Always mistaking cause and effect.

From Misleading plots on secular trends and education by Emil O. W. Kirkegaard.

He is tackling the confusion arising from changing category populations when represented in data.  

It seems an obvious conclusion from the graph that one way to improve the lives of children, as measured by percentage in a single parent home, is to help young adults stay in school and increase their human capital in terms of years of education.

This is closely related to the common fallacy that assumes that consumption attributes create behaviors.  Back in the early 2000s, across two administrations, there was strong support for policies to increase the number of people who owned their own home.

The thinking was that by adopting consumption patterns of the middle class, those new owners would adopt the culture and behaviors of the middle class.  That fallacy ended in the Great Recession.

This was a classic Butterfield Effect where cause and consequence are confused.  

It is, of course, middle class behaviors (thrift, industry, work ethic, reliability, futurity, executive control, etc.) which create the capital to purchase and maintain a home, not the ownership which creates the behaviors.  

Confusion about cause and consequence is rife in public policy.

What Kirkegaard is pointing out is that the compelling graph is misleading because of education attainment certification inflation.  We have relaxed the standards for HS graduation, for junior college graduation, for college graduation, etc.  At each level of education attainment the average IQ is markedly lower today than it was forty years ago.  

So the desired interpretation here is to show that among mothers with low education, the proportion of single-parenthood has skyrocketed. The problem here is that the meaning in psychological terms of "high school or less" is not constant. That's because the distribution of education changes over time. This plot therefore cannot show what they purport to show. A simple way to show this is using the GSS, a public dataset of Americans. Audacious Epigone already made this point in 2019 of sorts (Average IQ of College Undergrads and Graduate Degree Holders by Decade), but here we will redo it a bit better.

From Audacious Epigene is the raw data.

Click to enlarge.

By 2020 it is not improbable that the average college graduate IQ will be the same as that of the average HS graduate in 1960.

Back to Kirkegaard, commenting on the declining IQ by level of education attainment.

The reason all the [IQ] lines decline is that all degrees are increasing in prevalence. Thus, the selection threshold to reach the next degree is decreasing for each of them, which causes lower mean human capital levels inside every degree group (including the bottom, since it is getting depopulated). The overall mean is not actually declining, that's a Simpson's paradox/fallacy. In these data, because we standardized the values within each year of testing, the mean across years is exactly the same i.e., 0. With that in mind, we see that high school graduates (but not higher) used to be about average intelligence in 1970s (z score is about 0), but now it's somewhat below average at -0.35 or so. Less than high school is now considerably below average at about -1.30, whereas it used to be about -0.75.

It is not education attainment which is associated with single-parent status but IQ.  HS graduates are a smaller percentage of the population than earlier and they have markedly lower IQs than a generation or two ago.  Lower IQ is associated with a range of tribulations in terms of decision-making, productivity, income, etc.  And lower levels of marital stability.

Those looking at the first graph would plausibly leap to the conclusion that if we wish to improve the lives of children (through reduced exposure to single-parent conditions), then we need to ensure that more people complete HS and at least some college.

The reality is that that mistakes cause and effect.  Education attainment does not cause stable families.  It is a blend of IQ and behavioral values which create stable families.  There is not much which can be done to increase IQ but there is quite a bit which can be done to change behaviors.  But changing behaviors is a difficult task in an individualistic, freedom oriented, culturally heterogenous and religiously heterogeneous society.  

Far easier, even though ineffective, to invest in the kabuki theater of inflating education attainment.