i read through once quickly. can someone explain the difference between model 1 and model 2?

Model 2 includes "only people under the age of 40 from mid-century." This partition is useful because of the ambiguity as to whether the large B-W gap "among persons born in the 1920s and earlier reflect a widening B-W difference among those tested at older ages rather than a difference that was greater in the early 20th century."

thanks --

"Models 1 and 2 imply very different interpretations of changes in the B–W difference on the Woodcock–Johnson."

this is where i got tripped up.

it makes more sense realizing that the difference in the models is just a matter of censoring some of the data.

it would be interesting to plot this data alongside other anthropometric measurements for which we have ratio scale data. Height in particular would be great -- what does the plot of the Euro - Chinese height difference look like over the last century?

As this is real data we're dealing with, it would be more informative to plot the full distributions rather than just one parameter (the SD). I'm thinking something along the lines of Gapminder, where you can visualize exactly this kind of thing for the Chinese vs. American income distributions as a function of time:

http://www.gapminder.org/downloa...ution- 2003.html

Select "China" and "USA", deselect "World", and then move the scrollbar along the bottom. You can see much more information than just the time evolution of the standard deviations.


You could also represent this as a static graphic, where the x-axis is the value of time, the y-axis is the IQ, and the z value at a given (x,y) point is the probability density of individuals in the given time period with the given IQ. You would want to actually draw two such density estimates for each cross sectional time point to look at how the densities evolved with time.

As evidence for the interesting kinds of things that can be seen with the time-dependent plot of conditional density estimates (for which the underlying data is a 3 column table w/ first column = continuous time, second column = categorical population, third column = continuous income/IQ/height/etc.), try looking at the USA vs. Brazil.

You see that something happened in Brazil around like 1986 or so which just completely halted economic growth in the country.

You can also see that the US, unlike Japan, is a multimodal society. I bet that you could do a Griffe style regression in which the height of the second, lower bell curve in the US distro is a function of the percentage of low IQ groups in the country.

How a bimodality emerged in the chinese income around 1986 was interesting.

To gc:

What happened to Brazil was the Latin American debt crisis of the 1980s, starting 1981-1982. After that, Brazil had about ten years of enormous irregularity in GDP trends (up 5% in one year, down 3% the next, up again, down again...) and was generally close to hyperinflation. In the 1990s, Brazil solved the inflation conundrum, but as a result of the medicine applied has been regularly growing very little ever since.

If we wanted to explain the jump in the B-W gap from the 1950 to 1960 birth cohorts as something other than an artifact of data collection, I'd suggest lead poisoning as a possible explanation. Urban black populations were disproportionately impacted by lead poisoning prior to the introduction of unleaded gasoline and unleaded paint...

So the conflict between the evidence showing that the BW gap has narrowed, and the evidence that the BW gap is still 1 SD is resolved by the new data showing that the BW gap was actually 1.5 SD for much of the 20th century. It narrowed .5 SD for blacks born between 1950-1970*, but hasn't narrowed further for blacks born in any of the 37 years since that time.

We're right back where we started, the 1 SD difference reported by Shuey.

The NLSY dysgenic evidence Murray reported for blacks in his Flynn debate, along with Flynn's remark that social conditions haven't even improved for blacks suggests that widening or at least a continuing stagnation of the gap should be reasonably expected whether you have a "liberal" or "conservative" outlook.


* with the 1.45 gap for those born in 1959 being a possible artifact.