by Ian J. Deary
But it is precisely at this point that one’s head begins to spin: do less demanding textbooks and low-level TV programs raise intelligence while lowering verbal skills; do declining standards in schools sharpen the mind while undermining study habits; does student absenteeism mean students are engaged in mentally demanding tasks while missing out on knowledge; does a demoralised family environment boost IQ while lowering motivation? (p. 38)
Perhaps, though, it’s not as bad as that. It could just be that the test companies are not testing appropriately representative groups of people in their attempts at making up their norms. They might be going out generation after generation and getting it wrong by testing evermore biased, more clever samples for their norms tables, making it harder for those being tested to do well by comparison. Or perhaps the contents of the tests are steadily leaking out over time to the public so that people in successive generations have had more experience with the items? Thus, at the end of his first large-scale study, James Flynn came up with three points that might explain the ‘massive gains’ that successive American generations were attaining in IQ scores.
1 Artefact. The gains might be ‘not real, but an artefact of sampling error’. That is, the groups recruited to provide norms might, over time, become more biased toward containing cleverer people. This is very unlikely to have occurred in such a systematic way as to make all later normative samples brighter than all earlier ones. But even if this is the explanation, it still makes scores across IQ tests non-comparable.
2 Test sophistication. Successive generations might not actually be more intelligent; they might just be scoring better on the tests for some reason that we have to go and find. This leaves us the large, additional problem of explaining the reason for SAT test scores declining.
3 A real intelligence increase. If the test score differences represent real increases in intelligence, they are very hard to explain. Flynn tried to examine the most likely candidate: that socio-economic improvements accounted for the IQ gains across generations. However, the gargantuan alterations that would be needed in living standards to account for the IQ changes were just not plausible.
James Flynn wanted more definitively to identify the source of the rising IQ scores. Broadening out from the USA, he sought examples of IQ test scores that had been collected across generations. Here’s how he described that search:
The method used to collect data can be simply put. Questionnaires, letters, or personal appeals (usually a combination of all three) were sent to all those researchers known to be interested in IQ trends on the basis of scholarly correspondence and the exchange of publications. One-hundred sixty-five scholars from 35 countries were contacted. They came from Europe (every nation except Albania, Denmark, Greece, and Portugal); Asia (Japan, India, and Israel); Latin America (Argentina, Brazil, Chile, Cuba, Mexico, and Venezuela); the Caribbean (Barbados and the U.S. Virgin Islands); and the Commonwealth (Australia, Canada, and New Zealand). American data were available from a previous study. Military authorities in charge of psychological testing were contacted in every European country, plus Australia, Canada, Greenland, Iceland, and New Zealand, as were 21 educational research institutes in Western Europe and the Commonwealth. (p. 171)
Key dataset 11
This is typical of James Flynn. He does nothing by halves, and he has thrown over 20 years of his academic life into scouring the world for data to address the ‘rising IQ’ problem. Some of Flynn’s strongest data came from military samples, in those countries where nearly all young men were given IQ tests at entry to compulsory military service. Figure 25 illustrates some of Flynn’s data.
Here’s how to look at Figure 25. The vertical scale at the left-hand side is an IQ scale. Along the horizontal are some different countries from which Flynn got good data. In each country the most recently available data have been set at an arbitrary IQ score of 100. These appear at the top of the 5 vertical lines. An IQ of 100 is, by an arbitrary definition, the population average. For each of the 5 countries in the Figure there were earlier testings of the same population. The dates down the dotted
vertical lines shows just how much lower the populations’ IQs were in earlier testings. Note the dots on each vertical line with a date against them: these dates are when the IQ testings of the population took place. If you read across from these dots/dates to the left you can see what the average IQ of the population was at that date, compared with the 100 score of the most recently tested population. Please note that we should expect that all of these testings would give rise to average IQs of 100. They do not. Whenever a population was tested at an earlier date the imputed IQ average was lower. The effect found among American whites was also found in many other countries, leading Flynn to name his 1987 Psychological Bulletin article ‘Massive IQ gains in 14 nations’.
25. Generation by generation, nations are scoring better on IQ tests.
Take the example of the Netherlands. Since 1945 the Dutch military have tested almost all young men in the Netherlands on 40 of the 60 items of Raven’s Progressive Matrices. This is a non-verbal mental ability test and is supposed to be quite good at testing general intelligence. Flynn examined these data and reported the percentage of young men who were scoring more than 24 of the 40 items correct. The percentages were:
31.2% in 1952
46.4% in 1962
63.2% in 1972
82.2% in 1981/2.
Setting the 1982 scores to a mean IQ of 100, one can work back and ask the question: what was the mean IQ score of the earlier generations based on the percentage that achieved the pass rates? Figure 25 reveals that the Dutch men in 1972 averaged around IQ 90, the 1962 population around 85, and the 1952 population below 80. Additional proof of this increase arose from a comparison of over 2,800 men tested in 1981/2 and their fathers tested in 1954. The sons were 18 IQ points higher than their fathers who had been tested 27 years earlier. Thus, as I hinted in the first paragraph of this chapter, we see this puzzling effect in people who are genetically related and who have lived in the same culture, where we would expect similar average IQ scores.
Look again at Figure 25. Norwegian data for approximately the same period show gains for later generations too, but they are smaller than those of the Dutch. Belgian military data showed a rise of 7 IQ points over the relatively short period from 1958 to 1967. New Zealand children gained an average of 7.7 IQ points between 1936 and 1968 (data not shown). Two further sets of data from Flynn’s large number of comparisons are shown: Israelis gained 11 IQ points over the 15 years from 1970 to 1985 and people in the United Kingdom went from a putative mean IQ of 73 in 1942 to 100 in 1992.
This last increase makes a good illustration of the impact if these changes were real alterations in intelligence levels. Compared with a mean of 100 in 1992, the mean for the population in 1942 would be almost at a level that indicated mental handicap for the average person. (It is this consideration that makes me very sceptical about the veracity of these supposed ‘IQ gains’.) In the end, Flynn found reasonable data on 14 nations and for a generation (30 years) he found IQ gains between 5 and 25 points, with an average of 15. These data are stunning, and very challenging for researchers in the field of intelligence.
One key fact to focus on when thinking about the ‘Flynn effect’ of rising IQ scores is that the biggest effects tend to occur in so-called culturally reduced tests. That is, the rises occur most markedly in those tests that do not seem to have contents that can easily be learned. For example, Raven’s Progressive Matrices is among the tests that show the highest gains. Yet Raven’s Matrices involves finding the correct answer that completes an abstract pattern. It has no words, no numbers, nothing really that can be taught so that the later generation will do better than the former. Flynn’s review of his massive datasets confirmed this.
A consensus about the significance of generational IQ gains depends, therefore, on whether they manifest themselves on culturally reduced tests like the Raven’s. These tests maximise
problem-solving and minimise the need for more specific skills and familiarity with words and symbols. [There are] strong data for massive gains on culturally reduced tests: Belgium, the Netherlands, Norway, and Edmonton show gains ranging from 7 to 20 points over periods from 9 to 30 years; when the rates of gain are multiplied by 30 years, they suggest that the current generation has gained 12–24 points on this kind of test. Tentative data from other nations are in full agreement. This settles the question at issue: IQ gains since 1950 reflect a massive increase in problem-solving ability and not merely an increased body of learned content. (p. 185)
The Flynn effect is well established. Its importance is reflected in the eponymous title, and in the interest it has attracted since the late 1980s. The American Psychological Association had a full meeting on the issue, and published a book in which many experts sought an answer to it. It is easy, and accurate, to summarize by saying that experts are dumbfounded. There are two broad responses to the Flynn effect.
The first response is to suggest that the Flynn effect is real, marking an actual improvement in brain power in successive generations across this century. People who opt for this account suggest that we have a good exemplar in height. Human height has increased across the century as a result of better nutrition and general health, so why not intelligence? Flynn himself seemed not to favour this option. He worked out that, in countries such as the Netherlands and France, where there have been high IQ gains across generations, teachers should now be faced with classes in which 25% are gifted and where geniuses have increased 60-fold! ‘The result should be a cultural renaissance too great to be overlooked.’ (p. 187) Flynn searched French and Dutch newspapers, especially periodicals relating to education, from the late 1960s to the present and found no mention of any great increase in intellectual achievements by newer generations.
The second response suggests that the Flynn effect is an artefact. It is not the case that people are more intelligent. Instead, what has happened is that people have become more familiar with the test materials. Children’s toys, magazines, television programmes, computer games, and so forth might contain materials that have IQ-item-like properties, and so people do better on the tests when they come across them. This might be termed the ‘Early Learning Centre’ theory.
There is one thing to note about the Flynn effect that Flynn himself has been keen to emphasize. Though the effect is clearly important, it does not compromise the validity of mental test scores within generations. Mental test scores, despite the ‘massive gains’ through time, do retain their reliability, their ability to predict educational and job successes, and their heritability, but only within each generation. The key point is that something in the environment (many researchers believe that it has to be the environment because some of the across-generation samples tested fathers and sons) of many countries across the middle years of the 20th century has led to ability scores increasing substantially.
Flynn makes a telling point when he asks us to reflect on the fact that being born a generation or so apart can make a difference of 15 IQ points. We have no good account of the causes for this change; it is officially mysterious. Given, though, that he could find no evidence for the present generation’s genius in achievement over former generations, Flynn says that IQ tests like Raven’s do not measure intelligence, but only some correlate of intelligence, which he calls ‘abstract problem-solving ability’. Further, he insists that differences in this ability are 15 points between successive generations, and these differences must arise from some environmental factor. He concludes that IQ test differences cannot be used to make trustworthy comparisons of the intelligence of different generations or of different cultural groups.
The reader might like to reflect on the Flynn effect and its causes, not least because some fresh thinking on this matter might offer psychologists a foothold on a slippery problem. If there was a prize to be offered in the field of human intelligence research, it would be for the person who can explain the ‘Flynn effect’ of the ‘rising IQ’.
To follow this area up…
James Flynn’s oeuvre consists of three rather stunning research papers. The last of these is the most accessible, giving a general, popular summary of his findings. The second half of it is about Flynn’s view of social justice and how intelligence differences fit into that view: it’s intelligent, humane, and worth looking at.
Flynn, J. R. (1984). The mean IQ of Americans: Massive gains 1932 to 1978. Psychological Bulletin, 95, 29–51.
Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 95, 29–51.
Flynn, J. R. (1999). Searching for justice: the discovery of IQ gains over time. American Psychologist, 54, 5–20.
The former two papers are full of details about datasets from around the world. Had they been written by a psychologist I feel sure they would be as dry as dust. Flynn, probably because of his background, makes the accounts readable, and he tries to spell out even the technical stuff so that one need not be a psychometrician to understand them.
The American Psychological Association’s book which addresses the continuing conundrum of the Flynn effect contains a good range of opinion: from those who think there has been a real rise in IQ in recent decades (usually citing better nutrition as the key factor) to those who think the Flynn effect is an artefact (more educational toys and TV programmes, etc.) or something more complex. What I can tell you is that this book assembled an impressive list of relevant international researchers and none has a convincing explanation of the ‘rising IQ’.
Neisser, U. (ed.) (1998). The Rising Curve. Washington, DC: American Psychological Association.
Chapter 7 Eleven Twelve (not-so-) angry men (and women)
Psychologists actually agree about human intelligence differences
A key working party
As an interested layperson, it can’t be easy or satisfying trying to get to grips with some straight facts about human intelligence differences. The highly visible experts in the area tend to represent one extreme or the other in advocating IQ-type testing. Media coverage reflects this, tending to put one side of the debate, or just the two extremes, or merely reporting the slanging match. Many years ago a proponent and antagonist of human intelligence testing, Hans Eysenck and Leon Kamin respectively, jointly wrote a book about intelligence. Its title was The Battle for the Mind. They wrote separate sections on the research as they saw it, and they responded to each other’s sections. The result was heat rather than enlightenment for the reader. The writers were further apart at the end than they were at the start. What hope is there for the general, curious reader when the cognoscenti are inhabitants of this Babel?
In fact, it took a big furore to knock some heads together and for psychologists to come to a clear realization that there was a strong consensus about the research findings on human intelligence, right across the spectrum of researchers. The result was one of the most useful accounts of intelligence research ever to become available for the non-specialist.
26. The cover page of the American Psychologist for February 1996, which featured the American Psychological Association’s Task Force report ‘Intelligence: knowns and unknowns’.
The furore: in the mid-1990s a book called The Bell Curve rewrote the rules for academic book distribution. Close on 900 pages long, almost 300 of which were statistical analyses, detailed footnotes, and academic journal references, it sold in the USA in the hundreds of thousands. It brought just about every dispute concerning IQ freshly to the pages of newspapers and magazines, and got the Western world (at least) and the psychological research community in a turmoil over the impact that mental ability has on our destinies. It excoriated the body of intelligence research by addressing IQ scores in the context of social outcomes and social policy. The resulting tintinnabulation from the chattering classes alerted the professional psychological associations: if people were arguing about IQ, shouldn’t they at least have some undisput
ed facts as a basis for commenting on The Bell Curve’s contents?
The response: the American Psychological Association (APA), the largest and most authoritative professional psychological society in the world, became fed up with uninformed argument on intelligence. Not prepared any longer to stand on the sidelines, they decided they had a responsibility to put on record the findings about human intelligence that attracted very wide consensus among psychologists. Their Board of Scientific Affairs (BSA) appointed a Task Force to collect together what researchers did and did not know about human intelligence differences. My aim in this chapter is to show that the report from this Task Force is the best available, unbiased summary of the topic. It will add variations to the themes raised in this book and is a good first source of further reading.
The Task Force’s report comprehensively and concisely tells the wider world what is and is not known about human intelligence (IQ) differences. Here’s how they introduced their report:
In the fall of 1994, the publication of Herrnstein and Murray’s book The Bell Curve sparked a new round of debate about the meaning of intelligence test scores and the nature of intelligence. The debate was characterised by strong assertions as well as strong feelings. Unfortunately, those assertions often revealed serious misunderstandings of what has (and has not) been demonstrated by scientific research in this field. Although a great deal is now known, the issues remain complex and in many cases still unresolved. Another unfortunate aspect of the debate was that many participants made little effort to distinguish scientific issues from political ones. Research findings were often assessed not so much on their merits or their scientific standing as on their supposed political implications. In such a climate, individuals who wish to make their own judgements find it hard to know what to believe.