Standard protocol for withdrawing published paper that has wrong (or erroneous) data (published without wrong intentions)

Question

Suppose a research group publishes an article after thorough data analysis, statistics based on repeatability and reproducibility of experimental data. Now I consider three cases:

a). Published article had erroneous uncertainties associated with measured data

b). Published article had erroneous measured data notwithstanding repeated experiments

c). Published article had wrong interpretation of measured data

Now, in such cases, what is the standard protocol for withdrawal of published paper, and whether it is really necessary? If the same research group publishes another article giving more accurate experimental data with revised uncertainties with more reliable and logical interpretation, does this in any way puts a question mark on the credibility of the research group?

What according to the strictest ethical rules of publishing should be done in order to accept mistake and provide justification on why the earlier interpretation was wrong and what makes the research group change its view on the same experimental data (in this case, revised data)?

Answer 1

Retraction is usually only used when there was misconduct. If you made an honest mistake, it's sufficient to contact the publisher and ask to publish a corrigendum to the paper (different publishers have different procedures for doing this). If you just no longer believe the conclusions you made are correct, that doesn't even need a corrigendum, it's just part of the normal scientific process; next time you write a paper on that topic you can spell out your criticism of the previous work and why you don't agree with the conclusions. This is part of why science is supposedly "self-correcting".

I certainly don't think publishing a corrigendum or a new paper criticizing your previous work would cast doubt on your credibility. To the contrary; being able to criticize one's own work and recognize problems in it is an important part of doing science, so doing this would not reflect poorly on you, it would reflect well on you.

Answer 2

Not every published paper is right. Publications are a snapshot of authors' understanding at a given time. If they don't stand up to history, the literature corrects itself. Descartes's publications on the pineal gland as the generator of motor actions have never been retracted!

If a paper is wrong because new data or new understanding has changed the authors' outlook, perhaps the best course of action is to publish a new paper, specifically discussing the previous paper.

If there is a real MISTAKE or MISREPRESENTATION in the original paper, the editor should probably be contacted about the possibility of publishing an ERRATUM. Retraction should be reserved for the most severe of cases.

Answer 3

Some papers can't be "retracted" if they have been printed and widely distributed. Corrections can be issued in some journals.

But, another way to correct the record is to attempt at least to publish another paper updating and citing the old paper. Spell out the errors in the original. You may or may not want to say why this happened.

However, such a paper is only likely to be accepted (I think) if you provide a correct interpretation. I think this is your situation here.

Whether it puts a negative slant on the group or not, leaving it uncorrected will probably do more harm.

You might want to try to work this out with the editor of the original journal if possible. But, as Scott Seidman implies, people make mistakes. It goes along with being human.