Having agreed (again) that the data smells, let me nevertheless point out one place where I think it doesn't smell quite as badly as you say.
More confusing is that she only had 46 people with IQs under 120. This is only 22% of her total population, yet an IQ of at least 120 occurs in 1 out of every 11 people. At a minimum, nearly everyone in her random-kids group (70 kids, right?) should have had an IQ below 120, yet even if all 46 were in the non-gifted group, they'd only comprise about 2/3 of it.
The ones you're calling random kids were not random; they were matched for SE level and school class with children who were parentally identified as highly-gifted. Factor in that IQ is not (and was not) nearly as socially acceptable a topic in the UK as in the US - and "giftedness" as a phenomenon is identified much less - it wouldn't be surprising if the children whose parents identified them as gifted, against this social pressure, tended to be quite extreme. Then factor in that IQ is correlated with SE level and with educational background, and it's not surprising that children from the same background and schools as the identified-gifted children were themselves far from average. If in addition to all this the numbers were on a test with outdated norms, that might do the rest for the paucity of children with IQs under 120. I don't see how to stretch this argument to account for the crazy numbers she has at the very top, or for the apparent ability to match raw scores on Raven's.
