Many years ago I wondered if it was possible to estimate a person’s IQ from physical and demographic traits. A member of the Prometheus society suggested I use a fascinating technique called multiple regression and I am indebted to him for introducing me to the math. I suggested we try to co-write an article on the subject in the prestigious peer reviewed academic journal Intelligence which I have dreamed of publishing in since childhood (I’m now in my thirties). He thanked me for asking, but was not comfortable putting his name on something that was not well researched, and so for well over a decade, the concept stayed on the shelf. However when I realized that others in the field were already using multiple regression to predict IQs from demographic variables, or simple regression to predict IQs from a biological variable, I decided it was time to publish our method, which combined both approaches.
Several commenters here have expressed an interest in Angela Merkel’s IQ, including blogger AKarlin, and blogger Alcoholicwisdom so in part 1 of this series, I use a biodemographic approach to estimate her IQ from her three most extreme demographic and physical traits: Academic success, power, and physical coordination.
Expected IQ of an elite academic: 137
Merkel worked and studied at the Central Institute for Physical chemistry of the Academy of Sciences in Berlin-Adlershof from 1978 to 1990. After being awarded a doctorate (Dr. rer. nat.) for her thesis on quantum chemistry,[29] she worked as a researcher and published several papers.
According to Wikipedia, The German Academy of Sciences at Berlin (abbreviated AdW) was the most important research institution of East Germany, and in the 1980s (near the time Merkel was there), had no more than 176 East German members. Thus, the median ranked member of the AdW was roughly the 88th most academically accomplished person in East Germany. Let’s say there were around 17.068 million East Germans when Merkel was at AdW, and 2/3rds were adults, and Merkel was the 88th most academically accomplished of these. Thus, relative to East German adults, Merkel had a normalized Z score of +4.33 (one in 128,000 level).
Academic success seems to correlate between 0.5-0.6 with IQ. In the United States, since by the late 1970s, IQ correlates 0.57 with IQ, and the correlation between IQ and grades appears to be 0.54. In a recent post, I very crudely estimated the correlation between IQ and college attended to be 0.5, and that was probably an underestimate. Overall, the correlation between IQ and academic accomplishments is probably 0.57 in the United States. Assuming the correlation is similar in Germany, then Merkel’s academic Z score of +4.33 implies an IQ Z score of 0.57(+4.33) = +2.47 or roughly 137 (white norms). Based on academic success alone, we can expect with 95% certainty that Merkel’s IQ is between 113 and 161.
Expected IQ of the most powerful person in Germany: 133
U.S. Presidents tend to have IQs only 40% as far above average as you’d expect if IQ and power were perfectly correlated, suggesting that the correlation between IQ and “power” is about +0.4. Germany has about 81 million people and assuming about 2/3rds are adults, we’d expect the most powerful German to have a power Z score of +5.53 (one in 53 million), and thus have an IQ Z score only 40% as extreme (+2.21), equating to an expected IQ of 133 (95% confidence interval: 106-160)
Expected IQ of a klutz: 90
According to an in-depth profile of Merkel, she might be kinesthetically challenged:
Angela was physically clumsy—she later called herself ‘a little movement idiot.’ At the age of five, she could barely walk downhill without falling. ‘What a normal person knows automatically I had to first figure out mentally, followed by exhausting exercise,’ she has said.
According to technical studies by the U.S. Department of Labour, IQ and physical coordination correlate 0.35. Scholar Arthur Jensen has put the figure at a potent 0.7, but that sounds impossible given the validity of the “dumb jock” and “clumsy nerd” stereotype. We should expect physical coordination and intelligence to be correlated because they’re both functions of the brain, but since only intelligence is a function of the mind, the correlation should be modest.
However even small correlations can have big effects. If Merkel is kinesthetically impaired, then by definition she’s at least 2 standard deviations below the mean in coordination. Assuming a 0.35 correlation with intelligence, this would predict an IQ that is 2 SD(0.35) = 0.7 SD below the mean, or roughly 90 (95% confidence interval: 63 to 117)
Expected IQ of a hyper-educated, supremely powerful klutz: 135
Statistically, we’d expect Merkel to have an IQ of 137, 133, or 90, depending on whether we estimate her IQ from her education, power, or physical coordination respectively. All of these predictions carry large margins of error, so it’s interesting to ask what happens when we combine them.
As the Promethean once told me, it’s not enough to know each variables correlation with IQ, we must also know each variables correlation with one another, and thus we get the following correlation matrix.
| academic success | power | coordination | iq | |
| academic success | 1.00 | 0.44 | 0.20 | 0.57 |
| power | 0.44 | 1.0 | 0.14 | 0.40 |
| coordination | 0.20 | 0.14 | 1.00 | 0.35 |
| iq | 0.57 | 0.40 | 0.35 | 1.00 |
Many of these correlates are very crudely estimated. For example the correlation between academic success and power is derived by noticing that U.S. presidents are about 44% as academically successful (in Z score units) as they are powerful. The correlation between academic success and coordination is derived from assuming the only link between these variables is their shared correlation with IQ. Similarly for the correlation between power and coordination. Such assumptions are admittedly speculative.
Nonetheless, the correlation matrix, speculative as it is, gives the following equation:
IQ Z score = 0.44(academic success Z score) + 0.17(power Z score) + 0.27(coordination Z score)
[Edit, June 7, 2016: The regression coefficients in the formula above are smaller than each variable’s correlation with IQ. This is because the intercorrelation between predictors means no two predictors are completely independent, so each predictor in the multiple regression formula must be adjusted for the predictive power already captured by the other two predictors]
Entering Merkel’s Z scores into the formula gives:
IQ Z score = 0.44(+4.33) + 0.17(+5.53) + 0.27(-2)
IQ Z score = 1.91 + 0.94 – 0.54
IQ Z score = 2.31
Now converting IQ Z score into IQ is simply a matter of multiplying by 15 and adding 100.
IQ = IQ Z score(15) + 100
IQ = 2.31(15) + 100
IQ = 135 (white norms)
We can say with 95% confidence that if given an updated German version of the WAIS, Merkel would score anywhere from 112 to 158 (white norms).
Considering we used three predictors, this is not as much precision as I would expect. Part of the problem is that Merkel’s two most extreme traits (academic success and power) are moderately correlated, thus they do not add enough independent information.
Since the biodemographic prediction is uncertain, we search for psychometric data in part two.
Woah.
Could this; “IQ Z score = 0.44(academic success Z score) + 0.17(power Z score) + 0.27(coordination Z score)”,
because of the fact that it is not merely the factor to IQ correlation, relate to the relationship of each variable to each other (that’s the whole schtick, right?)
Could you just go into a little more info on exactly how you go those numbers?
*be because of the fact*, I mean.
So is it,
Academic Success to Power
Academic Success to Coordination
Power to coordination
?
https://www.google.com/search?q=relationship&biw=1920&bih=963&source=lnms&sa=X&ved=0ahUKEwjDkID4qZXNAhWCPz4KHdkoDQAQ_AUIBigA&dpr=1
https://www.google.com/search?q=relationship&biw=1920&bih=963&source=lnms&sa=X&ved=0ahUKEwjDkID4qZXNAhWCPz4KHdkoDQAQ_AUIBigA&dpr=1#q=correlation
So one is indirect, relying on something else, and one is mutual?
The key point is the higher the correlation between variables, the less INDEPENDENT information each one provides.
So if we’re predicting IQ based on power alone, power would have a predictive strength of 0.4, but when we include it in an equation that also has academic success and coordination, its predictive strength drops to 0.17
That’s because most super educated people become powerful as do some coordinated people, so the fact that powerful people are smart is largely just a redundant by-product of educated coordinated people being smart.
The other two variables already indirectly imply you’re powerful, so power doesn’t bring much NEW information about your IQ. Hence the huge drop in its predictive strength
I understand it conceptually, but if given a problem of it I could not quite find the exact numbers for myself.
“IQ Z score = 0.44(+4.33) + 0.17(+5.53) + 0.27(-2)”
I thought the correlations were .57 .40 and .35? where did you get those numbers?
Because the predictors are not 100% independent, their predictive power drops when you use multiple predictors
For example once you credit someone with being educated, you can’t give them full credit for beig powerful because part of the reason they’re powerful is their education. So there’s a redundancy that needs to be removed from the numbers
Hence when you use multiple predictors, the predictive power of each predictor drops to avoid crediting (or deducting) people for the same thing twice
Oh that makes sense. So how do you figure out how much the predictive power drops for each value? What I mean is what math did you do to get .44.17.27 from .57.40 and .25?
sorry i mean .35* instead of .25
I too would like to know the math. I feel some matrix multiplication and division here.
Seems that way….
A little off topic but I was wondering if I did this right. Since I got in the 95 percentile on my GED score that translates to about 1 in 20 people which is equivalent to an IQ score of about 124 according to this chart http://www.iqcomparisonsite.com/iqtable.aspx
so assuming a correlation between education and IQ is about .57 my IQ should be:
124-100= 24
24/15= +1.6
+1.6*.57= + .912
.912*15= 13.68
13.68+ 100= 113.68
so about 114. Now if I average that with my bio-demographic score of 103 I get:
108! Lol look at that my IQ is the same as Racerealists. XD
Merkels quant is prob elite level.
The autism, barren personality and physical clumsiness betray a low verbal (relatively).
Hence her iq may be 135.
But thats not worthwhile knowing.
”The autism”
What*
She’s not autistic, even in the broad spectrum, don’t mistake with sociopathy.
I say that relative to “normals” or I suppose someone on the far end of the other side of the spectrum.
Then again, my perception may be warped as even beta bux office workers appear on the spectrum to me haha.
DEEP DIVE
Looking at wiki my (larger than normal) eyes see “fluent in Russian” – verbal intelligence is quite high…majored in physics (physicists have very good verbal)…early devotion to politics (yes, yes).
On the negative side she supported the war in Iraq, which in Germany without neocon media propaganda, is tantamount to retardation…or political power playing with the Brain Trust.
Interest in football and lack of children indicates lesbianism (like Clinton). Marriage to person who also specialises in quantum chemistry indicates sham marriage perhaps (women don’t like nerds, particularly quant nerds).
Overall I retract my assertion she is on the spectrum.
Also, I think Merkel’s IQ is far higher than 135.
Her eyes are very telling in that regard. But not necessarily enough to indicate she is a sociopath like Hilary which I believe further proof is needed.
On the negative side she supported the war in Iraq, which in Germany without neocon media propaganda, is tantamount to retardation…or political power playing with the Brain Trust.
Actually the German intelligentsia reads the New York Times religiously.
”women don’t like nerds” I heard about this a lot but it doesnt make sense. Women might not like to ‘date’ nerds but there is nothing about nerds that makes women not want to ‘marry’ them. Nerds in general make more money, have higher status in society….two factors which women give a lot of importance to when marrying. The only thing that may make women dislike nerds is probably they might not be interesting enough to talk to. (I might be wrong about this too as i get this image of nerds being not interesting for shows like big bang theory not real life).
Are you guys just butthurt over the Syrian migrants?
versus, say;
White Geno-Lie.
That’s not the point.
These populations are extremely clannish, and have a lower genetic IQ than Whites.
We share 98% of our genes with Chimps, should we welcome thousands of Chimps every month in our countries ?
Your logic is retard, and your sources are probably not even reliable.
Is your Presidente, Michel Temer, of Lebanese descent, an evil Thur Wurld Rapist?
I thought you would adore him for deposing Dilma;
http://media.gettyimages.com/photos/michel-temer-brazils-vice-president-listens-during-an-interview-in-picture-id483256925
In my opinion female political leaders are generally terrible of both ‘left’ and ‘right’.
I include Thatcher, Gillard, Pelosi or anyone else you’d care to name.
Once you understand female psychology, you have no alternative but to be deeply skeptical a woman could be a leader or visionary. I think the quantitative track record of female CEOs of listed firms is also indicative of that fact.
Being lesbian helps though as it affords more testosterone, seratonin and the ego is less externally referenced. Hence the recent study showing why a disproportionate proportion of female managers are lesbian.
That being said, if one is capable, either by mutation, diligence or circumstance, one should be given the chance to succeed.
Well that’s why they say she doesn’t have charisma, but she is quite competent in other ways, and is the best Germany’s got, regardless.
Speaking aside from (our) Mathematical discussion- I would have expected her IQ to be a good deal higher. I mean, I’ve known three people in my close family with IQs at that level or higher, and I just see an influential world leader- a regional leader, as much much smarter than them.
The PhilosopherAlso, I think Merkel’s IQ is far higher than 135.
Her eyes are very telling in that regard.
I think that my genetic potential IQ is much higher than my phenotype IQ. The reason is that I have been under circumstances growing up under abuse and neglect. My mother is high functioning autistic. And my dad hit me on the head when I was one, black and blue. Malnutrition can cause huge drops in IQ. In my case I remember we always had food yet my sister and brother and I were never taught how to regulate our emotions. I found a video of how 4 year olds can figure ice skate professionally. Mozart learned the violin at age 4. The brain is plastic and water. Practically, energy regulation is emotion. I just spent a week in the hospital because of a mental breakdown dealing with my emotional pain. When I came home yesterday I felt much better because of mood stabilizers and the doctor confirmed to me that OCD is caused by problems with dopamine.
In the hospital I met a man at IQ 140. It was the first time I could talk to a person who understood me. I showed him my note I was taking from my head and said he remembered everything on all 16 pages. And because he did remember all 16 pages I was then able understand processing speed. Processing speed is how fast materials sing into you subconscious. Because mine is at 86 and verbal is at 132 this means I comprehend 512 time faster talking in person rather than reading, and this is why my general intelligence is 130 as my genetic potential. I may read slow or have to reread but I do comprehend. The way I go about understanding is through language. Language is the vehicle to theory of mind and this explains why Jews average IQ 110.
Language requires a good understanding of emotions. Jews understand emotions and adapted by reflecting on the motivations of others. Understanding motivation allows you to construct a self model (meta-cognition). And that helps in understanding interactions between other people. And this is why some groups are suspicious of them. They listen too well. And they emote from the perspective of others. They appeal to the others motivation. And in the times before 70 A.D. they questioned God’s motivations. That put them above other nations. That put them as equal to a God above the gods. They were not slaves but children of the most high. People like Richard Dawkins don’t understand that without Abraham child sacrifice may never have stopped. And without Jesus, crucifixion would have never stopped.
When you stop something you inhibit it. The brain is an attractor network. In the hospital I told my appointed representative that the frontal lobes are the main rudders and that it steers the other atractores / rudders. The anterior cingulate cortex is the rudder for conflict monitoring. If you are making no progress you must change your behavior. Thus you need to change direction just like a boat rudder changes the ship’s direction (cybernetics). This requires energy regulation in and out of the subconscious. The cool thing about the brain is that it finds the optimal way of creating a world model by jiggling the fitness landscape (neuron trees) with noise to reach a global minimum to increase reaction time. 95% of what you see is created by your brain.
I think my verbal comprehension is much higher than the IQ test I took measures. And I believe the reason is my theory of mind. I am not a grammar nazi. I read between the lines. IQ tests do not measure your theory of mind, nor does it measure cultural knowledge. Nor does it measure synthesis, only analysis. IQ tests do not measure your ability to produce results under uncertainty or probability. I think Pumpkin Person is right that DNA tests would be able to measure the potential phenotype of an attractor network as energy regulation. I also think I have a great trove of cultural knowledge that IQ tests do not measure. Everything I learn is intrinsically self motivated.
Since artificial intelligence is growing exponentially I do believe IQ tests will become obsolete as a.i. will be able to increase the flynn effect to the point where norming does not work. I study these things. I understand what it means for people to be 512 times faster than me. Because I understand theory of mind. IQ tests are going to be the 90’s pager compared to the iPhone 6. Facebook already usages A.I. – I still have this picture that can be labeled as such:
Energy
Theory of mind
Intrinsic motivation
When I took my IQ tests in Nov 2015 I was really tired. I have insomnia and I almost fainted the last ten minutes of the test. I got 113.
Sometimes the smartest people are not the ones who did well on a test, but are the ones who survived the harshest conditions and still reproduced. The book “The Secret” got it right. Everything is energy. A gallon of gas is a gallon of gas no matter how damaged a can become from life stress. It still passed on its genes. Angela Merkel most likely had help with her emotions growing, that is how you can tell she is smart.
Artificial Intelligence is going to explode in the next four years. I hope it will help me find some scientists to help me with the brain damage I have. I want to process more data faster.
To PP: cats love pumpkins 🙂
Anime kitty, Internet IQ tests are extremely inaccurate. I got 155 once and I answered every answer randomly. Also I’m glad to see a fellow anime lover 🙂
It was not a test, I answered no questions.
It was a data analysis of my entire internet activity by deep learning.
Deep learning is now super human at classification. Facebook analysed my links to friends and what they post on their wall. People for clusters so deep learning understands by calculation, my associations with smart people. I visit this blog allot so I get extra points for reading material that the average person does not read. It is like the Google rank algorithm of websites accept it is about people. I post lots of links about developmental psychology on my Facebook wall.
I posted these three pictures on my wall, so when I clicked on the analyser it gave an accurate assessment of my personality. Deep learning is why voice search works. And deep learning is why personality clusters have been discovered without the need for human involvement. All you need to do is be on the internet and you already have a profile. Artificial intelligence will only become more accurate and independent as time passes.
my personality profile as deemed by the A.I.
I’m incredibly skeptical especially sense My facebook does not really reflect my current* personality. I barely get on it, I used to a lot. But I’ll still check it out.
http://mtweb.mtsu.edu/stats/regression/level3/multicorrel/multicorrcoef.htm
I found this on the multiple regressions subject- and it seems that although the variables do fit quite well in the independent portion of the formula, that only one correlation is given…do you average the z-scores.
Obviously the math cant be shown here,
but
I get a higher dimensional “r” of 0.81 (only the case when none are that high because it’s 3d) and an average Z-score (I used -0.35 for coordination) of 3.95
100+3.95(0.81)(15)= 148.
If, for coordination you use 0.35 and -2, you get 0.69 and an average Z-score of 2.69
100+2.69(0.69)(15)= 128.
I’ll use 128 since it’s closer to the 136 you got, but it’s still well off.
Can you go into more detail about your methods from this?
Or did you use a computer program?
also….
you can use the patterns in the formula to work for almost any variable number (as long as all but one are independent).
do you average the z-scores
Oh God no. Some variables have far more independent predictive power than others so if you just averaged Z scores you would lose all that rich information.
I thought that you could use R and multiply by the average of the SDs, since the different strengths, etc. were accounted for through the formula, sort of “making it flow in one direction”…as you can see.
But I guess that’s really way to much of a gross oversimplification if not really possible.
average of the Z-scores I mean.
I.E. the higher R is, the better the general direction of the score is.
I suppose it should theoretically be accurate in determining whether something is above or below the median.
The formula you’re citing is not intended to make predictions based on multiple variables because doing so requires multiple coefficients as my post showed.
Instead the multiple R is intended to tell you how well predictions made from a multiple regression equation would correlate with the value being predicted.
I see.
What I would like to know, and it could use a more 2+ dimensional approach, is how the numbers in the table were found (beyond the variable to IQ correlation)
i.e. some are easy like coordination to power (0.35×0.4)= 0.14
but what about academic success to power, 0.57×0.4 does not equal 0.44,
I estimated the correlation between academic success and power by noting that incredibly powerful people seem to be about 44% as extreme in academic success as they are in power.
I looked at the recent U.S. presidents and estimated their average academic success in normalized Z score units. It was something like +2.38.
But since U.S. presidents are +5.4 in power, this implies a 0.44 correlation between academic success and power (2.38/5.4 = 0.44)
So it was just mainly based on outside data, not anything given here…
the rest after that seem to be just multiplied together.
So, I was reading one of Margaret Thatcher’s books recently, and was wondering what her IQ was, so as an example, I could say,
she was influential,academically as a Chemist (part of a team that invented something) that is now a $13.7 Billion dollar industry, maybe around +4.5 SDs for academic success.
And the power z-score would also be +5.5
while,say she was about -2SDs on ideology.
Academic Success to power: 0.44
Academic Success to conservatism: 0.65 x 0.3= 0.195
conservatism to power: 0.12
0.44(4.5) + 0.195(5.5) +0.12(-2)= 2.8 SDs
2.8(15)= 42
100+42= 142
Does this sound plausible (she is very similar to Merkel in biography, and the scores were very close so I do suppose it is).
I ask because this blog has largely taught me statistics…:)
“conservatism to power” should be ideology to power, the conservatism comes into play as a negative Z-score.
oops.
You seem to be confusing the predictor coefficients with what you estimate to be the predictors correlation with each other.
Also, Thatcher was not 2 SD from the mean in conservatism or ideology. She had mainstream acceptance in her day, suggesting her politics were close to the mean of her time and place,
“You seem to be confusing the predictor coefficients with what you estimate to be the predictors correlation with each other.”
So, do you subtract the ind.-ind. correlation from the Ind.-Dependent correlation?
I.e. to explain the overlap?
Great analysis! Thank you! I just read it first time. You rarely give high estimations. 🙂
Especially in politics women are positively discriminated, so one could subtract some points…
When you have time, please estimate also Israeli Prime Minister Benjamin Netanyahu. In internet there are lists showing him having 180 IQ, I do not know where that number comes from and I do not believe those estimations.
Yes, that would be interesting. And rank him among other world leaders as well.
Use your chronometric technique as well.
please 😉
Highly doubt his IQ is that high. Probably the same as the “estimates” for Stephen Hawking and Albert Einstein.
I’ll add him to my waiting list. The 180 figure sounds like propaganda, or a childhood ratio score, though it would be cool if it were true.