scifisam
feck! arse! girls! drink!
They have error bars showing the confidence interval, and the number of samples involved. This provides some sense of what they think in terms of reliability of different results, ie an attempt to quantify the sort of thing you are talking about.
Well, they used to have the numbers involved in that earlier report, but the format of the age-based antibody reporting graph has changed, this is the latest one (still has error bars but they can be hard to see clearly):
View attachment 231910
from https://assets.publishing.service.g...COVID19_Surveillance_Report_week_39_FINAL.pdf
The error bars are longer for the oldest age group but not by enough to erase my original point.
I'm sure - but honestly it really will almost all be the early 70s. It looks like a pretty rigourous study but they can't change the initial self-selecting nature of the cohort.