My comrade in blogging Ed Realist uses the phrase “The Voldemort View” to describe the following proposition:
The Voldemort View: Mean differences in group IQs are the most likely explanation for the academic achievement gap in racial and SES groups.
Someone in comments asked if I subscribe to the Voldemort View, the View that Shall Not Be Named. The short answer is yes, as stated. The longer answer is something like this.
The Efficient Market Hypothesis (EMH) is an idea in economics that asset prices (stocks, bonds, real estate) probably reflect all available information. The EMH comes in three flavors, weak, semi-strong, and strong.
- Weak EMH: Prices reflect all market information. You can’t beat the market by looking at what happened to the price yesterday.
- Semi-Strong EMH: Prices reflect all publicly available information. You can’t beat the market by reading up on the company.
- Strong EMH: The market is efficient, reflecting all public and private information. The asset price is a full and complete measure of the “true” value of something, including expectations about the future.
The Weak EMH says you couldn’t, in 2006, beat the housing market merely by noticing what had happened to prices since 2000 where you were planning to buy; this is by and large true. Consider that Las Vegas, Phoenix, Brooklyn and San Francisco all increased in price greatly between 2000 in 2006, but while buying in Vegas or Phoenix in 2006 was a terrible idea, buying a single family home in Brooklyn or San Francisco in 2006, if you were willing to grit your teeth through a few bad years, would’ve been a great idea.
The Semi-Strong EMH says you couldn’t do some research, look at demographics of the people buying in each place in 2006 and tell that Brooklyn was a better bet than Phoenix (or that stocks were a better bet than real estate.) This is probably false, but not certainly false. Sure, a lot of people buying in Brooklyn had the kind of money where they could put down $800,000 cash, while a lot of people buying in Phoenix were getting Countrywide loans with zero down payment and zero documentation. On the other hand, some places that looked demographically similar (about half white, 1/3 black, and 1/6 other) and had a higher median income than Brooklyn have done a lot worse since 2006; for example, Chicago:
So, call the semi-strong EMH a draw.
The strong EMH says there was no housing bubble, there was just a “black swan” unpredictable event that eradicated what would’ve been endless growth in home prices in Phoenix, Vegas, and the Inland Empire east of LA. Sounds wrong to me.
So, I’ll propose the weak, semi-strong, and strong version of the Voldemort View (VV).
Weak VV: Mean differences in group IQs at the time of test taking explain the academic achievement gap in racial and SES groups.
Semi-Strong VV: Mean differences in group IQs that K-12 schooling cannot significantly affect explain the academic achievement gap in racial and SES groups.
Strong VV: Mean differences in group IQs are fully representative of “intelligence, broadly defined,” and of the value of schooling to different groups.
The Weak VV is, as Ed Realist puts it, not a hypothesis but a fact. You can completely explain differences in 4th grade test scores by 4th grade IQ gaps, 8th grade test scores by 8th grade IQ gaps, and 12th grade test scores by 12th grade IQ gaps. In fact, the IQ gaps are larger than the test score gaps, in many cases; this is because the elementary school tests are often not that g-loaded (though that may be changing) and because schooling in minority-majority schools has been highly focused on academic subjects and test preparation.
The Semi-Strong VV is somewhere between a hypothesis and a consensus interpretation of the evidence. For example, James Heckman is widely hailed by liberals for providing evidence in favor of pre-K, but he doesn’t really believe that schooling (even pre-K) can greatly influence IQ and cognitive ability. Instead, he thinks early investments like pre-K and Nurse Family Partnership influence non-cognitive abilities, which show later benefits (completing school, not going to jail, sticking with jobs) independently of IQ. Here, for example, from one of his presentations on the Perry Pre-School Project, are the IQs of the Treatment and Control groups over time, showing-complete fade-out:
So the Weak VV is indisputable and the Semi-Strong VV is endorsed even by people you wouldn’t expect. What about the Strong VV?
The Strong VV is something of a metaphysical question. What is “intelligence, broadly defined” if it’s not IQ? Ed Realist quotes this passage from Stephen Pinker about the validity of IQ as a measure:
Pinker: I think you’re wrong about IQ tests in general. They’ve been shown to predict (statistically, of course) a vast array of outcomes that one would guess require intelligence, including success at school, choice of intellectually demanding professions, income (in a modern economy), tenure and publications in academia, and other indicators, together with lower crime rates, lower infant mortality, lower rates of divorce, and other measures of well-being. The idea that IQ tests don’t predict anything in the real world is one of the great myths of the intellectuals.
I’ve said before that I disagree with the Strong VV (“Against Unidimensionality”) because smart people are so very dumb so often and because while abstraction and inference are very valuable things, they aren’t everything.
For example, I spent the majority of my time as a teacher teaching 7th grade life science in a very diverse school, racially and socioeconomically (roughly 15% white, 25% black, 40% Hispanic, and 20% Asian.) The performance of kids in my class followed the distribution you would expect from differences in mean group IQ only very, very roughly. This makes sense, because 7th grade life science is a pretty concrete topic and I was doing my very best to make it more concrete. On the other hand, I taught an after-school course preparing kids for the New York Specialized High School admissions test (the test to get into Bronx Science and Stuyvesant), and while I had a lot of success preparing kids for that test by the standards of students citywide, it was exactly the kids you’d expect: upper-middle class white kids and the children of Chinese and South Asian immigrants. The test is a very g-loaded measure, it’s all logic puzzles and math problems much more difficult than those on the GRE. (In fact, I didn’t need to study for the GRE at all later on because the SHSAT, while given to 8th graders, is a much harder test.)
But does that mean that my 7th grade life science class didn’t involve any intelligence? It certainly seemed like the kids were using their minds, designing experiments and arguing about evolution and explaining how the circulatory system worked.
Arthur Jensen, one of the founding fathers of IQ research, described these differences in terms of “Level I and Level II” intelligence as follows:
- Level I ability consists of rote learning and primary memory; it is the capacity to register and retrieve information with fidelity and is characterized essentially by a relative lack of transformation, conceptual coding, or other mental manipulation intervening between information input and output. Level II ability, in contrast, is characterized by mental manipulation of inputs, conceptualization, reasoning, and problem-solving; it is essentially the g factor common to most complex tests of mental ability and standard tests of intelligence.
- Level I tasks were those that required little or no mental manipulation of the input to arrive at the correct output. A clear example of Level I ability is Forward Digit Span in which people recall a series of digits in the same order as that in which they are presented. Level II tasks, however, require some mental manipulation of the input in order to arrive at the appropriate response. A clear example of Level II ability is Backward Digit Span in which people recall a series of digits in the reverse order to that in which they are presented.
Jensen’s explanation is that IQ gaps between groups are mostly in “Level II” intelligence; lower average IQ groups can perform “Level I” tasks (rote retrieval, essentially) but not more complex manipulation.
This actually doesn’t match my observations all that well. It’s not about “rote versus complex,” it’s about “concrete versus abstract” and “socially-embedded versus individually-achieved.” The kids who could fix the classroom air conditioner or figure out why the fish were dying in the fish tank weren’t the kids who did best on written tests. The kids who learned the bones of the body when they were part of a dance with pulsating techno-background music everyone did together weren’t the kids who could go home and memorize them easily on their own. (In fact, the kids who could go home and study the bones on their own often hated doing the dance.)
This doesn’t mean the kids who fixed the air conditioner or liked doing the bones dance are going to go on and be huge successes in college and get high paying jobs. By and large, they aren’t, because college (the academic side, anyways) and the modern workplace are abstract, individualistic places. But it does, to me, say something about whether IQ represents “intelligence, broadly defined.”
One idea I’ve had is that, instead of Jensen’s distinction between Level 1 (rote) and Level 2 (complex) reasoning, a better distinction is one that became very well known recently in a different context: psychologist Daniel Kahneman’s separation of cognition into “System 1” and “System 2”:
• System 1 operates automatically and quickly, with little or no effort and no sense of voluntary control.
• System 2 allocates attention to the effortful mental activities that demand it, including complex computations. The operations of System 2 are often associated with the subjective experience of agency, choice, and concentration.
System 1: Fast, automatic, frequent, emotional, stereotypic, subconscious
System 2: Slow, effortful, infrequent, logical, calculating, conscious
Kahneman became famous and revered in large part because he identified a set of System 2 tasks that were extremely difficult for even very intelligent individuals to accomplish. Kahneman’s work is considered important and was the foundation of behavioral economics (showing why the Strong EMH is wrong, and providing the scientific justification for broader paternalism) because it appears to show departures from rationality by everybody– even super-smart people often make the mistakes he predicts. This finding has been used to elide over the fact that dual process cognition is every bit as much a theory of individual and group differences as it is a theory of universal departures from rationality. Many studies suggest that the ability to accomplish System 2 tasks in general, if you don’t just isolate those mistakes that everyone makes, is in good part the same thing as IQ:
System 2 function should be related to measures of general intelligence, although System 1 function should be independent of such measures. Stanovich and West have demonstrated in a recent series of studies [6,26,36,37] that mthe ability of participants to find normatively correct solutions to a range of inferential and decision making tasks was consistently associated with those who were high in cognitive ability as measured by SAT scores.
In some cases, indeed, as my experience teaching suggested, individuals can accomplish a task if it is concrete and relevant to prior information in a way that they can’t if it is purely abstract:
One of the most investigated reasoning problems in the literature is the Wason selection task (see Box 3). In an abstract, indicative version (Figure Ia in Box 3), it is known to be very difficult. However, in a realistic, deontic version (Figure Ib), is it is quite easy. On the dual-process account, the former is difficult because it requires explicit and abstract logical reasoning of the kind that only System 2 can provide. With the latter task, however, the correct answer is strongly cued by relevant prior knowledge, reflecting System 1 processes .
This matches my experiences much more than Jensen’s theory: that different groups are predisposed to abstract versus concrete reasoning to different degrees, and to using cognitive heuristics versus figuring everything out for themselves from scratch. Kahneman describes heuristics as a source of bias and error, and indeed in the modern, increasingly abstract and statistically constructed world it no doubt often is. But as Gerd Gigenrezer has repeatedly pointed out, heuristics aren’t always wrong, they’re just ill suited to certain tasks:
As reflected in the amount of controversy, few areas in psychology have undergone such dramatic conceptual changes in the past decade as the emerging science of heuristics. Heuristics are efficient cognitive processes, conscious or unconscious, that ignore part of the information. Because using heuristics saves effort, the classical view has been that heuristic decisions imply greater errors than do “rational” decisions as defined by logic or statistical models. However, for many decisions, the assumptions of rational models are not met, and it is an empirical rather than an a priori issue how well cognitive heuristics function in an uncertain world. To answer both the descriptive question (“Which heuristics do people use in which situations?”) and the prescriptive question (“When should people rely on a given heuristic rather than a complex strategy to make better judgments?”), formal models are indispensable. We review research that tests formal models of heuristic inference, including in business organizations, health care, and legal institutions. This research indicates that (a) individuals and organizations often rely on simple heuristics in an adaptive way, and (b) ignoring part of the information can lead to more accurate judgments than weighting and adding all information, for instance for low predictability and small samples. The big future challenge is to develop a systematic theory of the building blocks of heuristics as well as the core capacities and environmental structures these exploit.
Does that mean that IQ isn’t a valid measure, doesn’t predict a lot of things (especially things very similar to IQ tests), and that disparities in certain types of accomplishment aren’t quite possibly here to stay? Of course not.
But it does mean that our baseline intuition- that who is dumb and who is smart depends on when it is and where you are– isn’t a total bust.
So sign me up for the Weak Voldemort View, and I’ll defend the Semi-Strong Voldemort View 80 percent of the way. But I think the jury is out on the Strong VV, and the human world remains a complex place, where a single number only gets you so far.