Has Ron Unz Refuted “Hard Hereditarianism”?

In the current issue of The American Conservative, Ron Unz attempts to take on “hard hereditarianism”: the belief that observed IQ differences between ethnic groups are largely genetically based. [Race, IQ, and Wealth | What the facts tell us about a taboo subject, July 18, 2012]

Paradoxically, Unz relies on Richard Lynn and Tutu Vanhanen’s IQ and the Wealth of Nations, which is usually (and only) cited by race realists. But he analyzes the data presented in the book and comes to the conclusion that those who worry that mass Hispanic immigration is lowering the American national IQ are misguided.

Before getting into the substance of Unz’s piece, it is worth pointing out how much has changed regarding the race and IQ debate.

Two decades ago, Stephen Jay Gould argued against the existence of both race and intelligence. Despite devastating criticism, Gould has been widely and uncritically cited by anti-hereditarianism writers ever since.

But today, those who argue that racial intelligence gaps are not genetic are generally more honest. They have mostly stopped using postmodern obfuscation to reject even the possibility of a link between ethnicity and IQ.

For example, University of Michigan Professor Richard Nisbett [Email him] in his book Intelligence and How to Get It, accepts that IQ predicts life outcomes and acknowledges that there is no a priori reason to assume that all ethnic groups have, on average, equal cognitive ability. He instead tries to prove that they do through evidence.

(Charles Kenny’s recent attack on John Derbyshire, Richard Lynn and IQ in general in Foreign Policy is an embarrassing exception to this laudable trend. But read the derisive comment thread! [Dumb and Dumber |Are development experts becoming racists? April 30, 2012])

In his American Conservative piece, Unz notes that Lynn and Vanhanen’s data reported wide variations in IQ between genetically indistinguishable populations. He claims to have found a pattern that indicates that greater wealth causes higher IQ, rather than vice versa.

For example, the Greek IQ increased by 7 points in less than two decades and Ireland saw an 11-point rise between 1972 and 1979. These increases in IQ both occurred during periods of economic expansion. Similarly, we find IQs in the low 90s for countries that were at the time of testing dirt poor and Communist such as Bulgaria and Romania. Wealthier Communist states scored in the high 90s:

Countries with Large IQ Increases

Country

IQ at Point 1

IQ at Point 2

IQ Change

Annual GDP Per Capita Increase

Time Gap Between Tests (years)

East Germany

90

99

+9

$769 (using Czech figures)

11

Greece

88

95

+7

$6047

18

Ireland

87

98

+11

$1191

7

 

Of course, these IQ scores are necessarily based on only one or two studies at each point in question—Lynn and Vanhanen simply used whatever tests were available. So any such anomalies could be the result of random noise in the data.

One indication that this is the case: in Sweden we find an increase in IQ of 7 points over an 11-year period, even though at Point 1 Sweden had already had a high GDP per capita.

Interestingly, we also see countries that have become richer while at the same time losing IQ points:

Countries with Large IQ Drops

 

Country

IQ at Point 1

IQ at Point 2

IQ Change

GDP Per Capita Increase

Time Gap Between Tests (years)

Israel

97

90

-7

$5276

14

Poland

106

92

-14

$561

10

Portugal

101

88

-13

$1073

8

France

99.5 (average of two studies)

94

-5.5

$9630

17

 

So if Unz had wanted to make the argument that greater wealth actually reduces IQ, he would also have been able to find a basis for this in Lynn and Vanhanen’s data.

Of course, the idea that getting richer makes you dumber seems absurd, and contradicts everything we know about the Flynn Effect.

And it’s not the case that the poorer countries studied by Lynn and Vanhanen only saw increases in IQ over time. At the time of each country’s first testing, Poland and Portugal both had lower annual GDPs per capita than Ireland. But Unz finds Ireland’s later rise in IQ significant, while the declines in Poland and Portugal apparently need no explanation.

All of this indicates that the wild variations in scores we see are random noise. You can only find a pattern by cherry-picking the data.

Unz also claims to find evidence consistent with the “wealth causes IQ” hypothesis in European Communist countries that did not see an IQ swing: the poorer ones have lower IQ than the richer ones.

Communist Countries without an IQ Swing

Country

IQ

Annual GDP Per Capita

Year of Testing

Bulgaria

94

$1,315

1979

Romania

94

$1,364

1972

Bulgaria

91

$1,428

1982

Croatia

90

$2,324 (1969)

1952

Slovakia

96

$3,368

1983

Hungary

99

$3,767

1979

Czech Republic

98

$3,791

1979

Czech Republic

96

$4,262

1983

 

Here, we do indeed seem to see a pattern. The states with annual GDPs per capita over $3,000 all score over 95, while those below that threshold have IQs between 90 and 94.

However, if this pattern reflects some underlying reality, it contradicts other data presented in favor of Unz’s theory. At the time that their IQs were supposedly depressed, Greece and Ireland had GDPs well above the $3,000 per annum threshold; Irish citizens were over two and half times richer.

It seems hard to believe that the IQ of the Irish was depressed because they only had an annual GDP per capita of $7,790, while the Hungarians, Slovaks, and Czechs were able to realize their full potential by becoming no more than half as wealthy.

“Soft hereditarianism” looks good only if we selectively focus on the poorest countries of all: those whose citizens earned less than $2,500 a year at the time of testing. There are five studies from these nations, and four of them show average scores of 94 or below (the exception is Poland’s score of 106). This does suggest that, at the worst levels of poverty that Europe has seen in the post-World War II era, IQ can be depressed significantly.

Still, the data set is small. And after a country reaches an annual GDP per capita of about $2,500-$3,000, it is quite difficult to find any patterns at all in any of Lynn and Vanhanen’s data.

Furthermore, even if we accept that the numbers show extreme deprivation depresses IQ, it is another stretch to apply this finding to the current U.S. immigration debate.

For decades, Hispanic Americans have been as wealthy as some of the richest countries studied in IQ and the Wealth of Nations. Thus the U.S. Census measures annual “total money income per household member” by race, which appears to be the equivalent to an annual GDP per capita figure.[Historical Income Tables: Households (table H-15)] In 1980, Hispanics earned $7,720 per household member, and in 1990 that number increased to $14,197. This is obviously well beyond the level of income where we find any kind of support for the “wealth increases IQ” hypothesis.

More directly pertaining to his point, Unz cites blogger Inductivist’s Wordsum analysis, The Flynn Effect among Mexican Americans [February 10, 2008] relying on data from the 2000s, that shows a native-born Mexican American score that corresponds to a 95 IQ. As the same group’s Wordsum-IQ was 84-85 in the early 1980s, Unz concludes that

almost two-thirds of the IQ gap between American-born Mexican-Americans and whites disappeared in two decades, with these results being based on nationally-representative American samples of statistically significant size.

Unfortunately for this claim, individual data points on Hispanic American IQ have fluctuated wildly in the past. In Race Differences in Intelligence , Richard Lynn lists 20 Hispanic American IQ studies going back to the 1920s (pages 164-165). A few have shown scores in the mid to high 90s.

Hispanic American IQ Scores

Year

IQ

Sample Size

1974

95

608

1986

94

434

1986

99

111

 

But all 17 other studies, going up to 2001, show scores between 83 and 93.

In other words, results from IQ studies, even those conducted in the United States, tend to show a great deal of random fluctuation. While the recent Wordsum scores are encouraging, we need more data points.

One possibility: the SAT, which has a 0.81 correlation with IQ scores (By comparison, Unz cites a 0.71 correlation between Wordsum and IQ.).[Scholastic Assessment or g? The relationship between the Scholastic Assessment Test and general cognitive ability, by Frey and Detterman, Psychological Science, June, 2004] From 1980 to 2010, the SAT gap between whites and Hispanics has held steady at between 0.6 and 0.8 standard deviations.[The Unsilenced Science: Racial Amplitudes of Scholastic Aptitude, April 11, 2012]

Of course, using the SAT data may be more problematic than Inductivist’s Wordsum scores, since the latter was able to look at only native-born Mexican Americans whereas the former must include immigrants.

But, presumably, a higher portion of Hispanic Americans taking the SAT in 2010 would be native-born compared to 1980. If their IQs were rising, the U.S. born Mexican Americans would be expected to bring up the SAT scores of all Hispanics. Yet, over thirty years, we do not see any convergence between Hispanic and white SAT results.

(And it should also be noted that a higher percentage of Hispanics drop out of high school, which means that some of the lowest performing students never get around to taking the SAT).

Ron Unz is to be commended for soberly looking into such an important and much neglected issue. It would indeed be ironic if someone used Lynn and Vanhanen’s book to refute “hard hereditarianism.” Despite Unz’s triumphant tone, however, his analysis comes up short.

David Sanders [email him] is the name a young American academic is living under.

Private: Has Ron Unz Refuted “Hard Hereditarianism”?

In the current issue of The American Conservative, Ron Unz attempts to take on “hard hereditarianism”: the belief that observed IQ differences between ethnic groups are largely genetically based. [Race, IQ, and Wealth | What the facts tell us about a taboo subject,  July 18, 2012]

Paradoxically, Unz relies on Richard Lynn and Tutu Vanhanen’s IQ and the Wealth of Nations, which is usually (and only) cited by race realists. But he analyzes the data presented in the book and comes to the conclusion that those who worry that mass Hispanic immigration is lowering the American national IQ are misguided.

Before getting into the substance of Unz’s piece, it is worth pointing out how much has changed regarding the race and IQ debate.

Two decades ago, Stephen Jay Gould argued against the existence of both race and intelligence. Despite devastating criticism, Gould has been widely and uncritically cited by anti-hereditarianism writers ever since.

But today, those who argue that racial intelligence gaps are not genetic are generally more honest. They have mostly stopped using postmodern obfuscation to reject even the possibility of a link between ethnicity and IQ.

For example, University of Michigan Professor Richard Nisbett [Email him]  in his book Intelligence and How to Get It, accepts that IQ predicts life outcomes and acknowledges that there is no a priori reason to assume that all ethnic groups have, on average, equal cognitive ability. He instead tries to prove that they do through evidence.

(Charles Kenny’s recent attack on John Derbyshire, Richard Lynn and IQ in general in Foreign Policy is an embarrassing exception to this laudable trend. But read the comment thread!) [Dumb and Dumber |Are development experts becoming racists? April 30, 2012

In his American Conservative piece, Unz notes that Lynn and Vanhanen’s data reported wide variations in IQ between genetically indistinguishable populations. He claims to have found a pattern that indicates that greater wealth causes higher IQ, rather than vice versa.

For example, the Greek IQ increased by 7 points in less than two decades and Ireland saw an 11-point rise between 1972 and 1979. These increases in IQ both occurred during periods of economic expansion. Similarly, we find IQs in the low 90s for countries that were at the time of testing dirt poor and Communist such as Bulgaria and Romania. Wealthier Communist states scored in the high 90s:

Countries with Large IQ Increases

Country

IQ at Point 1

IQ at Point 2

IQ Change

Annual GDP Per Capita Increase (dollars)

Time Gap Between Tests (years)

East Germany

90

99

+9

$769 (using Czech figures)

11

Greece

88

95

+7

$6047

18

Ireland

87

98

+11

$1191

7

 

Of course, these IQ scores are necessarily based on only one or two studies at each point in question—Lynn and Vanhanen simply used whatever tests were available. So any such anomalies could be the result of random noise in the data.

One indication that this is the case: in Sweden we find an increase in IQ of 7 points over an 11-year period, even though at Point 1 Sweden had already had a high GDP per capita.

Interestingly, we also see countries that have become richer while at the same time losing IQ points:

Countries with Large IQ Drops

 

Country

IQ at Point 1

IQ at Point 2

IQ Change

GDP Per Capita Increase

Time Gap Between Tests (years)

Israel

97

90

-7

$5276

14

Poland

106

92

-14

$561

10

Portugal

101

88

-13

$1073

8

France

99.5 (average of two studies)

94

-5.5

$9630

17

 

So if Unz had wanted to make the argument that greater wealth actually reduces IQ, he would also have been able to find a basis for this in Lynn and Vanhanen’s data.

Of course, the idea that getting richer makes you dumber seems absurd, and contradicts everything we know about the Flynn Effect.

And it’s not the case that the poorer countries studied by Lynn and Vanhanen only saw increases in IQ over time. At the time of each country’s first testing, Poland and Portugal both had lower annual GDPs per capita than Ireland. But Unz finds Ireland’s later rise significant, while the declines in Poland and Portugal apparently need no explanation.

All of this indicates that the wild variations in scores we see are random noise. You can only find a pattern by cherry-picking the data.

Unz also claims to find evidence consistent with the “wealth causes IQ” hypothesis in European Communist countries that did not see an IQ swing: the poorer ones have lower IQ than the richer ones.

Communist Countries without an IQ Swing

Country

IQ

Annual GDP Per Capita

Year of Testing

Bulgaria

94

$1,315

1979

Romania

94

$1,364

1972

Bulgaria

91

$1,428

1982

Croatia

90

$2,324 (1969)

1952

Slovakia

96

$3,368

1983

Hungary

99

$3,767

1979

Czech Republic

98

$3,791

1979

Czech Republic

96

$4,262

1983

 

Here, we indeed do seem to see a pattern. The states with annual GDPs per capita over $3,000 all score over 95, while those below that threshold have IQs between 90 and 94.

However, if this pattern reflects some underlying reality, it contradicts other data presented in favor of Unz’s theory. At the time that their IQs were supposedly depressed, Greece and Ireland had IQs well above the $3,000 per annum threshold; Irish citizens were over two and half times richer.

It seems hard to believe that the IQ of the Irish was depressed because they only had an annual GDP per capita of $7,790, while the Hungarians, Slovaks, and Czechs were able to realize their full potential by becoming no more than half as wealthy.

“Soft hereditarianism” looks good only if we selectively focus on the poorest countries of all: those whose citizens earned less than $2,500 a year at the time of testing. There are five studies from these nations, and four of them show average scores of 94 or below (the exception is Poland’s score of 106). This does suggest that, at the worst levels of poverty that Europe has seen in the post-World War II era, IQ can be depressed significantly.

Still, the data set is small. And after a country reaches an annual GDP per capita of about $2,500-$3,000, it is quite difficult to find any patterns at all in any of Lynn and Vanhanen’s data.

Furthermore, even if we accept that the numbers show extreme deprivation depresses IQ, it is another stretch to apply this finding to the current U.S. immigration debate.

For decades, Hispanic Americans have been as wealthy as some of the richest countries studied in IQ and the Wealth of Nations. Thus the U.S. Census measures annual “total money income per household member” by race, which appears to be the equivalent to an annual GDP per capita figure.[Historical Income Tables: Households (table H-15)] In 1980, Hispanics earned $7,720 per household member, and in 1990 that number increased to $14,197. This is obviously well beyond the level of income where we find any kind of support for the “wealth increases IQ” hypothesis.

More directly pertaining to his point, Unz cites blogger Inductivist’s Wordsum analysis, The Flynn Effect among Mexican Americans [February 10, 2008]  relying on data from the 2000s, that shows a native-born Mexican American score that corresponds to a 95 IQ. As the same group’s Wordsum-IQ was 84-85 in the early 1980s, Unz concludes that

“almost two-thirds of the IQ gap between American-born Mexican-Americans and whites disappeared in two decades, with these results being based on nationally-representative American samples of statistically significant size.”

Unfortunately for this claim, individual data points on Hispanic American IQ have fluctuated wildly in the past. In Race Differences in Intelligence , Richard Lynn lists 20 Hispanic American IQ studies going back to the 1920s (pages 164-165). A few have shown scores in the mid to high 90s.

Hispanic American IQ Scores 

Year

IQ

Sample Size

1974

95

608

1986

94

434

1986

99

111

 

But all 17 other studies, going up to 2001, show scores between 83 and 93.

In other words, results from IQ studies, even those conducted in the United States, tend to show a great deal of random fluctuation. While the recent Wordsum scores are encouraging, we need more data points.

One possibility: the SAT, which has a 0.81 correlation with IQ scores (By comparison, Unz cites a 0.71 correlation between Wordsum and IQ.).[Scholastic Assessment or g? The relationship between the Scholastic Assessment Test and general cognitive ability, by Frey and Detterman, Psychological Science, June, 2004][4] From 1980 to 2010, the SAT gap between whites and Hispanics has held steady at between 0.6 and 0.8 standard deviations.[The Unsilenced Science: Racial Amplitudes of Scholastic Aptitude, April 11, 2012] 

Of course, using the SAT data may be more problematic than Inductivist’s Wordsum scores, since the latter was able to look at only native-born Mexican Americans whereas the former must include immigrants.

But, presumably, a higher portion of Hispanic Americans taking the SAT in 2010 would be native-born compared to 1980. Yet we have seen no convergence between their scores and those of whites.

(And it should also be noted that a higher percentage of Hispanics drop out of high school, which means that some of the lowest performing students never get around to taking the SAT).

Ron Unz is to be commended for soberly looking into such an important and much neglected issue. It would indeed be ironic if someone used Lynn and Vanhanen’s book to refute “hard hereditarianism.” Despite Unz’s triumphant tone, however, his analysis comes up short.

David Sanders [email him] is the name a young American academic is living under.