Sunday 14 June 2015

Is France sinking even further?

 

There are few territories more blessed with natural treasures, agreeable climates and regional cuisines than the Grand Hexagonal, that gorgeous chunk of Europe called France. By every environmental theory this part of the planet should breed a race of super-folk: Asterix the Gaul on natural geographical steriods, a fraternal band of clever Gallic communards. Instead, if we are to believe French self-perceptions, they have fallen into a morose morass of despondency. They have lost their joie de vivre, esprit d'aventure, esprit de corps, sens de la nationalité, and in a mangling of Bonnie Tyler’s song, they are truly Perdu en France.

What ails our French cousins? One of their problems seems to be that they are cursed with a reverse Flynn Effect, a degeneration of the national intellect. So argue Edward Dutton and Richard Lynn in “A negative Flynn Effect in France, 1999 to 2008–9” Intelligence 51 (2015) 67–70.

https://drive.google.com/file/d/0B3c4TxciNeJZZFhLc1NuRHFRcTA/view?usp=sharing

They have looked at a small, probably representative sample used by the Wechsler team in their 2011 French standardisation of the adult form of their general intelligence test. They say: The results of the French WAIS III (1999) and the French WAIS IV (2008–9) are compared based on a sample of 79 subjects aged between 30 years and 63 years who took both tests in 2008–2009. It is shown that between 1999 and 2008–9 the French Full Scale IQ declined by 3.8 points.

Please draw hard on your Galloise, then bite the bar of Menier chocolate in your baguette, sip your Carte Noir coffee, and have a look at the results.

The Wechsler Adult Intelligence Scale III (WAIS III) was standardized in France in 1999 (Wechsler, 2000) and the Wechsler Adult Intelligence Scale IV (WAIS IV) was standardized in France in 2008–9 (Wechsler, 2011). The two tests were administered to 79 subjects (a separate sample from the 876 subjects who composed the broader French WAIS IV) who were aged between 30 years and 63 years (mean age 45 years), approximately half of whom took the WAIS IV first and half took the WAIS III first, in order to control for practice effects. The time between the administration of the two tests varied from between 6 and 76 days, with an average of 27 days' gap. The manual does not state whether there were significant differences in the test spacing between the two groups. However, the sample of 79 was a means of comparing the norms yielded by the two standardized samples. As such, if there were significant differences in test spacing between the two groups this would substantially undermine the purpose of administering the tests in this way.

There are a couple of problems. As the authors themselves make clear, 79 subjects is a rather small sample from which to draw any conclusions about the Fall of the Fifth Republic. The Wechsler team haven’t said much about the standardisation sample, not even the proportion that are recent immigrants. More important, in my view, is the whole business of giving contemporary subjects a new and an old test, and then making judgments about intellectual levels then and now. “Then” has been digested and is part of current collective knowledge, “Now” is still inchoate, fugitive and capable of surprising.

Giving the same subtests isn’t even a totally straight comparison: the constituent items can vary, and the actual responses (in terms of number correct per subtest) have to be converted to standardised scores using a table which attempts to bundle the raw results into ranges which are then expressed as a single standardised subtest score. For example, if a person gets a raw score of 12 out of 28 on a test, then the examiner looks up the conversion table to establish what the scaled score should be (mean score always 10, standard deviation always 3). Depending on the particular subtest, a range of raw scores may share the same standardised score. For example, raw scores of 12 or 13 might both have the same standardised score of, say, 9. The standardisation procedure loses fine detail, and the Flynn effect should be based on the most accurate measures possible.

image

The comparison of subtests is shown above. If subjects do better on the more recent test than the one designed a decade before, this suggests that the early test was based on norms for brighter persons, and that the current test has been dumbed down to a new, lower, average. Sure enough, save for the rather dull and mechanical Symbol Search information-processing task (find a particular letter in an array of letters) all the subtests are now a bit harder for contemporary test-takers. Their vocabularies have shrunk considerably. Either that, or Wechsler screwed up the choice of new words (impressive fall only if they were the same words). Even if only one or two are wrong in terms of contemporary word frequency then the large change is explicable. Digit span (which could have been shown in raw scores, because it it a real ratio measure) is unchanged. Arithmetic, which follows standard, eternal rules is effectively unchanged. Comprehension and Information, which are hard to craft precisely in terms of difficulty, show the biggest apparent changes.

All that said, there is no particular reason that Matrix Reasoning should have dropped a bit, nor Picture Completion, nor Block Design. I would need to see both versions to see what, if anything, had been changed in all the subtests. In UK standardisations there are always perturbations in the new tests when compared with the old. Often the newer ones are less good in clinical practice.

The authors do well to draw this unremarked finding to our attention. They consider various explanations, but do not find any strong candidates. While it would be good to know how many recent immigrants were in the sample they judge it unlikely to have been so many as to influence the results greatly, and although it might be due to dysgenic fertility, drops of this magnitude have not been found in other countries. As a corollary, one cannot help but conclude that if indeed national intellect is leaking away so quickly, it may explain why French intellectuals are revered in their home country, and less admired elsewhere.

All this remains a French puzzle, best not mentioned whilst on holiday in that estate, lest it deepen the all-encompassing Gallic gloom.

16 comments:

  1. "While it would be good to know how many recent immigrants were in the sample they judge it unlikely to have been so many as to influence the results greatly"
    A demographic shift seems the most likely and obvious explanation. The authors perhaps "judge it unlikely" for political reasons.

    ReplyDelete
    Replies
    1. No, for statistical reasons, compared with Finland. However, no hard data on this particular sample.

      Delete
  2. By every environmental theory this part of the planet should breed a race of super-folk: Asterix the Gaul on natural geographical steriods, a fraternal band of clever Gallic communards.

    But they were! In recent memory, the French produced more eminent mathematicians per capita, per year than any other nationality, ever.

    A partial listing: Descartes, Pascal, Laplace, Lagrange, Fourier, d'Alembert, de Moivre, Argand, Condorcet, Legendre, de l'Hopital, Parseval, Rolle, Fermat, Vernier.

    All of these men were giants! There are so many that I've probably left off some extraordinary figure or another.

    ReplyDelete
    Replies
    1. Poincare, Poisson, Galois, Hermite, Carnot, Binet, Jordan, Laurent, Lebesgue, Levy, Sturm and Liouville.

      Et cetera et cetera.

      Delete
    2. Assuming that the Flynn Effect probably began in earnest around 1870 at earliest, we are looking for outstanding achievement since then. Even if we say that the last 50 years are too contemporary to allow us to judge, we would still expect French science Nobels to be significant in the 1980 to 1965 period which they are not, compared to UK. Indeed, Trinity College Cambridge has more Nobels than France. If good food and education lead to genius, then the French should be ahead.

      Delete
    3. "Good food": try plotting the decline of IQ in France with the rise in McDonald's restaurants.

      Delete
    4. France is way ahead on Fields Medals:

      http://www.trin.cam.ac.uk/medallists-and-prize-winners

      And that is even if you decide to not count Grothendieck, Werner, Châu, Kontsevich and Avila...

      Delete
    5. At a quick glance, by affiliation at the time of award of Medal, US 26, France 16, UK 8. So US way ahead, France well ahead of rest.

      Delete
    6. Alright, let's pretend France and the US have anywhere near the same population size.

      Delete
    7. Of course, of course. More interesting, where do the best mathematicians move to when they leave their country of birth and/or country of maths education? US tends to hoover up global intellects. Also of interest: big split between STEM type achievements, where France does well, and the arts type subjects, where success outside France is more limited.

      Delete
    8. Oh I agree with our host, there aren't anywhere near as many eminent Frenchmen now as there were in 1870. Some of them obviously retain the problem-solving knack, but it is unlikely we'll see another Laplace. The decline is startling in its quickness.

      Delete
  3. As an aside - my top 3 favourite operas were composed by Frenchmen.

    Hippolytus and Aricia (Rameau)

    The Trojans (Berlioz)

    Pelleas and Melisande (Debussy)

    ReplyDelete
  4. In my opinion it's a general problem which concerns every single country new technologies are responsible since creativity and thinking are more and more "the web". But "Paris sera toujours Paris" ...

    ReplyDelete
  5. There may be a brain drain at the very highest levels of achievement.

    http://news.sciencemag.org/2010/11/why-french-scientists-find-joie-de-vivre-u.s.

    http://www.ft.com/cms/s/0/72627106-9318-11e3-8ea7-00144feab7de.html#axzz3d9Cqb2rX

    A quick search turned up "brain drain" in the scientific & tech fields. I wonder if it's present in other fields as well? And yet, over a population of millions, would the loss of a few thousand each year make a significant difference in the entire nation?

    ReplyDelete
  6. Sample size too small, lack of necessary statistics to draw an inference. The authors don't even try.

    ReplyDelete
  7. If it doesn't result from the inclusion of lower IQ immigrants, in other words a failure to match the samples, I will vote for the increasing mutation load, or whatever "x" factor has been causing decreases in intelligence in the West for decades.

    ReplyDelete