I agree a small truly random sample can often be reliably extrapolated - but in this case the 136>7m figure was based on research paid for by the BMI (who lobbi

I agree a small truly random sample can often be reliably extrapolated - but in this case the 136>7m figure was based on research paid for by the BMI (who lobbied heavily against file-sharing) in a report still not fully published, based on two numbers of which one differs from the official ONS number without explanation. The fact that the lobbyist organisation paid for the original report was obscured by misattributing the source, which also seems dubious to me. I don't doubt that political misuse of statistics is common, but it should always be called out.