I think he was trying to say it was evidence of an unrepresentative sample, which is an accurate observation. His counterpoint isn't very sensible though - why not just lead with the U.K. study?

Then it gets into this: "It’s common knowledge in psychometric circles that reading comprehension tests are statistically equivalent to IQ tests..." and that's it for justification of projecting welfare recipient IQs. Okay then. Stopped reading there.