[burger]

The authors need to publicly and loudly retract the preprint and admit their errors.  It's painful to see Stanford's name dragged through the mud, but it's not too late to avoid the worst damage.

No, I could not disagree more.  If every methodologically imperfect study was retracted, the medical literature would never advance.  In fact, there would be no medical literature.  Unless the authors are downright falsifying their data, they should not retract it.  Hopefully, they are addressing criticisms of their work and are working to address them and they are working to make the data and underlying methodology more transparent (and also, undergoing the peer review process).  But science only works if you accept that some "wrong" articles are published.  For one, sometimes the results of methodologically flawed papers are confirmed through more rigorous analysis.  Secondly, the scientific debate is healthy.  But it can only work if scientific papers are evaluated on their merit, and not because someone doesn't like their conclusions.  Does the paper have flaws?  Absolutely.  Have I seen papers with worse flaws get published in high impact journals?  Absolutely.  The paper helps us better understand the pandemic and as such it should absolutely be out there, even if the message challenges our understanding of the facts and there may be issues in how it was carried out.

BC

I think you're wrong for a few reasons.

1) The flaws in this preprint are so egregious that it has no legs left to stand on.  All of the positives could be false positives, and the authors erroneously did not account for this.  This isn't one of those "opinions differ on how to do this analysis" sorts of criticisms.  It's a fundamental error that makes the conclusions invalid.  Someone once told me that the worst thing you can say about a scientific paper is that the conclusions are not supported by the data. That's what we have here.

2) Getting an analysis wrong does not make us better understand the pandemic.  In fact, in the worst case scenario (already maybe happening), this faulty analysis may kill people by encouraging governors to open up their states to soon.

3) It's a preprint, and retracting now will actually be good for the entire research process going forward.  The media and public needs to realize that a certain proportion of all of these preprints floating around here are wrong to some degree, and some are very wrong.  It's not a bad thing for "science" to correct itself before publication.

4) I don't think you understand how peer review is supposed to work.  Articles that have fundamental mistakes should never be published, period.  Science is ultimately self-correcting, but there's a big difference between, say, a new theory that makes old studies invalid and publishing a paper full of mistakes.  Just because some papers get through with mistakes doesn't mean that that's a good thing.  Quite the opposite.  And during a pandemic, we don't have time to wait months for a better paper to come along and correct the erroneous one that decision-makers are already basing their actions on.

Edit: forgot to mention the non-random sampling issue which is as bad as the statistical issues, if not worse.

[burger]

The authors need to publicly and loudly retract the preprint and admit their errors.  It's painful to see Stanford's name dragged through the mud, but it's not too late to avoid the worst damage.

No, I could not disagree more.  If every methodologically imperfect study was retracted, the medical literature would never advance.  In fact, there would be no medical literature.  Unless the authors are downright falsifying their data, they should not retract it.  Hopefully, they are addressing criticisms of their work and are working to address them and they are working to make the data and underlying methodology more transparent (and also, undergoing the peer review process).  But science only works if you accept that some "wrong" articles are published.  For one, sometimes the results of methodologically flawed papers are confirmed through more rigorous analysis.  Secondly, the scientific debate is healthy.  But it can only work if scientific papers are evaluated on their merit, and not because someone doesn't like their conclusions.  Does the paper have flaws?  Absolutely.  Have I seen papers with worse flaws get published in high impact journals?  Absolutely.  The paper helps us better understand the pandemic and as such it should absolutely be out there, even if the message challenges our understanding of the facts and there may be issues in how it was carried out.

BC

I think you're wrong for a few reasons.

1) The flaws in this preprint are so egregious that it has no legs left to stand on.  All of the positives could be false positives, and the authors erroneously did not account for this.  This isn't one of those "opinions differ on how to do this analysis" sorts of criticisms.  It's a fundamental error that makes the conclusions invalid.  Someone once told me that the worst thing you can say about a scientific paper is that the conclusions are not supported by the data. That's what we have here.

2) Getting an analysis wrong does not make us better understand the pandemic.  In fact, in the worst case scenario (already maybe happening), this faulty analysis may kill people by encouraging governors to open up their states to soon.

3) It's a preprint, and retracting now will actually be good for the entire research process going forward.  The media and public needs to realize that a certain proportion of all of these preprints floating around here are wrong to some degree, and some are very wrong.  It's not a bad thing for "science" to correct itself before publication.

4) I don't think you understand how peer review is supposed to work.  Articles that have fundamental mistakes should never be published, period.  Science is ultimately self-correcting, but there's a big difference between, say, a new theory that makes old studies invalid and publishing a paper full of mistakes.  Just because some papers get through with mistakes doesn't mean that that's a good thing.  Quite the opposite.  And during a pandemic, we don't have time to wait months for a better paper to come along and correct the erroneous one that decision-makers are already basing their actions on.

Edit: forgot to mention the non-random sampling issue which is as bad as the statistical issues, if not worse.