Transcript for:
Understanding Intelligence and Validity in Psychology

Okay everybody, let's get started. So once again, I am behind where I wanted to be, so we will probably not finish the chapter on intelligence today. We'll have to start Tuesday's class by wrapping it up. And then Tuesday we'll be starting chapter 12 on development. developmental psychology.

And after that it's it's not in a random order. After 12 we just go straight to the end. That's what we'll be doing the rest of the semester.

A couple of you were really interested in what the right answer was. I don't I don't have a key, but I looked at it after class, I'm pretty sure it's seven. But I'm not Einstein, maybe I'm getting it wrong.

But anyway, those of you who are obsessed with that, pretty sure that's the right answer. Okay, so we ended last class by finishing the chapter's discussion of reliability, and now we'll be talking about different types of validity. I'm going to be adding a few to what the book says, and you will thank me later if you take future psychology classes. So, remember that reliability... and wrap up your conversations even if you're special.

Remember that reliability is about how consistent a test is. How consistent are the results you get from the test. Whereas validity is about, is the test measuring what it's supposed to measure? And reliability is kind of easier to define. Validity we have various more indirect ways of getting at it.

The first is one the book doesn't mention, but in the early stages of test creation, it is very important, and that is face validity. It's not in the book, but it's good to know. Face validity is just subjectively, when you look at the test, does it look like it's measuring the right thing? And when you're first creating test items, this is basically what you use.

We haven't done any fancy statistics yet, we haven't done any rigorous theory, but we just ask, okay, does this look like it's measuring the right thing? So, if it's an IQ test, okay, you got a test, has a bunch of math problems, has some vocabulary problems, some spatial problems. It looks like it's measuring intelligence.

You measure someone's left pinky toe, that really does not... look like it's measuring intelligence. That seems like a pretty dumb way to do it.

This relates to a term you'll see in law, which is prima facie. You don't need to know that term, but prima facie means at first face or first glance. And that's basically what it says.

You just look at the test. It doesn't seem like it's measuring what it's supposed to measure. I'm going to go in a different order than the book.

The next one is content validity. Content validity is about, is the test measuring the full range of what it's supposed to measure? As opposed to just a narrow subset.

So, you know, at the end of this course, there will be a cumulative final exam. If all the questions come from Chapter 13, the test would not have content for you. It's only measuring a narrow subset of what it's supposed to measure, whereas you want it to measure full range. The fancy way of saying this is the test items should be a representative sample of the universe of content, but you don't need to know that.

If you want to sound really pretentious, you can say that. The next one is criterion validity, which is about... Does the test correlate with something in the real world that it should correlate with? So, something in the real world would be like a criterion.

So, for example, let's say there's a measure of aggression, some kind of aggression scale. If it's really measuring aggression, it should correlate with your likelihood of winding up in prison, right? If there's a measure of extroversion, it should correlate with how many friends you have. Not perfectly, the criterion validity will never be 100%, but it should have some correlation with those real world measures.

So the big one that's used in IQ is grades, right? If IQ tests are measuring intelligence, they should be correlated with grades. Not perfectly, but if they don't correlate with grades at all, well, kind of silly to say that it's measuring intelligence. This further breaks down into two subtypes, and these are not in the book, but these will help you... with some terms later in this chapter.

And those are concurrent and predictive validity. Really simple. Concurrent validity is about, does the test correlate with something in the real world right now?

And predictive validity is, does it correlate with something in the real world in the future? In other words, is it predicting something? So, I give you all an IQ test, and I look at what your grades are right now, and I see if they're correlated, that would be concurrent validity. If I give you an IQ test now, and then measure your grades in three years, that's predictive validity. Okay.

The last one is one that is in contemporary psychology probably the most important, but it's also the most controversial. And that's construct validity. Construct validity is about whether the test is measuring the right theoretical construct.

So in the case of intelligence tests, we have a theory about what intelligence is. Like we said at the beginning of the chapter, we mean roughly, by intelligence, your ability to reason effectively, to acquire knowledge, and to behave in the environment in a way that's adaptive. Okay, so that's what intelligence is in psychological theory. Okay, do IQ tests measure that?

Do they measure that theoretical construct or not? An imperfect analogy would be one with temperature, which I mentioned last class. So in physics, temperature is defined as molecular motion, so essentially how much kinetic energy do the atoms have, like how much are they moving.

So, you know, if you heat something up, the atoms are moving more. So that's the theoretical construct, if you like. That's how, theoretically, temperature is defined.

And for a thermometer, the question would be, is the thermometer measuring that? Is it measuring how fast the atoms are moving? And the way psychologists think about it is these other types of validity can be indirect pieces of evidence for this kind.

If you have a test that satisfies these, that gives you more confidence that the test is really tapping into the right theoretical construct. As I say, this is controversial. One of my professors in undergrad was very critical of this notion.

He thought it was kind of too theoretical and we should focus on the types of validity that are more precise. But in contemporary psychology, construct validity is kind of what we're aiming at. We're aiming at having tests that measure the right theoretical construct. Okay, so I'm going to talk now about how IQ does on some of these counts, but any questions about what these types of validity can mean?

Okay, the type of validity that I'm going to focus on in the case of IQ is this last one, criteria of validity. In my experience, this is the one that psychologists are most impressed with. Even if in the back of their head what they're really aiming at is this one, the one that you can't really argue with is this one.

This one's not theoretical. This is just like, okay, does the damn test correlate with X or not? So in the case of grades, in the case of high school grades, the correlation between IQ and GPA is about 0.6, is what the book says, which is pretty good for a psychology test. Correlation's that higher. That's on the higher end.

For university grades, apparently it's a bit lower, maybe more like 0.5. So again, for psychology that's pretty good. And I'll actually show you what that looks like. So here's a data set, this is real data, looking at the correlation between IQ and grades. So as you can see there's a lot of randomness, there's people with lower IQs that still have high grades, but there's a general trend.

that the higher your IQ, the higher your GPA is on average. For things that are less directly related to intelligence, there's still a relationship, but a smaller one. So this is a correlation of about 0.3. between IQ and income. And this one's point three, you kind of have to squint at it a bit to see the relationship.

Like if you look at it, like around here, there's kind of a tighter cluster. There's a higher concentration of dots down here in low IQ and low income, but there's still people with low IQs and high income, and people with high IQs and low income. So there, the relationship is weaker, though still not zero. Interestingly, the correlation between IQ and wealth is basically zero. So income is like how much you make every year, and there there is a relationship, which kind of makes sense, right?

Some people will be better at getting cognitively demanding jobs and therefore make more money. But since wealth is inherited, basically no relationship with IQ. Now I want to pause at this point and clarify something.

I think it's... I think that these numbers might be scaring some of you, and I want to make them a little less scary, because two things are true at the same time. On the one hand, IQ is probably the best measure we have in psychology, certainly one of the best, but it still doesn't explain that much.

The fact that IQ is the best measure is really just a reflection of the fact that we don't know very much in psychology. So let's take the correlation with grades. which as I say is around 0.5, 0.6, kind of depends upon the study, but it's typically around there. Let's take 0.6, which is the number that the book gives.

What a correlation of 0.6 means is that For every, and by the way, I won't test you on this super specifically, this is just to give you kind of an intuitive sense of what these numbers mean. It'll help you in your statistics classes. What this means is, for every standard deviation increase, in this variable, there is a 0.6 standard deviation increase in the other variable.

So you go up one standard, let's say it's IQ and GPA, which is what we're talking about. standard deviation you go up in IQ on average you'll go up 1.6 standard deviations in GPA. So to make that concrete I looked up what the standard deviation for a GPA is and it's about 0.65 like on a scale of up to 4. So that means for every standard deviation increase in IQ on average you'll go up one letter grade like from a B plus to an A.

lines. Okay, that's something. But also it's not that big of an effect. It's been for psychology, but it's not the be-all and end-all. Another thing that you can do, this is really useful to know if you're looking at correlations in a study.

So this is, the term for correlation is R, like Pearson's R. That's a really bad R. But what you can do with an r is you can square it. For some reason they capitalize the r, but r squared, so you square whatever the correlation is, and you get something called the coefficient of determination. Again, I will not test you on this, but this may help us all relax a little bit about it.

So if you square a decimal, it gets smaller. So r squared in this case is 0.36. Here's what that means.

R squared is how much or what percentage of the variation is explained by the relationship. So in this case, what this means is 36% of the variation in grades is explained by its relationship with IQ. So, okay, so of all the variation in grades, from the lowest to the highest, there's all this variation, about 36% and maybe less, because some studies find it's like 0.5, about 36%...

60% of that variation is explained by IQ, which means 64% of the variation is stuff other than IQ. So IQ is real, it does measure something that gives us predictive validity. but it's not everything. You'll sometimes hear people say that IQ tests only measure your ability to take IQ tests. That is not true.

Or if it is true, it's true of literally every test in psychology. By the standards of psychology, it is measuring something real, but it's not everything. So hopefully that is helpful. Another thing the book mentions that may help you think about it, a correlation of 0.5 is about the correlation between height and weight.

Taller people are on average heavier than shorter people, but you've all lived in the world, there's obviously a lot of variation. There are short people who are heavy and tall people who are not heavy, but there is that general trend. Okay, that's all I have to say about validity. Any questions about that?

Actually, I sort of want. This is sort of related to validity. This is another.

So if you look at occupations, this is the 10th percentile to the 90th percentile for each job, kind of ranked by cognitive difficulty. So you may not be able to see this, but you'll see a general trend where the cognitive difficulty is ranked by the cognitive difficulty. more demanding jobs have a higher range, but for each job there's still a huge range.

So even like, you know, like lawyers, there's around 10% of lawyers have IQs of under 100, if you zoom in, that's what's going on. So that also reinforces the point I was making. As I mentioned last class, you should note these numbers.

So the next section is on standardization. Every few years, IQ tests are re-standardized. And they are always standardized such that the mean is 100 and the standard deviation is...

They just force the test to be like that. But you have to change the test every few years to make sure it still looks roughly like this. And as you can see, most people fall between 85 and 115. That's where the means of the...

So this is called a normal distribution. Also sometimes called a bell curve, informally. And it's just kind of an interesting fact about the world that for many traits both in humans and actually animals too Variation tends to work out to roughly this. It is actually an equation for the normal distribution And it is typically a pretty good approximation for how things fall out so personalities like this various other traits But one really surprising thing, oh and like I said last time, don't worry about the one in the book, there's one, well you probably can't see it from here, but the one in the book breaks it down by 90, 100, 110, etc. Just ignore that.

This is more useful to know, so for the test you should know these percentages, how many people fall in the various ranges. But now remember that I said that this is re-normed every few years. So every few years, they change the test. But a very important psychologist, well, he's kind of a psychologist, he actually did his PhD in philosophy. So I like this guy, he's like made his book, Psychology and Philosophy, named James Flynn.

who actually just passed away recently. He looked at, well, what if we forget about the re-standardization? What if we just compare the raw scores on specific tests across the generations? And what he was interested in was, is IQ going up or down? If you ignore the re-standardization and just look at raw scores.

And it turns out, very surprisingly, IQ has gone up pretty dramatically. over the last hundred years. How much? Since we've been doing it, around three points per decade, which works out to about 30 points, which is a huge effect. So 100 is the average.

70 is typically used, at least by convention, as the cutoff for having an intellectual disability. So 30 is a huge effect. And that's just in 100 years.

There's debate about what that means. means. It seems kind of strange to say everyone born in 1900 had an intellectual disability. But whatever the case may be, there's no question this is true, that IQ has gone up very dramatically. And one thing that James Flynn used this to argue for is that IQ is substantially influenced by the environment.

We will be later seeing in the chapter evidence of a genetic component as well, but evolution... Evolution does not take place over decades. It takes a very long time for gene frequencies to change in response to selection pressure. 100 years is simply not enough time for, you know, IQ genes, quote, to change very significantly. So the Flynn effect, as it's called, so the term for the fact that IQ has gone up over the last 100 years is the Flynn effect.

That's how you know you're an influential scientist, when you have a good vet named after you. And the Flynn effect has to be due to the environment. Now, what is it in the environment that's caused it?

It's probably multiple things. There's ongoing debate about it. Probably nutrition has something to do with it.

If you are malnourished, your brain... simply does not develop as well. And over the last 100 years, it's been improving nutrition.

Not as many kids are malnourished. Education certainly contributes. And Flynn himself argues that cultural shifts have also been responsible.

So we'll actually watch a TED Talk, or some of a TED Talk by Flynn, explaining what he thinks the increase is due to. Okay, so well said by Flynn. That TED Talk's on YouTube if you want to see the rest of it.

So what he emphasizes in his explanations is, as he said, various habits of mind. Habits having to do with taking the hypothetical seriously, using logic, and so on. Okay, but the key takeaway from the Phelan effect is what I mentioned earlier.

It conclusively demonstrates that IQ can be influenced by the inquires. There's one thing we know for sure the Phelan effect is not due to, and that's genetic changes. The book briefly mentions the difference between static and dynamic testing. I'm not going to say too much about that.

The difference is, most of the tests you have taken in your life are static tests. It means you're given the test, you respond a certain way. and you get a score. In dynamic testing, it's done live and the test giver can give feedback and part of the test is about how you use the feedback. But as I said, I don't have much to add to that, but that's the distinction there.

The next section is about the issue of cultural bias and cultural fairness on the tests. One early allegation against IQ tests is that they were culturally biased, particularly the verbal items. For starters, you have to be able to speak English if the IQ test is given in English. If English is your second language, that would be a case of potential bias. But there were also specific items that seemed to...

least prima facie be tracking class more than intelligence. One famous example is a question something like this. You were given various pairs of words that had some kind of connection, and then you had to choose which, for multiple choice, which of the pairs best corresponded to it.

So the question was, a runner is to a marathon. What a blank is to a blank. And you had a bunch of options like, I think I wrote them down, an envoy is to an embassy, a martyr is to a massacre, and a bunch of things.

And the right answer was an oarsman is to a regal. A regatta is a boat race. And oarsmen are people who are racing the boats.

It's kind of a rich man's game. If you grow up in poverty or are part of an upper class culture, there's no reason why you know what the hell a regatta is. So people pointed to questions like this and said, this isn't really measuring intelligence, this is measuring exposure to white upper class culture. This is something that the creators of IQ tests take seriously.

There is an effort to get rid of items like this. I would say the consensus nowadays is that modern IQ tests are significantly better on this score, but this is something that has been brought up. Another possible reaction.

is to say, well, forget about the verbal items. Just focus on things like the Raven's Progressive Matrices. Surely those are more culture fair. And at least prima facie, it seems like this would be something that doesn't require exposure to a particular culture to be able to understand what's going on.

And that seems right, but this is complicated by the fact that, believe it or not, the Raven's Progressive Matrices score... on the Raven's Progressive Matrices have actually improved more than other items. You may have suspected that the types of IQ tests that people have been getting better on were like, you know, how much stuff do you know, like crystallized intelligence, but if you look at this, this is the Flynn effect over a narrow window, so 47 to 2002, the one that's improved the least is information arithmetic and vocabulary.

That has hardly improved at all. If you look at the top, Raven's Progressive Matrices has improved. There's no consensus as to why that is.

I just point this out to say it's not always clear which parts of the IQ test are more influenced by the environment. You can't just trust your intuition. Okay, so that's some information about environmental contributions to IQ. We'll now talk about some of the studies into biological correlates.

And the first bit there is on different brain correlates. One idea that goes back to Francis Galton... Is that IQ is in part a measure of neural efficiency.

Just how efficient is your brain at dealing with problems? And indeed, if you measure neural efficiency in different ways, so one way you can measure it is you hook someone up to EEG, you show them visual stimuli, and you can measure how fast the electrical activity occurs after the presentation of visual stimulus. And that really crude measure does correlate positively with IQ. You can also measure how much glucose the brain uses when solving a problem, and the less glucose your brain uses to solve a problem, the less glucose your brain uses to solve a problem.

a problem, the higher your IQ on average, which at least prima facie seems to indicate that if your brain is more efficient, more kind of quick at dealing with a problem and requires less resources to do it, that enables you to solve problems more effectively. There's also evidence that brain plasticity is related to IQ, so how good your brain is at growing new connections and adapting to change. And probably the most controversial way of measuring, or measure that's poorly used, is brain size.

There were some early, really crude attempts at doing this, but if you use more sophisticated measures today, it depends what kind you use. So, if you use MRI, The correlation between brain size and IQ is about 0.35 to 0.45. The book mentions another study that found it was 0.19. So, you know, it depends what study you look at. But not especially high, even if you take the upper range.

However, if you look specifically at the G factor, remember that the G factor is... tests all tap into, right, general intelligence. The correlation is more like 0.6 between G and brain size as measured by MRI. One paradox, though, that I think I mentioned in passing back in Chapter 9, is that male brains are larger than female brains on average, but the average IQ... is the same.

One possible explanation for this that the book mentions has to do with the difference between gray matter and white matter. So grey matter, if you look at the brain when the blood is out of it, you'll see grey sections. Grey matter consists of clumps of neurons, like cell bodies. White matter is clumps of... axons, so connections between regions.

On average, men have more gray matter and women have more white matter. The difference is not especially big, but what that means is there's more dense interconnections between regions in women, whereas more raw number of cells for men. And so maybe that's what explains the fact that there's a difference in size, but no difference in average IQ.

Okay, next, another issue of biological contributions to IQ is the section on heritability, so genetic influences on IQ scores, and here I'm going to have to say not as much as I'd like, so I mentioned that I covered this in more detail in 104 in the section on behavior genetics. So we're going to be taking a more cursory look at this, but it will be enough for you to do well on the test. So heritability is one of the three Sources of human variation in behavior genetics. So the way behavior genetics has studied human variation is they apportioned the variation to three sources.

Heritability, the shared environment, and the non-shared environment, sometimes called the unique environment. So, heritability is genes. So, how much genes contribute to human variation in traits.

The shared environment is the environment that you share with siblings. So, how nice your parents are. are, how much money they have, what neighborhood you live in, what school you go to, those kinds of things that you share with siblings will fall under the shared environment.

The non-shared environment is everything else. Everything else that contributes to differences in psychological measures is the non-share. So, you're a particular group of friends, a particular teacher you had who had an effect on you, random stuff, like you fall off a bike and you hit your head really hard, you know, that's not genes and it's not the shared environment, so technically, that's part of the non-share.

So I gave three caveats to this in 104, and I'm going to focus on two. For heritability, two really important caveats about this. So heritability, all of these are measured from 0 to 1. They just add up to 1. So if a train is 50% airable, we would say this is 0.5, for example. But crucially... 1. Heritability numbers are about individual variation in a trait.

It's about differences between people. So take height. Height is very heritable. Something like.8 to.9.

80 to 90% of the differences between people in height are due to genes. But that does not mean like 80% of my height or this guy's height or whoever. is due to genes.

That doesn't really make sense. It's not like this much of my height is genetic. What it means is if we were to take this class and rank and order everyone from shortest to tallest, 80% of the differences between people are due to genes.

That's what heritability is about. And number two, heritability can change based on the environment. In other words, heritability is not fixed. It's not like a law of nature or something.

So again, take height. Height is very heritable in most studies. But if you go to a part of the world that's very malnourished, the heritage...

Heritability will actually be lower. So I'm aware of studies in rural India that find that the heritability of height is more like 0.4 as opposed to 0.8. Because if it's a region that's very malnourished, people are not reaching their full genetic potential and therefore there's less genetic contribution to the variation in the trait. This is also true within Western countries.

You can just look at socioeconomic status in Western countries and specifically in the case of IQ, IQ is more heritable for wealthy people and middle-class people than it is for poor people. That might sound kind of strange but what the way to think about that is if you're more poor you're not reaching your full genetic potential and therefore there's less genetic contribution to variation in the trip. Okay, so very important caveats. So how heritable is IQ?

There's many ways of estimating it, but the two most common are two different types of twin studies. How many of you know about twin studies? Remember twin studies from 104. So, the easiest one to understand is you look at identical twins who are raised in the same family.

in separate families. There's problems with this method but it's kind of actually pretty compelling. So what you do is you look at identical twins who are adopted and then raising different families and you look at how similar their IQ scores are. If genes make no contribution to IQ the correlation should be zero.

They should have nothing in common. But in fact, identical twins raised in separate families are highly similar for IQ. Another method is to look at twins raised in the same family, but to compare the similarity of identical twins with the similarity of fraternal twins, or non-identical twins. If genes contribute to a trait, then identical twins should be more similar than non-identical twins.

that trait because they share all the environment, right? Identical and non-identical twins share their parents and everything else, but identical twins have the bonus of also sharing twice as many genes, right? Non-identical twins share half of their genes.

Identical twins share all of their genes. So when you use these methods, here's what you find. In childhood, IQ is around 50-70% heritable, meaning 50-70% of the variation in IQ is due to genes.

Funny enough, it actually goes up with age. In adulthood, it's as much as 80% heritable, meaning approximately 80% of the differences between people in IQ are due to genes. That might seem strange to you, given that I just finished saying that the Flynn effect shows there's a very large environmental component, but here's the way to think about it.

So let's say this is the bell curve for IQ in 1920, and this is the curve now, something like that. If you just freeze at 1930 and ask how much of the differences between people in 1930 are due to genes, you'll get a high number. But remember, heritability is only about how much of the differences between people at any time. given time are due to genes. And the same is true for math.

If you just freeze today and you say, OK, this guy has a higher IQ than that guy, why is that? Most of the answer is genes. But the reason that the average has gone up is due to the environment.

That's the way to keep that consistent. One analogy that might help you. Something very similar is true for weight. So weight is also very heritable.

Maybe 50-70% heritable, depends on the study, but certainly the heritability is not zero. But weight has also been going up over the last hundred years. And that changed, the change in the average is due to the environment, but if you freeze today and ask why is there this much variation, much of the answer is genetic. Oh, I forgot to mention, you will not be tested on this, but if you're curious, the way you calculate an... An estimate of heritability from twin studies is this equation.

Again, this is just if you're interested in this. You take the correlation for monozygotic twins, meaning identical twins, so R and Z, just meaning R means correlation, monozygotic twins, minus the correlation for dizygotic twins, and multiply that by 2. And in the case of IQ, what it is is the correlation for mono-zegar twins raised in the same environment is 0.8. And for fraternal twins, it's 0.4. So 0.8 minus 0.4 is 0.4.

multiplied by 2, that's 0.8. Sorry if people here can't see that, but you won't be tested on this. If you kind of work through it, you can get an intuitive feel for why that's the case. It's based on a very simple model with very simple assumptions, but that's how you get an estimate from twin studies.

With twins raised apart, it's easy. You just look at the correlation between identical twins raised apart, because they share none of their environment, but all of their genes. Okay, any questions about what I've said about Herod a bit?

Okay. The next section is probably the single most controversial topic in all of psychology, arguably the single most controversial topic in all of the social sciences, and that's ethnic group differences in intelligence. This is obviously a very sensitive topic, but I will go through it carefully and might not be as scary as it seems at first glance.

So if you give large samples of people IQ tests, and most of these studies are done in the United States, you will find average differences between F4s. Now just to show you what that looks like. Crucial word there is on average. They look something like this, right? Overlapping bell curves.

So for any ethnic group, there will be people at every level of IQ, but there is a difference in the needs. What the book says is, so again, most of the study found in the United States, for African Americans around, yes? Well, you can just calculate it.

You just calculate the mean. So in this case, yeah, the peak of the distribution is typically where the mean is, but I mean it's easy to do. You just get a big sample out of everyone's scores divided by the number of scores.

It's just calculating an average. So for African Americans, the average used to be around 85. There's more recent studies finding it's more like 90. The gap has narrowed to some degree, but still persists at around 90. For Americans of European descent, it's around 100. For Americans who are East Asian, around 103 to 105. And for Jewish Americans, there's fewer studies, but around 107 to 115. So that's roughly where it falls. The book mentions that apparently in Japan...

least for children it might be as high as 111 so there are these average differences so no one disputes that that's just something you can find doing studies the controversy surrounds is this due to test bias and if not what is it that causes these differences so I'll start with the issue of test bias The book mentions two different definitions of bias. Two different types of bias. Predictive bias and outcome bias.

And I'll start with predictive bias because it's easier to define. Predictive bias connects back to that term we saw earlier, which is predictive validity. Does the test predict things that it should predict? And what you can ask is, does the test predict performance for members of different ethnic groups equally well? Or is the test under-predicting the performance of certain ethnic groups?

So as an analogy, there is some evidence... But this is true of the SAT for gender, not for race or ethnicity. Men score slightly higher than women on the SAT, but women get better grades in college. So the whole point of the SAT is to predict how well you'll do in college, and in the case of gender, it seems that the SAT, and there's debate about this, but you can argue based on that, that it's predictively biased against women, because they have a lower average on the SAT, but they get better grades in college. So it's not doing as good of a job at predicting how well they'll do in college.

So you can measure whether that's the case for ethnic groups as well. Is it under predicting performance? And in the case of IQ it seems to be that's not the case. It seems to be that the predictive validity is equally good for different ethnic groups. However, there's a very good case to be made that the tests have what the book calls outcome bias, which means that given that people are exposed to different life experiences and have different degrees of deprivation, the test may be not measuring someone's true intrinsic ability.

They may have not reached their full genetic potential due to deprivation required. Now, this is not a sociology class, but... So in the case of the black-white gap, for example, many have argued that it's due to systematic deprivation of the environment in the United States, that there is this average difference. So one thing that I'm sure many of you know about is, for a long time in the United States...

States it was impossible or nearly impossible for African Americans to get loans for housing and if you can't get loans for housing and you can't live in a neighborhood that has any degree of economic stability schools are funded by local property taxes so then you go to a bad school with not as much resources and also your family can't accrue wealth over time. I lived in Atlanta for almost a year. I was supposed to live there for two, but COVID happened, so I did the second half of my master's on one.

And you can see the long-term effects of this in the United States, especially in the South. Atlanta was around 50% black, 50% white, but the homeless population was almost 100% black. Anytime you went downtown, that's what you saw.

So if things like that are negatively affecting people's ability to achieve well or do well on IQ tests, that would be an example of what the book calls outcome bias. It's not getting at someone's true ability. or at least they haven't reached their full potential.

Okay, so those are two types of bias. As I said, there's no controversy about whether these differences exist, but a controversy that persists is what explains the difference. So there's a history of people talking about this. One book that came out in the 1990s, the text was... mentions is the bell curve this was very controversial and in there they did argue that a proportion of the black-white difference was genetic but the book does not mention that the APA the American Psychological Association actually published a critique of the bell curve.

It was written by a panel of experts. And what they argued in that paper, it's still worth reading, I think it's called Intelligence, Knowns and Unknowns, something like that. And they agreed that IQ is, in general, heritable, but they said that the evidence points against the conclusion that... the ethnic group differences are heritable.

They mention two studies that I'll briefly describe. These are relatively small studies, but they certainly point against the idea of ethnic group differences being genetic. One was done in Germany. It was done right after World War II.

So, as, I don't know if you guys are history buffs or not, but World War II, Germany was taking over much of Europe, and the Americans, Canadians, the British, and the Russians invaded Germany from both sides, and overthrew the Nazi government, and for a while after they occupied Germany. And as happens sometimes, some of the local German women are hooked up with some of the American soldiers. And some of the soldiers were white and some of them were black. And then once their tour was over, they went back home and these German mothers were left single with these children.

And this is a nice study because these kids are all raised the same way. They're just raised by German single mothers in Germany. The only difference between the kids who had a black father from the United States and the kids who had a white father from the United States is the genes.

Otherwise, they're raised in the same environment. And in that study, you don't find a difference. That study found no difference between... like half black kids and white kids.

Additionally, there were some studies of degree of white ancestry in the United States. So most African Americans in the United States have some percentage of white ancestry. And what you can do then is you can look at, does percentage of white ancestry affect IQ scores. Because if it's, you know, quote, European genes and not the environment, then having more European ancestry should increase your IQ. And there were two studies that found no relationship there.

So these are the studies that the APA pointed to as evidence against the idea of a genetic contribution. As I say, and actually as the EPA says too, these studies are small and not conclusive, but they point against the idea of a genetic contribution to the, specifically the black-white difference in the United States. So that's still somewhat uncertain, but I want to close this section with four things from the book that we know for sure about these differences.

One, which I've already mentioned, but I'll say it again just to hammer it home, these differences are only on average. Every group will have people at every level of IQ, the differences only exist, as I say, on average. One example I used to use with my AP psych classes to hammer this point home is think about the last three presidents of the United States.

Someone tell me the names. Who are the last three presidents? Including the current.

Joe Biden, Donald Trump, Barack Obama. If you have not listened to the three men speak, I encourage you to do so. And I'm not going to make any political point about whose policies are right, but as far as intelligence, I don't think even someone who's conservative would deny that Obama is much smarter than the other two.

And if you don't believe me, you have to listen to them speak for about five minutes. You'll be convinced. Which reinforces the point, right?

Every group will have people at every level. Number two, and this is one that the book has a very good discussion of. It's a little bit tough to get at first, but very important. Heritability of a trait does not imply that a group difference does not imply. ...that a group difference is paramount.

Because some of you might be thinking, well wait, don't these ethnic differences have to be genetic? Because you just finished saying that twin studies and other studies show... that there's a genetic contribution to IQ differences, but that does not follow. Twin studies are not about showing that group differences are due to genetics, they're about showing that individual differences are due to genetics.

The fact that we can find heritability for a given trait does not imply that group differences are. And here's a classic example, the book gives this too. Suppose I have two patches of soil, good soil and bad soil. I take a fistful of seeds and I randomly put them into these two environments.

So the fistful of seed, it's the same stalk of seed. There's not any genetic difference between the two groups. And unsurprisingly, the good soil will have taller blades of grass.

I would say it's grass seed. And the other one will have shorter blades of grass, long-ass. Now, if I look within any given flower bed, I will find that heritability of height is very high. The reason this one is taller than this one is going to largely be due to the genetics of the seeds. But the average...

difference between these two is, in this example, totally due to the environment, even though height in blades of grass is heritable. That's an example to illustrate this. Heritability of a trait does not necessarily imply that the group difference in that trait is heritable.

Yes, epigenetics is just one possible way that the environment can influence a trait. It's not necessary for this point. For those of you who don't remember, epigenetics is a way that the environment can influence a trait. geneticist means that gene expression can be modified by the environment and that gene expression can also be passed on so you can have a case where the letters of the DNA code don't change but the way they're expressed is changed and that gene expression can also pass down but you will not be tested on that.

If this point seems a little subtle, spend some time with it. I have a discussion of it in the PowerPoint notes. And there's a discussion of it in the textbook, but it's a very long one.

Point number three, we know that the environment can produce... changes as large as the average. And one way we know this, we've already talked about it, is the Flynn effect.

The Flynn effect is entirely due to the environment, and it produced a change of around 30 points over the course of 100 years. James Flint himself has actually written about the black-white difference. That's the one that there's been the most written on. And has said that all you have to assume in order for the black-white difference to be entirely environmental is that the environment for African Americans today is as enriched as it was for whites like 40 years ago. Which is not that difficult to believe.

The book also mentions that there are various European groups who used to have IQs of around 85 that no longer do because now they're part of the mainstream culture and don't have as much poverty. So one example is Italian Americans. Apparently they're like used to be about a standard deviation below the average, so around 87, and now it's actually slightly above average because early Italian immigrants were very poor and were kind of culturally separated from the rest of the United States.

And then the last point, which I'll close off this section with, is we know that genetic differences... Between ethnic groups are small. The book mentions this. So I believe I mentioned this in my 104 class.

All humans on earth are 99.9% genetically identical and if you zoom in on the 0.1% of differences I've seen different estimates from geneticists. I'm not a geneticist. But something like 5-10% of all, since this would be all of human genetic differences, all of the 0.1%, 5-10% of that is between group differences. So differences between different continental groups.

So most of human genetic variation exists in every group. The way this is sometimes being put, if everyone on Earth, except for people in Kenya, or whatever country you want, died, more than 90% of all human genetic variation would still exist. Most of the variation is within groups, not between groups.

As I say, I am not a geneticist, so you don't have to take my word for it. We're going to watch a video by the biologist Richard Dawkins, who is arguably the most important living evolutionary biologist. talking about the point that genetic differences between human groups are actually small.

And I think this is, it doesn't disprove that there's any genetic contribution to these differences, but it should make us skeptical that there is such a contribution. And that will probably take us to the end. So watch this video, and then I'll quickly close this off. Okay, so we'll stop there.

I got farther than I thought I would, so next time we'll finish the chapter at the beginning and then start the developmental side.