Murray's core arguments and the implications he draws do not depend upon genetic determinism.
No, but they do depend on these factors being beyond manipulation by other means, a point he himself makes.
Only beyond massive and relatively feasible manipulation sufficient to override the less controllable factors. There is little evidence that such manipulations are feasible and much evidence that the relative differences are highly stable across decades. This particular study is perhaps the most extreme example, comparing the relative scores of 11 year olds (before the nuerological changes of puberty) to their scores 70 years later at the age of 80 (after aging effects began and a half century outside and formal schooling).
The correlation is quite strong, despite the deck being stacked highly against it. Not only are the timepoints separated by major periods of natural (and genetically influenced) neural development, in addition to 60 years of differential education, jobs, experiences, exposure to toxins, etc., but also the fact that the study began in 1932 means that their measure has far more random error and less reliability than the ways of measuring g used to today.
All that needs to be assumed is that factors outside of direct control combine to impact early brain development in a manner that creates relatively stable differences in various cognitive abilities that have an impact on cognitive performance (either directly and/or via effects upon rates of learning from practice and experience). If, the causes of these developmental differences were factors impacting fetal development, how exactly would that change his arguments?
It would demolish them entirely. If it all turned out to be based on external factors influencing fetal development then his call to reform society would entirely spurious - all we'd need to do is adopt best practice in early stage pregnancy and the entire problem disappears.
That is incredibly naive to think we could directly control every aspect of every person's life that impacts fetal brain development. Not only are we not even close to knowing what most of the factors are, but even if we knew many would turn out to be uncontrollable, such as stressful events the mother experiences while pregnant interacting in complex ways with how the mother responds to stressful events, which in turn are responses shaped by her own genetics interacting with a lifetime of experiences. You are talking about making every single pregnant women identical in what they experience and how they respond to their experiences. It is impossible to reduce any more than a tiny fraction of the variance in these factors, and thus impossible to reduce more than a fraction of any variance in fetal brain development and thug g-factor that results.
The reality is that the only plausible way of controlling variance in the general intelligence would be if nearly all the variance were from some very finite and confined aspect of peoples life that were easily manipulable by state-level policies. That excludes about 99% of the factors that could be responsible for a portion of that variance.
-It has to establish that IQ is a measure of inherent ability or potential.
IT only has to establish that there are measures showing relative stability in a related basic cognitive processes that impact performance and rates of improvement in performance across many intellectual tasks, such as reading comprehension, logical reasoning, arithmetic, scientific argumentation, etc..
Not so. There are plenty of existing factors that already have these characteristics. Parental income, for example. Parental income is relatively stable, and relates to basic cognitive processes that impact performance. It's also highly heritable. But there's no logic to the idea that we should structure society around family wealth.
Family wealth can be redistributed to reduce variance due to its influence. In fact, Murray argues for wealth distribution. Despite your naive suggestion, general intelligence cannot be plausibly redistributed to any notable degree by any feasible social policy, even the variance due to environmental factors.
It does not matter if other factors influence the same things as general intelligence, only that general intelligence has an independent impact upon these things and it is not plausible raise everyone's intelligence to be equal to those at the top of the distribution. Those are really the extent of the scientific assumptions upon which most of their recommendations rest, but all policy also rests upon ethical values and subjective goals.
BTW, my defense of the Bell Curve largely stops when they begin to recommend policy. Policy is not science and must always go beyond science to rely upon assumed subjective values and goals. So, I basically don't intend for anything I say to apply to anything beyond the reasonableness of their assumption that general intelligence exists, is stable and not feasible "redistributable", and in anything resembling a free society will have an impact on profession and related financial success (and avoidance of failures, which is equally important). What we decide to do about that is a worthwhile discussion, but not one that I have a strong opinion on, unlike my very scientifically grounded opinion that people who claim that the notion of general intelligence and its reliable measurement have been discredited are talking out of gross ignorance and the overwhelming majority of people with Ph.D.s in the relevant fields of Cognitive Psychology and Cognitive Neuroscience disagree with such notions.
-It has to establish that these differences are meaningful on a practical level
With the number of intellectual tasks that the g-factor predicts performance on, the only way it is not practically meaningful is if essentially all of formals education (which seeks to develop skills on these very tasks) has no practical meaning.
Again not true. IQ could be the result of formal education, rather than a predictor of it, thus making it meaningless on a practical level, while formal education remains important.
If higher IQ is the result of a formal education, then it measures cognitive reasoning and judgment processes that are impacted by some of the more basic mental processes one develops via formal education. If the most basic mental processes that are developed by formal education are "meaningless", then what does that say about formal education? In addition, there are studies showing that early IQ scores predict later grades, after controlling for prior grades. Also, when taught highly novel information with little connection to prior knowledge, people with higher IQ learn it better and more quickly. These favor a bi-directional causal arrow, at minimum. In addition the studies showing test-retest stability in relative scores with years of differential education in between, suggest that formal education has limited impact upon relative IQ scores. But no matter the underlying causal pattern giving rise to the relationship between them, only a minority of the possibilities would allow for one to reflect important cognitive skills and the other meaningless skills. In addition, a growing mountain of research supports the ideas that the variance represented by the g-factor is variance due to some of the most basic and general features of how the brain processes information such as the manner in which the frontal lobes control the focus of attention and switching between mental tasks, and the efficiency of glucose metabolism that impacts neural transmission. Thus unless "real life" reasoning does not involve these features of basic information processing, it is theoretically implausible that it wouldn't be impacted by differences in general intelligence.
Even then, IQ test still have meaning in determining success within formal education and thus upon everything that is in turn impacted by one's schooling, both in terms of grades, graduation, college attendance, degree type, etc..
'Determining'? Careful, you're in danger of assuming causation from a finding of correlation - a fairly critical statistical error. IQ is correlated with academic success. That doesn't mean it causes or determines it.
There is a large body of research supporting a causal impact of intelligence upon rates of learning new information, and thus upon the academic success. IF you don't realize that, then your knowledge of the field is about 3 decades out of date.
Togo said:
But the biggest gap by far is in the area of demonstrating that these values represent a practical difference. Gould's criticisms are the most famous, mainly because he goes into the most detail. It boils down to:
-The differences being described between groups are very small.
-The individual differences between people in the groups are extremely large.
In nearly every behavioral or medical experiment the group differences are a fraction of the individual differences among people within a group. This does nothing to undermine inferences about the contributors to the differences.
No, but it does undermine the social recommendations that Murray is making, since they depend not on group differences, but on individual performance. Which is not significantly effected by membership of the group. What he's doing
is measuring a group difference, assuming this implies an effect on the individual, and then recommending social change to cater for these individual differences that he has not managed to measure.
Huh???The group differences are a statistical by-product of individual differences. The data most certainly is composed of measurement of IQ and other variables at the individual level. The group means are a statistical artifact of individual variance around a central tendency, and the mean difference just reflects that the groups are composed of different proportions of individuals who vary along the continuum of IQ scores. In addition, I am fairly sure that he does not limit himself to group comparisons, but shows that even ignoring group differences, the relationship between IQ and other variables holds (such as the strong correlation between IQ and income, between individual whites). Showing that relationships hold across levels of analysis and different ways of grouping the sample is a compelling way to reduce the plausibility of third variable influences, because most third variables will only be able to account for the relationship at one level of analysis. For example, showing that level of education and religiosity are negatively correlated between countries, between states in the US, and between individuals within states, rules out a lot of third variables and suggests a more direct and non-context dependent relation between the variables.
In addition, if biology were the cause of general intelligence, then even if those biological factors had some relation to race, one would still expect larger within group differences in IQ than between group differences. This is because the difference in the biological factors will almost always be greater between the most extremely different members of a single group, then between the most prototypical and statistically "central" members of different groups. Nothing about this in any way undermines the central points of that book.
It undermines his recommendation that society be restructured to cater for differences in IQ.
It does no such thing. The recommendation is not merely about equalizing average group outcomes. It is about individual variability in "luck" factors that impact outcomes. So, the within group variance is as supportive of the recommendation as the between group variance. The relative proportions of those variances has no logical relevance. All that matters is that both the within and between group variance in IQ impacts the corresponding within and between group variance in outcomes.
It also undermines his discussion of how various groups within US society have fared as being driven by IQ.
No it doesn't, it supports it. There is a lot of variance in IQ and outcomes that has nothing to do with race. Yet, if IQ differences between races related to outcome differences, then IQ differences within a race should also relate to differences outcome differences within a race. They do.
In addition, the racial differences in income are also small relative to the differences within group. So, that pattern mirrors that of IQ, just as it should if IQ is a causal factor.
It further undermines his contention that the increasingly unequal nature of US society is a result of employment increasingly relying on cognitive processes.
How? That argument has to do with the idea that a given difference in IQ (say 20 points) has more impact on income now than it used to. I am not arguing for that claim, but it isn't at all undermined by (its totally unrelated to) the between/within group variance ratio.
You can probably rescue g as a concept, but that's not what the book is about.
I don't need to rescue g as a concept, it is alive and well in the most rigorous journals and cognitive science with a mountain of empirical support for its validity and theoretical importance. I am just pointing out to those ignorant of the past 30 years of cognitive science that the rumors about the decrediting of g have been greatly exaggerated.
-The authors, intentionally or otherwise, blur the differences between differences between groups and differences within groups.
Yes, because the issue is not merely group level differences but individual differences that can sometimes manifest as differences at the group level if the factors that determine the differences are partially related to the factors that define group membership. The group differences are not the point. They are used just as an illustration and because most people (especially critics of his books) view society entirely in terms of groups and their different outcomes rather than variance at the individual level. Most of their critiques and those who focus everything on race care only about making the aggregate outcomes between groups the same by any means neccessary, no matter what injustices must be done at the individual level. To them, it does not matter whether an individual white person actually got advantages over an individual black person. All that matters is that giving special treatment to the black person over the white (even if the black person already had more advantages) will make the group level aggregate outcomes more equal. The Bell Curve authors do not see group level unequal outcomes as the sole problem to be addressed, and refer to them in service of a larger point about individual differences.
If he could detect individual differences based on group membership, as distinct from other factors that Murray claims don't properly manipulate IQ, why didn't he publish them?
I don't understand what you are saying here. The individual differences are in IQ itself, and those differences have and impact and create other differences in life outcomes. He shows plenty of data supporting this (and doesn't show the additional mountain of data over the past 20 years further confirming it). He merely shows that groups known to differ on these life outcomes also differ in IQ in the same way that individuals do. The point is the reliable relationship among the variables, at both the group and individual level of analysis.