I Took an IQ Test to Find Out What it Actually Measures

• In popular culture, the term IQ is everywhere.
• Do IQ exams do that?
• You probably need 120 points of IQ.
• Don't know what my IQ is.
• And IQ.
• IQ.
• IQ.
• Low IQ individual.
• People who boast about their IQ are losers.

When people say IQ, what they mean is intelligence, an objective, rigorous measurement of intellectual ability. But does it actually work? Well, in this video, I want to find out: Where IQ came from. What does it actually measure? What can it predict about your life? And I guess, what is my IQ? I have never taken an official IQ test before. Honestly, I don't think I'm terribly smart. I've always kind of considered my IQ to be maybe just a little above average.

• Exothermic or endothermic?
• I feel like that should be exothermic.
• Good job, science guy.

There are a lot of IQ tests online, but I am very skeptical about their accuracy. Still, I figured some of them may be good practice for the real thing. Tomorrow I'm going to do an IQ test for real. Before I do that, I want to try to improve my score, and so I'm going to try to do a whole bunch of practice tests. I think this test is trainable. But tomorrow we're going to see whether that's true or not.

The idea of intelligence testing goes back hundreds of years, but the first concrete breakthrough occurred in 1904. English psychologist Charles Spearman was studying students' grades in different subjects, and he wondered how their performance in one subject, like English, would relate to their performance in another, like Math. One option would be that the better a student did in math, the worse they would do in English, maybe because they spent more time on their math work and so had less time to devote to English. So performance in different subjects would be negatively correlated.

Another option was that performance in one subject would be completely unrelated to performance in another. After all, different subjects require different skill sets, so maybe marks would be totally uncorrelated. The third option was that the better a student did in math, the better they would do in English. In other words, their marks would be positively correlated. A correlation coefficient can vary anywhere from negative one to positive one. A correlation coefficient of negative one indicates a perfect negative correlation, meaning an increase in one variable corresponds to a precise, predictable decrease in the other variable.

Similarly, a correlation of positive one indicates a perfect positive correlation. A correlation of zero indicates no relationship between the two variables. And any value between zero and one indicates a positive correlation, but the data has some random spread. The square of the correlation coefficient tells you the amount of variation in one variable that can be explained by variation in the other variable. For example, if the correlation coefficient is 0.5, then 25% of the variation in one variable can be explained by the other.

When Spearman analyzed his data, he found a clear positive correlation. Students who did better in math also tended to do better in English, and the correlation coefficient was 0.64. But Math and English weren't the only subjects the students studied. They also took Classics and French. And when Spearman looked at the correlations between all of these subjects, he found the same pattern. Students who did well in one subject tended to do well in them all.

So how do you explain this observation? Well, Spearman proposed that each person has some level of general intelligence, what he called the g-factor. This construct was meant to capture how quickly students could learn new material, recognize patterns, and think critically regardless of the subject matter, which explains why students' scores across subjects are correlated. Those with high g score well on all subjects and those with low g score poorly on all subjects. Spearman published his conclusions in a paper titled "General Intelligence Objectively Determined and Measured."

But the correlations weren't perfect. So on top of the g-factor, Spearman proposed subject-specific factors, or s-factors. A student's performance in math, for example, would depend on their general intelligence plus their subject-specific factor for math. Subject-specific factors could increase or decrease performance on that particular subject. Spearman believed that specific factors could be trained, but general intelligence was fixed. So he wanted to find a way to reliably measure general intelligence.

At around the same time in France, Alfred Binet was tasked with figuring out which kids needed more help in school. Together with Theodore Simon, he developed the Benet-Simon test. Students were asked to name what's missing in the drawing, define abstract terms, and repeat back sentences. And there was also this question asking which face is prettiest. There were 30 tasks in all. Their performance was benchmarked against other students of different ages in order to assign them a mental age. For example, if a student performed about as well as the average eight-year-old, their mental age would be eight. This mental age was then divided by their actual age and multiplied by a hundred to arrive at a so-called intelligence quotient, and IQ was born.

So the Binet-Simon test was the world's first IQ test. It was translated by Goddard into English and brought to the U.S. At Stanford, Lewis Terman standardized it using a large American sample, and with some modifications it became the Stanford-Binet test, and for decades it was the most widely used test in the United States. But this was just the start. Many other IQ tests were developed, and they all had the same goal of measuring the g-factor. The way they did this was by assessing many different mental abilities, including memory, verbal, spatial, and numerical skills.

Each one of these areas might have a subject-specific shift. But by averaging them all together, the idea was the subject-specific effects would cancel out, leaving a decent approximation of g. Of course, there would always be some error. But that's why psychologists designed IQ tests with upwards of seven to ten sections with distinct tasks to try to minimize subject-specific distortions. All the different IQ tests differed in the number of questions and their difficulty.

So to standardize the scoring system, each test was given to a large sample of the population. Raw scores were normalized, usually so the mean was a hundred and the standard deviation was 15, and that's how it's still done to this day. This is known as IQ, and it's meant to be a measure of an individual's g-factor in comparison to the rest of the population. The way it's scaled, 68% of people have an IQ between 85 and 115. Only around 2% score over 130 or under 70.

(tranquil music)

11 lions, four cats, and seven crows have a total of... Oh, boy. As I was studying for my IQ test, I practiced all the different types of questions that appear on modern tests. One section will almost certainly be on vocabulary. They give you one word like sanguine, and you have to pick which of the multiple choice options is most similar in meaning. Is it gloomy, asinine, recalcitrant, optimistic, or reflective? They might also ask you to pick a word with the opposite meaning. So what is the opposite of perspicacious? Is it canny, obsequious, dull, fanciful, or sagacious?

Another section tests your ability to spot patterns with numbers. So pick the number that best completes the pattern. 3, 5, 8, 12. What comes next? Originally, I was looking for complicated patterns, but as I familiarized myself with the online tests, I discovered the patterns were usually pretty simple. A good technique is to find the difference between adjacent terms. So in this case, the first two terms are separated by two, the next by three, and then four. So the logical next term should be five more than 12, so 17. The answer is C.

Sometimes the numbers grow rapidly, like in the sequence three, 15, 60, 180. What comes next? In cases like this, I look at the ratio of one number to the one before it. In this case, the second number is five times the first. The next number is four times bigger, and the next one is three times larger. So the answer should be two times the fourth term, which is 360, answer B. One of the best-known types of IQ test questions are Raven's progressive matrices. These involve a three-by-three grid with symbols in each of the cells, and you have to select the ninth cell which follows the pattern.

I found that the bulk of these puzzles obey one of only a few different logical rules. One is translational motion. So the symbols move from one cell to the next in a predictable fashion. The second is rotational motion. One or several objects rotate from one cell to the next. The third is missing symbols, where in each row or column each symbol appears once. So to figure out which symbols appear in the final cell, you just have to spot which ones are missing. And the fourth is addition, where the first cell plus the second cell equals the third cell. In this case, lines that overlap cancel out, but a line plus nothing equals a line.

In most modern IQ tests, all the questions are completed under time pressure. You may have only around 10 to 30 seconds per question.

Okay, this morning I'm taking an official IQ test, and I got to say that I'm pretty nervous. I always want to do well on tests; it's something I pride myself on. But at the same time, who knows how this is going to go? I'm not allowed to take you in there because obviously people don't want the questions getting out, and they don't even want video of what it looks like in there. They're very strict about these things. So I'm going to go in, do the test, I'll come out, and I'll let you know how it went. Wish me luck.

What's remarkable about IQ tests is that an hour or two of questions on vocabulary, numbers, and arbitrary shapes can predict a surprising amount about your life. For one thing, the higher your IQ, the larger your brain is likely to be. A large meta-analysis from 2005 estimated a correlation of 0.33 between IQ and brain size. So high IQ is literally big brain. IQ is also predictive of school success. In 2007, Scottish psychiatrist Ian Deary measured the IQs of 13,000 eleven-year-olds. And five years later, when these students completed national school examinations, Deary compared their exam marks to their IQs.

• Their performance on an IQ test when they were 11 correlated with their performance five years later on the GCSEs, about 0.8. That's an extremely high correlation.

It means about two-thirds of the variation in national school examination scores could be predicted by IQ tests taken five years prior. Now, the correlation coefficient of this study is at the high end of the 0.2 to 0.8 range found in similar studies. But research supports the claim that IQ is a good predictor of school success. It also predicts how much schooling a person will complete. Maybe this shouldn't be so surprising since some school tests are essentially IQ tests. It's been argued that tests like the SAT, ACT, and the GRE are basically IQ tests. They correlate with standard IQ tests at around 0.8.

Now, on my SATs I got a score of 1,330, which corresponds to an IQ of around 130. So it'll be interesting to see if my official IQ score matches that, or if I was able to increase my score by familiarizing myself with IQ-style questions. I don't know. But IQ also has predictive power outside of school. One of the most robust findings is that IQ can predict job success.

• Particularly in technical or high complexity jobs. -

How do you measure occupational success?

• You ask people's bosses to rate them. You ask what people's income is. You measure productivity in ways that economists use about the output generated. -

The correlations typically range from 0.2 to 0.6, and the effect is most notable for more complex jobs, which makes sense. The highest effect is for military training. In fact, the U.S. military will not accept anyone with an IQ under 80. They also limit to 20% the number of recruits with IQs between 81 and 92. During the Vietnam war, in order to increase the pool of applicants, they relaxed this last requirement. But what they found was that those below the threshold were 1.5 to three times as likely to fail recruit training, and they required between three to nine times as much remedial training.

Taken together, this added so much strain that the military ran more efficiently without the extra recruits. In total, 5,478 people recruited under this initiative died at a fatality rate three times higher than ordinary recruits. So the military reinstated their requirements, and today anyone with an IQ less than 80, that is about 30 million Americans, would be ineligible to join the military.

Even outside the military, IQ seems to play a role in how long you live. In a Scottish study, scientists uncovered IQ tests from kids when they were 11 years old. Now, 65 years later, they checked to see who from the sample was still alive at age 76. And they found that on average, for every 15-point increase on the IQ test, you would be 27% more likely to still be alive at age 76. A large meta-analysis confirms that people with higher IQs have a lower risk of dying during the timeframe investigated in each study.

The last major thing that IQ seems to predict is income. This study shows a clear tendency for income to increase with IQ, and it found a correlation coefficient of 0.3. But the variance is huge. In fact, the top three earners in this study all had IQs below 100. A large meta-analysis of 31 studies found the correlation between IQ and income to be 0.21. That is significant but small. It means that only 4.4% of the variance in income is explained by IQ.

Maybe one of the reasons why we don't see as high a correlation for income is just because economically intelligence is not necessarily that highly rewarded, in that maybe there are jobs, like just doing a real estate-type scheme, maybe that doesn't require a huge amount of intelligence. Simultaneously, you have all these very highly intelligent people who maybe become college professors, but that doesn't necessarily pay very well.

• Yeah.

A lot of people who have very high intelligence scores don't have the same interest in accumulating money. The relationship with net worth is even weaker. It hardly seems to correlate with IQ, even though people with higher IQs are supposedly more intelligent and on average they make more money each year. But this apparently doesn't translate into saving or accumulating more wealth overall.

But if IQ correlates with school achievement, job performance, income, and longevity, why don't we hear more about it? Why aren't more people tested? I think it's because IQ has a dark history. When Henry Goddard brought Binet's test to America, its use and interpretation shifted dramatically. In France, Binet believed intelligence could be improved through education. He designed his tests so that struggling students could be given more help to catch up.

But in the U.S., the modified test was given to adults to rank them by intelligence. And researchers like Spearman believed that g was unchangeable, that whatever general intelligence you were born with you would have for the rest of your life. And many thought g was inherited, passed down from parents to children. These days we would say it has a genetic basis. There is some evidence to support these assertions.

IQ appears fairly consistent over one's lifetime.

• So they had tests done when people were 11 years old. They found all those tests in a filing cabinet and followed those people up and gave them the same test when they were 90 years old, much, much later. -

Their scores 80 years apart were correlated at around 0.5 to 0.6. There's also evidence for a genetic basis to IQ.

• You find, for instance, that if you get two identical twins and give them an IQ test, they have a very strong correlation. It's actually about the same as giving the same person the test a few weeks apart. -

Henry Goddard used the claims that intelligence was inherited and unchangeable to put IQ at the center of the American eugenics movement. Eugenicists wanted to prevent those with undesirable traits from having kids. In many states, laws were passed to enable forced sterilization of people who failed to meet a certain threshold on an IQ test. The constitutionality of these laws was upheld by the Supreme Court in 1927.

• Even words that we now use as insults, moron, idiot, imbecile, were used as scientific terms. -

In his judgment, Justice Oliver Wendell Holmes wrote, "It is better for all the world if instead of waiting to execute degenerate offspring for crime, or to let them starve for their imbecility, society can prevent those who are manifestly unfit from continuing their kind. Three generations of imbeciles are enough." In total, over 60,000 people were forcibly sterilized as a result of these laws. In fact, they served as a model for Nazi Germany. Hitler himself claimed to be inspired by American eugenicists.

• These have been used for horrific things in the past. -

At the Nuremberg trials after the war, some Nazis quoted from the American Supreme Court decision. Given this awful history, I think it's understandable that many people completely disregard IQ today. On the science of intelligence, there are a number of things those early researchers got wrong. One is that IQ is not entirely determined by genetics. Can you quantify the effects of genetics versus environment?

• When you look at twin studies, on average, across the whole lifespan, it's about 50/50 to heritability and environment. -

You simply can't, for ethical reasons, estimate it in humans with a reasonable degree of certainty or accuracy. Given my reading of that literature, it's a pretty broad range, probably somewhere between 40% and 70%.

• Okay.

And since education can improve IQ, it is not completely fixed over a lifetime. Plus, intelligence might not be a single construct as initially imagined. These days, scientists recognize two forms of intelligence, fluid and crystallized. Fluid intelligence is your ability to learn, process information, and solve novel problems. Whereas crystallized intelligence involves the knowledge you've accumulated over your lifetime.

Both types of intelligence increase throughout childhood, but fluid intelligence peaks in early adulthood and then declines. Whereas crystallized intelligence remains more stable. But IQ has been further misused to promote the idea of racial differences in intelligence. There is, for example, an observed gap between the average IQ of black and white Americans. Articles have also been published on the IQs of different nations around the world. Many of these nations are purported to have average IQs below 70. That's the cutoff for intellectual disability.

How could this be? The conclusion that some draw is that there are genetic differences between races or nations in intelligence. But I think that's a gross misrepresentation of the data. The problem, I'd argue, is that IQ tests don't necessarily measure what you think they're measuring. And the proof is that there's a representative sample of white Americans whose average IQ is 70. Who are these people? Just ordinary Americans who lived around a hundred years ago.

Researcher James Flynn studied the average results of IQ tests over the past century, and every so often, the tests get updated and re-normalized to keep their average at a hundred. Now, what Flynn noticed was that each time they got re-normalized, the scores had to be shifted down a bit more by about two or three IQ points per decade. And if they didn't do this, what we would see is that the average IQ of the whole population was increasing at a steady rate for the last 100 years, adding up to around a 30-point increase.

This is known as the Flynn Effect.

• Were our immediate ancestors on the verge of mental retardation? Because 70 is normally the score for mental retardation. Or are we on the verge of all being gifted? Because 130 is the cutting line for giftedness. -

Now the genetics of the population haven't really changed over a hundred years. So what caused the increase? Well, there is some debate about the true causes, but one of them is probably improving childhood nutrition and health.

• You know, height also increased across that time period. People got taller, and taller, and taller. -

Another cause is better education.

• There's lots of evidence that school makes you more intelligent. -

You become better at problem-solving if you have more knowledge. Because it's easier for you to make associations if you have more things to make the associations with.

A third proposed cause is a shift in the types of work that most people do, from mostly manual labor a hundred years ago to much more abstract thinking these days. And that shift may have made us better at answering the types of questions that are asked on IQ tests. Rotate again.

The point is that IQ tests appear to objectively measure intelligence, but they don't. Even in the same country, separated only by time, cultural changes can affect the average scores on IQ tests. So why shouldn't we expect cultural differences between groups at the same time to have the same effect?

Some tests go so far as to label themselves culture fair, meaning the questions should be equally valid for all cultures. But the truth is, it's impossible to construct such a test.

Does that work?

• No. -

Okay.

• No. I mean, that's just a title, right? That's just a marketing term. I don't think there is such a thing as a completely culture-free or culture fair test. -

Culture fair tests assess visual relations, geometric shapes, and patterns, ignoring the fact that cultures differ in, for example, whether they have words for shapes or spatial relations. These differences influence how people think about and use categories. It's also debatable whether cultures without printed materials even perceive them in the same way that we do.

What culture fair tests don't assess is ethnobotanical knowledge, or training dogs to hunt, or surviving alone in the rainforest. Arguably, these forms of intelligence are more important for survival than knowing, say, the next number in the sequence. But since they're less common in our culture and we don't have good ways of measuring them, we see IQ puzzles as the definitive way to quantify intelligence.

And the people who make these tests agree. There are stringent requirements before a test validated for one population can be used with a very different population. Even in the limited forms of intelligence that IQ attempts to assess, there are factors other than g which affect the final IQ, like motivation. How much someone is incentivized to complete the test can have a marked impact on their score.

Many studies have tried paying subjects to complete an IQ test. In some studies, they're offered a little, say around a dollar. Other studies offer between $1 and $10. And the real high rollers offer more than $10. A large meta-analysis showed that motivating people in this way increased IQ. And the larger the dollar amount, the greater the average increase. At the high end, IQ increased by up to 20 points. The effect is largest for those with below average IQs.

So in addition to g, IQ tests also measure motivation. But it doesn't stop there. They do rotate. Training and coaching for an IQ test can boost scores by up to eight points.

I just completed the test in some random notice. (mumbles) I can barely talk after that. It seemed pretty fair. There were lots of different sections. The math section in particular, I feel like I killed. Those questions were easy. I would say having done the test, I feel like that should be trainable, like you should be able to train someone to do that well.

Test taking strategy is also important. Some people are just better at taking tests under time pressure than others. I think the hardest thing about the test was the time limits. Looking for the patterns in a series of shapes, I mean, it just normally takes me a little while, and so I feel like I didn't finish those. You have to know when to skip questions, how to eliminate clearly wrong answers, and when to guess.

Anxiety also plays a role. Apparently, a small amount of anxiety is good, but past a certain point, it negatively impacts performance. I guess the overall review is, I think I did okay, and I think the training actually really, really helped. That's my prediction.

Let us fast forward to the future and see how I actually did. I actually got my results from the author of the IQ test I took.

• Those are three areas, three specific areas. -

For the math one, for the numbers ones, I think that's where I felt really comfortable, and got there before the time was done and then I could go back and look at a few things.

• Yes. You blew the roof off the quantitative one. On the quantitative, it was 143. Whereas on the crystallized intelligence index, it was 132. The fluid intelligence index was 118.

• Hmm.

Which still is a higher score than 88.5%.

• It's not bad. But it's interesting that that one is significantly lower, I guess. -

And that's not an unusual difference. Around that concept of g, people have strengths and weaknesses. If we were to look at the best estimate of g for you on this set of tests. And it'd be different if you took a different test. It was at 134, which is higher than 98.8% of the population.

• Wow.

Hopefully, you're not disappointed with any of that.

• No. I wanted to do well. I feel like my motivation was high, possibly higher than the average person.

So what is IQ good for?

• My clinical practice now is forensic neuroscience, and about 90% of my cases are death penalty. One of the most common issues is what's referred to as the Atkins defense after the name of the US Supreme Court case that eliminated the death penalty for people with intellectual disability.

• Can't the criminal just throw the IQ test? Can't they just intentionally -

No.

• answer every question wrong?

• We'd know that. I mean, we include, just like in the test that you took, there are embedded measures of invalidity. It's detected using various mathematical algorithms. We're better than 95% accurate in detecting people who are attempting to fake poor performance.

• Oh, wow.

One thing that we may be interested in doing is to boost people's cognitive ability early in life so that it takes them, even if they go into cognitive decline, it takes them longer to reach the point where they'll have sort of functional actual everyday problems where they lose independence, whether it comes to dealing with their money, or whether it comes to dealing with reading labels, whatever it is that people struggle with when they get into later stages of cognitive decline. If we could discover a way to lastingly boost people's intelligence, that would be massively helpful.

• Maybe its best use is in identifying individuals with strong intellectual abilities who haven't otherwise been able to demonstrate them.

• Teachers would recommend that a kid gets put in the gifted and talented program because generally, they'd observed them doing well in the classroom. But if you replace that with a standardized test, an IQ test, you find a higher proportion of poorer kids and kids from minority ethnic backgrounds in the gifted and talented program when you use an IQ test. And the reason is that you're using an objective measure. You're not just relying on some teacher's opinion. Getting into a good school was about who you knew, or who your parents knew, or how much money your parents had, not so much about how you were doing.

The idea that you could try to develop an objective-ish measure that would try and iron out all those social biases was clearly a well-meaning idea.

• IQ is something that not only psychology, but the general public has a love-hate relationship with.

• Tell me about that.

• Psychologists hate to talk about intelligence and people's intelligence test scores and that kind of stuff. And I've had parents, when I've included intelligence as part of a neuropsychological evaluation of their kid, they'll say, "Well, yeah, I'd like to know his IQ, but you know, we don't really care about that. What was it?" (both laughing)

And then I also think you have the debate about IQ, extremes on both sides, which I think doesn't help. You have the extreme of people who say this is the most important thing ever, people's IQ is a majorly important factor that we must know about them, and then we can classify them into particular schools, or particular ways of education or whatever. That's one extreme, and I think that's totally unproductive. But there's another extreme.

The other extreme is the kind of blank slate view, which is that these tests are completely useless, they don't tell us anything, that they're only a tool of racism and prejudice and so on. I think that's wrong as well. And there's just this massive firestorm on both sides happening around them. And the people in the middle just get forgotten, the people have more moderate views on these sort of topics. So I'd recommend that people look for the more moderate views on this.

• I think the big mistake is thinking that IQ in some way determines someone's worth. What's much more important, in my opinion, is how you interact with and help the people around you. Which is why I think Stephen Hawking said, "People who brag about their IQ are losers." While IQ tells us something, it doesn't tell us how our lives will turn out. We have the ability to dramatically improve our outcomes by building knowledge and analytical skills.

And if you're looking for a free and easy way to do that, then you should check out this video's sponsor, brilliant.org. With Brilliant, you can master key concepts in everything from data science and math to programming and technology. Just set your goal and Brilliant designs the perfect learning path for you, equipping you with the tools you need to succeed. They also helped me understand how IQ and net worth are related by creating some amazing data visualizations.

Want to learn how to think like a data scientist? Then Brilliant's latest course, Exploring Data Visually, is a fast and easy way to get there. It requires no coding skills, and instead, you'll focus on the whys behind the data and analyze real data sets to learn things about the world. In today's world, data skills aren't just nice to have, they're essential. Brilliant is more than just a learning platform. It's your path to understanding the world of math, data, and computer science.

And what's great about Brilliant is that all the lessons are hands-on. So you build real intuition so you can put what you've learned to good use. One thing I didn't include in the main part of the video is that the only aspect of personality that correlates with IQ is openness to experience. The more you're willing to give new things a try, the more you'll learn and the smarter you'll be.

So to try everything Brilliant has to offer for free for a full 30 days, visit brilliant.org/veritasium. I will put that link down in the description, and the first 200 of you to sign up will get 20% off Brilliant's annual premium subscription. So I want to thank Brilliant for sponsoring this video, and I want to thank you for watching.