Below are three pie charts (click for larger view).  The first shows the racial distribution of homeless Americans.  The second shows the racial distributions of all Americans.  And the third shows the racial distribution of billionaire Americans.


It’s fascinating how well money mirrors IQ.  For example, blacks (who average the lowest IQs) are nearly half of the homeless, but only 0.25% of the 400 richest Americans.  And the only black on the Forbes 400 has arguably the largest black brain size in America, if not the world.

By contrast, Jews (who average the highest IQs) are only 2% of America, but 36% of American billionaires.

Now on a scale where U.S. whites average 100 (SD = 15), the U.S. population as a whole now has a mean IQ of about 96 (SD=15.8).  In a previous post, we imagined an incredibly evil IQ test that had the same distribution of scores as other IQ tests, yet your score correlated perfectly with your race. Below are the IQs the races would have if IQ and race correlated perfectly in America.  In parentheses are the IQs these races actually do have:

Black Americans: IQ 70 (IQ 85)
Native Americans: IQ 72 (IQ 86)
Hispanic Americans: IQ 79 (IQ 89)
Southeast Asian Americans: IQ 86 (IQ 92)

ArabAmericans: IQ 106??(IQ 100??)
White Americans: IQ 106 (IQ 100)
East Asian Americans: IQ 115 (IQ 104)
Indian Americans: IQ 130 (IQ 110)
Jewish Americans: IQ 130 (IQ 110)

Since the homeless are 45% black, 5% Native American, 12% Hispanic, perhaps 1% Southeast Asian, 35% white, perhaps 1% East Asian, and perhaps 1% South Asian,  their average IQ on this evil IQ test would be:

0.45(70) + 0.05(72) + 0.12(79) + 0.01(86) + 0.35(106) + 0.01(115) + 0.01(130) = 85

An IQ of 85 would be the average IQ of the homeless if race and IQ correlated perfectly, but in reality, race and IQ only correlate 0.43 in contemporary America, and race and general intelligence (g) only correlate 0.48.

Because life is partly an IQ test, the homeless are selected for low general intelligence, and thus regress upward, to the U.S. mean of 96, on crude proxies for g, like race.  So since the homeless score IQ 85 on this racist IQ test; 11 points below the U.S. mean of 96, we must divide 11 by the 0.48 g loading of race to estimate how low they are in g:

11/0.48 = 23 points below the mean

However since even the best IQ tests only correlate 0.9 with g, they would regress a bit to the mean if given the WAIS-IV, so multiply 23 by 0.9, which tells us they should score 21 points below the U.S. mean of 96.  Or roughly an IQ of 75.

A recent study found that the WASI full-scale IQ distribution of the homeless has a mean of 84.3 and an SD of 15.7. The WASI was published in 1999, and the study was published in 2011, so we should probably subtract 3.6 points for old norms which expire at a rate of 0.3 points a year, bringing the average homeless down to 81.  But this mean is further inflated by the fact that the WASI norms include non-whites which inflates low scores.  When we adjust the norms to the scale where U.S. whites have a mean of 100 and an SD of 15, we find that homeless have an IQ of 77, remarkably close to the IQ 75 I estimated from their racial composition.

It’s also worth noting that on any given night circa 2013, there were 450,000 homeless adults in America.  Since there are  242,470,820 adults in America, being homeless puts you in the bottom one out of 538 in financial success, which means the median homeless would be in the bottom one in 1,077 level.  If you force financial success to fit a normal curve, as a member of Prometheus once advised to do, that equates to being  3.1 SD below the mean in money.

Given the 0.4 correlation between IQ and money, simple regression predicts the homeless are 0.4(3.1 SD) = 1.24 SD below the U.S. mean in IQ.  Because of demographic shifts, the U.S. mean has shrunk to 96 and the SD has widened to 15.8, which puts the average homeless at 96 – 1.24(15.8) = 76.

So here we have three completely different methods all agreeing that the homeless have a mean IQ in the mid 70s which is excellent cross validation!

The IQs of self-made billionaires

Now let’s turn to billionaires.   Of course not all of billionaires are self-made, but people who inherit money are typically the same race as their kin who earned it, so the inherited money are stand-ins for the self-made wealth.  Forbes list of the 400 richest Americans is 0.25% black (Oprah), 0.5% Hispanic, 60.5% white, 2% Arab American, 2% East Asian, 1% Indian, and 35.75% Jewish, which means that if IQ and race were perfectly correlated, it would have a mean IQ of:

0.0025(70) + 0.005(79) + 0.585(106) + 0.02(106) + 0.02(115) + 0.01(130) + 0.3575(130) = 115

So if IQ and race were perfectly correlated, billionaires (at least the self-made ones) would have an average IQ 19 points above the U.S. mean of 96.  But since race only correlates 0.48 with g which correlates 0.9 with IQ, we must divide those 19 points by 0.48 then multiply by 0.9 to see that self-made billionaires are 36 points above the U.S. mean of 96, giving them an average IQ of 132.

An average IQ of about 130 is supported by three other lines of evidence:

1) About 1% of the self-made Forbes 400 (Bill Gates and Paul Allen) reportedly scored 1590-1600 on the old much harder SAT.  That’s equivalent to an IQ of about 170.  Since the smartest 1% of the U.S. is 37 IQ points smarter than the average American, it’s reasonable to think the smartest 1% of U.S. self-made billionaires are 37 points above the average self-made billionaire, which gives the average self-made billionaire an IQ of 170 – 37 = 133.

2) About 56% of the Forbes 400 is self-made (236 people), and by definition, the median self-made Forbes 400er would be the 118th richest self-made person in America.  Given that there are over 242,470,820 adults in America, the median self-made super-rich is above the top one in 2 million level in money.  which when you normalize the distribution, is about +5 SD.  Given the 0.4 correlation between IQ and money, the expected IQ would be:

0.4(5 SD) = 2 SD above the U.S. mean

Given that the U.S. mean is about 96 with an SD of 15.8, that works out to an IQ of 2(15.8) + 96 = roughly 130

3) Lastly, scholar Jonathan Wai notes that 43% of self-made American billionaires attended schools requiring test scores in the top 1% of the U.S. ability distribution, which would be equivalent to an IQ of 133.  Although he realizes that many non-brilliant people attend elite schools, he feels they are roughly canceled out by all the brilliant minds who did not attend elite schools, so on balance the 43% figure might be significant.  If the top 43% of self-made billionaires have IQs around 133, it implies that the average self-made billionaire has an IQ around 130.


When people are not explicitly selected for race (i.e. affirmative action, tokenism, geographic region) or non-cognitive variables linked to race (i.e. athletic talent), the racial distribution of a group of exceptionally accomplished or exceptionally unaccomplished people is an excellent proxy for IQ when statistically adjusted for the very imperfect correlation between IQ and race.  Even when some races succeed or fail because of non-IQ reasons (i.e. ethnic networking, language barriers), these other factors tend to cancel out when all the races in America are aggregated into the statistical analysis, as was done above.