The truth is that no study that would generate these numbers plausibly has ever been conducted. There is no country-by-country, apples-to-apples international IQ test database. IQ tests are a diagnostic measure, so if you're going out into the field looking for lots of test results, you're either drafting off large-scale psychometrics work done in and by wealthy countries, or you're pulling from mental and developmental health institutions where tests were administered to diagnose patients.

That's exactly what Lynn did: he has "good" IQ data generated for-purpose by the research communities in developed countries, and just uses IQ data from, essentially, mental health institutions elsewhere. He presents these data sets as apples-apples equivalent.

It gets worse: for what should be obvious reasons, there are lots of countries where it isn't easy to get any such data. Lynn doesn't try; instead, his team takes data from neighboring countries and extrapolates.