3 ms·
Use of blood donors is itself questionable and open to selection bias. Particularly if blood donors were getting free antibody testing and seeing the results,
by lamontcg 6y ago
Use of blood donors is itself questionable and open to selection bias. Particularly if blood donors were getting free antibody testing and seeing the results, but even if they weren't. The authors seem to only have considered the demographics of the sample selection, and somewhat poorly.
- timr 6y agoOh, definitely. I don't dispute the need for the adjustments that the authors were applying. It's just that...after a certain amount of "adjustment", what good is your prediction, anyway? Correcting the seroprevalence by a few percent is one thing. "Correcting" it by over 200% is quite another. Basically, folks are going back to this paper (or more realistically: not reading it at all, and simply parroting headlines), ignoring the intense massaging of the data, and claiming that reality disagrees with the predictions from the paper, therefore reality must be wrong.
- lamontcg 6y agoRight, I would argue there's factors that might call for a 200% correction the other direction. You just can't validate their tea leaf reading here of what are the correct set of corrections.