Tuesday, June 19, 2018

Why is IQ declining in Norway?

Norwegian conscripts (Wikicommons: Soldatnytt) – in Norway, the conscript register provides invaluable information on IQ. This information can then be linked to other registers of the Norwegian population registry

IQ scores have been gradually increasing since the 1930s at the rate of 3 points per decade (Rindermann 2018, pp. 85-89). This increase, named the Flynn effect, has been much debated. Are we becoming more intelligent because of better nutrition and more stimulating learning environments? Or are we simply becoming more familiar with tests and test-taking?

Whatever the cause, this increase seems to be slowing throughout the West (Flynn 2007, p. 143). In Scandinavia, mean IQ peaked during the late 1990s and has since declined (Teasdale and Owen 2005).Why? There is plenty of speculation. Perhaps the poor are outbreeding the rich. Or perhaps lower IQ immigrants are replacing higher IQ natives. Whatever the cause, it seems to be something that is more advanced in Scandinavia than elsewhere in the West. 

Two Norwegian economists, Bernt Bratsberg and Ole Rogeberg, have tried to pin it down. 

Population intelligence quotients increased throughout the 20th century—a phenomenon known as the Flynn effect—although recent years have seen a slowdown or reversal of this trend in several countries. [...] Using administrative register data and cognitive ability scores from military conscription data covering three decades of Norwegian birth cohorts (1962-1991), we show that the observed Flynn effect, its turning point, and subsequent decline can all be fully recovered from within-family variation. (Bratsberg and Rogeberg 2018)

"Within-family variation" means that IQ is declining even among people with more or less the same genetic background, i.e., siblings. More to the point, this within-family decline seems to account for most, if not all, of the IQ decline among Norwegians. So this is not a matter of the poor outbreeding the rich or immigrants outbreeding natives. In fact, it doesn't seem to reflect any sort of genetic change.

All of this assumes, of course, that the genetic differences between siblings are the same now as in the past. That assumption may appear reasonable, but appearances can deceive.

Siblings in Norway are increasingly half-siblings. Among Norwegian women with only two children, 13.4% have had them by two or more men. This figure rises to 24.9% among those with three children, 36.2% among those with four children, and 41.2% among those with five children (Thomson et al. 2014). This multiple paternity is produced not so much by family size as by the passage of time. The probability of a relationship breaking up will increase over time. Brothers who are born farther apart are more likely to have different fathers.

Stepfamilies typically arise when a woman divorces and has a child by a second husband. Or she may divorce, have a child out of wedlock, and then marry a man who provides for the family without ever fathering any children. In this chain of sexual partners, the second or third man is qualitatively different, on average, from the first man. The stepfather, in particular, is less able to compete in the marriage market for one reason or another. In general, he has been less successful in life, and life, in itself, is an IQ test.

So, within a stepfamily, one would expect to see an IQ decline between older and younger siblings. The methodology of Bratsberg and Rogeberg (2018) is especially sensitive to this stepfather effect because the IQ decline is a function of time, i.e., the decline is most visible between brothers born farther apart. As this birth interval increases, however, so does the probability that the younger brother is a half-brother.

How strong is this stepfather effect? Bratsberg and Rogeberg studied the recent IQ decline in Norway by looking at pairs of brothers. The IQ data come from the military conscript register, and only men are subject to conscription. Now, to produce a pair of brothers, a woman has to have, on average, three children. Among Norwegian women with three children, 36.2% have had them by two or more men. So this is not a trifling matter.

There would be no problem if the word "father" in the registries means "biological father." It is now common practice, however, for a stepfather to adopt his wife's children. At that point, he becomes the "father" for all intents and purposes. This has been the case since 1986, when Norway passed The Adoption Act: "If a spouse or cohabitant has adopted a child of the other spouse or cohabitant, the said child shall have the same legal status in relation to both spouses or cohabitants as if he or she were their joint child." The Adoption Act (1986), Chapter 3, Section 13.

This paternity issue affects different registers to different degrees. The family register is most vulnerable. It is updated annually and thus identifies the current legal father as the “father” (Black et al. 2011, note 4). The birth register is least vulnerable. The mother will usually identify the biological father as the "father." The exceptions are cases of infidelity or cases where a new relationship has formed during the pregnancy and the stepfather wishes to be recognized as the child's father.

Bratsberg and Rogeberg (2018) were most interested in the conscript register. This was where they obtained the IQ data. They then used the family register to locate the conscript's brother or brothers, at which point they returned to the conscript register to find that person's IQ. The family register, however, will most likely identify a stepfather as a "father." One would have to go back to the birth files and double-check, but this doesn't seem to have been done.

When I discussed this issue with one of the co-authors, Ole Rogeberg, he replied that very few of the brothers could be half-brothers because the correlation between brothers for IQ was 0.47, and this figure is similar to a previously published estimate of 0.49 for inter-sibling correlation (Paul 1980). That estimate, however, is largely based on American whites, who are a more heterogeneous population than native Norwegians. The inter-sibling correlation should be higher in Norway, and higher figures have in fact been estimated from British samples. Record et al. (1969) found a correlation of 0.55, based on 5,054 pairs of English siblings. 

As for the correlation between paternal IQ and child IQ, this is because most of the fathers are, in fact, biological fathers. Probably about two-thirds of them. That is enough to produce a father/son correlation, but it is not enough to prevent a within-family decline in IQ between older and younger brothers.


The Norwegian population registry has long been popular with population geneticists. It is actually a collection of different registers with information on different life events, but if you know a Norwegian's unique personal identifier you can easily navigate from one register to another to collect information on that person and on all related individuals. Black et al. (2005) have described these registers at length:

Our primary data source is the birth records for all Norwegian births over the period 1967 to 1997 obtained from the Medical Birth Registry of Norway. All births, including those born outside of a hospital, are included as long as the gestation period was at least 16 weeks. The birth records contain information on year and month of birth, birth weight, gestational length, age of mother, and a range of variables describing infant health at birth including APGAR scores, malformations at birth, and infant mortality (defined as those who die within the first year).

[...] Using unique personal identifiers, we match these birth files to the Norwegian Registry Data, a linked administrative dataset that covers the entire population of Norwegians aged 16-74 in the 1986-2002 period, and is a collection of different administrative registers such as the education register, family register, and the tax and earnings register.  These data are maintained by Statistics Norway and provide information about educational attainment, labor market status, earnings, and a set of demographic variables (age, gender) as well as information on families.

[...] Another source of data is the Norwegian military records from 1984 to 2005 which contains information on height, weight, and IQ.  In Norway, military service is compulsory for every able young man. Before entering the service, their medical and psychological suitability is assessed; this occurs for the great majority between their 18th and 20th birthday.  For the cohorts of men born from 1967 up to 1987, we have information on height, weight, and Body Mass Index (BMI), all of which were measured as part of the medical examination. We also have a composite score from three speeded IQ tests -- arithmetic, word similarities, and figures.

At first sight, this collection of registers seems to be a gold mine of information. Unfortunately, the quality of the information has suffered from a social trend that has been stronger in Scandinavia than elsewhere in the West, i.e., the redefinition of the family. As a result, the word “father” no longer has a consistent meaning. In some cases, such as birth records, it usually means the biological father—the man who provided half of the child’s genetic makeup. In other cases, such as the family register, it means the man who provides the family with at least some economic support. Increasingly, the two roles are no longer played by the same person.

This is the nub of the problem. The family unit is no longer defined as a vehicle for procreation. Its members no longer have to share a biological commonality. Increasingly, it is an administrative entity, and as such it can be dissolved and reformed in many ways.

This redefinition of the family calls into question the value of this population registry. Some of its components, particularly the birth records, still provide reliable information on biological relationships, but the same can no longer be said about information that is collected at later times in a person's life and which is continually updated. As a result, we know less and less about the people who are providing the genetic material of the next generation.


Black, S.E., P.J. Devereux, and K.J. Salvanes. (2005). From the Cradle to the Labor Market? The Effect of Birth Weight on Adult Outcomes. IZA Discussion Paper No. 1864

Black, S.E., P.J. Devereux, and K.J. Salvanes. (2011). Older and Wiser? Birth Order and IQ of Young Men. CESifo Economic Studies 57(1): 103-120.

Bratsberg, B., and O. Rogeberg. (2018). Flynn effect and its reversal are both environmentally caused. Proceedings of the National Academy of Sciences Jun 2018, DOI: 10.1073/pnas.1718793115

Flynn, J.R. (2007). What is Intelligence? Beyond the Flynn Effect. Cambridge University Press.

Government of Norway. (1986). The Adoption Act. Act of 28 February 1986 No. 8 relating to adoption,

Paul, S.M. (1990). Sibling resemblance in mental ability: a review. Behavior Genetics 10(3): 277-290.

Record, R. G., McKeown, T., and Edwards, J. H. (1969). The relationship of measured intelligence to birth order and maternal age. Annals of. Human Genetics 33: 61-69.

Rindermann, H. (2018). Cognitive Capitalism. Human Capital and the Wellbeing of Nations. Cambridge University Press.

Teasdale, T.W., and D.R. Owen. (2005). A long-term rise and recent decline in intelligence test performance: The Flynn Effect in reverse. Personality and Individual Differences 39(4): 837-843.

Thomson, E., T. Lappegård, M. Carlson, A. Evans, and E. Gray (2014). Childbearing across partnerships in Australia, the United States, Norway, and Sweden. Demography 51(2): 485-508


Luke Lea said...

As usual, well argued and well documented.

Anonymous said...

Do you think the Scandinavian welfare state has anything to do with it as well as the undermining of the traditional family unit caused by bastardry?

Anonymous said...

The paper only reproduces a trend within families that was already noticed to operate between families. The IQ of firstborn sons also rose until birth year 1975 and then started to drop. Your theory demands that there are two different things going on, one for firstborn sons and another for pairs of brothers, that lead to exactly the same changes at exactly the same time. That's not very plausible.

I applaud your creativity but I think this time you missed the mark. But here is another puzzle piece to exercise that creativity on: Take another look at the graph. It basically rises linearly until 1975 and then turns on a dime and drops linearly. Most potential causes of a decline would not result in that kind of abrupt reversal. Instead they would weaken the positive trend until it slowly reverses. I think your proposed solution would also not cause a abrupt reversal. Instead we are looking for a cause that probably started from scratch between 1975 and 1985, was immediately strong enough to reverse the positive trend and proceeded linearly.


Sean said...

Conscription age in Norway was lowered from 20 to 19 years of age in 1976. Maybe those 19 year olds were trying less hard for some reason of their own.

An bit of an IQ decline between older and younger siblings would be expected, the first born has a definite biological advantage--younger your mother the better. The mother would be appreciably older by the time she had children in a different relationship.

Anonymous said...

Norway - Infant mortality rate

Born Too Soon

Preemies and underweight babies more likely to have low IQs

Peter Frost said...


Norway's welfare state was created in the late 1930s and the postwar era. Initially, it helped to strengthen the family, and it has only been since circa 1980 that an effort to redefine and weaken the family has come into play.


First, before "damning me with faint praise," please provide references for your arguments. As far as I can see, there is no contradiction between your argument and mine. I agree that something happened between 1975 and 1985. There was an ideological shift away from old-line leftism and toward a new kind of leftism that was less interested in the working class and more interested in dismantling traditional structures, like the family.

I agree that other deleterious changes have been going on. In general, the "deregulation of sexual life" has had negative effects on everyone concerned. However, Bratsberg and Rogeberg (2018) argue that the decline in IQ can be largely, if not wholly, explained by within-family effects. It certainly looks like the major cause of the decline is within-family.


In that case, there would have been a one-time drop circa 1976. That's not what we see.


This could be another within-family effect. Infant mortality and miscarriages tend to remove children or fetuses that already have developmental problems

Anonymous said...

@Peter Frost: Page 3 in the paper under discussion has three graphs comparing within-family trends with between-family trends.The caption includes the explanation that the between-family trend is calculated on the basis of the IQ scores of firstborn sons: "... The dashed line depicts the trend for firstborn sons ...".

Your "low IQ second husband"-explanation cannot account for the fact that firstborn sons show exactly the same trend as pairs of brothers.

By the way, for your thesis it is not enough to show that second husbands are less intelligent (do you have a reference for that?), after all divorce also frees up the high-IQ first husband to have more kids with other women. Given the existence of menopause, the divorced first husbands may actually have more kids on average than the divorced wife. That is certainly the case in my social circle, which of course may not be representative.

Anyway, I didn't mean to "damn you with faint praise", I'm just somewhat tone-deaf and arrogant, sorry about that. I really do think that you have very creative insights, which is the reason why I read your blog.

@Sean: You mix birth year (drop started with birth year '75) and conscription year (which is 19 years later). That conscription age lowering probably even lies outside the time interval covered by the data.


Peter Frost said...

The paper argues that within-family variation can account for all of the IQ decline. The declining IQ of first-born sons over time is therefore due to whatever is causing IQ to decline from first to second sons within each family. This decline would be consistent with an overall genetic change in the population. In other words:

1. The second son is being fathered by a segment of the Norwegian population that has, on average, a lower IQ than that of Norwegians in general.

2. This low-IQ segment is thus contributing disproportionately to the next generation.

3. Therefore, from generation to generation, the first-born son should, on average, have a lower IQ.

"for your thesis it is not enough to show that second husbands are less intelligent (do you have a reference for that?), after all divorce also frees up the high-IQ first husband to have more kids with other women."

I will discuss this at greater length in my next post. The short answer is that in Norway stepfathers are most likely to be men with the lowest level of education (10 years of schooling). Second place goes to men with college or university education (14 to 17 years of schooling), and third place goes to men with upper secondary (11 to 13 years of schooling).

It is difficult to calculate overall reproductive success because many men are little more than sperm donors, i.e., there is no good record of all of the women they inseminate. If we look at childlessness, the highest rate is among men with the lowest level of schooling and the lowest rate is among men with the highest level of schooling. But the children are not necessarily fathered by the legally defined "father." When a man adopts the children of his spouse, they become his children.

Many men from the lowest level of schooling go through life without every being in a stable relationship, but this doesn't mean they are childless:

"[…]the propensity to have children with more than one woman is most pronounced among those with low education. Becoming a father is thus more of a selective process for men with low education than for men with higher education, but having become fathers, low-educated men are much more likely to have another child with a new partner. Obviously, multi-partner fertility is closely linked to union dissolution, but we should underline that some of these men have never been in a stable relationship with the mother."


Anonymous said...

Ridiculous. Men who are competitive in marriage markets aren't the one with High IQs! Many of today's incels are educated and intelligent.

Anonymous said...

@Peter Frost: Of course, if second sons are fathered by lower IQ men, this would also show up in the IQ trend of firstborn sons. But only in the next generation, not at exactly the same time. These two trends reverse both exactly in 1975, and I still don't see how you would get that within your theory.


Sean said...

Anders Brievik's father was a diplomat, his mother remarried--an army officer. Beivik was not conscripted. It would seem that the proportion of people who avoid serving has been rising for decades. It was illegal to cohabit in Norway up until the 50's. Oil wealthy Norway has had excellent welfare and maternity benefits and rates of pay for some time now.

"UNMARRIED cohabitation has increased steadily since 1977. Of all women aged 20–44, 5% cohabited in 1977, 18% in 1987 and 24% in 1996".

Martin Van Creveld says that when women enter any occupation or institution in numbers it declined in prestige and men don't want to make it their career. Norway is a very feminist country and currently it even conscripts women.

Peter Frost said...


"at exactly the same time" is an overstatement. It would be more correct to say "within a span of perhaps ten years"

In any case, the Flynn effect stopped in Norway probably for the same reason it has been slowing down elsewhere in the Western world. We have largely exhausted the possibilities for improving IQ scores within the existing genotype. There is still debate on the causes of the Flynn effect, but the general view is that it is partly due to people becoming more familiar with tests and test-taking and partly due to better nutrition and a more stimulating learning environment.

That in itself would not cause a subsequent IQ decline. It would simply cause the increase to level off. The IQ decline seems to have begun long before, perhaps in the early 20th century. The Flynn effect has simply masked it, and now that the Flynn effect has lost steam, we're now seeing this underlying decline.

Unknown128 said...

Dear Dr.Frost

first I wanted to say that I have been a long time admirer of your work and am very happy to see you posting again.

I wanted to ask if there was any way I could contact you via E-Mail?

I would really like to ask you some questions and to offer the assistance that my training as a historian allows me.

Thank you in advance and sorry for my bad English.

Anonymous said...

The graph in the paper indicates that it is in fact at exactly the same time. Are you saying that they manipulated the numbers until the curves perfectly overlapped? Also, as I mentioned earlier: A slowing Flynn effect would not show the linear rise we see in the data. Within the time interval covered by the data, there is no slowing. Just a sudden reversal.


Cerberus said...

I would caution against using military IQ tests, because they tend to have perverse real life incentives. Score too high and you are more likely to be selected for NCO training of for reserve officer school. In Finland since 1995 most conscripts have to serve only six months, but those selected for wartime leadership positions are expected to serve 12 months. Before 1995 the difference was only between 8 months and eleven months.

I am not familiar with the Norwegian system, but there may well be similar factors at play.