Race and IQ

I think Otto T. Goat is stalking me:

According to him I am incapable of comprehension, as regards the paper he posted supporting his race realist position.

Let’s test that theory, shall we?

This link should at least take you to the right blog post, if not the right comment. But first, a definition for those who were not party to the discussion, or who are unclear on the terminology. (Also, there is a glossary at the end for some of the statistical terms that might be unfamiliar to some readers.)

Race realism (also called (pseudo-)scientific racism), or just racialism is defined by Rational Wiki, here:

Racialism is the idea that humanity can be easily divided into well-defined groups (or “races”) that are both broad (each “race” should include many humans, such as entire continents) and specific (the categorization method should almost never misidentify someone’s “race”). Racialism implies that these races are substantially different from each other and that these racial differences strongly determine the abilities and behavior of individuals and peoples. Essentially, racialism argues that human populations are substantially different from each other to a degree which necessitates biological classification below the species level. In addition, most racialists maintain there is also a ‘racial hierarchy’ of best and worst races.

Rational Wiki goes on to put Scientific racism in its historical context (I’ve included the original citations):

Racialism was virtually unchallenged until the 1960s, when it was shown to be erroneous by genetics. Simply put, modern racialism is pseudoscientific bullshit.[i] [ii] [iii] [iv] As Lieberman et al 2003 write:

“The rejection of the race concept […] in the 1960s[…] was based on the genetic evidence reviewed earlier. Conformity to political correctness was not the cause of these changes[.]”

I started to write a response in the thread that this arose in, but it quickly became apparent that it was too long. It’s probably too long for a blog post, too, so if you can’t be bothered, here’s RationalWiki’s TL;DR.

The paper in question

All following quotes are from the study that Otto linked to unless otherwise stated:

Beaver, K. M., DeLisi, M., Wright, J. P., Boutwell, B. B., Barnes, J. C., & Vaughn, M. G. (2013). No evidence of racial discrimination in criminal justice processing: Results from the National Longitudinal Study of Adolescent Health. Personality and Individual Differences, 55(1), 29-34.

…as expected, African-American males are more likely to be arrested, incarcerated, and receive longer criminal sentences than White males. Importantly, however, the results of the t-tests in Table 1 also reveal significant racial differences with African Americans self-reporting more violent behavior over their life course and Whites scoring significantly higher on the composite IQ measure.

Of course, this is what Otto wants me to read from this paper, and to feel duly chastised about my position, right? And in particular Otto wants to make this point, “African Americans score about one standard deviation below Whites on standardized IQ tests.”

Let’s look a little deeper

Overall, the final analytic sample size ranged between N = 1308 and 3506 and varied as a function of missing data and the unique restrictions placed on the data for some of the statistical models (more detail about the modeling strategy provided below).

What this paper doesn’t tell us is the Ns for White and African American participants as separate populations. If the authors do their best to rule out confounding variables, they will presumably do their best to age-, SES-, and IQ-match participants, at a minimum. At least that is what would be necessary to support Otto’s racialist claim.

African-Americans make up around 13% of the total US population – non-Hispanic whites, just under 62% – this may be why an initial sample of 90,000 participants, was whittled down to 3,506. To be clear, 15,000 (one sixth of the original 90,000) participated in waves 2, 3 and 4 of the survey. That 13% to 62% (almost 1:5) disparity may be one of the “unique restrictions placed on the data.”

It is of note that Hispanic background is a feature of an individual’s spoken language. Not even a race realist can claim that language clearly defines race. When including Hispanics, in the white population it swells to more than 77%. However, I will not include the vast majority of white Hispanic in my analysis (I don’t need to); I will merely use this anomaly to justify rounding the ratio up to a mathematically easier 1:5 ratio of African-Americans to whites.

Now, did the authors of this study use matched pairs, which is to say one white and one African-American matched on Socio-Economic Status (SES), or allow that there are more unaccounted for confounding variables?

The paper does not say whether there were matched pairs, and it seems unlikely. As for socio-economic status, it appears that “lifetime violence” and “IQ” are, in combination, being used as proxies for socio-economic status. It is not explicitly stated, probably because it would be a red flag for begging the question. You will see the same statistical trick later on. The direction of causality for low-IQ/lifetime violence and low-SES are generally understood, or at least recognised, however, the relationship between being African-American and having these traits is the very topic of discussion.

Furthermore, Tables 4 and 5 (and thus the graph in figure 3 of the orginal paper), have Ns of below 1200. So, was the “range in analytic sample size [1308-3506]” misreported, or the Ns for those tables and graph?

The context of other work in related fields

What other challenges for data collection for this study were there?

Well, according to the APA:

African American children are three times more likely to live in poverty than Caucasian children. American Indian/Alaska Native, Hispanic, Pacific Islander, and Native Hawaiian families are more likely than Caucasian and Asian families to live in poverty (Costello, Keeler, & Angold, 2001; National Center for Education Statistics, 2007).

We know that the African American population of the US is about 13%. We now also know that “African American children are three times more likely to live in poverty than Caucasian children.” The conclusion must be that for 15 whites there may only be one SES-matched African American, particularly when looking for those matching on “low” socio-economic status. Whether the issue is Otto’s claim that low African-American IQ leads to their low-socio-economic status, or mine, that their lower average socio-economic status leads to low IQ, which reinforces low socio-economic status – this is a problem for studies and statistics.

Did the investigators control for SES at all?

After controlling for life-time violence and verbal IQ, however, the effect of race on the probability of being arrested dropped from statistical significance.

No, it seems not.

What about age- and IQ-matching the participants? We know that the disparity in arrest and incarceration rates disappeared “after controlling for life-time violence and verbal IQ.” As I will argue shortly, the additive effects of IQ and its expression mean that merely including a covariate in the analysis, without also matching the participants for it is problematic. These are significant confounding variables, and thus we must know how the investigators accounted for them. But we don’t. Further, we have a sample that has a 1 in 15 chance of matching a white to an African-American on SES. And if the investigators did match the participants, then African Americans were over-represented in the sample, by a factor of around five. The paper gives no information on how they corrected for either of these issues.

Socio-Economic Status (SES) and intelligence

Otto says that African-Americans are more likely to be low-SES BECAUSE their average intelligence is 1 Standard Deviation (15 IQ points) below that of whites. The following articles (and many, any others) refute that:

In a series of experiments, the researchers found that pressing financial concerns had an immediate impact on the ability of low-income individuals to perform on common cognitive and logic tests. On average, a person preoccupied with money problems exhibited a drop in cognitive function similar to a 13-point dip in IQ, or the loss of an entire night’s sleep.” Morgan Kelly, Princeton [v]

The mental strain could be costing poor people up to 13 IQ (intelligence quotient) points and means they are more likely to make mistakes and bad decisions that amplify and perpetuate their financial woes, researchers found.” Kate Kelland, Reuters [vi]

I included the second quote to highlight the additive nature of how lower IQ impacts and perpetuates low-SES, even when the lowered IQ is a state induced by the money worries themselves, or the on-flow effects, such as lack of sleep, poor nutrition, and so on. Furthermore, “…children from low-SES families scored on average 6 IQ points lower at age 2 than children from high-SES backgrounds; by age 16, this difference had almost tripled”[vii].

Hmm, so low-SES appears to affect individuals by more than one standard deviation. Additionally, making poverty salient can reduce IQ by as much as 13 points. How does one make poverty salient? Oh, I don’t know, how about having a psychologist turn up to your home to ask questions relating to race, violence, and socio-economic status… on three separate occasions! And how about having the last of those occasions be when you’re in your 30s, and life is manifestly not turning out the way you’d hoped.

What I’ll call trait-IQ is relatively stable. “The heritability of intelligence increases from about 20% in infancy to perhaps 80% in later adulthood”[viii]. The reason for this increase in heritability seems to be the increasing likelihood that individuals will alter their environment to suit their intelligence.

Genotype-environment correlation seems the most likely explanation in which small genetic differences are magnified as children select, modify and create environments correlated with their genetic propensities. This active model of selected environments—in contrast to the traditional model of imposed environments—offers a general paradigm for thinking about how genotypes become phenotypes. – ibid.

If an individual has the agency to manipulate their environment in line with their genetic predilections, this is correct. I think that it is fair to say that people of low-SES can still do this but within a narrower range of available opportunities (and let’s not forget, many studies are still done on students engaging in tertiary education, thereby skewing the sample). If the population, as a whole, was wealthier and thus had greater agency, those gene-intelligence correlations would be higher still.

I’m not aware of any papers that explicitly focus on prison populations, and so the above quote can’t be as fully applied to that population as to those outside of jail. I would argue that including the prison population would probably weaken the strong effects of genes on intelligence – especially if done in the US. Of course, this goes to the heart of the racialist argument, and the likely response will be that imprisonment is indeed a selection of environment by African-Americans because of low IQ.

Blacks, and to a lesser extent Hispanics, suffer much higher rates of robbery and homicide victimization than do whites. Homicide is the leading cause of death among young black males and females. These differences result in part from social forces that ecologically concentrate race with poverty and other social dislocations. – Sampson & Lauritsen, 1997.

A digression into the criminalization of drugs

In the States, the war on drugs has, since the 1980s, disproportionately impacted blacks [ix] [x]. One response to this could be, ‘So? If you can’t do the time, don’t do the crime!’ However, notice that the drugs that are legal – alcohol and tobacco – are favoured amongst Europeans. Drugs preferred by non-Europeans: marijuana and opium, for example, are criminalised.

Actually, no. Whilst that may have been true when the laws were drawn up (I tend to think it was), it isn’t anymore (see figure 21, below – numbering taken from the source).

https://www.washingtonpost.com/news/wonk/wp/2013/06/04/the-blackwhite-marijuana-arrest-gap-in-nine-charts/

Notice, however, the arrest rate aligns with the belief that African-Americans use more marijuana (see figure 22).

https://www.washingtonpost.com/news/wonk/wp/2013/06/04/the-blackwhite-marijuana-arrest-gap-in-nine-charts/

The question becomes why does targeting of African-Americans for marijuana offences persist in line with the stereotype, and not the reality?

…the primary focus of the war on drugs has shifted to low-level marijuana offenses. During the study period, 82% of the increase in drug arrests nationally (450,000) was for marijuana offenses, and virtually all of that increase was in possession offenses. Of the nearly 700,000 arrests in 2002, 88% were for possession. Only 1 in 18 of these arrests results in a felony conviction, with the rest either being dismissed or adjudicated as a misdemeanor, meaning that a substantial amount of resources, roughly $4 billion per year for marijuana alone, is being dedicated to minor offenses. – King & Mauer, 2006 [xi]

Why does the ineffective war on drugs persist despite the massive cost to the taxpayer, a fact that should have the “big government” goons howling from the rooftops? The answer seems obvious:

…the remaking of ‘race’, the redefinition of citizenry via the production of a racialized public culture of vilification of criminals, and the construction of a post-Keynesian state that replaces the social-welfare treatment of poverty by its penal management. – Wacquaint, 2001[xii]

Note, here, that “criminal” is someone who is engaging in a behaviour that only affects themselves, and as such is criminal in a conservative religious moral concern sense. The action is not a crime in any meaningful sense of the word.

We have two things going on, here. According to those graphs (based on FBI and Census data, before Otto complains about Washington Post being the source), African-American arrests are at between 2.5 and 3.5 times the rate of whites. But that’s only in gross numbers. Given the 1:5 ratio of African-Americans to whites, African-Americas are, on average, 15 times more likely to be arrested for marijuana offences, something that they engage in slightly less often than whites (unless you persist in believing the stereotype rather than the reality). The issue shows up in the imprisonment rate but is reduced by between half and three-quarters, depending upon the state, when it comes to convictions. While this means that courts are not putting as many African-Americans in prison as are being arrested – by quite a margin – it indicates that police targeting of African-Americans is endemic. The arrest rate does not reflect the offence rate, so the courts are (more or less) correctly interpreting the law with an incorrect sampling of the population.

From page 17 of ‘The War on Marijuana In Black and White: Billions of dollars wasted on racially biased arrests.”

note: the arrest rate is individuals per 100,000 of population.

Back to IQ

Trait-IQ is one standard deviation down due to the effects of being raised in a low-SES environment. Being in a low-SES environment is potentially due to the pernicious incarceration of one or more family members, and thus reduced income. If someone raised in such an environment is matched to participants in a study, there is no reason why the 13 points that IQ drops when socio-economic status is made salient could not be additive to the already reduced IQ due to being raised in a low socio-economic environment. Additionally, an African-American (especially a young African American), in such a study, is around three times more likely to have a lower IQ due to socio-economic status rather than genetic endowment. In combination, these effects could account for as much as a two standard deviation drop, in extreme cases.

Depending on how the participants in the study are selected, African American individuals are 15 times more likely to fit this depressed IQ criteria than whites, by some combination of long-term and current socio-economic status, and to have this aggravated by being asked poverty-salient questions.

In addition to even all of this, the likelihood of being able to age-, IQ-, and SES-match participants reduces still further when one takes into account that African-Americans are 15 times more likely to be in jail and between four and eight times more likely to be in prison at any given time. These are significant barriers to a representative sample.

Racial Identity

With all of that out of the way, there’s just one more thing to bring up. At no time does the paper state the definition of “African-American”, and that is pivotal to the race realist claim about the impact of genetics on intelligence.

According to Bryc, et al (2015[xiii]), the average African-American has 24% European DNA. In an extreme 2% of cases, individuals that self-identify as African American had less than 2% African DNA, around 88% European DNA, but, importantly, 10% other ethnicities. By contrast, Baharian, et al. (2016[xiv]) suggests that European DNA makes up 16.7% of the African-American genome. Whatever the actual amount of European DNA, it seems that the results are going to be skewed by where in America the sample is taken. The further South and East the higher the endowment of African DNA. Even to the point that there is more African DNA in African-Americans in Florida than in Kentucky (ibid.).

The researchers propose an interesting answer. They argue that the Great Migration of African Americans out of the South was genetically biased: African Americans with a higher fraction of European ancestry, who often have lighter skin, had better social opportunities and were thus in a better position to migrate to northern and Western states. Though it will take further evidence to show this definitively, the McGill researchers’ results imply that, even after the end of slavery, discrimination that varied with shades of skin color continued to influence the genetic history of African Americans. – White, 2016 [xv]

(Taking this idea on a different direction – darker skinned individuals may have preferred sunnier climes from a health point of view, specifically vitamin D-production.)

Furthermore, paler skinned African slaves (whether due to European DNA, or just being phenotypically paler) were often assigned to easier tasks, while darker skinned Africans were made to work harder, and usually outside, having the effect of making them darker still (Hill, 2002[xvi]). In a similar vein, “Traffic-safety stops, the authors conclude, are based on ‘how people drive,’ whereas investigatory stops are based on ‘how they look’”[xvii].

Dark-skinned blacks in the United States have lower socioeconomic status, more punitive relationships with the criminal justice system, diminished prestige, and less likelihood of holding elective office compared with their lighter counterparts. – Hochschild & Weaver, 2007 [xviii]

And if all of that wasn’t confusing enough, there’s one final issue with some racialists’ view of race. The majority of slaves came to America from West Africa. The delightfully named Carleton S. Coon would have us believe that West Africa was part Congoid, and part Caucasoid. Meaning that some portion of African slaves were Caucasoid, albeit that those who seem to define race by how someone looks, would call them African-American.

Breakdown of geographical source of race, from ‘The Races of Europe’ (1939) by Carleton S. Coon, president of the American Association of Physical Anthropologists from 1930 to 1961.

Summary

Issues with the single paper that Otto is using as evidence of his contention:

1. It is unclear what methodology was used to account for confounding variables.
2. Without that information, we have to allow for the fact that the study had only a 1 in 15 chance of getting reasonably matched populations, and even then, only on SES.
3. If the populations were matched-pairs, the data is then skewed by not being representative of the US as a whole, unless corrected for.
4. There is no indication as to which of these disparities is the case, and how it was taken into account.
5. The paper misstates the “analytic sample size,” or it mislabels two of the tables, and one of the figures.

Serious issues with the paper given other papers cited above:

6. When poverty is made salient a 13-point drop in state IQ is noted. Wave 4 participants were in their 30’s – no longer juveniles living with a parent.
7. When raised in poverty individuals may be more than one SD (>15 IQ points) behind peers by age 15.
8. These two potentially additive effects could cause as much as two SDs to be wiped off what would genetically be expected – but African-Americans are only behind by one SD according to Otto’s cited paper.
9. African-Americans are arrested for marijuana possession at 15 times the rate of whites, and imprisoned at between four and eight times the rate, despite the fact that whites use marijuana slightly more than African-Americans.
10. African-Americans have, on average, around 20% European DNA.
11. 2% of self-defining African Americans may have less than 2% African DNA, but 10% non-European DNA.
12. Whose definition of “African” is used by the study, or indeed by race realists?
13. The impact of paler skin on access to Socio-Economic improvement has been the case for 400 years and is not necessarily linked to European DNA, given that relative darkness caused individuals to be forced to work in the sun thereby getting darker.
14. Some Africans were possibly Caucasoid, but are nevertheless now considered African-American by racialists.

So, for Otto to make his race realist case, he has to account for all 14 issues (some obviously easier than others).

The fundamental problem Otto has is that African-Americans have, on average, around 20% European DNA, and one has no way of knowing whether the relevant SNPs that make up that 20% impact intelligence, and whether they do so positively or negatively, likewise with the 70+% that actually is African.

Otto will have difficulty claiming that the genes for melanin production and intelligence are linked. So, whether he assigns someone an African-American identity (with around 20% European DNA), or they self-identify as African American (with a 2% chance of only having 2% African DNA) there are some serious statistical issues for making the case stick.

The likelihood that 400 years of racism continues to relegate African-Americans to disproportionately low Socio-Economic Status, and thus lower IQ, irrespective of gene-based intelligence, cannot be discounted by the paper, or Otto’s rhetoric.

Glossary

N = statistical symbol. The Number of participants in a study, overall. Also used to report the Number of participants in a subsample within the study, where that subsample is the focus of a statistical test. In this particular case “N = 1308 and 3506” implies that the total number of participants was 3506. In some analyses, there were only 1308.

SES = Socio-Economic Status. Societal position usually taken as some combination of education, occupation, and income.

Standard Deviation (SD) = A standardised measure of difference from a population mean. In the case of IQ, the mean is 100, and the Standard Deviation (or SD) is 15. Plus or minus one SD of the mean contains 68.2% of the population. So 68.2% of the population has an IQ between 85 and 115. One further SD out (70-130) contains 95.4% of the population.

Footnotes

[i] Edgar, Heather and Keith Hunley, “Race reconciled?: How biological anthropologists view human variation”, American Journal of Physical Anthropology, May 2009, Volume 139, Issue 1, Pages 1–4. DOI: 10.1002/ajpa.20995. Full quote:

The points of agreement in the following articles reflect a shared evolutionary perspective that focuses on describing and interpreting the apportionment of biological variation between individuals both within and among groups (see also Lee et al., 2008). We agreed that:

• There is substantial variation among individuals within populations.
• Some biological variation is apportioned between individuals in different populations and among larger population groupings.
• Patterns of within- and among-group variation have been substantially shaped by culture, language, ecology, and geography.
• Race is not an accurate or productive way to describe human biological variation.
• Human variation research has important social, biomedical, and forensic implications.

[ii] Lieberman, Leonard, et al. “The decline of race in American physical anthropology”. Przegląd Antropologiczny – Anthropological Review. 2003, Vol. 66, 3-21. Full quote: TLDR: The reason racialism is challenged is perhaps because of political correctness; the evidence on which it is challenged is purely scientific.

Occasionally a student will suggest that the race concept has been rejected because of political correctness arising from the horrified reaction to the holocaust in World War II Nazi Germany. We must acknowledge that political correctness is a generic aspect of human behavior. Without conformity to group expectations human societies would be anarchies devoid of cultural patterns. An example of the negative consequences of political correctness is seen in the widespread belief that there were superior and inferior races. Colonialism, involving the conquest, exploration and slaughter of millions of people in the Americas, Africa and parts of Asia was rationalized by the belief in the inferiority of the alleged races inhabiting those areas [COCKER 1998]. The rejection of the race concept by most anthropologists beginning most recently in the 1960s, was based on the genetic evidence reviewed earlier. Conformity to political correctness was not the cause of these changes; rather awareness of the uses of race in colonialism, slavery, segregation, and in the holocaust stimulated reexamination of the race concept using the new genetic data that was accumulated throughout the 20th century. The presence of new genetic data does not guarantee that the data will be given careful consideration. This consideration came about as a new generation of anthropologists with new biographical experiences entered the discipline and examined the new data that developed during the century. Those who characterize these developments as political correctness are using simplistic reductionism, and a naive conception of science in an ivory tower. Scientists must struggle with and develop new data in the context of biography and history.

[iii] Wagner, Jennifer K., et al. “Anthropologists’ views on race, ancestry, and genetics.” American Journal of Physical Anthropology (2016).

[iv] Ruvolo, Maryellen, and Mark Seielstad. “The apportionment of human diversity: 25 years later” in Thinking about Evolution: Historical, Philosophical, and Political Perspectives, edited by Rama S. Singh et al. (2001). Cambridge University Press. ISBN 0521620708.

[v] https://www.princeton.edu/main/news/archive/S37/75/69M50/index.xml?section=topstories

[vi] http://www.reuters.com/article/us-poverty-brain-idUSBRE97S10W20130829

[vii] Von Stumm, S. & Plomin, R. (2015). Socioeconomic status and the growth of intelligence from infancy through adolescence. Intelligence, 48, 30-36.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4641149/

[viii] Plomin, R., & Deary, I. J. (2015). Genetics and intelligence differences: five special findings. Molecular Psychiatry, 20(1), 98–108. http://doi.org/10.1038/mp.2014.105

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4270739/

[ix] Sampson, R. J., & Lauritsen, J. L. (1997). Racial and ethnic disparities in crime and criminal justice in the United States. Crime and justice, 21, 311-374.

[x] Wacquant, L. (2001). Deadly symbiosis: When ghetto and prison meet and mesh. Punishment & Society, 3(1), 95-133.

[xi] King, R. S., & Mauer, M. (2006). The war on marijuana: The transformation of the war on drugs in the 1990s. Harm Reduction Journal, 3(1), 6.

https://harmreductionjournal.biomedcentral.com/articles/10.1186/1477-7517-3-6

[xii] Wacquant, L. (2001). Deadly symbiosis: When ghetto and prison meet and mesh. Punishment & Society, 3(1), 95-133.

http://home.ku.edu.tr/~mbaker/CSHS503/WacquantUSprisons.pdf

[xiii] Bryc, K., Durand, E. Y., Macpherson, J. M., Reich, D., & Mountain, J. L. (2015). The genetic ancestry of African Americans, Latinos, and European Americans across the United States. The American Journal of Human Genetics, 96(1), 37-53.

http://www.cell.com/ajhg/pdf/S0002-9297(14)00476-5.pdf

[xiv] Baharian, S., Barakatt, M., Gignoux, C. R., Shringarpure, S., Errington, J., Blot, W. J., … & Gravel, S. (2016). The great migration and African-American genomic diversity. PLoS genetics, 12(5), e1006059.

http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006059

[xv] White, M. (2016). How Slavery Changed the DNA of African Americans, Pacific Standard.

https://psmag.com/how-slavery-changed-the-dna-of-african-americans-c5ba430ebd90#.aluu7as77

[xvi] Hill, M. E. (2002). Skin color and the perception of attractiveness among African Americans: Does gender make a difference? Social Psychology Quarterly, 77-91.

http://www.jstor.org/stable/3090169?origin=crossref&seq=1#page_scan_tab_contents

[xvii] Ghandnoosh, N. (2014). Race and punishment: Racial perceptions of crime and support for punitive policies.

Race and Punishment: Racial Perceptions of Crime and Support for Punitive Policies

[xviii] Hochschild, J. L., & Weaver, V. (2007). The skin color paradox and the American racial order. Social Forces, 86(2), 643-670.

https://academic.oup.com/sf/article-abstract/86/2/643/2235030/The-Skin-Color-Paradox-and-the-American-Racial?redirectedFrom=fulltext