Post updated, 10/21/14. See below!
It’s general trope in the HBD community: people are getting dumber. The low IQ are outbreeding the high IQ, leading to a slow decline in genetic intellectual potential in the population. Indeed, my own analyses seem to have shown that there was a fair fertility advantage among the lower IQ over the higher IQ (seen most recently in my post Who’s Having the Babies?):
The key limitation is that most of these analyses left off in 1950s cohorts – the people who had their children in the 70s and 80s. This was my parents’ generation! We don’t know what people born later did. Does the apparent dysgenic pattern continue right up to the present day? I decided it was time to take a look:
These are average number of children had by White Americans aged 42 and older born during the 1960s, from the GSS (drawn from all GSS years, with those two parameters established), by WORDSUM score, a proxy for IQ. 95% confidence intervals shown, which should give an idea of sample sizes. I’ve collapsed the 0-3 score to make the distribution symmetrical on both sides in terms of number of subjects (there appears to be a significant rightward skew in the WORDSUM data).
Now this is interesting; unlike previous cohorts, fertility among the 1960s cohorts doesn’t look dysgenic for IQ. If anything, it looks slightly eugenic.
Let’s see further what’s going on, by looking at the sexes separately:
Here is fertility by WORDSUM for White males and females separately. Previously, we’ve seen that fertility is eugenic for men and dysgenic for women. For the 1960s cohorts, this appears to be case. But the eugenic fertility for men is strong enough to outweigh the dysgenic fertility for women, so the net effect is slightly eugenic (things were probably a bit more eugenic when you consider childhood mortality is more concentrated on the low end).
To confirm that net White fertility was eugenic, I ran the correlation between WORDSUM and average number of children (for all individuals 42 and older at time of survey):
|Cohorts||Total fertility-IQ correlation||Males only||Females only|
This is a fascinating finding. A big point of alarm in the HBD world (even noted by myself previously) is that people are getting slowly dumber with each generation. Aside from the fact that this process is, at best, very slow (see Greg Cochran here and here), as far as we’re concerned, it doesn’t even appear to continuing! At least for one period, it reversed somewhat. This seriously calls into question the practice of projecting fertility trends into the future on the assumptions current patterns will hold.
So why did dysgenic fertility halt for the 1960s cohort? That’s currently not clear. These were people born in the tail end of the Baby Boom, who would have been having children in the ’80s and ’90s. Economic conditions (which, as we’ve previous seen, can strongly affect fertility – see Another Tale of Two Maps and A Tale of Three Maps) – while clearly being not as good as during the Baby Boom – were not particularly bad, nor particularly good. This was during the “Rust Belt” epoch – “deindustrialization”– where many manufacturing jobs across the Midlands and Greater New England left the region for other parts of the country and overseas. The erosion of earning ability for low-ability males may have stymied their child-bearing prospects compared to earlier decades. I will return to this point shortly.
What about the 1970s cohorts? Well, this is a bit harder to call at the moment, because that generation is still having children (I should know). Here is a look at fertility for 1970s White Americans:
These are only individuals age 38 and over. I didn’t bother with error bars, because sample sizes are all really small here. I collapsed score 0-4 and 9-10 to make the distribution symmetrical. Sample sizes were too small for me to look at the sexes separately. But, at first look, it would seem fertility appears to have returned to a dysgenic pattern.
Though before we go too far with that conclusion, let me show you something else:
This is average age of having the first child for the marked cohorts, for White Americans, from all GSS years. To ensure every respondent had a chance to contribute their likely lifetime datum, I included only individuals age 48 and older at the time of the survey. As we see, there is a fairly consistent pattern for smarter individuals to have their first child systematically later than dumber individuals. The pattern reverses a bit at the lowest IQ levels, it appears (note: this doesn’t appear to an artifact of small sample size).
Here are males and females separately:
For the 1970s (and younger) cohorts, the smartest individuals are likely not done having children, due to the high average age of first child, which is in the 30s for these folks. Indeed, see here [Edit, 10/21/14: I’ve inserted a chart that includes all sampled individuals from the 1970s cohorts, to increase the sample size. The samples are still pretty small
(age 38 and older) –forewarned: samples are tiny, sometimes in the single digits]:
For the record, I did look at other races. Sample sizes are much smaller here, so charts would have much less value. For Blacks, fertility is generally much more strongly dysgenic throughout, increasing with time. The correlation between fertility and WORDSUM is remains in -0.19 to -0.25 range. However, the correlations are smaller in magnitude for the oldest cohorts, though it has to be expected that this is partly due to attrition through death at these ages. While samples are small, preliminarily for the 1970s cohort, the correlation is -0.30. The male-female difference was present, but small (see also Dysgenic Fertility Among Blacks? Apparently, Yes).
Samples sizes for Hispanics were too small to do anything useful.
Edit, 10/21/14: [On the advice of Greg Cochran, I also looked at whether the above pattern was just an artifact of unreliable WORDSUM scores by looking at fertility by education. Here at the results:
This is average number of children had by non-Hispanic Whites, age 44 or older, 95% confidence intervals shown. Here again we what looks like a neutral to perhaps very slightly negative relationship between education and fertility for the 1960s cohorts.
Here are males and females separately:
We see the classic pattern: fertility appears to be eugenic for men and dysgenic for women. Indeed, a look at the correlation between education and number of children shows very near neutral total fertility for the 1960s cohorts:
|Cohorts||Total fertility-Education correlation||Males only||Females only|
(Indeed, the correlation for the 1960s people becomes positive, r = 0.02, if I set the cutoff at age 48. This is driven by an increase in the correlation for men, from 0.14 to 0.20. This indicates that older fathers may be driving the relationship.)
Here are the 1970s cohorts (included only those age 40 or over) – sample sizes are very small, so I forwent the confidence intervals (0-1 = high school and below; 2 = some college; 3 = bachelors; 4 = graduate):
This suggests a return to a dysgenic pattern, as seen from the correlations. However, the as seen before, the small samples and the possible age effect makes this hard to call.
Either by the IQ (as gauged by the WORDSUM) or by reported education, the GSS data shows that dysgenic breeding seems to have significantly stalled or even reversed for those born in the 1960. For what it’s worth, Audacious Epigone has noted a decreasing correlation between the WORDSUM score and reported education in the GSS:
the correlation between wordsum scores and educational attainment by decade of birth among all native-born Americans who have participated in the GSS:
Born prior to 1950: .536
Born in the 1950s: .507
Born in the 1960s: .469
Born in the 1970s: .419
Born in the 1980s: .373
One serious issue would be this: the current neutral to slightly eugenic fertility reported in the GSS stems from a serious sex difference: eugenic for men and dysgenic for women. This leads one to suspect that low-IQ men may just be systematically underreporting the number of children they have, perhaps unwittingly so. To check for this, I looked at the total average number of children for men and women separately. I did find a slight difference (1.88 for men vs 1.96 for women), however it’s not statistically significant. ***End Edit***]
Looking at total fertility rates over time and age of first child underscores a pattern I found earlier (as noted in my post Some guys get all the babes – not exactly). Specifically, the Baby Boom was a brief period (at least during the past 100 years or so) where the total fraction of individuals who contributed to the gene pool increased:
In the era before the Baby Boom (the people who gave birth during the Great Depression), ~20% of individuals, male and female, had no children. That fraction fell to less than 10% during the Boom. It has since returned to its pre-Baby Boom size of ~20-25% (higher for males). During the Baby Boom, all sorts of individuals (about 10% more of the population) were having children who previously wouldn’t have. Since we see that in the pre-1900 cohorts, the fraction of the childless was about the same as it was for the people born at the start of the 20th century (~20%), this doesn’t appear to be solely a product of the generation trough during the Great Depression.
This may explain something noticed by blogger “Agnostic“: homeless individuals, who are very often mentally ill (especially schizophrenic) appear to be disproportionately Baby Boomers. If a certain segment of the population who, in earlier epochs, normally didn’t reproduce as much, suddenly increased their fertility due to a time of easy living, then you would expect an uptick of those sort of individuals in the following generation. If that 10% of people who bred more during the Boom were in the top 10% of those with genetic load, say, then the following era would witness a significant increase, at especially at the extreme ends of the distribution – which schizophrenia, for example, may represent.
These generational effects in fertility, with boom and bust cycles, represent the effects of the population cycle as described by Peter Turchin (see here for a good description of the process). Population growth sows the seeds of its undoing, by decreasing the share of resources (be it food, land, or these days, well-paying jobs) available to the up-and-coming generation. In short, the more people, the smaller slice of the pie each individual gets. The fewer people, the larger piece of the resource pie each family can acquire, typically boosting fertility. Immigration exacerbates these trends (see Turchin on it here). The whole process represents one of the most reliable “environmental” effects I have examined.
Coupled with the good times during the Baby Boom, we see that age of first child fell a bit (though it was fairly low before), as you’d expect. Afterwards, it rose significantly, especially for smarter women. The smartest individuals (WORDSUM 10 – roughly IQ 120+) now typically have their first child after their 30th birthday.
These and other factors makes many individuals (you know who you are) want to return us to the Baby Boom-like era, where labor was scare, and anyone who wanted a well-paying (though often gungy) job could have one. Restricting immigration, as Turchin discusses, is most likely to trend things in that direction. But it’s starting to look more and more questionable that the Baby Boom was an unadulterated good. Sure, the living was easy for those during the Boom, but its products haven’t necessarily been the best.
Fertility rose among everybody, even the smartest were breeding well over replacement then. However, the dumbest were breeding much more, more than they otherwise would, apparently. Perhaps excessive good times aren’t really all that great in the long run.
But, the following period appears to have given us an epoch of eugenic breeding, if ever slightly. Regardless, the important thing this demonstrates is that, at least in the U.S. anyway, we can forestall the coming of the supposedly inevitable idiocracy. We had a long way to go to get there anyway, but even that required a sustained dysgenic trend, and it’s unclear if that can be taken as a given. Razib Khan was right; population projections 50 years into the future are fantasy. Demographic trends have a nasty habit of changing quite a bit over time, enough to mess with the predictions of the most enlightened prophet.