We all make mistakes…

There’s a storm in a beaker today about the Chemistry blogosphere and twitter regarding a paper in which a comment in the supplementary information seems to imply that a student ‘just make up’ some data. You can read about it over at ChemBark who first reported the comment, and who intends to update the post with the corresponding author’s response to a query on the issue if/when it comes.  Make sure you read the comment thread though because it is being pointed out that the term ‘make up’ has variable meaning, particularly to us British people. I regularly as students to ‘make up an NMR’, colloquially meaning to prepare then run the appropriate sample. Perhaps I shall be more linguistically precise.  In the Pipeline has found it now as well.

Without wishing to join the masses piling on to these particular authors because the full story is yet to appear in this instance,  this issue is an absolute pet hate of mine. Poor data management and verification makes me crazy and creates an environment where errors can be made (and falsifications/undue manipulations can pass through undetected). I work on the basic ideas that any author of any paper should be able to produce original (e.g. instrument specific file types/print outs with date/time) data to back up any claim if required by reviewers and retain such original data for a reasonable period of time afterwards. I’d suggest that the supplementary information is an ideal place for scanned copies of such data to be presented, allowing the reader to make their own interpretations should they wish (and note that there are NMR spectra presented in the current case) . As scientists, we should have nothing to fear from post-publication peer scrutiny of our work and potential alternative hypotheses that may result.  Of course, I’m thinking of chemical synthesis, my logic may not apply to other disciplines, or indeed the data sets may require too much specialist processing to be useful in their raw form.

We try to drum into our undergraduates that they should use as many methods of characterisation as instructed (in the undergraduate teaching laboratories) or that they might reasonably obtain (in the final year project laboratories). I cannot imagine why a scientist would deem it appropriate to claim the existence of a compound with only one type of analytical data unless they can make a detailed statement as to why that was the only viable technique (and there are some instances where that would be reasonable). But how far should corresponding authors go in verifying their work before submission? Should they check each original data file themselves? Should they trust their researchers and collaborators to act in accordance with the highest ethical standards? Personally I’d prefer that the contribution of each author is clearly detailed in the publication, including which data they were responsible for generating and ‘checking’. That also helps in cases where a named author may not have had sight of a manuscript prior to submission* and would have included different/additional analysis or would have interpreted things a little differently if they had seen it.

As per usual I see I have far more questions than answers. Sigh.

*and that’s another can of worms in itself – is it/when is it appropriate/ethical to pop someone’s name on a manuscript and submit it if they haven’t seen it?

3 thoughts on “We all make mistakes…

Comments please!