Highest IQ Majors

College Majors
21

They don’t explain methodology, but looking at the list, I believe it is solely based on the average GRE scores by planned graduate major grouping as listed at https://www.ets.org/s/gre/pdf/gre_guide_table4.pdf . The article came out in 2017, so they are using an older version of the GRE scores by major than the linked table, but the general order is still extremely similar . I believe they are using a direct conversion of GRE VR score + QRE QR sections, ignoring the AW section. If you also include AW, the results are more in favor of humanities since only 1/3 is quantitative. Their conversion of what GRE VR + GRE QR converts to what IQ seems completely arbitrary. The scale appears to be something simple, in the form IQ = GRE VR + GRE QR - k, where k is a constant. Obviously this is nonsensical.

If anyone is curious, the highest GRE scores by major for the current 2015-18 sample are below. I listed them in order from highest to lowest. I only included majors with at least 1000 students. I realize one cannot assume all majors are reflective of the average for small portion of students choose to take the GRE.

Highest GRE VR Score

  1. Philosophy* – 160
  2. Creative Writing – 158
  3. European History – 158
  4. Religion & Theology – 158
  5. Art History, Theory, and… – 157
  6. English Language – 157
  7. Political Science – 157
  8. Humanities Arts: Other – 157
  9. Library Sciences – 157
  10. Various Small Submajors – 157
    *Philosophy sub-grouping of philosophy broad category

Highest GRE QR Score

  1. Statistics – 164
  2. Applied Mathematics – 163
  3. Mathematics – 163
  4. Materials Science – 163
  5. Materials Engineering* – 163
  6. Physics – 162
  7. Chemical Engineering – 161
  8. Finance – 161
  9. Electrical Engineering – 161
    *ME sub-grouping of ME broad category

Highest GRE AW Score

  1. English Literature – 4.3
  2. European History – 4.3
  3. Philosophy* – 4.3
  4. Public Policy Analysis – 4.3
  5. American History – 4.2
  6. Creative Writing – 4.2
  7. English Language – 4.2
  8. History: Other – 4.2
  9. International Relations – 4.2
  10. Neurosciences – 4.2
  11. Political Science – 4.2
  12. Religion & Theology – 4.2
    *Philosophy sub-grouping of philosophy broad category
1 Like
22

Adding to my earlier post, I believe I found the original source of the article. It appears to be from the following table that a random guy on the Internet made in 2011 – https://web.archive.org/web/20120217122331/https://www.statisticbrain.com/iq-estimates-by-intended-college-major/ . The author of https://thetab.com/ copied this table, without listing the source.

As I expected, the IQ estimate is based solely on the sum of ETS reported average GRE VR + average GRE QR, without considering AW. He calls the GRE “SAT” in the column headers for some reason, but the actual scores match the 2009 GRE exactly and are not SAT scores. The conversion formula appears to be the following. I believe all listed IQ values match exactly when using this formula. Note that the more recent GREs use a different scale.

IQ = ROUND((2009 GRE VR + 2009 GRE QR) / 12.5) + 31

1 Like
23

Are you serious? You are comparing vaccinations which have been proven to save millions of lives with the preudoscience of IQ testing?

I have been teaching stats and science for decades, and I don’t know whether to laugh or cry in response to this comparison. As somebody who has used and has taught statistics and research methodology at the graduate level, I really wish that you had attended one of my courses.

Of course, this comparison is no more than a red herring, since you use it without actually responding to my arguments.

As for “worldwide evidence”, the claim “everybody uses it, so it must be valid” is the argumentum ad populum fallacy. The fact that many people believe something does not validate it. Everybody believed that the earth was the center of the Solar System, yet that did not make it so.

So, rather than using red herring and argumentum ad populum fallacies, perhaps you could actually explain how IQ tests the issues I raised?

@Data10 that demonstrates exactly what I wrote above - people in a specific field will do well on tests which measure the primary skill needed by that field. If you weigh the tests in favor of Mathematics, engineers will do best. Logic will favor mathematicians and philosophers, while verbal will favor humanities. If you test the ability to solve a complex social issue in real time, social workers will have the highest scores.

These just monitor how well gifted kids are dealing with school work which is above their grade level. Some can identify gifted kids, but miss many. So they tend to have few false positives and many false negatives. They are also weak at determining anything but whether at kid is gifted in a few areas.

I was involved in our elementary school district’s gifted program when D19 attended K-8, so I became somewhat familiar with the system and the testing. We tested our kid for starting school early, for being part of the gifted program, and for some other programs which she joined. Being who I am, and my wife being who she is, we learned a lot about testing and giftedness.

As a stats and scientific methodology geek I looked into IQ testing, and the more I learned, the less I accepted the validity of these tests.

of course, since my kid aces the tests (as do my wife and I), I should really tout them as True Measures Of Intelligence. Unfortunately my intellectual honesty will not me to do so.

24

As is often the case, left out of this discussion is the importance of EQ. Most I’ve seen rise in the ranks in business, regardless of degree, are great with people, leading, etc. Of course they need a certain level of intelligence, but beyond that, it’s about getting others to buy in and do things.

25

It’s a given that success requires a good combination of IQ, EQ, grit and opportunities but just like some fields are largely EQ dependent, others are mainly IQ dependent. EQ can only go so far in dealing with Applied Maths or Quantum Physics.

As they say, everyone is good at somethings and ignorant at others. Lucky ones find what they are good at and with grit they can excel.

26

That’s very generous of you to correct my statistical failings. But clearly, you would have a much bigger impact correcting my professors at MIT, since apparently they have been teaching everyone incorrectly. Should I extend your offer to correct their teaching?

I get angry at anti-intellectual views like yours because I have seen the damage the lack of testing can do to others, through interactions with my son’s peers. My daughter is very bright, but my son is off the charts. Just one example: He taught himself Algebra by age 7, to the point where he could solve multiple equations with multiple unknowns. By the time we discovered this, he was actually starting to explore concepts in linear algebra before we redirected him towards non-math stuff, because there is more to life than just math.

I know this comes under the category of “good problem to have”, but raising a child like him is a lot of work because school is not remotely challenging, and that’s with us living in a very high performing school system where learning is encouraged. Because of this, we started looking at many types of enrichment. One of them was Johns Hopkins CTY SET, and the other was Davidson Gifted, both of which target roughly the top 0.1% of students by IQ. SET requires an SAT test before age 13, and 700+ on either section (the cutoff for normal CTY is lower). Davidson Gifted accepted either IQ tests or an SAT. My son took the SAT at age 11, and qualified for both. That’s the only “IQ test” he has taken.

The following summer, I dropped off my daughter to a CTY course, and my son to a Davidson Gifted program. And the two communities could not have been more different. The CTY Intensive Studies courses were filled with quite bright kids who come from well-off backgrounds and high performing school systems. This isn’t a surprise because CTY courses are expensive.

Davidson was free and was filled with super-bright kids, mostly from poorer families with kids that attended schools where learning was ridiculed. And despite this, many kids often under-performed because they were so mentally disengaged. But at Davidson, the kids found their intellectual peers and were happy, playing and generally acting like kids but with highly atypical conversations. In a parent session, a mother whose child attended for the first time broke down and cried because it was the first time she saw her child fully immersed with other kids who had similar interests and abilities. My eyes well up whenever I remember that.

Another group of high school students had been coming for a long time and were an incredibly tight group. During the school year they would cheer each other on for their academic achievements, even though nobody in their school would. Even though these people lived across the country, this was one event nobody would miss.

Davidson’s IQ testing and identification literally saved many of these kids from depression and underachievement. Not that many people know about its programs so there remains is a large group of wasted talent. And the lack of knowledge of the anti-testing crowd only makes it worse.

I relate this story because many people don’t realize who the real beneficiaries of a high ceiling testing are. It’s not privileged people like my family as my kids test scores are the least impressive thing about them. It’s not the people who have several thousand dollars in test prep to get a marginal increase in test scores. It is kids like the Davidson crowd, who get identified so they can get the support they need.

And lest you think that I am only talking about the top 0.1%, I am actually talking about anyone student who is considerably stronger than their peer group. My hypothesis is that a student that is is less than +2SD above their peers does really well in a school system. But any student much stronger than that could become a subject of ridicule. In poor performing school systems, that bar could be much lower than the top 0.1% that Davidson and SET target.

27

A few comments:

  1. Intelligence is primarily innate. Environmental factors are secondary.
  2. No single test, including IQ test, can accurately and completely capture human intelligence. They're, at best, partial measures of intelligence. However, testing, especially the tests that're less influenced by mechanical test preps, is still important to partially identify intelligence.
  3. More difficult majors naturally have higher concentrations of more intelligent students, simply because intelligence is required to succeed in those majors.
28

However, environmental factors will determine whether someone can make use of his/her innate intelligence, and how s/he makes use of it.

Other personal factors can also affect whether and how intelligence can be used. For example, motivation versus laziness can determine whether an intelligent person achieves something useful with it, or merely scores high on some kinds of tests.

1 Like
29

How are you defining “more difficult” majors and majors in which “intelligence is required to succeed” vs majors that do not require intelligence to succeed?

The major distribution seems as correlated if not more so with skew between section scores than total score. For example, the GRE by major table that was used to generate the IQ table from the first post shows the follow skew between the VR and QR sections. I only included broad major categories in this example, not smaller sub-majors. Having a high quant skew seems to be well associated with quant oriented majors, which are often called “tough majors.” Having a verbal skew seems to be well associated with verbal oriented majors, which are often called “easy majors.” If you swapped the math major students with the English major students, I expect both groups would have worse academic outcomes than the natural distribution, regardless of which one is “more difficult” and which major is one where “intelligence is required to succeed.”

Highest QR GRE - VR GRE Delta

  1. Electrical Engineering: +0.9 SDs
  2. Computer Sci – +0.8 SDs
  3. Civil Eng.: – +0.8 SDs
  4. Finance: – +0.8 SDs
  5. Ind Eng: – +0.8 SDs
  6. Mathematics – +0.8 SDs
  7. Mech Eng: – +0.8 SDs

Highest VR GRE - QR GRE Delta

  1. English – +1.2 SDs
  2. Library Sci – +1.2 SDs
  3. History – +1.1 SDs
  4. Religion – +1.1 SDs
  5. Arts – +1.0 SDs
  6. Anthropology – +1.0 SDs

A similar pattern occurs with SAT scores, so it’s not just a matter of the limited and biased sample group of students who choose to take the GRE or the varied curriculum providing better prep for specific sections (For example, an English major who never took a math course during college might have the best prep for the more math-oriented quant section. Similarly a Math major who avoided all non-STEM classes that were not required by the college might not have the best prep for the more reading oriented verbal section.) Some specific numbers are below based on the 2016 report (didn’t see by major list past 2016).

I suspect much of this effect relates to students choosing what a field that they are relatively good at compared to other alternatives, and what they are good at have a non-zero correlation with standardized test subsection scores. Some studies have observed similar effects with peer subsection scores. For example, if you only increase peer CR score and hold peer math score constant, students as a whole become slightly more likely to choose “tough” STEM majors and slightly less likely to choose “easy” humanities majors.

Intended College Major with Highest Average SAT CR

  1. Multi/Interdisciplinary – 578
  2. English – 576
  3. Library Science – 572
  4. Social Sciences – 561
  5. Physical Sciences – 561

Intended College Major with Highest Average SAT Math

  1. Mathematics – 621
  2. Physical Sciences – 584
  3. Multi/Interdisciplinary – 583
  4. Engineering – 574
  5. Computer Science – 563
1 Like
30

Theoretical physics and pure mathematics would top my list. They’d require [higher] intelligence to succeed in their respective fields.

31

Success in high level math/physics is far more complicated than a single score. Yes, you need to have some degree of math/quant ability, but that’s only one small piece of the puzzle. For example, the study at https://arxiv.org/pdf/1803.00595.pdf reviews what factors contribute to success of students in mathematics PhD program at UCI. The math PhD students had a GRE distribution similar to the overall average for math majors as listed in the ETS table – 152 mean on the VR section (45th percentile) and 161 mean on the QR section (76th percentile), so skewed in favor of QR. The VR scores had a very wide range across the full spectrum, including as little as 1st percentile. While the QR section had a narrower range, with a minimum of 30th percentile.

The regression analysis found the following variance explained in prediction of success (defined as completing PhD within 7 years). The increase from 4 to 8% with GRE was statistically significant, but GRE score was not a particularly influential predictor of math PhD success with the limited range (VR started at 1st percentile, QR started at 30th percentile). Tier of undergrad college was a far more powerful predictor than GRE score, which the author suggests may relate to rigor of undergrad preparation.

Predicting Success in Math PhD Program
Gender + Citizen + BGPA – Explains 4% of variance in Math PhD success
Gender + Citizen + BGPA + GRE – Explains 8% of variance in Math PhD success
The Above + Undergrad Tier – Explains 23% of variance in Math PhD success

1 Like
32

High IQ students tend to score perfect or near perfect on both verbal and quant sections on standardized tests like SAT or GRE and without putting too much effort into it. Ones requiring intense prep or scoring high on one section only aren’t quite in same tier. They are good at math or English and have good IQ but not necessarily very high IQ.

33

I never said a single score can predict a person’s intelligence… In fact, I stated the opposite.

34

FYI That is not actually how “the saying goes.”

That is a common misquote. In fact it goes exactly the opposite - the plural of anecdote IS data. And if you stop and think about it, multiple anecdotes are in fact data.

I don’t normally do community service, but when I do it’s to correct that misquote. :smile:

Carry on.

1 Like
35

Perhaps that is related to why math (specifically) PhD programs are said to place a high importance on undergraduate school (presumably their opinion about how well its math department prepares students for advanced math study).

1 Like
36

If high IQ students tend to score near perfect on VR, then it follows that high IQ doesn’t tend to be required for high level math. It may actually be uncommon. For example, in the linked study sample above, math PhDs averaged a lower VR score than the overall test taking population. Many successful math PhDs had a very low VR, and VR score only had a weak correlation with math PhD success (4% variance explained without GRE VR increases to 6% variance explained with GRE VR). Other studies have come to similar conclusions. For example, Arcidiacono found that chance of graduating with science degree in 5 years at UCs increased as a student’s verbal SAT score went down. That is math SAT - verbal SAT score skew was more predictive of graduating with a science degree than was total SAT score .

The author of the table linked in the original post appears to define IQ entirely based on average GRE scores by planned field of study, with no apparent reason for the particular scaling (VR+QR sum divided by 12.5). There are some obvious flaws to this methodology. If we instead say intelligence is a nebulous concept that we can’t measure well with scores or other methods, but we know it is required for high level math; then it is a circular definition that cannot be verified.

What appears to be more clear is that a high QR score (not VR or AW) score is well correlated with choosing quant heavy fields, such as math. A high (QR - VR) score skew also appears to be well correlated with choosing quant heavy fields, such as math.

1 Like
37

For your information, my kid was a Davidson Young Scholar until she aged out at 18, so I am familiar with the requirements of the Davidson program She has also regularly has taken courses with CTD (what they have in the Midwest instead of CTY - based in Northwestern instead of Johns Hopkins). My kid is in that 0.1%, aside from her talents in stuff which is not tested in “IQ” tests (like her dance and art).

So I am intimately familiar with the wonders and terrors of raising an HG/PG kid.

Moreover, I was also active in an advocacy group for gifted children for most of her elementary school years as well, so I am well aware of all the issues of which you wrote, but in a context beyond my own personal experiences. I am well aware of the need to identify gifted kids, and to provide them with services, since those were the very things for which we advocated. That is where I learned the weaknesses and biases of the standard set of “IQ” tests, and of the term “IQ”. So, despite your claims, I do not lack knowledge.

Had you read my post, you would have read the part in which I wrote that the tests are good at identifying some gifted kids, but miss out a very large number of others. That is because they do not measure innate intelligence. They measure academic levels. Gifted, HG, and PG kids do extremely well in many academic fields, most often when they are able to do it on their own. THAT is why “IQ” tests can identify them.

HG/PG kids often read, do math, etc., at a much higher level than the kids around them, so you can test for giftedness by testing for this ability IN THAT CONTEXT. However, no matter how gifted a kid is, if their native language isn’t English, or if they did not have access to higher level books, there is no way that such a test would provide you with an indication of their actual potential.

If a kid hasn’t learned formal mathematics, they may not be able to solve problems which assume that a kid has learned the formalized symbols that are used at schools in the USA. Moreover, many of the questions on standardized tests are based on the assumption that students have done things like memorizing multiplication tables.

If we take an exaggerated case - if you tried to test a kid who has not had any formal education, they will likely fail to answer “5x3=?”. This is not because they cannot multiply five by three, but because the would have no idea what the 5, the 3, the “X”, or the “=” actually mean. Yet IQ tests include such questions without taking such things into consideration.

Yes, a PG kid who grew up in a middle class educated family will do a lot better on math and reading than a non-PG kid who grew up in such a family, but to rely on these tests as being objective indicators as to the innate intellectual capabilities of any child or any adult anywhere in the USA or in the world is ridiculous.

To claim that one must accept the validity of these tests as dogma is also ridiculous.

Davidson also accepts portfolio applications, in which you demonstrate that a kid is doing classwork at least two years above their level. They consider that to be on par with scoring highly on “IQ” tests. That is because it tests exactly the same thing that IQ tests do - how well these kids perform academic tasks at their age level or above.

So you are conflating using these tests to identify gifted kids with using these tests to measure innate intelligence of adults. These are two VERY DIFFERENT things.

I would like to point out that, in your entire very long post, you still did not even mention any of my points. You used many arguments to emotion (“Think of all the kids”). You used ad hominem arguments (“anti-intellectual”), and you focused on your own experience, which is no more than anecdotal.

PS. Did you remember what you learned about implicit and explicit bias in sampling in your MIT statistics courses? Did you learn about the pitfalls inherent in assuming that any data collected using questionnaires are independent from the cultural background of the person who created the questionnaire? Did you learn to be suspicious of any pattern which matches your own biases? Did you learn how to avoid logical fallacies in reasoning?

1 Like
38

Bear in mind, the reason variation in QR didn’t explain much is likely because there was very little variation in QR to begin with. The QR scores were all mostly pegged out at the top end. Had there been lots of low QR scores add in the sample the effect of QR would have likely been more easily detected.

So GRE score may not have been as predictive as it otherwise would have been if they included a full range. As it was, almost everyone had sufficient QR to meet the requirements of the program.

(Further, just skimming the article it looks like the QR curve was NOT normally distributed, violating a prime assumption of regressions, but that’s another matter.)

39

As I mentioned in my post, there is indeed a range restriction issue on QR (no range restriction on VR). I agree that the QR influence would be greater had math PhD students had a greater QR range. If I am reading the graphs correctly, it’s certainly nothing approaching a normal distribution, but the majority (~60%) were not “pegged out at the top end” and had a <165 score. The mean QR of the full sample of math PhDs was 161. 161 is a 76th percentile score in the previously linked 2015-18 sample of all test takers.

40

Absolutely true.

But it is equally true that people are locked out of career choice based on IQ. Don’t try to be a mathematician or research physicist - or systems trader on Wall Street for that matter - unless you are a bona fide triple 9.