Lynn cherrypicks his samples according to racial biases, and then extrapolates this to countries for which there is no testing based on proximity.
For Africa, Lynn ignored the majority of recently available IQ studies, and focused instead of unrepresentative, old IQ tests such as one conducted for a tribe islated in the jungles of the Belgian Congo in the 1950s, a study of 17 South African children that were deliberately chosen for their illiteracy (it was a study of the impact of illiteracy on IQ, Lynn uses this as his sample for South Africa's national average), and a sample of Ethiopian Jews living in Israel is used as an average for Ethiopia.http://racialreality.blogspot.com/2011/08/devastating-criticism-of-richard-lynn.html?m=1
As for China, Lynn's data comes from samples of Raven's Progressive Matrices and WISC-R that were taken over the internet. Most Chinese do NOT have internet access, particularly when these tests were taken, so this is already a biased sample that focuses on the upper classes living in the coastal cities, and moreover it could have been tampered by the CCP since we have no proof school children actually took these tests.
Pic related shows internet access in 2015 by percentage of population, you can see why conducting tests over the internet in China is a flawed measure... even moreso considering Lynn's data comes from the 1990s and 2000s.
Actual IQ tests on the ground in rural China report IQs in the range of 70 to 90.