If IQ were skin color, how visible would the racial gap be?

I often describe the differences between groups in terms of standard deviations or Cohen's d, but maybe we can get a clearer picture of what these mean with a black-and-white example; namely, skin tone.

Skin color is a very noticeable racial difference and tends to dominate discussions of race even though it is just one characteristic. People focus on it because it is obvious, and I suspect race deniers like it because they can treat it as something superficial, something only shallow people would focus on. People like Dinesh D'Souza who I often agree with will say idiotic things like race is nothing but a "coat of paint." Never mind that skin is our largest organ and carries out many important functions, and that skin color appears to have important functions as well. And anyone who claims that aesthetics aren't important doesn't know human nature.

Anyway, General Social Survey interviewers were asked to rate the skin tone of respondents with a number ranging from 1 (lightest) to 10 (darkest). The lightest ethnic group is Americans of Danish descent with a mean of 1.33. (I'm weird because I'm all NW European and 1/8th Danish but look like an Arab). We'll use Danes as the comparison group. What I show below is the gap between the Danish mean and the mean for the focal group in terms of standard deviations:

Degree to which group is darker than Danish Americans (in sd units)

Blacks  2.2
West Indians  2.2
Asian Indians  1.3
Mexicans  0.8
Filipinos  0.8
American Indians  0.8
Puerto Ricans  0.7
Chinese   0.6
Japanese   0.6

Total Sample  0.6

Spanish   0.5
Arabs   0.5
Jews  0.3
Portuguese  0.2
Italians  0.2
Greeks  0.2

A gap of 2.2 standard deviations between Danes and blacks is so huge, it's the reason why people decided long ago to define groups as the opposite of each other: black and white. If a study reports a 2 sd gap, be impressed.

Another way of describing the gap goes like this: not a single black in the sample is lighter-skinned than any Dane, and 97% of Danes are lighter than 94.5% of blacks.

If you want a visual that matches the size of the black/white IQ gap (1 sd), think of the color gap between Danes and a typical Mexican and typical Asian Indian blended together--a very large difference. Another way of saying this is that the black/white IQ gap is only half the size of the color gap.

Even a difference that is only 10% of the black/Dane gap is still visually easy to detect: the Italian and Greek means are 0.2 standard deviations darker than Danes--0.2 considered a very small difference--but southern Europeans are known for being darker than northern Europeans. The higher Spanish mean is probably due to darker Hispanics claiming Spain as their homeland.

Which racial groups think science does more harm than good?

While I am not a science worshipper, I do think it offers answers to many of our problems, at least eventually. Unfortunately, science doesn't happen magically. It requires very talented people and a society with the means and culture to support scientific work. 

Americans vary a great deal in their attitudes toward science. The General Social Survey (GSS) asked participants the following: "How much do you agree or disagree that, overall, modern science does more harm than good?" Answers ranged from "strongly agree" (1) to "strongly disagree" (5), so higher scores indicate greater belief in science. Here are the means by ethnic group (sample size = 6,226; I included religious affiliation as well):

Mean belief that science is good

Finnish  4.16

Jewish  4.08

Russian  3.99

Japanese  3.95

Lithuanian  3.87

Hungarian  3.86  

Buddhist  3.84

Austrian  3.83

Greek  3.83

English/Welsh  3.82

Danish  3.81
Scottish  3.81

Swedish  3.80

Norwegian  3.78

No religion 3.78

Italian  3.76

Irish  3.72

French  3.71

Hindu  3.67

Orthodox Christian  3.65

German  3.64
Swiss  3.64
Asian Indian  3.61

Dutch  3.61
Spanish  3.61

Total Sample  3.60

Muslim  3.60

Catholic  3.59

Chinese  3.53

Arab  3.50

Immigrants  3.37

Mexican  3.26

Black  3.20

American Indian  3.00

West Indian  2.82

The pattern is clear: whites groups have much more faith in science than non-whites. Japanese Americans, a smart, successful group, are an exception.  Among whites, Finns and Eastern/Central Europeans tend to be the most positive about science. Among religions, Jews come out on top. The gap between Finns at number one and West Indians on the bottom is well over one standard deviation--a very large difference. 

I included immigrants as a group: they are toward the bottom of the list. If we want to have a country fully in support of science, our immigration trends are counterproductive. 

It looks like some of the explanation for the pattern we see is IQ, but I wouldn't be surprised if non-whites have a tendency to see science as an evil white thing. Think Tuskegee. It's BS, but this might help explain negative attitudes. 

Are the Scots-Irish leading the way toward race consciousness?

I'm thinking that one way to measure an ethnicity's meaningfulness as a group is the extent to which they vote as a bloc. For example, why do blacks, whether wealthy or poor, urban or rural, religious or not, vote overwhelmingly Democratic? Because race is very important for them, and the vast majority see the Democratic party as best serving their interests.

The numbers shown below were calculated like this: 1) Look to see whether more of the ethnic group voted for Hillary or Trump; 2) subtract from that number the percent of all Americans who voted for that candidate; and 3) give a positive sign for those skewed toward Hillary and a negative sign for those skewed toward Trump. I included major racial and ethnic groups, and religions as well since they can also be cohesive groups (General Social Survey, sample size = 1,128).

Voting bloc index

Asian  46.5
White Southern Baptists  -45.0
White--American only  -43.3
Black   41.3
Puerto Ricans  32.2
Protestant Irish  -25.2
Scottish  -23.1
Southern Baptist  -22.0
Jewish   18.9
Mexican  17.5
Catholic Irish  -16.2
German  -14.1
Italian  -10.9
Whites  -9.9
Scandanavian  -9.4
English/Welsh   -9.2
American Indian  8.4
Slavic (non-Jewish)  -8.4
Episcopalian  -7.3
Catholic  -3.5
United Methodist   -2.9

Not surprisingly, non-whites tend to be the most ethnocentric. Asians include people whose families came from China, Japan, and India--they score even higher than blacks.

Hispanics and Jews are in the mid-range, but the striking numbers are for: 1) white Southern Baptists; 2) whites who say their ethnicity is American--Scots-Irish are likely to say this; 3) the Protestant Irish; and 4) Scots. The numbers for white Southern Baptists and for "Americans" are second and third highest on the list (!) and the scores for the other two groups are higher than that of either Jews or Mexicans.

I interpret this as an inchoate sense of identity. For other white groups or religious denominations, there appears to be weaker group cohesiveness. It's higher than zero for all groups, but the Scots-Irish seem to be leading the way on a path toward race consciousness.

Data: Do Americans think genes or environment are more important?

Do Americans think people act like they do because of environment or genes?  The General Social Survey gave respondents four scenarios:

1) Carol is a substantially overweight White woman. She has lost weight in the past but always gains it back again.

2) David is an Asian man who drinks enough alcohol to become drunk several times a week. Often he can't remember what happened during these drinking episodes.

3) Felicia is a very kind Hispanic woman. She never has anything bad to say about anybody, and can be counted on to help others.

4) George is a Black man who's a good all-around athlete. He was on the high school varsity swim team and still works out five times a week.

Respondents were asked to estimate the influences of environment versus genes for each scenario. Answers ranged from 100% genes (scored as a 1) to 100% environment (scored as a 21). 50/50 would be scored as an 11.

I summed the scores for these four questions, and list the means for ethnic groups below (sample size = 1,842):

Mean environmentalism score

Asian Indians  56.7
Scottish  54.9
Czech  52.6
Italy  52.1 
Filipinos  50.9
American Indians  50.9 
Norwegians  50.9
Russians  50.4
Chinese  50.2
Polish  50.1
English/Welsh  49.9
Jewish 49.4
Irish  49.3

All Americans  49.3

German 49.0
Dutch  47.9
Swedish  47.3
Mexican  47.1
Puerto Ricans  46.3
Danish  46.1
Blacks 45.2
French  44.2
Spanish  43.9

The mean score for all Americans indicates that the average person thinks the traits described are 55% due to environmental influences, and 45% due to genes.  This is not far from the truth as indicated by genetic research. Americans seem to be ignoring their environment-is-all social science teachers. On the other hand, I'm sure by "environment" most people are thinking of experiences which siblings share -- families, schools, etc -- but research clearly shows that these factors are not important.

The highest scoring group, Asian Indians, thinks the traits are roughly 65% environment.  The lowest group, people of Spanish descent, put the estimate at about 50/50.  Hispanics, in general, tend to think genes are more important than other groups. This tendency is also true for blacks, but other non-whites -- Chinese, Filipinos, Asian Indians, and American Indians -- think the environment is more important.  It's surprising that several poor minority groups think genes are so important.  The gap between Asian Indians and the Spanish is eight-tenths of a standard deviation -- a large difference.

Data: White Americans seem fairly homogeneous (so far)

I recently showed that white Americans whose families are from east of the Hajnal line tend to be more clannish than those from west of the line.

In addition to the issue of clannishness, hbdchick also presented evidence that Western Europeans have a number of positive traits at higher levels than Eastern Europeans. Do these differences hold up in the US?

So far I've seen little evidence that they do. Their mean IQs are not different, nor are their chances of an arrest, nor their frequency of donating blood. The one difference I've detected so far is that W. Europeans are slightly less approving of cheating on their taxes--not even 1/5 of a standard deviation difference.

At least so far, American whites seem fairly homogeneous. European differences seem to have faded.

White voting by ethnic group

According to this video by Stefan Molyneux, white people tend to prefer smaller government. I would agree that whites might be unusual in wanting to protect the individual against the state, but Europeans and all those white Bernie Sanders voters are certainly comfortable with confiscating lots of your money.  Whites seem to be more live-and-let-live, but many of them want to soak the rich just like everyone does.

Whites are so diverse, let's get a little more precise and see which American white groups are majority Republican currently. Based on GSS data:

Percent voting for Romney in 2012

Dutch  76.2
English/Welsh  56.5
Scottish  54.5
French  52.2
German  52.0

Polish 42.4
Irish  42.2
Swedish 42.1
Italians  36.2
Jewish  32.4

Protestants and groups that have lived longer in the United States are more likely to vote Republican. Folks like me (English in my case) might feel a stronger connection to the country's traditions than more recent groups.

I suspect we would be more collectivist like Europe if we were a racially more homogenous country. If the recipients of government assistance look like they could be a cousin, we might be more in favor of it. But if we drum up an image of someone unconnected to ourselves when we think of welfare, we think, why help them? We sense that we are sharing with a competitor.

Prussian Americans most likely to serve in the U.S. military

Memorial Day is a good day to look at military data, and it happens to be an opportunity to echo Razib's call to mend, not end, the American Community Survey. The survey asked respondents their ancestry, and if they are now serving or ever served in the military. I limited the sample to native-born Americans and to people ages 20 to 39 in order to focus on a single cohort. I list the percentages who answered yes by their ancestry (sample size = 1.74 million) and since there are as many categories as countries in the world I only list groups with the highest and lowest numbers. The mediocre middle is left out.


Percent serving in the military now or in the past

Above-average groups

Prussian 12.7
South African 12.3
Guamanian 12.2
Germans from Russia 11.4
Okinawan 11.1
Panamanian 10.2
West Indian 9.9
Belizean 9.8
Indonesian 9.8
Scottish 9.4
Dutch West Indies 8.7
Pacific Islander 8.6
Sicilian 8.6
Scots Irish 8.1
Australian 8.0
African 8.0


Below-average groups

Haitian 3.8
Turkish 3.7
Arab 3.4
Egyptian 3.3
Vietnamese 3.0
Israeli 3.0
Iranian 3.0
Armenian 2.8
Palestinian 2.6
Chinese 2.4
Hmong 2.0
Asian Indian 1.9
Taiwanese 1.8
Pakistani 1.3
Macedonian 1.2
Iraqi 0.0

It's great that Prussians come out on top, but I suspect that some military folks are being selective in which part of their ancestry they identify with. The high-serving groups are an interesting assortment. Using GSS data, I reported in an earlier post that West Indians like Colin Powell are more likely to serve than most groups. Pacific Islanders have high numbers as do your well-known tough guys: the Scots Irish and Sicilians.

On the low end, you have Asians and Middle-Easterners. It looks like Muslims (and Israelis for that matter) don't like to serve, but does anyone know the religion of most American Indonesians?  Are they Muslims?

Also--Mexican-Americans, contrary to the claims of some, do not serve in high numbers. Between 4.4 and 5.3 percent serve or have served (depending on whether you focus on the "Mexican" or "Mexican-American" category). That's below the national average of 6.3 percent.

The monster ethnic group

According to the 2010 Census, Hispanics in America have now surpassed 50 million. That is 16.1 percent of the total population. I've listed the 2000-2008 GSS percentages for large ethnic groups (sample size = 11,840) :

Percent of total U.S. population

Hispanic 16.1 (2010 Census)
German 15.1
Black 14.3
Irish 12.3
English/Welsh 11.4
Italian 5.5
American only 3.0

Hispanics are now the largest group. Demographically speaking, America is more Hispanic than it is German, black, Irish, English, etc. On the other hand, "Hispanic" is a contrived category. If we focus on Mexican Americans only-- a real ethnic group--they are two-thirds of Hispanics which would put them currently at 10.6 percent of the U.S. population. That is larger than any other ethnic group except for Germans. And as big as they are, Germans do not stick out in American society; Mexicans do.

Demographers project that Hispanics will be one-third of the population by 2050. (That will only happen if we let it.) Considered as an ethnic group, they will dwarf every other ethnicity. They might become the monster ethnic group, but then again, their growing presence, along with that of other non-white immigrant groups, might forge whites into a real, monster ethnic group.

More on ethnicity and voting among whites

In the comments of the last post, The Cold Equations raises a good point that region should be controlled in a model of how whites vote. Here are the results with region added to the model (all the regions are being compared with New England, chosen as the reference group because it is a particularly blue region):

Logistic regression coefficients

English .39*
Scottish .16
Scots Irish .17
Dutch 1.18*
German .57*
Swedish .35
Income .09*
Town size -.01
Mid-Atlantic .27
East North Central .61*
West North Central .35
South Atlantic .95*
East South Central .84*
West South Central .1.28*
Mountain .75*
Pacific .28

*statistically significant


Even after adjusting for region (and income and population size), Americans of English/Welsh, Dutch, and German descent are significantly more likely to vote Republican, compared to other whites. Implicit ethnicity at work? 

White ethnic groups more likely to vote Republican than whites in general

Looking at GSS data, I noticed that there were a few white ethnic groups who were more likely than whites in general to vote for Bush in 2004 (the most recent year available): English/Welsh, Scottish, Scots-Irish, Dutch, German, and Swedish. Why?

I checked to see if perhaps income or small-town residency might explain the tendency. Here are the logistic regression coefficients (sample size = 1,773):

Logistics regression coefficients

English .53*
Scottish .22
Scots Irish .30
Dutch 1.23*
German .62*
Swedish .43
Income .10*
Town size -.01

*statistically significantly

Even after adjusting for income and population, Americans of English, Dutch, and German descent are more likely than other whites to vote Republican. (All groups have positive coefficients). I wonder if these people tend to be attracted to conservatism because it is more backward-looking, and many from the groups have American ancestors that go way back. They might feel a stronger connection because of having deep American roots to the Founding Fathers who are associated with limited government and conservative values.  Their voting tendencies might be subconsciously influenced by their ethnicity. One the other hand, perhaps its just the default position for Americans who have been completely stripped of ethnicity.

America still seems to do a good job of winning over its people

MIDUS Study participants were asked the following: "How closely do you identify with being an American, in the sense of being a U.S. citizen?" Answers ranged from "not at all closely" (1) to "very closely" (4). I calculated the means for each ethnic group with at least 30 respondents:
 
Mean American identity score (N = 4,561)
 
Lebanese 3.86 (n = 14)
Polish 3.86
Amerindian 3.86
Norwegian 3.86
Mexican 3.85
Czech 3.81
Irish 3.81
Swedish 3.80
French 3.79

All Americans 3.79

German 3.79
England 3.78
Italian 3.78
Russian 3.77
Scottish 3.77
Black 3.75

All groups score very high, with no significant differences among them. I included Lebanese Americans even though there were only 14 respondents. Most individuals from all ethnic groups closely identify with being American. This is true even of groups with many immigrants and connections to the Old Country (e.g., Mexican Americans) and for groups that were mistreated historically (i.e., Amerindians and blacks). Surprisingly, somehow, America still seems to do a good job of winning over its people. God bless her.

Race, ethnicity, and attitudes toward inter-ethnic marriage

The Midlife Development in the United States Study asked Americans how important it is for members of their ethnic group to marry inside the group. Here are the percentages who said "very important" or "somewhat important" (N = 4,881): 

Jewish 47.4
Asian 42.3
Amerindian 37.3
Black 34.6
Hispanic 33.8
English 28.4
French 19.7
Italian 19.1
Scottish 18.5
Irish 18.3
Polish 16.3
German 16.0
Swedish 15.8
Norwegian 15.2

Non-whites and especially Jewish Americans have high numbers. 

The ethnicity of scientists

The General Social Survey asks respondents (American) about occupation and ethnic background. Here are the percent of all natural scientists in each ethnic group (those who are less than 5% of the total are excluded):

The ratio refers to the share of natural scientists compared to the ethnic group's share of the total population. You can see that Americans of English, Irish, Italian and especially Scottish descent are natural scientists in disproportionate numbers.

The picture changes for mathematical and computer scientists. Americans of Polish and especially Asian Indian descent are more likely than other groups to have this kind of job. 

The only two large American groups that did not each produce at least 5 percent of natural, mathematical, or computer scientists were blacks and Mexican Americans. (I was surprised that Jews were only 1.0 percent of the natural scientists and 2.9 percent of the math/computer scientists--small sample size does not help).  

More on Jewish ethnocentrism

In the comments in the last post, some readers contended that Jews are not ethnocentric. Using the same question I used in the comments section--how important is your ethnicity for your sense of who you are--I calculated the mean score for all American ethnic groups with at least 30 people. Answers ranged from "very important" (1) to "not at all important" (4), so high scores indicate low ethnocentrism.

The table shows that of the 13 groups big enough to measure, Jewish Americans are the third most tribalist group, behind only blacks and Mexicans Americans. The difference between Jews and those of English descent (the comparison group) is statistically significant at the 95 percent confidence level. The standard deviation is 1.14, so the English/Jewish gap is slightly over 1/2 of a standard deviation.

Irish and Italian Americans, thought by some in the previous post to be as ethnocentric as Jewish Americans, do not differ significantly from Americans of English descent. The gap between Jews and the Irish is 39% of one standard deviation; the Jewish/Italian gap is 35% of one SD.

Jews are significantly less ethnocentric than blacks, but their mean does not differ significantly from that of Mexican-Americans (p < .05, two-tail test).  (I wouldn't be surprised if this comparison would become significant if N were increased).

Feelings toward Jews: GSS respondents (Americans) were asked how warmly and favorably do they feel toward Jews on a scale from 0 to 100, 100 being the warmest. Here are the means by ethnic group (with Jews removed):


Mean warmth toward Jews (N =3,499)

Czech 66.6
Italian 64.5
Irish 63.7
Polish 63.6
English/Welsh 63.4
Scottish 62.9
Dutch 62.5
German 62.3
Danish 61.9

All Americans 61.6

Norwegian 60.5
Swedish 59.5
Black 59.5
French 58.2
Puerto Rican 57.4
Japanese 56.8
Spain 56.4
Mexican 54.3*
Amerindian 53.8
Greek 53.1
Russian 52.3
Asian Indian 52.2*
Swiss 51.9
Chinese 42.6*
Arab 34.2

* p < .05, two-tail test, compared with all Americans


While many Jews seem to feel an affection for racial minorities, the favor is not generally returned. Immigrants from Mexico, India, China, and Arabic countries are moving the country in a more anti-Jewish direction.

Ethnicity and IQ heterogeneity: I was curious about IQ diversity among American ethnic groups. Selecting GSS groups with at least 100 respondents, I calculated the coefficient of variation, a normalized measure of dispersion, for Wordsum scores. Here are the numbers:


Coefficient of variation

Puerto Rican .491
Mexican .415
Black .381
Amerindian .359
Spain .338
Dutch .315
Polish .312
French .307
German .306
Italian .304
Irish .302
Scottish .294
English/Welsh .289
Czech .281
Swedish .280
Norwegian .280
Danish .270
Russian .264
Jewish .239

Racially mixed groups are the most IQ-diverse, and Scandinavians and Jews are the most homogeneous (many of those Russians are Jews).

With Mexican Americans especially, we should avoid talking about them like they are a monolithic group.

Candor!

THIS is the type of candor about ethnic groups I was talking about! Fun stuff. Never mind whether it's accurate or not. It's honest.

UPDATE: Skip to about 8:45 if you only have a minute. If you like that and have time, watch parts 2 and 3.

Honest talk about ethnic groups: CNN can't seem to get enough of the Holocaust museum shooting. I've been learning all about the layout of the building, the details of the incident, and have been treated to White Nationalism 101. If folks are concerned about this type of violence, I recommend we ignore those who suggest we look to Europe and its punishment of bad speech.

How about a fresh idea? The culture offers Americans who are curious about Jews two options: Stormfront or nothing. The topic has been driven underground, and nothing but fungus grows in the dark.

Growing up in a small town, I learned all sorts of things about blacks and Hispanics, but the only thing I was ever told about Jews was that they are cheap--one, shallow stereotype--and that millions were killed by Nazis. And the few words that were uttered were in hushed tones.

The most basic facts are that American Jews are an incredibly accomplished group who are about as ethnocentric as Mexican Americans. Scary as hell to read in print, isn't it? My response to such facts is, so what? I've already expressed my disapproval of ethnocentrism and my wish that Jews found conservatism more persuasive (their embrace of abortion bugs me) but these folks have enriched the country in many ways, and even if they hadn't, they are citizens who are entitled to believe whatever they want.

All the whispers and taboo only feed paranoid thinking. How about some honest talk about ethnic groups? Let's try candor, for once.

Stop hatin', start participatin': Not all homosexuals are white! My dad knows I analyze data and asked me if it was true that there are very few homosexual non-whites. He's picked up the stereotype that all gays are Caucasian. (If I were asked, I'd say English). I told him that the prevalence of homosexuals is at least as high among blacks and Hispanics. According to the General Social Survey, it's 3.0% of white men, 5.4% of Hispanics, and 3.8% of blacks. I suggested that non-white homosexuals are more invisible since their ethnic groups are small, and there is less acceptance of the orientation among these groups.

According to The National Longitudinal Study of Adolescent Health, 7.0% of white adolescent boys say that they have been romantically attracted to another boy. The corresponding number for blacks is 8.2% and 9.0% for Hispanics (N = 3,101). Interesting how the ranking is the same for the two surveys.

Perceived weight by ethnic group:  General Social Survey interviewers recorded perceptions of their respondent's weight, ranging from "below average" (=1) to "considerably above average" (=4).  In contrast to scales, measuring weight in this way allows us to see how people are perceived by others.  I calculated the means for all ethnic groups with at least 30 respondents in the sample:

Mean weight score

 

Mexican 2.37

Blacks 2.33

Italian 2.32

French 2.32

American Indian 2.29

English/Welsh 2.22

USA 2.21

Irish 2.23

German 2.19

Polish 2.14

Scottish 2.10

Jewish 2.02

Chinese 1.83

SD 0.63

As we saw in a post on scale-measured weight, Mexican Americans are the heaviest group. The objective and subjective match up.  Blacks, the other large NAM, are next; American Indians are fifth.  Next heaviest are those of southern European descent.  Northern and eastern Europeans are next, followed by Jews.  Chinese Americans--no surprise--are the skinniest.  The pattern follows social class ranking. The difference between Mex-Ams and Chinese folks is almost one standard deviation.