, 2000s=6.90

LC+S: 1970s=6.79, 1980s=6.67, 1990s=6.76, 2000s=6.76

I focused on these three categories because the contrast was the starkest, leaving out the small towns, etc.

Anonymous said...

Ron Unz wrote,

"I focused on these three categories ***because the contrast was the starkest***, leaving out the small towns, etc."

Well that is not very scientific, is it?

Farm population is something less than 2% of the American population, which means this is an extreme form of cherry picking.

When you say the difference was starkest when you tossed out the "small towns etc.", what you are saying is that the small towns would have brought up the rural average and thus needed to be removed.

Mr. Unz, I am not accusing you of doing this intentionally, but if you do not correct yourself by showing what the complete rural numbers really are including small towns and whatever else you took out, one can only assume the deception was intentional.

Jason Malloy said...

The variable I used was COHORT, which is birth year.

I did not adjust for Flynn Effect; IQs are simple transformations of the WORDSUM mean.

Here's a visualization of what the Flynn Effect looks like in the GSS across age and cohort.

WORDSUM and Flynn Effect

This is for whites born in the country. WORDSUM scores increase slightly in the birth cohorts leading up to 1946, but there is an anti-Flynn Effect for cohorts born since 1965.

Baby Boomers have the highest WORDSUM scores in the GSS, and scores have been lower for subsequent generations. I'm not the first to show this.

Ron Unz said...

Anonymous: Well that is not very scientific, is it?...Farm population is something less than 2% of the American population, which means this is an extreme form of cherry picking...Mr. Unz, I am not accusing you of doing this intentionally, but if you do not correct yourself by showing what the complete rural numbers really are including small towns and whatever else you took out, one can only assume the deception was intentional.

I guess my view of the scientific method is different than yours...

In science, people tend to look for interesting patterns or intriguing empirical results. I found such a pattern in the comparison of white Wordsum between Farm vs. City+Suburb, and suggested a possible explanatory hypothesis. Obviously, I didn't also describe all the non-patterns or non-intriguing results, since these would be much less interesting to people.

Incidentally, the "Farm" value for RES16 is 12.4% of the total white population, hardly negligible. And my stratification was using YEAR rather than COHORT, which may or may not help to explain the divergence between my results and Jason Malloy's.

Jason Malloy said...

If we combine suburban and urban (which mostly cluster) and take the results by decade instead of by cohort we get this:

1970s
rural 91
non 101

1980s
rural 93
non 99

1990s
rural 95
non 100

2000s
rural 96
non 100


Here the gap narrows from 10 to 4, with the narrowing coming from the inflation of rural scores. However the cohort analysis suggests that the narrowing is due to a change in the age distribution of residence patters, where rural areas are getting smarter only because they are getting older. Not only do WORDSUM scores inflate with age, but older generations were smarter even when adjusted for age.

Jason Malloy said...

Whatever the explanation for the yearly changes (and we should keep digging), the cohort analysis is the appropriate one for tracking developmental changes. There have been no changes in the WORDSUM scores of whites raised in US rural environments for 100 years. While there has actually been a decline in WORDSUM scores for whites raised in non-rural areas since the 1970s.

Audacious Epigone said...

Here's a visualization of white IQ estimates by state taken from 2005 8th grade NAEP scores that I put together back in 2007.

Ron Unz said...

That's very interesting about Wordsum-IQ tending to generally rise with age, and also older generations have higher levels regardless. I guess I would have assumed that it rises until around 20-25, then levels off or gradually declines.

Has anyone taken these age/generational differences into account with regard to racial/ethnic analysis? For example, Hispanics and blacks have a much lower mean age that whites, and I'd think the various white ethnic groups might also have somewhat different age distributions.

Audacious Epigone said...

Ron,

This post might be of interest. Wordsum scores do increase with age up to a point, but around retirement, they start dropping again.

Jason Malloy said...

For those interested I've created a custom variable for the GSS called "IQ", which automatically transforms WORDSUM into an intelligence quotient. It's normed on whites from the 2008-2010 surveys, to account for secular changes.

Wordsum scores do increase with age up to a point, but around retirement, they start dropping again.

This may involve cohort issues as well. Here's the same chart from above, with the IQ variable. The Greatest Generation (1901-1924) shows declines which apparently begin in the 60s, but the Silent Generation (1925-1945) appears to have made gains even up beyond 78. We'll have to wait to see what happens with the boomers.

Jason Malloy said...

Here are the same Age/Cohort charts for Af-Ams and Mex-Ams.

TH said...

That's very interesting about Wordsum-IQ tending to generally rise with age, and also older generations have higher levels regardless. I guess I would have assumed that it rises until around 20-25, then levels off or gradually declines.

Generally, crystallized intelligence (i.e., accumulated knowledge such as vocabulary) tends to increase with age, while fluid intelligence (=novel problem solving ability) starts to decline as early as one's 20s.

Moreover, the Flynn effect seems to have little effect on crystallized intelligence tests, i.e., the amount of knowledge people have does not seem to have changed much over the last few generations, at least in America.

Anonymous said...

"That's very interesting about Wordsum-IQ tending to generally rise with age, and also older generations have higher levels regardless. I guess I would have assumed that it rises until around 20-25, then levels off or gradually declines."

Crystallized intelligence increases with age. Fluid decreases (after the mid 20s). Hence, IQ scores are age normed.

Anonymous said...

Crystallized intelligence increases with age. Fluid decreases (after the mid 20s). Hence, IQ scores are age normed.

The increase and decrease, also, are not equal in magnitude - crystallized intelligence only increases a little, while fluid decreases a lot.

Matt said...

Olave: "It's not clear exactly why US-born Hispanics are so much more crime prone, but the cultural conservative answer fits the facts. "

My theory would be that it is largely due to adolescence limited offending differences.

Migrants tend to be older than the populations they come from and go to.

Most offenders are adolescence limited.

This explains the criminality deflation in migrants relative to their children. It's not really to do with "Asian culture" or "Hispanic culture" which their kids lose, but just because, like Alex in A Clockwork Orange, they migrants are too old for crime.

Olave d'Estienne said...

Matt -

Your theory is plausible and simple. Sign me up.

For all I know, younger people crossing the border in from Mexico may actually immigrate (settle down--legally or illegally), much less often than older folks. As younger people they are probably more mobile (better at jumping fences) and less interested in steady work, thus more suitable for illegal activity beyond mere infiltration.

Anonymous said...

Migrants tend to be older than the populations they come from and go to.


I'm skeptical. Is that based on something or is it just your hunch?

Matt said...

I'm skeptical. Is that based on something or is it just your hunch?

I may not have phrased it as well as I might - I think the median and mean age is slightly older amongst migrants, but I'm not too sure and that's not precisely what I'm talking about either. What I'm really talking about is that there are comparatively few migrants who are aged in the peak offending years, compared to the native born.

Age structure of migrants in the EU -

http://epp.eurostat.ec.europa.eu/statistics_explained/index.php?title=File:Age_structure_of_the_national_and_non-national_populations,_EU,_2010_(1)_(%25).png&filetimestamp=20111125175202

http://migrationobservatory.ox.ac.uk/briefings/characteristics-and-outcomes-migrants-uk-labour-market

There's a big peak after 25, by which time adolesence limited aging should typically have ceased.

US:

http://pewresearch.org/pubs/261/who-are-the-immigrants

"The age structure of the foreign-born population was also markedly different from that of the native born. Only about 18% of the foreign-born population was under 25, compared with 37% among the native born. A majority of the foreign-born population (61%) was also concentrated in the prime working years, ages 25 to 54. By comparison, 41% of the native-born population was in this age group. The proportion 55 and older among the foreign-born population was only slightly smaller than among the native born (21% and 23% respectively).

Among the foreign born 25 and older, about 23% were high school graduates, compared with 31% among the native born. There were significant differences by region of origin, however. Among immigrants from South and East Asia, about 49% were college graduates or had advanced degrees, by far the largest of any group. About 42% of Mexicans and 32% of Central Americans had less than a ninth grade education."

Now, whether this is enough to explain the crime gap totally, where migrants are less criminal than their descendents, who although I didn't think about it in my original post, will actually tend to be more in the peak offending years than the native population, I don't know.

Audacious Epigone said...

Matt,

According to a recent CIS report, the median age for native-born Americans is 36.6 years compared to 42.4 years for immigrants. As you point out, that doesn't tell us much about distribution, but that's not a trivial difference even at the 50th percentile marks.

Ron Unz said...

Actually, you might want to separate your "Irish" category by Catholic vs. Protestant. A sharp-eyed commenter somewhere noticed that the "Irish" ethnicity seems to lump together the Irish-Irish (who are urban and have very high Wordsum-IQ) with the Scots-Irish (who are heavily rural and have very low Wordsum-IQ). When I did that stratification, I didn't bother with AGE or BORN, but I doubt that would impact the Irish too much, and my Irish Wordsum-IQ came out right near the top, very close to the Russian category.

To my mind, that's another bit of evidence for my Urban/Rural Hypothesis, since the Irish in Europe had a very low IQ when they were overwhelmingly rural, while in America, they're very heavily urban and have an extremely high Wordsum-IQ.

Anonymous said...

To my mind, that's another bit of evidence for my Urban/Rural Hypothesis, since the Irish in Europe had a very low IQ when they were overwhelmingly rural, while in America, they're very heavily urban and have an extremely high Wordsum-IQ.

Why on earth would it be better to cherry pick a particular group who happen to agree with your theory rather than look at the general structure as AE and Malloy have done?

Such behaviour seems tendentious and mendacious.

hbd chick said...

mr. epigone - please forgive me for haranguing ron on your blog:

" A sharp-eyed commenter somewhere...."

a sharp-eyed commenter somewhere? you know it was me, ron, and you know i wrote a whole post on it. you linked to it - twice. have you forgotten already?

Anonymous said...

Why on earth would it be better to cherry pick a particular group who happen to agree with your theory rather than look at the general structure as AE and Malloy have done?


Such behaviour seems tendentious and mendacious.


What makes it worse is Ron's dogged insistence that his ideas about Irish IQ are unquestionably correct, when in fact they are best described as "dubious".

Those Irish children who supposedly had an average IQ of 87 entered the Irish workforce in the late 1970's. By 1980 they were the age range 14 - 21. By 1990 they were 24 - 31 years of age.

How did the unfortunate Irish economy fare when this wave of semi-cretins entered the workforce? Between 1979 and 1990, the Irish economy grew at an average of 2.4% per year, outstripping the growth of the economies of virtually every other European county and the US.

The success of the Irish economy cannot be explained by Ron's answer - that later generations of Irish were smarter for some reason. Those later generations came too late to explain the rise in Ireland's standard of living.

Perhaps Ron believes that it was actually that very same generation of "87 IQ" people whose intelligence rose. That would answer the question of how people so stupid performed so well. And there appears to be nothing in the Unz conception of intelligence to preclude the IQ's of many individual people rising dramatically over time.

One you have rejected a genetic basis for IQ and embraced a social/cultural model, there is no reason to assume that IQ has to rise on a generational basis. You should be able to take a group of people with an average IQ of 85 who grew up in one social/cultural model, immerse them in a different social/cultural model, and see their IQ rise sharply and quickly.

Anonymous said...

Actually, you might want to separate your "Irish" category by Catholic vs. Protestant. A sharp-eyed commenter somewhere noticed that the "Irish" ethnicity seems to lump together the Irish-Irish (who are urban and have very high Wordsum-IQ) with the Scots-Irish (who are heavily rural and have very low Wordsum-IQ). When I did that stratification, I didn't bother with AGE or BORN, but I doubt that would impact the Irish too much, and my Irish Wordsum-IQ came out right near the top, very close to the Russian category.

To my mind, that's another bit of evidence for my Urban/Rural Hypothesis, since the Irish in Europe had a very low IQ when they were overwhelmingly rural, while in America, they're very heavily urban and have an extremely high Wordsum-IQ.




Your tendency to distort or overstate the facts continues unabated. Set aside the question of whether the rural Irish in Ireland once had a "very low IQ". The "extremely high wordsum IQ of urban Irish" in America is 101. That's not "extremely high".

On your blog you recently claimed that: .. the descendents of immigrants from Southern Italy, Ireland, Greece, and Yugoslavia hav(e) much higher IQs than Americans of British or Dutch ancestry.

The wordsum data does not support that conclusion. The wordsum IQ of the urban Irish is lower than the wordsum IQ of the urban English/Welsh and Scottish. All the evidence indicates that the descendents of immigrants from Ireland have slightly lower IQ's than the descendents of immigrants from Britain. It's a very small difference, but its there.

Anonymous said...

According to a recent CIS report, the median age for native-born Americans is 36.6 years compared to 42.4 years for immigrants.

A lot of "native born Americans" these days are the children of first generation immigrants, so that skews the numbers.

Anonymous said...

Need an edit or delete function ..

Rugby Fan said...

Poor old anonymous keeps getting himself in a twist at the slightest mention of the Irish.He must have had a run in with them at some stage of his bitter life.

mrcan said...

Urbanization has forced many formerly rural people into larger cities. The Irish-Catholic Americans have been heavily urbanized for a long time. Recent Irish-American arrivals from Appalachia for example, to an urban setting may have a disadvantage if they left rural poverty. Thus the combination of the two scores gives a inaccurate result. Separate Catholic-Irish (and their descendants regardless of religion / no religion) outcomes yield higher scores.