The Black-White IQ Gap: Charles Murray VS James Flynn (2006 Debate)
What this covers
What you’re about to listen to is a fascinating debate from 2006 between two leading scholars in intelligence research, Charles Murray and James Flynn. I recently found this in the deep dark corners of YouTube split up into 12 parts, as it had clearly been uploaded back when YouTube only allowed 10 minute uploads! So I’m sure many people haven’t heard this debate and having the whole thing in one place would make it more likely that more people will hear it. Of course, I want to stress that I don’t own the rights to this clip. I believe the event was sponsored by the American Enterpirse Institute, but I haven’t been able to find this recording anywhere else. I’ve cleaned up the audio and edited out some dead air. Unfortunately, there are some cuts in the original recording so you occasionally miss a few words. Enjoy.
My links:
Substack: https://ideassleepfuriously.substack.com
Patreon: https://www.patreon.com/Ideas_Sleep
Twitter: https://twitter.com/Ideas_Sleep
Source description (no synthesized summary yet).
Flynn and Murray debate whether the Black-White IQ gap is closing: Flynn argues environmental factors explain most of the gap and it can narrow further with social intervention, while Murray contends the gap has stalled since the 1970s and will persist due to heritability and dysgenic fertility patterns.
- Flynn's evidence from Germany shows Black and White children perform equivalently on IQ tests when removed from American Black subculture, suggesting environmental causation
- Murray presents data showing gap closure stopped for cohorts born after 1970s and points to genetic heritability of g-factor and reproductive patterns as persistent barriers
- Both scholars agree the gap narrowed significantly during 20th century but disagree sharply on future trajectory and causal mechanisms
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Height of average Japanese has increased by over one standard deviation in 40 years due to improved nutrition, and while residual difference from Caucasians persists and is unlikely to diminish further, this demonstrates that large environmental effects on heritable traits are possible.
“a height of average height of japanese has gone up by how many inches in the last uh 40 40 years a lot by well over a standard deviation yeah it's gone up several inches and that's obviously environmental better nutrition but it's also true that there is a residual difference between uh japanese and caucasians let's say that is very unlikely to diminish because we've gotten all the environmental effect we're going to get”
From child development research, professional-class parents read and discuss books in front of their children, while some parenting approaches (like encouraging physical activity over homework) may inadvertently signal lower priority for academic achievement, affecting children's cognitive development through modeling and cue-taking.
“my first question to them is how often do you read how often does your kid see you discussing books they take their cues from you”
West African runners (Kenyans and Ethiopians) dominate middle and long-distance running internationally, suggesting possible genetic advantage in oxygen conversion or physiological adaptation to altitude training accumulated over generations.
“the striking i was the complete dominance of individuals from west africa in middle and long distance running”
Vocabulary as a Wechsler subtest was nearly excluded from IQ batteries in the 1930s because psychometricians assumed it was too culturally determined to measure true intelligence, but later evidence showed vocabulary correlates with cognitive ability despite cultural loadedness.
“with the original vocabulary subtest in the wechsler back in the 1930s i guess they weren't even going to include that as a subtest in the iq battery because they assumed it was so culturally determined that it would be useless as a measure of iq and it was only over a period of time that the argument was first made and then evidence adduced for this argument”
Group differences in IQ are not evidence of individual differences: knowing that group A has lower mean IQ than group B tells you nothing about whether a specific individual from group A is smarter or dumber than a specific individual from group B, making group data irrelevant to individual assessment.
“it means differences in the means and the distribution around the means within large populations and is not a reference to all of the populations or to any in any particular individual”
Flynn's analogy: if we observed a female sample that appeared elite but actually was mildly elite, this would eliminate the mean height gap between men and women but would not eliminate the growth spurt pattern (girls peak at 14-15, boys continue to 18-20) or complexity differences in athletic tasks (women struggle more with triple axels), because those patterns are genetic; therefore, if Eyferth's sample was only mildly elite yet eliminated the g-factor gap, this suggests the g-factor gap is environmental.
“even if i first got a mildly elite sample of american blacks if the loss with age and if the g or complexity factor were genetic it would be very surprising if a mildly elite sample had eliminated those two things”
Flynn challenges Murray's interpretation of g-factor evidence: even if equal gains across all g-loadings occur (e.g., 5-point gain on all tests regardless of g-loading), the gap narrows on all subtests and eventually closes on all, regardless of whether the pattern correlates with g, because proportional gains eliminate differences regardless of initial correlation with g.
“let us imagine that at a certain time black kids are 15 points behind whites on all three of those tests and let's imagine they gain five points on all three and cut it to ten well richards would tell you then that the pattern of gains doesn't show a gain on the g factor and he's quite right isn't it...you know if that was a gain on the g factor it would be slightly higher on test two and slightly higher again on test three wouldn't it...and the gains are dead even there's no replication of the g factor whatsoever and yet between those two periods on all three tests the gap has fallen from 15 to 10 points”
Preventing discussion of race and IQ allows prejudice to thrive unchallenged, whereas open debate with high intellectual standards can actually reduce prejudice by replacing misconceptions with accurate information.
“you're not protecting blacks or chinese or irish if you make their plight undiscussable anything that's undiscussable you then leave to prejudice and opinion and ignorance and that's they're the only gainers when you ban discussions of this sort”
Reaction time tests (measuring cognitive processing speed vs. physical motor speed) show consistent Black-White differences in reaction time but Blacks perform better on movement time, suggesting motivation is not a primary factor in test performance differences.
“there's thing called reaction time tests where they're very simple you have your finger on a button and another other and a row of buttons another light goes on and you simply move your thumb from the home button to the target button everybody gets every item right what they're measuring electronically to thousands of a second is the reaction time to get your thumb off that button but they also measure movement time uh the reaction time is the cognitive process the movement time is a purely physical one and in this case what you find consistently is that there is a black white difference in reaction time but blacks are better in movement time”
Eyferth's study of Black and White children born to American occupation troops in post-WWII Germany found that Black children scored approximately 96.5 on IQ tests while White children scored 97, nearly eliminating the 15-point gap observed in America, suggesting environmental rather than genetic causation for the gap.
“he found that when he tested their iqs on the whisk which is an intelligence test of impeccable credentials the wechsler intelligence scale for children that they came out almost exactly even that is for the total samples i think the whites were 97 and the blacks were 96 and a half something of that sort”
Racial profiling in criminal justice (Black men more likely to be stopped, arrested, and shot by police; job applications with Black names receive 50% fewer callbacks than White names; both Black and White people assign harsher penalties to more Black-appearing individuals) creates statistical associations between Blackness and negative outcomes that influence behavior through anticipated discrimination.
“21 black police officers undercover have been shot by their white colleagues in new york city alone since world war ii now while no white undercover officer has ever been shot”
State-administered standardized tests under NCLB face severe penalties if proficiency targets aren't met, creating incentive to inflate proficiency measures and deflate reported gaps, making state-level data unreliable for tracking whether gaps are actually narrowing.
“state administered tests tend to significantly inflate proficiency levels and proficiency gains as well as deflate racial and social achievement gaps in the states as these states and school systems face severe penalties if they don't come up to the expectations of the act all sorts of games are being played with these numbers”
IQ tests measure not raw intellectual capacity but developed 'habits of mind' for handling complex cognitive problems, making them dependent on prior experience and cultural practice (e.g., crossword puzzle solving), not purely innate ability.
“when i took up uh crossword puzzles i tried but i am a man who has always put things as simply and as plainly as possible i wasn't used to looking for ancillary meanings tricks between verbs and nouns and so on and i had to develop the habits of mind to allow me to do well on them”
Vocabulary acquisition depends on the kind of mind that, upon hearing a word, wonders about its meaning and naturally fits it into sentence context, rather than mere exposure to words, suggesting children from less-educated homes aren't simply hearing fewer words but lack cognitive habits that develop vocabularies.
“the reason that kids develop a large vocabulary isn't because they hear the words because you have lots of kids from upper middle class homes with professor parents who don't acquire really good scores of vocabulary tests it's do you have the kind of mind which when it hears a word says i wonder what that means or which naturally fits that into the context of the sentence in other words acquiring vocabulary is very heavily loaded toward kids with brains”
Perry Preschool program is the archetypal example of the 'rule of one' in social science: a small-sample intervention (40 kids control, 40 experimental) showing positive effects, widely cited despite hundreds of other preschool interventions showing no effect, because it supports the desired conclusion.
“you have this library of studies which show no effect and you have poor little perry preschool with a control group of 40 kids or a experimental group of 40 kids very small the one that everybody cites all the time is prairie preschool”
The Dickens-Flynn model explains how environmental effects on IQ can be immediate rather than permanent: environments shape IQ continuously rather than leaving permanent imprints, so even if early environment doesn't have permanent effects, a succession of increasingly worse environments (school, teenage peer group) for Black children will produce cumulative IQ decline with age.
“the dickens flynn model is based on the transience of environment but that doesn't have anything to do with closing a gap between two groups environmentally let us imagine that the verbal difference in the black and white home before the age of four is a significant cause at that age it's not a matter that it puts an imprint in clay in the brain that remains forever uh that's quite true as charles murray says what happens is that when the kids go into school a new environment takes over that is slightly worse and then when they get to about 14 or 15 another new environment takes over that slightly worse”
Even if the direct influence of genes on brain physiology for IQ is as low as 0.3 of the variance (as the Dickens-Flynn model suggests), and you account for dysgenic mating patterns (losing perhaps 0.25-0.33 IQ points per generation), this is still not obviously concerning compared to environmental factors that could be modified.
“if you do the mathematics the and you look at the difference between the upper half and the lower half of the black iq curve they would be losing about eight tenths of a point per generation now the dickens flynn model says that the direct influence of genes on brain physiology in regard to iq is perhaps something as low as point three of the variants so you have to discount that i mean maybe they are losing something like a quarter of an iq point to a third of an iq point per generation to disgenic mating maybe a little more but uh even that is not obvious and that worries me a good deal less than the environmental factors”
Adoption studies show that while environment can have huge effects when children are severely deprived (locked in closets, not spoken to), once a nearly adequate environment is reached, environmental improvements produce minimal additional IQ gains, suggesting diminishing returns on environmental interventions.
“it is true that the environment can have a huge effect you lock a kid in a closet for the first couple of years their life or don't talk to them at all as jim said that can have an effect on iq but once you get to a nearly adequate environment even though it may not look very good to a lot of us probably you've gotten most of what you're going to get from environmental changes”
Income does not drop out of multivariate equations predicting offspring IQ when controlling for other variables, showing an independent relationship between parental income and children's IQ, though the relationship is not huge.
“income does not drop out of the equation uh in some circumstances uh when you are controlling for other things sometimes income is an important has an important relationship to uh to offspring iq not a huge one but but it's there”
The Black-White IQ gap closure has stalled because shared environment (basic education, nutrition, family socioeconomic status) has improved for Black children born in the 1980s and 90s compared to the 1960s, yet the gap did not decrease, suggesting shared environment improvements have exhausted their effect on closing the gap.
“the shared environment did get better for children born in the 1980s and 90s but the black white gap didn't decrease the environmental variables in the 1960s when the gap was going down were awful for blacks compared to the 1990s at this point there isn't that much difference on a variety of important environmental variables”
Black youth culture in America is described as cognitively counterproductive, with vocabulary circumscribed, limited exposure to reading and books, and use of non-standard English dialect, creating cultural barriers to academic achievement that schools alone cannot overcome, though not originating from genetic deficiency.
“when black kids go to school they become immersed in my opinion in a black youth culture where you don't see kids playing chess you don't see kids discussing books it wouldn't be bad if they spoke a different language but they speak a substitute for english that will not help them in the mainstream of american society and i think blacks have to realize there's a price to be paid for that”
Policy should focus on treating individuals as individuals rather than group members, stopping use of racial group statistics to judge schools and institutions, and assessing whether individuals have fair opportunity to develop capabilities, regardless of race.
“you judge people as individuals you stop judging people as members of group you stop asking do we have enough uh latinos or whites or asians or blacks in a given classroom you say are we giving a fair shot to everybody who can do the work”
The Eyferth study findings would be truly significant if the age-decline effect (the narrowing with age found in Germany data) had disappeared, which would demonstrate that environmental factors drive the age effect, but the sample sizes are too small to be confident about this pattern.
“if we could only be sure that in the german environment the tendency to drop with age had disappeared that would be truly significant... what a pity that i first didn't have 700 black children then we would probably be home free we would know that this age decline was an environmental factor”
g (general intelligence factor) is potentially sociological in origin rather than biological, as suggested by recent papers, with even modest biological origins, making the interpretation of correlations between g-loading and Black-White gaps ambiguous regarding causation.
“there is also an interesting paper in the mill i think it's by which isn't it yes in which he argues that g is fundamentally sociological in origin uh that you know it's uh it's not at all obvious that it has a biological origin i think it has a very slight biological origin i wouldn't discount that it has some you know physiological substratum”
If America could give all citizens a decent life, there would be far less interest in IQ differences, which are trivial compared to the human misery that drives group comparisons—the issue is not the gap itself but poverty and exclusion.
“all of these emotional angst about group differences is after all a product of human misery my group irish americans are not beating their heads on the floor that there's an iq gap between them and chinese americans you know we don't care much we sort of think look at them they're crazy rather than just wanting a nice civil service job and to talk politics at the local pub they want all of these glittering prizes and we don't”
Legalization and cheap supply of heroin might reduce drug economy profitability and thereby reduce the rate at which black males serve time in prison (currently about one-third), which is an environmental/policy intervention that could affect achievement motivation and family structure.
“one of the things that i left out as a thing i would do is legalize you know heroin and have it supplied cheap to take the profit out of it to destroy the black drug economy so that a third of blacks wouldn't males wouldn't serve sometime in prison and i think you'd actually have fewer addicts”
Birth year effects and test year effects in IQ data are mathematically confounded (birth year = test year minus age), so it is impossible to distinguish between birth cohort effects and period effects from raw trend data, though ancillary analyses of other variables can suggest which interpretation fits better.
“you cannot tell the difference between a birth order effect or birth year effect and a year effect because if you think about it what is uh what is a person's uh uh birthday it's the uh or if you take the uh the birthday is simply there's going to be the test year minus their uh minus their age okay so they're they're mathematically linear completely dependent on each other you can't distinguish the effects”
The Black-White gap on standardized tests has substantial sample size limitations: WISC standardizations have 100-300 Black samples (in some cases 135 for children ages 2-6, less than 200 for Stanford-Binet), making strong inferences about group means unreliable because the samples are designed to represent national distributions, not to measure precise group differences.
“sample sizes for blacks are small and that's because this is something that all of us who use these standardizations have to remember you choose a standardization sample so that the iq company can know that a given score is at a given place on the national distribution they are not intended to yield precise measures of group means”
Both Black and White mothers report having children with multiple partners rather than stable partnerships, and this family instability represents a deterioration in environment that should predict worse cognitive outcomes, yet hasn't prevented Black IQ gains in some areas.
“all of my data tells me that you have three points lower iq if you're in a solo parent home”
Spearman's Hypothesis states that the Black-White IQ difference is largest on tests most saturated with g (general intelligence factor), and this has been established empirically by Arthur Jensen and others as a fact, not merely a hypothesis, with implications that g-based differences may have genetic origins since g is highly heritable and correlates with physiological measures like frontal cortex gray matter volume.
“charles spearman discovered g back in the early part of the 20th century he formulated what subsequently became known as spearman's hypothesis he said that the black white difference would be the largest in the tests most saturated with g”
When Bill Dickens and I looked at how much the IQ gap diminished, we then looked at how much the g gap diminished and it was 90% of the IQ gap reduction, meaning the g gap is closing nearly as fast as the overall gap, contrary to implications that g-concentration would prevent closure.
“when bill and i looked at our data and we looked at how much the iq gap diminished we then looked at how much the g gap had diminished and it was 90 percent you know it was a little less than the iq gap but not much”
The government as employer of last resort (like the Israeli Army) could provide useful roles even for people with IQs as low as 70, since humans are more flexible than robots and can perform many tasks, reducing unemployment and its associated harms to achievement.
“the israeli army doesn't flunk people out with an iq of 70. they need manpower and they tell me they find useful jobs for people with iqs of 70 to do they may not put them on the front line but there are plenty of things you know even a person with that iq is a lot more flexible than a robot and a lot better than a lawn mower”
Chinese and Japanese Americans provide evidence that persistent ethnic subculture emphasizing academic achievement produces outcomes (IQ, professional attainment) beyond what would be predicted by measured IQ alone, suggesting cultural factors can substantially influence cognitive development and achievement.
“i wrote a book on the subject in 91 and they're one of the best examples of where a persistent ethnic subculture does make a difference and it's one that persists and persists and that is you can match chinese and caucasian white kids in american schools for iq and the chinese kid will score seven points higher on academic measures including the scholastic aptitude exam they'll become professionals at double the rate you would expect them to become professionals that is matched with iq as whites do”
If the Eyferth study was the only example showing environmental causation of the gap, and no similar results appear in adoption studies, other countries, or other time periods, then relying on this single study violates the 'rule of one' in social science where single anomalous findings (like Perry Preschool) are over-cited.
“i call the rule of one in issues involving all kinds of social policies in which people want the results to come out a certain way...you have this library of studies which show no effect and you have poor little perry preschool with a control group of 40 kids or a experimental group of 40 kids very small the one that everybody cites all the time is prairie preschool it's the rule of one”
Genetic evidence could be advanced through large-scale genetic marker studies that go beyond binary Black/White classification to measure continuous genetic ancestry, combined with cognitive and physical appearance measures, which would help separate genetic from physical discrimination effects.
“we can answer these questions now a lot of them if you're if you're interested in nature versus nurture we no longer have to treat race as a binary yes no variable we don't have to rely on very dubious blood group data or anything else we now have genetic markers that that correspond to self-identified ethnicity in 99.9 of the time”
The Flynn Effect (upward drift of IQ scores across generations) is qualitatively different from Black-White IQ differences in the US: Dutch psychometricians using complex methodologies found that the Flynn Effect appears to operate on different factors than those driving the Black-White gap, and therefore the Flynn Effect is unlikely to automatically close the racial gap.
“it appears therefore that the nature of the flynn effect is qualitatively different from the nature of black white differences in the united states and following that implications of the flynn effect for the black white difference differences appear small”
Both Flynn and Murray demonstrate unusually high intellectual collegiality and good faith in disagreement compared to typical public discussions of race and IQ, with both believing the other is arguing in good faith despite fundamental disagreements about data interpretation and causation.
“at the time of the bell curve when a lot of hateful and awful stuff was being said that jim flynn who was on the opposite side of some of these issues for me was from the very outset the one who was collegial who was interested in data”
Disbarment from society, mass incarceration of Black men, and public fear of police violence against studious Black youth are interconnected problems that make it difficult for Flynn to recommend Black academic aspiration as a solution without also addressing criminal justice disparities, because rational fear of discretionary police abuse makes academic visibility risky.
“if black social statistics improve the profiles will weaken we are uh come to the end of this seminar i'm going to give an opportunity for a last word or two”
Flynn's policy recommendations include drop-in centers for single parents offering cognitive stimulation, improved housing design, book clubs in Black communities, and government as employer of last resort to reduce unemployment, representing colorblind social interventions focused on environment rather than race-specific programs.
“i would like to see drop-in centers that single parents could go to and not to park their children but to be surrounded by sub-professionals to engage their kid and talk tapes books...you know things that would give single parents less isolation”
If the Black-White IQ gap at age 4 is about 5 points, find a good sample of 14-year-olds where the gap is less than 10 points and Flynn will examine it with interest, challenging those who doubt the age effect to produce contradictory evidence.
“if they agree that the iq gap is at present for four-year-olds something like five points find me a group of 14 year olds where it's less than 10 points and i'll look at that with interest i think they'll have an awfully hard time finding a good sample of 14 year olds where it's less than 10”
General Social Survey vocabulary test (10 items, unchanged for 20+ years) shows Black-White gap has been declining, but primarily because White performance has declined, not because Black performance has improved, raising questions about what is being measured by test score changes.
“the highest five year aggregation was 1945 to 1949. the only reason the black white gap has been diminishing is because whites have been doing worse and worse”
NLSY data (National Longitudinal Survey of Youth) provides the best data for gap closure assessment because: large samples (Black N > 2,500), identical tests over time, most children tested multiple times, sample not limited to school-attending children, and includes both academic achievement and verbal IQ measures plus detailed background information including mother's IQ.
“we now come to the best data for persons born from the 1970s onward...large samples black end of over 2 500 the tests are identical over time absolutely identical most children are tested more than once which has a bunch of advantages”
Proper educational policy should enable every child to develop their capabilities to the maximum regardless of race, and the No Child Left Behind Act—by imposing a one-size-fits-all solution that might work for worst inner-city systems onto all schools—was disastrous for American education and should be repealed.
“the proper educational policy has nothing to do with any difference between blacks and whites proper education policy is one which enables every child to develop his or her capabilities to the maximum and the no child left behind act is an attempt i think and i'm actually stealing a line from howard gardner here it says it's imposing on the whole nation a solution that might possibly make sense for a few of the worst inner city school systems it is i think as someone who had a child in high school during the first years of the no child left behind act a disaster for american education and ought to be repealed forthwith”
Black and North African occupation soldiers in Germany may have differed genetically (North Africans historically distinct from sub-Saharan populations), and inclusion of ~23% North African-fathered children in the Black sample potentially biases Eyferth results, though this confounding is difficult to untangle.
“i am told there are other problems besides the obvious one that that jim already described with the eye for the study which is that you did not have a random sample of blacks and you described that very well i've also been told and you can tell me whether i'm right or wrong that some 23 percent of the fathers were not actually um sub-saharan blacks they were north africans”
Three markers distinguish high-value vs. trivial claims: importance (depends on novelty, how unexpected/counterintuitive), specificity (concrete numbers or named mechanisms vs. vague generalizations), and actionability (whether knowing it changes what one should do), and the most valuable claims in intelligence research typically combine these—e.g., the Eyferth study combines geographical specificity, counterintuitive findings, and policy implications.
draft — not yet grounded
Arthur Jensen's recent work (by 2006) has converged with Flynn and Dickens' views on the existence of age-related increase in the Black-White IQ gap, with Jensen no longer disputing the decline with age that he disputed in his earlier 1970s work, though Murray notes this requires examining Jensen's later publications.
“jensen no longer disputes it when he and rushton wrote a reply to our article he didn't object to my characterization of his position now as believing that there is a drop off with age which was parallel to our own”