« "Anthropology" | Gene Expression Front Page | Carl on Hobbits - Update »
June 14, 2005

Lynn-Flynn Effect, R.I.P.

Teasdale & Owen and Sundet, Barlaugb, & Torjussen both have recent data from Northern Europe showing an end to the Lynn-Flynn Effect.

As a picture says 1000 words.....

Subdet et al got nationally representative (male) data from 1957 through 2002 on Norway's military's "IQ" instrument (r[WAIS FSIQ]=.73). If you use 1957 as the reference group, there was an initial LFE for all the subtests, but it stagnates around 1993 (Figure 1). If you make 1993 the reference, then Figures (think Raven's Matrices) stagnates, while Vocabulary and Math decline.

Figure 1.

Figure 2

Teasdale and Owen, in a to-be-published article, showed a similar effect in Denmark using a similar population as the Sundet group. If you use 1959 as the reference group, then there is an initial increase in IQ (actually the Børge Priens Prøve; r[WAIS FSIQ]=.82) (See Figure 3), but when you make 1993 the reference group, there is a brief stagnation, followed by a drop that shows no recovery.

Figure 3.

Figure 4.

Now, lest we think this effect is only in Europe, I have data from the mathematics section of the American College
Basic Academic Subjects Examination
showing the same effect:

First, the Mathematics section was used becasue its substests load well on g:

CBASE Math Skill:

g
Geometrical calculations .711
2- & 3-Dimensional figures .687
Equations & Inequalities .658
Evaluating Expressions .636
Using Statistics .770
Properties & Notations .673
Practical Applications .736

Second, from 1996-2001, there has been the same reverse LFE, using Classical Test and Item Response Theory models:

Year
Means
SD
CTT Standardized "True Score" Difference
IRT Scaled Latent Trait Difference
1996 36.214 10.219    
2001 34.393 10.929 -.178 -.222

The interpretation of the data is still up-for-grabs; I am in the Burt (1952) camp in thinking that the whole LFE (although, obviously, he didn't call it that) is just a psychometric artifact hiding a dysgenic trend1. Still no matter if Lynn et al. are correct and there really has been true rise, the effects appear to be leveling off or perhaps dissipating, at least in developed countries (in "still developing" countries the effect might still be going on).

References
Burt, C. L. (1952). Intelligence and fertility (2nd ed.). London: Eugenics Society.

1. As soon as my committee gives the thumbs up, I'll give the details of how this is quite possible.

Posted by A. Beaujean at 12:32 AM